Datasets:

keerthanapacific

/

demo-json

like 0

1043859

Perceptual and Neural Olfactory Similarity in Honeybees

The question of whether or not neural activity patterns recorded in the olfactory centres of the brain correspond to olfactory perceptual measures remains unanswered. To address this question, we studied olfaction in honeybees Apis mellifera using the olfactory conditioning of the proboscis extension response. We conditioned bees to odours and tested generalisation responses to different odours. Sixteen odours were used, which varied both in their functional group (primary and secondary alcohols, aldehydes and ketones) and in their carbon-chain length (from six to nine carbons).The results obtained by presentation of a total of 16 × 16 odour pairs show that (i) all odorants presented could be learned, although acquisition was lower for short-chain ketones; (ii) generalisation varied depending both on the functional group and the carbon-chain length of odours trained; higher generalisation was found between long-chain than between short-chain molecules and between groups such as primary and secondary alcohols; (iii) for some odour pairs, cross-generalisation between odorants was asymmetric; (iv) a putative olfactory space could be defined for the honeybee with functional group and carbon-chain length as inner dimensions; (v) perceptual distances in such a space correlate well with physiological distances determined from optophysiological recordings of antennal lobe activity. We conclude that functional group and carbon-chain length are inner dimensions of the honeybee olfactory space and that neural activity in the antennal lobe reflects the perceptual quality of odours.

Introduction Stimulus discrimination and generalisation constitute two major abilities exhibited by most living animals. Discrimination allows treating different signals as distinct, while generalisation allows treating different but similar stimuli as equivalents [ 1 , 2 , 3 ]. Similarity along one or several perceptual dimensions determines the degree of generalisation between stimuli [ 2 ]. Determining such dimensions is fundamental for defining an animal's perceptual space. This objective remains, however, elusive in the case of the olfactory modality in which the dimensions along which odours are evaluated are not well known. Characteristics such as the functional chemical group or the carbon-chain length of a chemical substance may influence olfactory perception. It is known that at least some features of odorant molecules influence olfactory perception. For instance, some enantiomers can be discriminated by humans and nonhuman primates [ 4 ]. If and how chemical group and carbon-chain length are integrated as inner dimensions into an olfactory perceptual space remains unknown. Vertebrate and invertebrate nervous systems show important functional as well as anatomical similarities in the way in which olfactory signals are detected and processed in their brains, particularly at the level of their first olfactory centres, the olfactory bulb in the case of vertebrates and the antennal lobe (AL) in the case of insects [ 5 , 6 , 7 ]. Insects are useful models for studying olfaction, as their behaviour heavily relies on the use of olfactory cues. The honeybee Apis mellifera is one such model in which behavioural and neurobiological studies have been performed to unravel the basis of olfaction [ 8 , 9 , 10 , 11 ]. Honeybee foragers are ‘flower constant' and learn and memorise a given floral species that they exploit at a time as long as it is profitable. Floral cues, among which odours play a prominent role, are then associated with nectar or pollen reward [ 12 , 13 ]. However, under natural conditions, the blends of volatiles emitted by floral sources vary widely in quantity and quality both in time and in space [ 14 , 15 ]. To cope with such changes in an efficient way, a ‘flower constant' forager should be able to generalise its choice to the same kind of floral sources despite fluctuations in their volatile emissions. In a pioneering investigation, von Frisch [ 16 ] trained freely flying bees to visit an artificial feeder presenting several essential oils (odour mixtures). Using a set of 32 odour mixtures, von Frisch observed that after learning that a blend was associated with sucrose solution, bees tended to prefer this odour blend, but they sometimes visited other blends that were similar (to the human nose) to the rewarded one. Olfactory generalisation in honeybees was mainly studied on restrained honeybees using the conditioning of the proboscis extension reflex (PER) [ 17 , 18 ]. In this paradigm, harnessed honeybees are conditioned to odours associated with a sucrose reward. When the antennae of a hungry bee are touched with sucrose solution, the animal reflexively extends its proboscis to reach out towards and to lick the sucrose. Odours presented to the antennae do not usually release such a reflex in naive animals. If an odour is presented immediately before sucrose solution (forward pairing), an association is formed and the odour will subsequently trigger the PER in a subsequent unrewarded test. This effect is clearly associative and involves classical conditioning [ 18 ]. Thus, the odour can be viewed as the conditioned stimulus (CS), and sucrose solution as an appetitive unconditioned stimulus (US). Bees conditioned to individual odours or to olfactory mixtures can generalise PER to a wide range of different olfactory stimuli. Using the PER paradigm, Vareschi [ 19 ] showed that bees generalise most often between odours with similar carbon-chain lengths and between odours belonging to the same functional group. However, Vareschi conditioned odours in a differential way, with two rewarded and many unrewarded odours, so that several generalisation gradients (excitatory and inhibitory) may have interacted in an unknown way to determine the generalisation responses exhibited by the bees [ 19 ]. Using a similar approach and a restricted (6 × 6) set of odour combinations, Smith and Menzel [ 20 ] confirmed that bees generalise among odours with the same functional group, but their analysis did not detail the results obtained with individual odour combinations, thus rendering impossible the analysis of generalisation between odours with similar carbon-chain lengths. Free-flying bees trained in a differential way to a rewarded odour presented simultaneously with multiple unrewarded odours also generalise between odours with similar functional groups [ 21 ]. As for Vareschi's study [ 19 ], such an experimental design makes it difficult to interpret the generalisation responses due to unknown interactions between excitatory and inhibitory generalisation gradients. Recently, optical imaging studies facilitated our understanding of how olfactory stimuli are detected and processed in the bee brain [ 22 , 23 , 24 , 25 , 26 ]. The first relay of the bee's olfactory system involves the ALs, which receive sensory input from the olfactory receptor neurons of the antennae within a number of 160 functional units, the glomeruli [ 27 , 28 , 29 ]. Within each glomerulus, synaptic contacts are formed with local interneurons and projection neurons (PNs). PNs send processed information from the ALs to higher brain centres such as the mushroom bodies and the lateral protocerebrum [ 30 ]. Stimulation with an odour leads to a specific spatiotemporal pattern of activated glomeruli, as shown, using in vivo calcium imaging techniques that employ fluorescent dyes to measure intracellular calcium in active neurons [ 22 , 24 , 31 ]. The odour-evoked activity patterns are conserved between individuals and constitute therefore a code [ 23 , 24 ]. Odours with similar chemical structures tend to present similar glomerular activity patterns [ 23 ]. Furthermore, it is believed that the neural code of odour-evoked glomerular patterns measured in the bee brain actually represent the perceptual code, although this idea was never tested directly. In the present work, we studied behavioural olfactory generalisation, using the PER conditioning paradigm, with 16 odorants varying in two chemical features, functional group and chain length. The odours belonged to four chemical categories: alcohols with the functional group on the first or second carbon of the carbon chain (henceforth primary and secondary alcohols, respectively), aldehydes, and ketones. They possessed therefore three functional groups (alcohol, aldehyde, ketone). Their chain length ranged from six to nine carbon atoms (C6, C7, C8, and C9). The pairwise combination of 16 odours defined a 16 × 16 matrix. These odours are well discriminated by free-flying bees [ 21 ] and give consistent odour-evoked signals in optical imaging studies [ 23 ]. Using a behavioural approach, we measured similarity between odours and calculated their perceptual distances in a putative olfactory space. These perceptual distances were correlated with physiological distances measured in optical imaging experiments [ 23 ]. The correlation between both datasets was highly significant, thus indicating that odours that are encoded as physiologically similar are also perceived as similar by honeybees. Although other studies have addressed the issue of perceptual correlates of neural representations [ 32 , 33 ], we show for the first time that neural olfactory activity corresponds to olfactory perception defined on the basis of specific dimensions in a putative olfactory space, a finding that is of central importance in the study of the neurobiology of perception. Results We trained 2,048 honeybees along three trials in which one of the 16 odours used in our experiments was paired with a reward of sucrose solution (conditioned odour). Afterwards, each bee was tested with four odours that could include or not include the trained odour. Acquisition Phase The level of PER in the first conditioning trial was very low (between 0% and 8.60%) for all odours ( Figure 1 ). All the 16 odours were learnt but not with the same efficiency. An overall (trial × odour) analysis of variance (ANOVA) showed a significant increase in responses along trials ( F 2, 4064 = 2215.50, p < 0.001) and a significant heterogeneity among odours ( F 15, 2032 = 8.80, p < 0.001). Responses to the CS in the last conditioning trial reached a level of approximately 70% for primary and secondary alcohols, 80% for aldehydes, and 61% for ketones. Figure 1 Acquisition Curves for Primary Alcohols, Secondary Alcohols, Aldehydes, and Ketones The ordinate represents the percentage of proboscis extensions to the training odour (CS). The abscissa indicates the conditioning trials (C1, C2, C3) and the test with the CS (T). The curves correspond to molecules with 6 (white triangles), 7 (white diamonds), 8 (black circles) and 9 carbons (black squares); ( n = 128 bees for each curve). As not all 128 bees were tested with the odour used as CS, the sample size in the tests was smaller ( n = 32). Different letters (a, b, c) indicate significant differences either between acquisition curves for different chain-length molecules (in the case of the ketones) or between test responses (post hoc Scheffé tests). In the case of aldehydes and primary and secondary alcohols, no significant chain-length effect within functional groups was found over the whole conditioning procedure (chain length × trial ANOVA; chain-length effect for primary alcohols: F 3, 508 = 0.18, p > 0.05; secondary alcohols: F 3, 508 = 1.47, p > 0.05; and aldehydes: F 3, 508 = 1.26, p > 0.05). In contrast, bees conditioned to ketones showed a significant chain-length effect in the acquisition (chain length × trial ANOVA; chain-length effect: F 3, 508 = 20.00, p < 0.005). Scheffé post hoc comparisons showed that acquisition was significantly better for nonanone (81.25% responses in the last conditioning trial) than for all other ketones. Octanone (68.75% responses in the last conditioning trial) was also better learned than hexanone and heptanone (45.31% and 48.44% responses in the last conditioning trial, respectively) ( Figure 1 , bottom right). The effect over trials was significant in all cases ( p < 0.05) as bees learned all odours. The analysis of acquisition for each chain length separately revealed that it varied significantly depending on the functional group (functional group × trial ANOVA; C6: F 3, 508 = 18.89; p < 0.005; C7: F 3, 508 =10.78; p < 0.005; C8: F 3, 508 = 3.84; p < 0.01; C9: F 3, 508 = 2.73, p < 0.05). Scheffé post hoc comparisons generally showed that this effect was mainly due to ketones being less well learned than aldehydes and alcohols. Generally, the longer the carbon chain, the lower the heterogeneity in acquisition between functional groups. Thus, apart from short-chain ketones, all odours were learned similarly (reaching a level of acquisition between 60% and 80% in the last conditioning trial). Test Phase When the conditioned odour was presented in a test ( Figure 1 , grey panels), the level of PER recorded corresponded mainly to that found in the last acquisition trial (McNemar tests [2 × 2 Table]: in all cases p > 0.05). To compare generalisation after conditioning, and because acquisition levels were heterogeneous between odours, we built a generalisation matrix in which only bees responding to the CS at the end of training (3rd conditioning trial) were considered ( Figure 2 ). The number of individuals included in the statistical analysis varied within each ‘training odour/test odour' pair. The number of bees completing the tests varied between 17 and 28 for primary alcohols, between 13 and 29 for secondary alcohols, between 23 and 30 for aldehydes, and between 11 and 31 for ketones. The responses to the CS in the tests ranged between 70% and 100% in the generalisation matrix. All further analyses were carried out on this matrix. In the following sections, we will use the matrix data to analyse generalisation within and between functional groups, within and between chain lengths, and the asymmetries in olfactory generalisation. Figure 2 Olfactory Generalisation Matrix The generalisation matrix represents the percentage of PER in the tests performed by bees that actually learned the CS, that is, bees that responded to the CS at the third conditioning trial ( n = 1,457). Upper part: percentages recorded. Lower part: colour-coded graphic display grouping the level of responses in ten 10% response categories. Red, maximal response; light blue, minimal response. Generalisation within Functional Groups Figure 3 A shows the percentage of PER to odours having different (white quadrants) or the same (grey quadrants) functional group as the conditioned odour. High levels of PER to odours different from the trained one correspond to high generalisation. In order to better visualise generalisation as depending on functional groups, we pooled all the observed responses within each quadrant of Figure 3 A (i.e., not considering chain length) and calculated the resulting percentage of PER ( Figure 3 B). Grey bars correspond to generalisation to the same functional group; white bars correspond to generalisation to different functional groups. Generalisation mainly occurred within a given functional group (grey bars). This pattern was clearest for aldehydes ( Figure 3 B, 3 rd row) because bees conditioned to aldehydes responded with a high probability to other aldehydes but showed lower responses to any other odour (see also the clear aldehyde “response block” in Figure 2 ). Figure 3 Generalisation Depending on Functional Groups (A) Data of the generalisation matrix (see Figure 2 ) represented as two-dimensional graphs for each conditioned odour. The right ordinate represents the CSs categorised in four functional groups, primary alcohols, secondary alcohols, aldehydes, and ketones (from top to bottom). The abscissa represents the test odours aligned in the same order as the conditioned odours (from left to right). The left ordinate represents the percentage of proboscis extensions to the test odours after being trained to a given odour. Each quadrant in the figure represents generalisation responses to one functional group after training for the same (grey quadrants) or to a different functional group (white quadrants). (B) Same data as in (A), but the observed responses within each quadrant were pooled and the resulting percentage of responses per quadrant was calculated. The abscissa and the right ordinate represent the four functional groups. The left ordinate represents the percentage of proboscis extensions to each of these groups after being trained to a given group. Grey bars correspond to grey quadrants in (A) and represent generalisation to the same functional group as the conditioned one. White bars correspond to white quadrants in (A) and represent generalisation to a functional group different from the conditioned one: 1-ol, 2-ol, al, and one mean primary alcohol, secondary alcohol, aldehyde, and ketone, respectively. Asterisks indicate significant differences along a row or a column ( p < 0.001) (C) Within-functional group generalisation, depending on chain length. The abscissa represents the functional groups tested. The ordinate represents the percentage of proboscis extensions to the functional groups tested after being trained to a given chain-length (lines). Thus, for instance, the first point to the left for C9 molecules (black circles) represents generalisation to 1-hexanol, 1-heptanol, and 1-octanol after conditioning to 1-nonanol. A significant heterogeneity was found in within-functional group generalisation for C8 and C9 but not for C6 and C7 molecules. (D) Generalisation within-functional groups. The figure shows results from pooling the data of (C) corresponding to each functional group. Each point shows the percentage of proboscis extensions to odours of the same functional group as the conditioned odour. Within-group generalisation was significantly heterogeneous (asterisks, p < 0.001). Pairwise comparisons showed that generalisation within aldehydes was significantly higher than within primary alcohols or ketones and marginally higher than within secondary alcohols (different letters indicate significant differences). We analysed within-functional group generalisation as depending on chain length (see Figure 3 C). To this end we represented generalisation from C6, C7, C8, and C9 molecules having a given functional group to the other compounds having the same functional group (e.g., Figure 3 C, black circle curve, first data point: generalisation to 1-hexanol, 1-heptanol, and 1-octanol after conditioning to 1-nonanol). A significant heterogeneity appeared for C8 and C9 molecules (χ 2 = 12.60 and 14.30, respectively, p < 0.01 in both cases, n = 67–85) but not for C6 and C7 molecules ( p > 0.05). In the case of C8 and C9 molecules, generalisation was significantly higher within aldehydes ( p < 0.05). When comparing within-group generalisation over all four functional groups ( Figure 3 D), a significant heterogeneity appeared (χ 2 = 14.40, df = 3, p < 0.01, n = 276–316). Pairwise comparisons (using a corrected threshold for multiple comparisons: α′ = 0.017) showed that generalisation within aldehydes was significantly higher than within primary alcohols (χ 2 = 11.80, df = 1, p < 0.0006) and ketones (χ 2 = 9.90, df = 1, p < 0.005) and close to significance in favour of aldehydes when compared to secondary alcohols (χ 2 = 4.40, df = 1, 0.017 < p < 0.05). Generalisation within Chain Lengths Figure 4 A shows the generalisation responses of bees to odours having different (white quadrants) or the same (grey quadrants) chain length as the conditioned odour. In order to better visualise generalisation as depending on chain length, we pooled all the observed responses within each quadrant of Figure 4 A and calculated the resulting percentage of PER ( Figure 4 B). Grey bars correspond to generalisation to the same chain length; white bars correspond to generalisation to different chain lengths. Generalisation was highest in the case of odours with the same or similar chain length. Figure 4 Generalisation Depending on Chain Length (A) Data of the generalisation matrix (see Figure 2 ) represented as two-dimensional graphs for each conditioned odour. The right ordinate represents the CSs categorised in four chain lengths, C6, C7, C8, and C9 molecules (from top to bottom). The abscissa represents the test odours aligned in the same order as the conditioned odours (from left to right). The left ordinate represents the percentage of proboscis extensions to the test odours after being trained for a given odour. Each quadrant in the figure represents generalisation responses to one chain length after training for the same (grey quadrants) or to a different chain length (white quadrants). (B) Same data as in (A), but the observed responses within each quadrant were pooled and the resulting percentage of responses per quadrant was calculated. The abscissa and the right ordinate represent the four chain-length categories. The left ordinate represents the percentage of proboscis extensions to each of these categories after being trained for a given chain-length category. Grey bars correspond to grey quadrants in (A) and represent generalisation to the same chain length as the conditioned one. White bars correspond to white quadrants in (A) and represent generalisation to a chain length different from the conditioned one: C6, C7, C8, and C9 mean chain length of 6, 7, 8, and 9 carbons, respectively. Asterisks indicate significant differences along a row or a column ( p < 0.001). (C) Within chain-length generalisation as depending on functional group. The abscissa represents the chain lengths tested. The ordinate represents the percentage of proboscis extensions to the same chain length after being trained to a given functional group (lines). Thus, the first point to the left for ketones (red circles) represents generalisation to 1-hexanol, 2-hexanol, and hexanal after conditioning to 2-hexanone; the second point represents generalisation to 1-heptanol, 2-heptanol, and heptanal after conditioning to 2-heptanone. A significant heterogeneity was found in within-chain-length generalisation for aldehydes and ketones. (D) Generalisation within-chain lengths. The figure results from pooling the data of (C) corresponding to each chain length. Each point shows the percentage of proboscis extensions to odours of the same chain length as the conditioned odour. Within-chain-length generalisation was significantly heterogeneous (asterisks, p < 0.001). Pairwise comparisons showed that generalisation within C9 molecules was significantly higher than within C7 and C6 molecules and marginally higher than within C8 molecules (different letters indicate significant differences). We analysed within-chain length generalisation as depending on functional group ( Figure 4 C). To this end we represented generalisation from primary alcohols, secondary alcohols, aldehydes, or ketones of a given chain length to the other compounds having the same chain length (e.g., Figure 4 C, red circle curve, first data point: generalisation to 1-hexanol, 2-hexanol, and hexanal after conditioning to 2-hexanone). Generalisation within-chain length was generally higher for longer than for shorter chain lengths. This effect was significant for aldehydes (χ 2 = 28.70, df = 3, p < 0.01, n = 75–80) but not for primary and secondary alcohols (χ 2 = 5.20 and 3.4, df = 3, p > 0.05, n = 67–73 and n = 61–66, respectively). For ketones, a significant heterogeneity was found (χ 2 = 10.00, df = 3, p < 0.05, n = 40–79), but generalisation was more important between C8 than between C7 molecules. The generalisation corresponding to other chain lengths fell in between. When comparing within-chain length generalisation over all four chain-length groups ( Figure 4 D, i.e., not considering functional group), a significant heterogeneity appeared χ 2 = 23.2, df = 3, p < 0.001, n = 247–293). Pairwise comparisons (using a corrected threshold for multiple comparisons: α′ = 0.017) showed that within-chain length generalisation was significantly higher within C9 than within C6 (χ 2 = 18.50, df = 1, p < 0.0001) and C7 molecules (χ 2 = 15.00, df = 1, p < 0.0001). Generalisation within C8 molecules was close to significance when compared to generalisation within C9 molecules (χ 2 = 5.00, df = 1, 0.017 < p < 0.05), and it was significantly higher than generalisation within C6 molecules (χ 2 = 4.3, df = 1, 0.017 < p < 0.05). Generalisation between Functional Groups To analyse generalisation between groups, we took into account the responses to functional groups different from the conditioned one (see white bars in Figure 3 B). Bees showed heterogeneous patterns of generalisation (all vertical and horizontal comparisons in Figure 3 B were significant: χ 2 > 37.70, df = 3, p < 0.001, in all eight cases). We found high between-group generalisation for primary and secondary alcohols: bees conditioned to secondary alcohols responded preferentially to primary alcohols, somewhat less to aldehydes, and even less to ketones (see Figures 3 A and 3 B, second row). A similar but less obvious response gradation was found for bees conditioned to primary alcohols Figures 3 A and 3 B, first row). In fact, the overall generalisation patterns were very similar for primary and secondary alcohols sharing the same chain length (see, for instance, the very close relationship between the two sets of blue [primary alcohol] and green curves [secondary alcohols] in Figure 4 A). As indicated before, bees conditioned to aldehydes generalised very little to odours belonging to other functional groups (see Figure 3 B, third row). Contrarily, bees conditioned to other functional groups highly generalised to aldehydes (see third column ‘al' in Figure 3 B). This shows that generalisation between aldehydes and odours belonging to other functional groups was asymmetrical. The topic of asymmetric generalisation will be considered below in more detail. Generalisation between Chain Lengths To analyse generalisation between chain lengths, we took into account the responses to chain lengths that differed from the conditioned one (see white bars in Figure 4 B). In general, responses to molecules with different chain lengths followed a clear decreasing gradient, depending on the difference in the number of carbon atoms between the molecules considered (see Figure 4 B; all horizontal and vertical comparisons were significant, χ 2 > 16.3, df = 3, p < 0.001 in all eight cases). For instance, when conditioned to a C9 molecule (see Figure 4 B, fourth row), bees responded in 53%, 31%, and 23% of the cases to C8, C7, and C6 molecules, respectively, while they responded to C9 molecules in 67% of the cases. This gradient was also evident when generalisation took place between functional groups: for instance, after training with 2-nonanol (see Figure 3 A, second row), the response of bees to odours of different functional groups (solid lines in white boxes) always followed a similar decreasing tendency with the same (C9) or similar (C8) chain length on top. Asymmetry in Olfactory Generalisation As previously mentioned, some groups like aldehydes induced asymmetrical cross-generalisation (i.e., bees responded less to other functional groups after training for aldehydes than to aldehydes after training for other functional groups). We analysed this asymmetrical generalisation and built an asymmetry matrix ( Figure 5 A). To this end, we calculated for each odour pair (A and B) the difference (in percentage) between generalisation from A to B and generalisation from B to A. Such differences were ranked in 10% categories from −55% to 55%. White boxes indicate no asymmetries. Blue shades in Figure 5 A indicate that cross-generalisation was biased towards odour A (i.e., conditioning to A resulted in lower generalisation to B while conditioning to B resulted in higher generalisation to A); red shades indicate that cross-generalisation was biased towards odour B (i.e., conditioning to A resulted in higher generalisation to B while conditioning to B resulted in lower generalisation to A). This representation showed that some odours induced generalisation while other odours diminished it. For instance, hexanal was well learnt but induced low generalisation to other odours, except to other aldehydes. On the other hand, bees conditioned to other odours very often generalised to hexanal. Thus, a clear blue row (or a red column) corresponds to hexanal in the asymmetry matrix. Conversely, 2-hexanone induced high generalisation to other odours but received few responses as a test odour. Thus a red row (or a blue column) corresponds to 2-hexanone in the asymmetry matrix. Most odours, however, showed little or no asymmetry. Figure 5 B presents the mean asymmetry found for each training odour. In six cases, the mean asymmetry deviated significantly from zero, which represents a theoretically perfect symmetry ( t -test). Two odours (red bars) significantly induced generalisation (2-hexanone and 2-hexanol, t -test, df = 14, p < 0.001 and p < 0.01, respectively), while four odours (blue bars) diminished it significantly (hexanal, heptanal, and octanal, and 2-nonanone, t -test, df = 14, p < 0.001 for the former and p < 0.01 for the three latter odours). Figure 5 Asymmetric Generalisation between Odours (A) The asymmetry matrix depicts asymmetric cross-generalisation between odours. For each odour pair (A and B), the difference (percentage) between generalisation from A to B and generalisation from B to A was calculated. Such differences were ranked in 10% categories varying from blue (−55%) to red (55%). Blue shades indicate that cross-generalisation was biased towards odour A (i.e., conditioning to A resulted in lower generalisation to B, while conditioning to B resulted in higher generalisation to A); red shades indicate that cross-generalisation was biased towards odour B (i.e., conditioning to A resulted in higher generalisation to B, while conditioning to B resulted in lower generalisation to A). For this reason, each odour pair (A and B) appears twice in the matrix, once in the upper-left of the black diagonal line, and once in the lower-right of the black diagonal line, with opposite values. See, for example, the two cells outlined in green for the pair 2-hexanone/2-octanol. (B) Mean generalisation induced or diminished by each odour A in (A). Each bar represents the mean asymmetry of the respective horizontal line in the asymmetry matrix. Red bars show that an odour induced more generalisation than it received, while blue bars show the opposite. Significant generalisation asymmetries were found in six out of 16 cases (**, p < 0.01; ***, p < 0.001). Olfactory Space In order to define a putative olfactory space for the honeybee, we performed a principal component analysis (PCA) on our data to represent in a limited number of dimensions the relative relationships between odorants in a 16-dimension perceptual space ( Figure 6 A). The first three factors represented 31%, 29%, and 15% of overall variance in the data (total of the first three factors: 75%). The analysis showed a clear organisation of odours depending on their chemical characteristics. First, chain length was very clearly represented by the first factor (see upper-right graph in Figure 6 A), from C6 to C9 molecules from the right to the left. On the other hand, the chemical group was mostly represented by factors 2 and 3. Whereas factor 2 separated mostly aldehydes from alcohols, with ketones falling between them, factor 3 segregated ketones from all other odours (lower-right graph, Figure 6 A). None of these factors separated primary and secondary alcohols. This analysis indicates that the chemical features of molecules (chain length and functional group), which are sometimes thought of as artificial perceptual (psychophysical) dimensions determined by experimenters [ 34 ] can be considered as true inner dimensions of the bees' perceptual space. Cluster analyses performed on the data segregated odours mostly according to their chain length. In the first group ( Figure 6 B, upper part), we found two subgroups, short-chain alcohols (C6 and C7, primary and secondary alcohols) and short-chain ketones (C6 to C8). On the other hand ( Figure 6 B, lower part), three clear subgroups were formed: short-chain aldehydes (C6 and C7), long-chain alcohols (C8 and C9, primary and secondary alcohols), and a last group with long-chain aldehydes (C8 and C9) and 2-nonanone. Very similar results were obtained using Euclidian or city-block metrics. Figure 6 A Putative Honeybee Olfactory Space (A) Left: The olfactory space is defined on the basis of the three principal factors that accounted for 76% of overall data variance after a PCA performed to represent the relative relationships between odorants. Primary alcohols are indicated in blue, secondary alcohols in green, aldehydes in black, and ketones in red. Different chain-lengths are indicated as C6, C7, C8, and C9, which corresponds to their number of carbon atoms. For each functional group, arrows follow the increasing order of carbon-chain lengths. Right: Chain length was very clearly represented by factor 1. C6 to C9 molecules are ordered from right to left. The chemical group was mostly represented by factors 2 and 3. Whereas factor 2 separated mostly aldehydes from alcohols, with ketones falling between them, factor 3 separated ketones from all other odours. None of these three factors separated primary and secondary alcohols. (B) Euclidean cluster analysis. The analysis separated odours mostly according to their chain length. Linkage distance is correlated to odour distances in the whole 16-dimension space. The farther to the right two odours/odour groups are connected, the higher the perceptual distance between them (odour colour codes are the same as in [A]). Correlation between Optophysiological and Behavioural Measures of Odour Similarity We asked whether optophysiological measures of odour similarity, obtained using calcium imaging techniques at the level of the honeybee AL [ 22 , 23 , 24 , 35 ], correspond to perceptual odour similarity measures as defined in our putative honeybee olfactory space. We thus calculated the Euclidian distance between odour representations in our 16-dimension “behavioural” space for all odour pairs (120 pairs). We then calculated distances between odours in optical imaging experiments, using the odour maps by Sachse et al. [ 23 ]. A correlation analysis was performed between both datasets. This analysis was possible because both the study by Sachse et al.[ 23 ] and our study used the same set of odours delivered under the same conditions. Figure 7 A presents the correlation obtained, including all 120 odour pairs. Both sets of data were highly significantly correlated ( r = 0.54, t 118 = 7.43, p < 2.10 –10 ), a result that shows that odours, which were found to be physiologically similar in the optical imaging study, were also evaluated as similar in behavioural terms. Note, however, that data points cluster quite broadly around the main trend line, showing that many exceptions were found. In order to use a more exact measure of physiological odour similarity, we used the correlation results between primary and secondary alcohol maps provided by Sachse et al. [ 23 ]. By correlating this more exact value of physiological similarity with our behavioural data, we also found a highly significant relationship between physiological and behavioural data ( Figure 7 B; r = 0.82, t 26 = 7.83, p < 7.10 –8 ). The correlation coefficient achieved with this second method was significantly higher than that achieved with the first method ( Z = 2.52, p < 0.05). A better fit between the two datasets was thus found, although outliers were still present in the data. These two analyses show that optophysiological and behavioural measures of odour similarity correlate well using the methods described here. Thus, in the case of the honeybee, olfactory neural activity corresponds to olfactory perception. Figure 7 Correspondence between Perceptual and Physiological Odour Similarity (A) Correlation between optophysiological measures of odour similarity (carried out using calcium imaging recordings [ 23 ]) and our behavioural measures of odour similarity. Euclidian distance between odour representations in our 16-dimension “behavioural” space for all odour pairs (120 pairs, x axes) and distances between odours in optical imaging experiments, using the odour category maps displayed by Sachse et al. [ 23 ] (also 120 pairs, y axes) were calculated. This correlation, including all 120 odour pairs, was highly significant ( r = 0.54, p < 0.001). Odours found to be similar in the optical imaging study were also similar in the behaviour. Data points cluster quite broadly around the main trend line, showing that many exceptions were found. (B) Correlation between measures of optophysiological similarity carried out using the optical imaging technique [ 23 ] and our behavioural measure of odour similarity. Using the exact data given for primary and secondary alcohols [ 23 ], a much better correlation between the two datasets was achieved than in (A) ( r = 0.82, p < 0.001), although outliers were still found in the data. Discussion In the present work, we have studied perceptual similarity among odorants in the honeybee, using an appetitive-conditioning paradigm, the olfactory conditioning of the PER [ 17 , 18 ]. We showed that all odorants presented could be learned, although acquisition was lower for short-chain ketones. Generalisation varied, depending both on the functional group and on the carbon-chain length of odours trained. Generalisation was very high among primary and secondary alcohols, being high from ketones to alcohols and aldehydes and low from aldehydes to all other tested odours; thus, in some cases, cross-generalisation between odorants was asymmetric. Some odours, like short-chain ketones or aldehydes, induced more asymmetries than other odours. Higher generalisation was found between long-chain than between short-chain molecules. Functional group and carbon-chain length constitute orthogonal inner dimensions of a putative olfactory space of honeybees. Perceptual distances in such a space correlate well with physiological distances determined from optophysiological recordings performed at the level of the primary olfactory centre, the AL [ 23 ] such that olfactory neural activity corresponds to olfactory perception. Previous studies have attempted to describe olfactory generalisation in honeybees and to study structure–activity relationships [ 19 , 20 , 36 , 37 , 38 ]. These studies generally supported the view that generalisation mainly happens when odours belong to the same chemical group. Moreover, they also suggested that the rules underlying olfactory learning and perception of different chemical classes [ 20 ] or of particular odorants (e.g., citral [ 20 , 37 ]) may vary. However, these studies used differential training, thus inducing several generalisation gradients (excitatory and inhibitory) that make the interpretation of generalisation responses difficult [ 21 , 36 ]. Furthermore, these studies were carried out on a rather discrete number of odour pairs [ 37 ], did not detail the results obtained with individual odour combinations [ 20 ], or used a very reduced number of bees per conditioned odour ([ 21 ]; two bees per odorant).Thus, the present study is the first one to provide (i) generalisation data based on absolute conditioning (i.e., only one odour conditioned at a time), (ii) a systematical test of all odour combinations, (iii) robust sample sizes for each experimental situation, and (iv) important generalisation gradients. These are in our view crucial prerequisites to describe odour perception and similarity in a precise way. Chemical Group and Chain Length Several studies in other species have shown the importance of functional group and carbon-chain length of the odour molecules for behavioural responses to odours. Differences in the response between molecules of diverse aliphatic and aromatic homologue odour classes (i.e., differing in functional group, chain length, and overall molecule form) were investigated in moths [ 39 , 40 ], cockroaches [ 41 ], rats [ 42 ], squirrel monkeys [ 4 , 43 ] and humans [ 38 , 44 , 45 ]. These studies show that both functional group and chain length affect the perceived quality of an odorant. Concerning chain length, the greater the difference in the number of carbons between odours, the easier the discrimination and the lower the generalisation ([ 21 , 40 , 42 , 44 ] and present study). In our study, both chemical group and chain length of odour molecules determined the bees' generalisation responses. Bees mostly generalised to other odours when these shared the same functional group. This effect was observed for all functional groups (see Figure 3 B) but was strongest for aldehydes. Other studies have found that aldehydes induced high within-group generalisation [ 20 , 21 , 36 ]. Thus, aldehydes may represent a behaviourally relevant chemical class for honeybees. Between-functional group generalisation depended on the functional group considered. It was high between primary and secondary alcohols, which appear therefore perceptually similar to the bees, and low between other chemical groups. Bees clearly generalised between odours that shared the same chain length. Increasing chain length promoted generalisation. Moreover, generalisation to other chain lengths decreased if the difference in the number of carbons between odours increased. This suggests a perceptual continuum between different chain lengths (but see below). Thus, the chemical structure of the odorants is critical for determining the amount of generalisation. A Putative Olfactory Space for the Honeybee We found that the two controlled physical characteristics of odour molecules used in this study, functional group and chain length, correspond to internal dimensions in the bees' olfactory perceptual space such as the three most important factors extracted in our PCA analysis, one mainly represented chain length and the other two were mostly influenced by functional group. Cluster analyses allowed separating odours in clusters according to their functional groups and their chain length. Interestingly, C6 and C7 molecules and C8 and C9 molecules were mainly grouped together, so that, for instance, all short-chain primary and secondary alcohols were grouped on one side, and all long-chain alcohols on the other side. The same happened for aldehydes, and in a different way for ketones (C9 separated from the rest). This discrepancy suggests that, although chain length appears mostly as a perceptual continuum in the PCA analysis, there may be a perceptual “jump” between short-chain and long-chain molecules. Neural Bases of Odour Perception Both in vertebrates and in invertebrates, studies quantifying the neural responses to structurally similar odours in the first relay of the olfactory pathway have been performed (olfactory bulb: e.g., [ 46 , 47 , 48 , 49 ]; AL: [ 23 , 50 ]). These studies show that activity patterns are more similar when the difference in the number of carbons between molecules is small. It was hypothesised that such a physiological similarity is the basis for olfactory discrimination and generalisation as measured behaviourally. This has indeed been reported for mucosal activity in mice [ 51 ], electrical mitral cell activity [ 42 ], and/or radiolabelled 2-deoxyglucose uptake in the rat olfactory bulb [ 32 ]. Also, in Manduca sexta, qualitative similarities were observed between the degree of behavioural generalisation according to chain length [ 40 ] and the degree of overlap between electrophysiological temporal patterns of activity across AL neurons [ 50 ]. Several correspondences, but also discrepancies, can be found between our behavioural results and the physiological results obtained at the level of the bee AL [ 23 ]. First, within the regions of the AL accessible to optical imaging (about 25% of the glomeruli), patterns of glomerular activity for different odours are highly dependent on chain length, but much less so on chemical group. Thus, most active glomeruli respond to several functional groups as long as the chain length corresponds, but respond differentially to different chain lengths. Glomeruli T1–28 and T1–52 are specialised in short-chain molecules (respectively C5–C7 and C6–C7), whilst glomeruli T1–33 and T1–17 are specialised in long-chain molecules (respectively C7–C9 and C8–C9). These glomeruli also respond to most functional groups but in a graded way. For instance, glomerulus T1–17 responds more to alcohols in the intermediate range than to aldehydes or ketones, whereas T1–52 generally responds more to ketones in the short range, more to aldehydes in the long range, and overall little to alcohols. No individual glomerulus was found that responds specifically to a chemical group. However, it should be kept in mind that some regions of the ALs are not yet accessible to calcium imaging techniques (about 75% of the lobe; see below). Thus, a possible explanation is that glomeruli responding to specific chemical groups (or with responses more dependent on chemical groups than on chain length) were not imaged. Second, primary and secondary alcohols induce extremely similar activation patterns in the AL, but subtle differences could be found, so that for a given chain length, the representation of a secondary alcohol was between that of the primary alcohol of the same chain length and that with one less carbon atom (see Figure 6 B in Sachse et al. [ 23 ]). We found a similar arrangement of alcohol representations, with primary and secondary alcohols alternating on a common axis (see Figure 6 A). Third, optical imaging data showed that higher chain lengths support more similarity between patterns (see Figure 6 C in Sachse et al. [ 23 ]). Our finding that longer chain lengths induce more generalisation agrees with the imaging data. These last two points suggest that the general rules governing odour similarity at the neural and the behavioural level are similar. The Correspondence between Perceptual and Physiological Odour Similarity We aimed at comparing behavioural and physiological data in a more precise way, using correlation analyses between our behavioural similarity matrix, in which distances between two odour points represent psychological distances between stimuli, and a physiological similarity matrix obtained from optophysiological recordings of glomerular activation patterns [ 23 ]. Comparing distances between odours in these two matrixes resulted in a good correlation. This means that glomerular activity patterns recorded in the brain could predict behavioural responses and vice versa. The optophysiological dataset of Sachse et al. [ 23 ] has nevertheless some limitations with respect to the objectives of our work: (i) bath application measurements of AL activity using calcium green as a dye [ 23 ] record the combined activity of several neuronal populations of the AL, among which primary-afferent activity seems to have the most important contribution [ 52 ]; (ii) such measurements survey only the dorsal part of the AL, which constitutes 25% of the neuropile studied; and (iii) learning alters odour representations in the AL [ 35 , 53 , 54 ] such that there could be a mismatch between our data collected after olfactory conditioning and the dataset of Sachse et al. [ 23 ], which was obtained from naive bees. With respect to the first point, it could be argued that the AL circuitry transforms the primary-afferent representations of odours [ 25 ] such that recordings where primary-afferent receptor activity is predominant are not very useful for evaluating optophysiological similarity. However the very fact that we found a significant correlation between our behavioural data and the imaging data by Sachse et al. [ 23 ], strongly suggests that the perceptual quality of odorants mostly appears at the peripheral level. Clearly, this correlation was not perfect, and odour quality is most probably refined by further processing within the AL, and/or at higher stages of the olfactory pathway, such as in the mushroom bodies or the lateral protocerebrum. In honeybees, new methods have been developed, which allow recording selectively the activity of the efferent PNs [ 25 ]. However, the two studies published using this method [ 25 , 26 ] do not provide an extensive odorant matrix as that provided by Sachse et al. [ 23 ]. In this sense the study on which we based our correlation analysis is certainly the only one of its kind published to date. However, in the future, a careful comparison of our behavioural data with both bath-applied imaging data emphasising receptor neuron input (as done here) and selective imaging of PNs would be extremely helpful in understanding to what extent AL processing shapes odour perceptual quality. With respect to the second point, calcium imaging recordings of AL activity are certainly limited to the dorsal part of the AL, which is the region accessible when the head capsule is opened in order to expose the brain for recordings. This is an inherent limitation of the method that the use of two-photon microscopy during calcium imaging measurements will soon allow us to overcome, as shown already by recordings obtained in the fruit fly Drosophila melanogaster [ 55 ]. Finally, with respect to the third point, it is known that learning alters odour representations in the AL, when bees are trained in a differential conditioning procedure, with one odour rewarded and another odour unrewarded [ 53 ]. This is not the conditioning procedure used in our work, which was absolute (only one odour rewarded at a time). In the bee, changes in the olfactory code due to absolute conditioning seem to be difficult to detect (C. G. Galizia, personal communication), such that this point may not be so critical for our correlation analysis. In any case, if there are changes in odour representations due to conditioning, recording glomerular activity patterns after conditioning would only improve our correlation analyses. Generalisation Asymmetries between Odours We have found a number of asymmetries in olfactory cross-generalisation, with bees responding more to odour B after learning odour A than in the reverse situation. Previous studies have observed such a phenomenon, but it was mostly related to olfactory compounds with pheromonal value (aggregation pheromone citral [ 20 , 37 ] and alarm pheromones 2-heptanone and isoamyl acetate [ 56 ]). In the present study, we found that six out of the 16 odours used induced significant generalisation asymmetries over the whole matrix; none of these six odours was related to any known pheromone (see Table 1 ). Generalisation asymmetries seem to be a general feature of honeybee olfaction. Table 1 Chemical and Biological Characteristics of the Odours Used The odours were listed by functional groups (primary alcohols, secondary alcohols, aldehydes, and ketones) and purity. Odour vapour pressure values (VP), pheromone characteristics and occurrence in floral scents (after Knudsen et al. [ 66 ]) are also given a Notation: *1, releases altering at hive entrance and stinging, repels clustering bees, inhibits scenting, repels foragers (sting chamber); *2, releases altering at hive entrance, inhibits foraging activity, repels foragers (sting chamber); *3, repels at hive entrance, releases stinging, encourages foraging activity (sting chamber); *4, releases stinging, inhibits foraging activity, repels foragers (mandibular glands) Odour concentration can affect stimulus salience. In our work, generalisation asymmetries could not be directly explained by differences in odour concentration (through differences in vapour pressure), because, for instance, the two odours with the highest vapour pressure in our sample (2-hexanone and hexanal) produced totally opposite results: 2-hexanone induced important generalisation, while hexanal strongly reduced generalisation. Also, although we used 16 different odours with a range of different vapour pressures, we found that acquisition was very similar for most odours, except for the short-chain ketones, which were less easily learned. This suggests that almost all odours used had a good salience for bees. Wright and Smith [ 57 ] studied the effect of odour concentration in generalisation in honeybees. They found that discrimination increased with concentration for structurally dissimilar odours but not for similar odours. Further experiments using odorants at different concentrations should be carried out to determine the effect of odour concentration on generalisation asymmetries. Generalisation asymmetries could be due to innate or experience-dependent differences in the salience of odours for honeybees, such that more salient odours would induce higher generalisation than less salient odours. This interpretation implies that most aldehydes (hexanal, heptanal, and octanal) are highly salient odours for honeybees, because aldehydes showed a clear “functional group” effect, which could reveal a certain bias of the olfactory system towards these odours. Ketones, on the other hand, showed a heterogeneous effect, as 2-hexanone seemed to have a low salience (it was not well learnt) and induced a high generalisation to other odours, while 2-nonanone consistently reduced generalisation to other odours. In the group of alcohols, only 2-hexanol induced generalisation to other odours. Therefore, only aldehydes showed a clear group effect on generalisation asymmetry. This effect could be due to innate odour preferences [ 58 , 59 ] or to previous odour exposure within the hive [ 60 , 61 ]. Innate odour preferences could be related to natural, floral odours that were more consistently associated with food resources [ 20 , 62 ]. It is thus important to investigate whether or not such ecological trends exist in the natural flora associated with the honeybee and whether or not other bee species also present such clear biases, in particular towards aldehydes. Conversely, asymmetries could be the result of the conditioning procedure. This would be the case if conditioning modifies odour representation in an asymmetric way. Indeed, experience-induced modifications of odour representations have been found at the level of the honeybee AL. Thus, odour-evoked calcium signals in the AL can be modified by elemental [ 53 ] and nonelemental olfactory learning paradigms [ 35 ] such that the representations of odours that have to be discriminated become more distinct and uncorrelated as a result of learning. In the fruit fly D. melanogaster , new glomeruli become active after olfactory learning [ 54 ], while in the moth M. sexta new neuronal units in the AL are recruited after olfactory learning [ 63 ]. These elements suggest that modifications of odour representation after learning two different odours could indeed be asymmetrical: if, for instance, the neuronal representation of A after conditioning becomes A′, which is slightly farther away from B than A in the bee's olfactory space, and if the perceptual representation of B becomes B′ after conditioning, which is closer to A than B, then bees would show less generalisation in behavioural tests from A to B than from B to A. On the level of the AL network, glomeruli are connected via lateral inhibitory interneurons [ 25 , 64 , 65 ]. Due to this, glomerular activation by an odour A will transiently inactivate parts of the network and possibly parts encoding a subsequent odour B. Optical imaging experiments have shown that inhibition between glomeruli may be asymmetric [ 25 ]. In our case, glomeruli activated by odour A may inhibit glomeruli coding for odour B, while glomeruli coding for odour B may not inhibit those coding for odour A. In this hypothesis, asymmetric cross-generalisation could reflect a sensory phenomenon. Nevertheless, we believe that inhibitions at the level of the AL are rather short-lived such that a purely sensory priming effect seems improbable. If, however, the strength of lateral inhibitions between glomeruli can be modified by learning as proposed by Linster and Smith [ 65 ], then asymmetrical generalisation would come from the fact that inhibitory lateral connections are modified. In order to determine the physiological mechanisms underlying asymmetrical cross-generalisation and the possible role of AL networks in it, future work will aim at visualising the evolution of glomerular activity patterns during and after olfactory conditioning with odours that showed asymmetries in our study. Conclusion We have shown that the two odorant physical dimensions that varied in our study, functional group and chain-length, correspond to internal dimensions of the bees' olfactory space. Generalisation was mainly due to these two characteristics with generalisation within functional group being more important. Such generalisation was particularly high for aldehydes, a fact that suggests that these odours may have an intrinsic value for bees. Generalisation between functional groups was mostly found between primary and secondary alcohols. Furthermore, a gradient in generalisation was found with respect to chain length. Asymmetric cross-generalisation was found in the case of certain odorants. Such asymmetries were neither strictly linked to chain length nor to functional group, but depended on particular odorants. The 16 odours used in our work represent a small part of the odorants that bees may encounter in nature (see Knudsen et al. [ 66 ]). For a complete description of the bees' olfactory perceptual space, more odours having other molecular features have to be studied. New dimensions in the bees' perceptual space could then be found. Finally, and most important, the perceptual distance between odours can be predicted on the basis of the differences in the patterns of glomerular activation in the first relay of the olfactory pathway: the AL, and vice versa. This emphasises the relevance of studying activity patterns in the brain in imaging studies and trying to relate them to perceptual tasks. Our work shows that this objective, which is at the core of cognitive neurosciences, can be achieved using an invertebrate model such as the honeybee. Materials and Methods Insects Every experimental day, honeybees were captured at the entrance of an outdoor hive and were cooled on ice for 5 min until they stopped moving. Then they were harnessed in small metal tubes in such a way that only the head protruded. The mouthparts and the antennae could move freely. Harnessed bees were left for 3 h in a resting room without disturbance. Fifteen minutes before starting the experiments, each subject was checked for intact PER by lightly touching one antenna with a toothpick imbibed with 50% (w/w) sucrose solution without subsequent feeding. Extension of the proboscis beyond the virtual line between the open mandibles was counted as PER. Animals that did not show the reflex were not used in the experiments. Stimulation apparatus The odours were delivered by an odour cannon, which allowed the presentation of up to seven different odours, and a clean airstream [ 67 ]. Each odour was applied to a filter paper placed within a syringe (see below) that was connected to the cannon. An airstream was produced by an air pump (Rena Air 400, Annecy, France) and directed to the relevant syringes with electronic valves (Lee Company, Voisins-le-Bretonneux, France) controlled by the experimenter via a computer. In the absence of odour stimulation, the airstream passed through a syringe containing a clean filter paper piece (clean airstream). During odour stimulation, the airstream was directed to a syringe containing a filter paper loaded with odour. After a 4-s stimulation, the airstream was redirected to the odourless syringe until the next stimulation. Stimuli Sixteen odours (Sigma Aldrich, Deisenhofen, Germany) were used in our work as CS and test stimuli (see Table 1 ). Racemic mixtures were used in the case of molecules that had chiral carbons. These odours are present in flowers and some in pheromones (see Table 1 ). Pure odorants (4 μl) were applied to 1-cm 2 filter paper pieces, which were transferred to 1-ml syringes, cut to 0.7 ml to make them fit into the odour cannon. Fifty percent sugar solution was used throughout as US. Experimental design Our work was designed to obtain a generalisation matrix with 16 different odours. Ideally, after conditioning each of the 16 odours as CS, the response to each odour (including the CS) should be measured (i.e., 16 × 16 = 256 cells). However, testing 16 odours implies presenting them without reward, a situation that may result in extinction of the learned response due to the repeated unrewarded odour presentations. Preliminary experiments were performed in which four groups of 180 bees were trained along three trials to 1-hexanol, 2-octanol, linalool, and limonene, respectively. Training was followed by tests with the four different odours, including the conditioned one. These experiments showed that after three conditioning trials, the response of the bees to the CS in the four tests remained at the same level, independently of the order of occurrence of the CS such that it was not influenced by extinction. We thus kept this protocol for the 16 × 16 matrix. Each of the 2,048 bees used in this study was thus subjected to three conditioning trials with their respective CS, and to four test trials, each with a different odour chosen among the 16 possible odours. Intertrial intervals of 10 min were used throughout. A randomisation schedule (detailed below) was developed for the test phase to reduce any possible day- and odour-combination effects. Conditioning trials One bee at a time was placed into the conditioning setup. The total duration of each trial was 37.5 s After 15 s of familiarisation to the experimental context, the CS was presented to the bee for 4 s. Three sec after onset of the CS, the antennae were stimulated with the US, leading to a proboscis extension. The bee was allowed to feed for 3 s. Stimulus overlap was 1 s (interstimulus interval, 3 s). The bee was left in the conditioning place for 17.5 s and then removed. Test trials The procedure was similar to that for conditioning trials but no US was given after odour delivery. After the four test trials, PER to the US was checked once again. Animals unable to show PER at this point were not considered for the analyses. Overall, less than 2% of the bees died during the experiment, and less than 1% of the survivors showed no US reaction at the end of the tests. Randomisation schedule On each day, two to three experimenters worked in parallel, each training 16 bees at a time. In the training phase, the 16 bees were divided into four groups of four bees, and each group was trained to one of the 16 different odours. In the test phase, four out of 16 odours were presented to each of the 16 bees. The combination of four odours tested together changed in each experiment, so that any effect of having particular odours in the same test combination was suppressed. The whole experiment was planned in such a way that in any of our experimental groups, two given odours appeared at least once, but a maximum of three times together in a test sequence. This was possible by carefully picking out eight of the 16! (2.1 × 10 13 ) possible experimental plans. Additionally, within each group, the testing order for the four test odours was determined randomly. Data analysis and statistics During the experiments, we recorded the response to the presented odour, that is, whether bees extended their proboscis after the onset of the odour and before the presentation of the sucrose solution in the case of reinforced trials, such that the anticipatory response recorded was due to the odour and not to the US. Multiple responses during a CS were counted as a single PER. The percentages of PER recorded during acquisition were used to plot acquisition curves (see Figure 1 ). To test whether bees learnt the different odours in a similar way, ANOVAs for repeated measurements were used both for between-group and for within-group comparisons. Monte Carlo studies have shown that it is permissible to use ANOVA on dichotomous data only under controlled conditions [ 68 ], which are met by the experiments reported in this study: equal cell frequencies and at least 40 df of the error term. The α level was set to 0.05 (two-tailed). To ensure that we analysed a true generalisation response in the tests, and hence built a true generalisation matrix, we kept only those bees which had actually learnt the CS (71% of the bees used in this work). We therefore performed new analyses that only included those bees that responded to the CS before the presentation of the US in the third conditioning trial. A lack of response to an odour in the tests could be due either to the fact that the bees had not made any association between CS and US or because their motivational level was low. For all odours tested, we observed that responses to the CS in the third conditioning trial were equivalent to responses to the CS in the tests (McNemar test; see Results). We represented the responses of the selected bees to the test odours (see Figure 2 ). As the numbers of bees were now heterogeneous in the different groups, we could not use ANOVAs to analyse the responses in the tests (see above). We thus used χ 2 tests for all further between-group comparisons. In the case of multiple two-by-two comparisons, the significance threshold was corrected using the Dunn–Sidak correction [α′ = 1 − (1 − α) 1/k where k is the number of two-by-two comparisons in which each dataset is used] in order to reduce the type I errors. Alpha values between α′ and 0.05 were considered as near significant. Olfactory space To observe the relationships between odours in a reduced number of dimensions, we performed a PCA, which identified orthogonal axes (factors) of maximum variance in the data, and thus projected the data into a lower-dimensionality space formed of a subset of the highest-variance components. We calculated the three factors, which accounted for most of the observed variance. Calculating distances between odours in the resulting putative olfactory space allowed the evaluation of their perceptual similarity, not only based on direct generalisation between these odours (i.e., generalisation from odour A to odour B and vice versa), but also including responses to these odours after conditioning to other odours (e.g., C, D, E, etc.). We performed cluster analyses to group odours, according to their respective distance in the olfactory space, using both Euclidian and city-block metrics, with Ward's classification method. Both metrics gave very similar results, so we later used only Euclidian metrics. Euclidian (i.e., direct) distances in the 16-dimensional space are defined as with i and j indicating odours, p the number of dimensions—that is, conditioning groups—and X ik the response of bees to odour i after conditioning to odour k. These distances were used in correlation analyses with optical imaging data (see below). Correlation analysis between perceptual and optophysiological similarity measures We studied whether or not physiological similarity between odours as determined by optical imaging studies of AL activity [ 22 , 23 , 35 ] actually reflects perceptual odour similarity for the bees. To this end, we performed correlation analyses between published optical imaging data that were obtained using the same set of odours as in our work [ 23 ] and our behavioural data. We used two sets of physiological data. First, to perform such a correlation on the whole dataset (including all 16 odours), we transcribed the activation maps presented by Sachse et al. [ 23 ] (see Figure 7 ) into activation levels for each glomerulus from zero to three, according to the following signal scale: dark blue (0%–20%) and light blue (>20%–40% activity), zero; green (>40%–60% activity), one; yellow (>60%–80% activity), two; and red (>80% activity), three. As the activity under 40% was less accurately separated from noise, activation levels between 0% and 40% were ranked as 0. Scaling the physiological data in this way instead of using the original imaging activation data, gave a good overview of physiological similarity between odours for imaging data ( see Results ). To provide a more precise correlation analysis between behavioural and imaging data, albeit on a more limited odour dataset (eight odours), we used exact correlation data ([ 23 ], Table 1 ). Each correlation value C, as calculated by Sachse et al. [ 23 ] between activity patterns for all pairs of primary and secondary alcohols, was converted into physiological distances by the operation 100 − C. All linear correlations were assessed by calculating Pearson's r, and using Student's t -test. Comparison between correlation coefficients obtained with the two methods was carried out statistically using a Z test as in [ 69 ].

D:\keerthana\PMC001xxxxxx\PMC1043859.xml

1043860

Principles of MicroRNA–Target Recognition

MicroRNAs (miRNAs) are short non-coding RNAs that regulate gene expression in plants and animals. Although their biological importance has become clear, how they recognize and regulate target genes remains less well understood. Here, we systematically evaluate the minimal requirements for functional miRNA–target duplexes in vivo and distinguish classes of target sites with different functional properties. Target sites can be grouped into two broad categories. 5′ dominant sites have sufficient complementarity to the miRNA 5′ end to function with little or no support from pairing to the miRNA 3′ end. Indeed, sites with 3′ pairing below the random noise level are functional given a strong 5′ end. In contrast, 3′ compensatory sites have insufficient 5′ pairing and require strong 3′ pairing for function. We present examples and genome-wide statistical support to show that both classes of sites are used in biologically relevant genes. We provide evidence that an average miRNA has approximately 100 target sites, indicating that miRNAs regulate a large fraction of protein-coding genes and that miRNA 3′ ends are key determinants of target specificity within miRNA families.

Introduction MicroRNAs (miRNAs) are small non-coding RNAs that serve as post-transcriptional regulators of gene expression in plants and animals. They act by binding to complementary sites on target mRNAs to induce cleavage or repression of productive translation (reviewed in [ 1 , 2 , 3 , 4 ]). The importance of miRNAs for development is highlighted by the fact that they comprise approximately 1% of genes in animals, and are often highly conserved across a wide range of species (e.g., [ 5 , 6 , 7 ]). Further, mutations in proteins required for miRNA function or biogenesis impair animal development [ 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 ]. To date, functions have been assigned to only a few of the hundreds of animal miRNA genes. Mutant phenotypes in nematodes and flies led to the discovery that the lin-4 and let-7 miRNAs control developmental timing [ 16 , 17 ], that lsy-6 miRNA regulates left–right asymmetry in the nervous system [ 18 ], that bantam miRNA controls tissue growth [ 19 ], and that bantam and miR-14 control apoptosis [ 19 , 20 ]. Mouse miR-181 is preferentially expressed in bone marrow and was shown to be involved in hematopoietic differentiation [ 21 ]. Recently, mouse miR-375 was found to be a pancreatic-islet-specific miRNA that regulates insulin secretion [ 22 ]. Prediction of miRNA targets provides an alternative approach to assign biological functions. This has been very effective in plants, where miRNA and target mRNA are often nearly perfectly complementary [ 23 , 24 , 25 ]. In animals, functional duplexes can be more variable in structure: they contain only short complementary sequence stretches, interrupted by gaps and mismatches. To date, specific rules for functional miRNA–target pairing that capture all known functional targets have not been devised. This has created problems for search strategies, which apply different assumptions about how to best identify functional sites. As a result, the number of predicted targets varies considerably with only limited overlap in the top-ranking targets, indicating that these approaches might only capture subsets of real targets and/or may include a high number of background matches ([ 19 , 26 , 27 , 28 , 29 , 30 ]; reviewed by [ 31 ]). Nonetheless, a number of predicted targets have proven to be functional when subjected to experimental tests [ 19 , 26 , 27 , 29 ]. A better understanding of the pairing requirements between miRNA and target would clearly improve predictions of miRNA targets in animals. It is known that defined cis -regulatory elements in Drosophila 3′ UTRs are complementary to the 5′ ends of certain miRNAs [ 32 ]. The importance of the miRNA 5′ end has also emerged from the pairing characteristics and evolutionary conservation of known target sites [ 26 ], and from the observation of a non-random statistical signal specific to the 5′ end in genome-wide target predictions [ 27 ]. Tissue culture experiments have also underscored the importance of 5′ pairing and have provided some specific insights into the general structural requirements [ 29 , 33 , 34 ], though different studies have conflicted to some degree with each other, and with known target sites (reviewed in [ 31 ]). To date, no specific role has been ascribed to the 3′ end of miRNAs, despite the fact that miRNAs tend to be conserved over their full length. Here, we systematically evaluate the minimal requirements for a functional miRNA–target duplex in vivo. These experiments have allowed us to identify two broad categories of miRNA target sites. Targets in the first category, “5′ dominant” sites, base-pair well to the 5′ end of the miRNA. Although there is a continuum of 3′ pairing quality within this class, it is useful to distinguish two subtypes: “canonical” sites, which pair well at both the 5′ and 3′ ends, and “seed” sites, which require little or no 3′ pairing support. Targets in the second category, “3′ compensatory” sites, have weak 5′ base-pairing and depend on strong compensatory pairing to the 3′ end of the miRNA. We present evidence that all of these site types are used to mediate regulation by miRNAs and show that the 3′ compensatory class of target sites is used to discriminate among individual members of miRNA families in vivo. A genome-wide statistical analysis allows us to estimate that an average miRNA has approximately 100 evolutionarily conserved target sites, indicating that miRNAs regulate a large fraction of protein-coding genes. Evaluation of 3′ pairing quality suggests that seed sites are the largest group. Sites of this type have been largely overlooked in previous target prediction methods. Results The Minimal miRNA Target Site To improve our understanding of the minimal requirements for a functional miRNA target site, we made use of a simple in vivo assay in the Drosophila wing imaginal disc. We expressed a miRNA in a stripe of cells in the central region of the disc and assessed its ability to repress the expression of a ubiquitously transcribed enhanced green fluorescent protein (EGFP) transgene containing a single target site in its 3′ UTR. The degree of repression was evaluated by comparing EGFP levels in miRNA-expressing and adjacent non-expressing cells. Expression of the miRNA strongly reduced EGFP expression from transgenes containing a single functional target site ( Figure 1 A). Figure 1 Complementarity to the miRNA 5′ End Is Important for Target Site Function In Vivo (A) In vivo assay for target site regulation in the wing imaginal disc. The EGFP reporter is expressed in all cells (green). Cells expressing the miRNA under ptcGal4 control are shown in red. Functional target sites allow strong GFP repression by the miRNA (middle). Non-functional target sites do not (right). Yellow boxes indicate the disc region shown in (B) and later figures. (B) Regulation of individual target sites by miR-7. Numbers in the upper left of each image indicate the mismatched nucleotide in the target site. Positions important for regulation are shown in red, dispensable positions in green. Regulation by the miRNA is completely abolished in only a few cases. (C) Summary of the magnitude of reporter gene repression for the series in (B) and for a second set involving miR-278 and a target site resembling the miR-9 site in Lyra [ 26 ]. Positions important for regulation are shown in red, dispensable positions in green. Error bars are based on measurements of 3–5 individual discs. In a first series of experiments we asked which part of the RNA duplex is most important for target regulation. A set of transgenic flies was prepared, each of which contained a different target site for miR-7 in the 3′ UTR of the EGFP reporter construct. The starting site resembled the strongest bantam miRNA site in its biological target hid [ 19 ] and conferred strong regulation when present in a single copy in the 3′ UTR of the reporter gene ( Figure 1 B). We tested the effects of introducing single nucleotide changes in the target site to produce mismatches at different positions in the duplex with the miRNA (note that the target site mismatches were the only variable in these experiments). The efficient repression mediated by the starting site was not affected by a mismatch at positions 1, 9, or 10, but any mismatch in positions 2 to 8 strongly reduced the magnitude of target regulation. Two simultaneous mismatches introduced into the 3′ region had only a small effect on target repression, increasing reporter activity from 10% to 30%. To exclude the possibility that these findings were specific for the tested miRNA sequence or duplex structure, we repeated the experiment with miR-278 and a different duplex structure. The results were similar, except that pairing of position 8 was not important for regulation in this case ( Figure 1 C). Moreover, some of the mismatches in positions 2–7 still allowed repression of EGFP expression up to 50%. Taken together, these observations support previous suggestions that extensive base-pairing to the 5′ end of the miRNA is important for target site function [ 26 , 27 , 29 , 32 , 34 ]. We next determined the minimal 5′ sequence complementarity necessary to confer target regulation. We refer to the core of 5′ sequence complementarity essential for target site recognition as the “seed” (Lewis et al. [ 27 ]). All possible 6mer, 5mer, and 4mer seeds complementary to the first eight nucleotides of the miRNA were tested in the context of a site that allowed strong base-pairing to the 3′ end of the miRNA ( Figure 2 A). The seed was separated from a region of complete 3′ end pairing by a constant central bulge. 5mer and 6mer seeds beginning at positions 1 or 2 were functional. Surprisingly, as few as four base-pairs in positions 2–5 conferred efficient target regulation under these conditions, whereas bases 1–4 were completely ineffective. 4mer, 5mer, or 6mer seeds beginning at position 3 were less effective. These results suggest that a functional seed requires a continuous helix of at least 4 or 5 nucleotides and that there is some position dependence to the pairing, since sites that produce comparable pairing energies differ in their ability to function. For example, the first two duplexes in Figure 2 A (4mer, top row) have identical 5′ pairing energies (ΔG for the first 8 nt was −8.9 kcal/mol), but only one is functional. Similarly, the third 4mer duplex and fourth 5mer duplex (middle row) have the same energy (−8.7 kcal/mol), but only one is functional. We thus do not find a clear correlation between 5′ pairing energy and function, as reported in [ 34 ]. These experiments also indicate that extensive 3′ pairing of up to 17 nucleotides in the absence of the minimal 5′ element is not sufficient to confer regulation. Consequently, target searches based primarily on optimizing the extent of base-pairing or the total free energy of duplex formation will include many non-functional target sites [ 28 , 30 , 35 ], and ranking miRNA target sites according to overall complementarity or free energy of duplex formation might not reflect their biological activity [ 26 , 27 , 28 , 30 , 35 ]. Figure 2 The Minimal miRNA Target Site (A) In vivo tests of the function of target sites with 6mer, 5mer, and 4mer seeds complementary to the first eight nucleotides of the miRNA. Sites were designed to have optimal support from 3′ pairing. The first 4mer seed site shows that extensive complementarity to the miRNA 3′ region is not sufficient for regulation in vivo. (B) Regulation of 8mer, 7mer, and 6mer seed sites lacking complementarity to the miRNA 3′ end. The test UTR contained one site (first column) or two identical sites (second column). To determine the minimal lengths of 5′ seed matches that are sufficient to confer regulation alone, we tested single sites that pair with eight, seven, or six consecutive bases to the miRNA's 5′ end, but that do not pair to its 3′ end ( Figure 2 B). Surprisingly, a single 8mer seed (miRNA positions 1–8) was sufficient to confer strong regulation by the miRNA. A single 7mer seed (positions 2–8) was also functional, although less effective. The magnitude of regulation for 8mer and 7mer seeds was strongly increased when two copies of the site were introduced in the UTR. In contrast, 6mer seeds showed no regulation, even when present in two copies. Comparable results were recently reported for two copies of an 8mer site with limited 3′ pairing capacity in a cell-based assay [ 34 ]. These results do not support a requirement for a central bulge, as suggested previously [ 29 ]. We took care in designing the miRNA 3′ ends to exclude any 3′ pairing to nearby sequence according to RNA secondary structure prediction. However, we cannot rule out the possibility that extensive looping of the UTR sequence might allow the 3′ end to pair to sequences further downstream in our reporter constructs. Note, however, that even if remote 3′ pairing was occurring and required for function of 8- and 7mer seeds, it is not sufficient for 5′ matches with less than seven complementary bases (all test sites are in the same sequence context; Figure 2 B). In addition, pairing at a random level will occur in any sequence if long enough loops are allowed. However, whether the ribonucleoprotein complexes involved in translational repression require 3′ pairing, and whether they are able to allow extensive looping to achieve this, remains an open question. Computationally, remote 3′ pairing cannot be distinguished from random matches if loops of any length are allowed. On this basis any site with a 7- or 8mer seed has to be taken seriously—especially when evolutionarily conserved. From these experiments we conclude that (1) complementarity of seven or more bases to the 5′ end miRNA is sufficient to confer regulation, even if the target 3′ UTR contains only a single site; (2) sites with weaker 5′ complementarity require compensatory pairing to the 3′ end of the miRNA in order to confer regulation; and (3) extensive pairing to the 3′ end of the miRNA is not sufficient to confer regulation on its own without a minimal element of 5′ complementarity. The Effect of G:U Base-Pairs and Bulges in the Seed Several confirmed miRNA target genes contain predicted binding sites with seeds that are interrupted by G:U base-pairs or single nucleotide bulges [ 17 , 19 , 26 , 36 , 37 , 38 , 39 ]. In most cases these mRNAs contain multiple predicted target sites and the contributions of individual sites have not been tested. In vitro tests have shown that sites containing G:U base-pairs can function [ 29 , 34 ], but that G:U base-pairs contribute less to target site function than would be expected from their contribution to the predicted base-pairing energy [ 34 ]. We tested the ability of single sites with seeds containing G:U base-pairs and bulges to function in vivo. One, two, or three G:U base-pairs were introduced into single target sites with 8mer, 7mer, or 6mer seeds ( Figure 3 A). A single G:U base-pair caused a clear reduction in the efficiency of regulation by an 8mer seed site and by a 7mer seed site. The site with a 6mer seed lost its activity almost completely. Having more than one G:U base-pair compromised the activity of all the sites. As the target sites were designed to allow optimal 3′ pairing, we conclude that G:U base-pairs in the seed region are always detrimental. Figure 3 Effects of G:U Base-Pairs and Bulges (A) Regulation of sites with 8mer, 7mer, or 6mer seeds (rows) containing zero, one, two, or three G:U base-pairs in the seed region (columns). (B) Regulation of sites with bulges in the target sequence or in the miRNA. Single nucleotide bulges in the seed are found in the let-7 target lin-41 and in the lin-4 target lin-14 [ 17 , 36 , 37 ]. Recent tissue culture experiments have led to the proposal that such bulges are tolerated if positioned symmetrically in the seed region [ 29 ]. We tested a series of sites with single nucleotide bulges in the target or the miRNA ( Figure 3 B). Only some of these sites conferred good regulation of the reporter gene. Our results do not support the idea that such sites depend on a symmetrical arrangement of base-pairs flanking the bulge. We also note that the identity of the bulged nucleotide seems to matter. While it is clear that some target sites with one nucleotide bulge or a single mismatch can be functional if supported by extensive complementarity to the miRNA 3′ end, it is not possible to generalize about their potential function. Functional Categories of Target Sites While recognizing that there is a continuum of base-pairing quality between miRNAs and target sites, the experiments presented above suggest that sites that depend critically on pairing to the miRNA 5′ end (5′ dominant sites) can be distinguished from those that cannot function without strong pairing to the miRNA 3′ end (3′ compensatory sites). The 3′ compensatory group includes seed matches of four to six base-pairs and seeds of seven or eight bases that contain G:U base-pairs, single nucleotide bulges, or mismatches. We consider it useful to distinguish two subgroups of 5′ dominant sites: those with good pairing to both 5′ and 3′ ends of the miRNA (canonical sites) and those with good 5′ pairing but with little or no 3′ pairing (seed sites). We consider seed sites to be those where there is no evidence for pairing of the miRNA 3′ end to nearby sequences that is better than would be expected at random. We cannot exclude the possibility that some sites that we identify as seed sites might be supported by additional long-range 3′ pairing. Computationally, this is always possible if long enough loops in the UTR sequence are allowed. Whether long loops are functional in vivo remains to be determined. Canonical sites have strong seed matches supported by strong base-pairing to the 3′ end of the miRNA. Canonical sites can thus be seen as an extension of the seed type (with enhanced 3′ pairing in addition to a sufficient 5′ seed) or as an extension of the 3′compensatory type (with improved 5′ seed quality in addition to sufficient 3′ pairing). Individually, canonical sites are likely to be more effective than other site types because of their higher pairing energy, and may function in one copy. Due to their lower pairing energies, seed sites are expected to be more effective when present in more than one copy. Figure 4 presents examples of the different site types in biologically relevant miRNA targets and illustrates their evolutionary conservation in multiple drosophilid genomes. Figure 4 Three Classes of miRNA Target Sites Model of canonical (left), seed (middle) and 3′ compensatory (right) target sites. The upper diagram illustrates the mode of pairing between target site (upper line) and miRNA (lower line, color). Next down in each column are diagrams of the pattern of 3′ UTR conservation. The vertical black bars show stretches of at least six nucleotides that are conserved in several drosophilid genomes. Target sites for miR-7, miR-4, and miR-10 are shown as colored horizontal bars beneath the UTR. Sites for other miRNAs are shown as black bars. Furthest down in each column the predicted structure of the duplex between the miRNA and its target site is shown; canonical base-pairs are marked with filled circles, G:U base-pairs with open circles. The sequence alignments show nucleotide conservation of these target sites in the different drosophilid species Nucleotides predicted to pair to the miRNA are shown in bold; nucleotides predicted to be unpaired are grey. Red asterisks indicate 100% sequence conservation; grey asterisks indicate conservation of base-pairing to the miRNA including G:U pairs. The additional sequence alignment for the miR-10 target site in Scr in Tribolium castanaeum, Anopheles gambiae, and Bombyx mori strengthens this prediction. Note that the reduced quality of 3′ compensation in these species is compensated by the presence of a better quality 7mer seed. A. ga, Anopheles gambiae; B. mo, B. mori; D. an, D. ananassae; D. me, D. melanogaster; D. ps, D. pseudoobscura; D. si, D. simulans; D. vi, D. virilis; D. ya, D. yakuba; T. ca, T. castanaeum. Most currently identified miRNA target sites are canonical. For example, the hairy 3′ UTR contains a single site for miR-7, with a 9mer seed and a stretch of 3′ complementarity. This site has been shown to be functional in vivo [ 26 ], and it is strikingly conserved in the seed match and in the extent of complementarity to the 3′ end of miR-7 in all six orthologous 3′ UTRs. Although seed sites have not been previously identified as functional miRNA target sites, there is some evidence that they exist in vivo. For example, the Bearded (Brd) 3′ UTR contains three sequence elements, known as Brd boxes, that are complementary to the 5′ region of miR-4 and miR-79 [ 32 , 40 ]. Brd boxes have been shown to repress expression of a reporter gene in vivo, presumably via miRNAs, as expression of a Brd 3′ UTR reporter is elevated in dicer-1 mutant cells, which are unable to produce any miRNAs [ 14 ]. All three Brd box target sites consist of 7mer seeds with little or no base-pairing to the 3′ end of either miR-4 or miR-79 (see below). The alignment of Brd 3′ UTRs shows that there is little conservation in the miR-4 or miR-79 target sites outside the seed sequence, nor is there conservation of pairing to either miRNA 3′ end. This suggests that the sequences that could pair to the 3′ end of the miRNAs are not important for regulation as they do not appear to be under selective pressure. This makes it unlikely that a yet unidentified Brd box miRNA could form a canonical site complex. The 3′ UTR of the HOX gene Sex combs reduced (Scr) provides a good example of a 3′ compensatory site. Scr contains a single site for miR-10 with a 5mer seed and a continuous 11-base-pair complementarity to the miRNA 3′ end [ 28 ]. The miR-10 transcript is encoded within the same HOX cluster downstream of Scr, a situation that resembles the relationship between miR-iab-5p and Ultrabithorax in flies [ 26 ] and miR-196 / HoxB8 in mice [ 41 ]. The predicted pairing between miR-10 and Scr is perfectly conserved in all six drosophilid genomes, with the only sequence differences occurring in the unpaired loop region. The site is also conserved in the 3′ UTR of the Scr genes in the mosquito, Anopheles gambiae , the flour beetle, Tribolium castaneum , and the silk moth, Bombyx mori . Conservation of such a high degree of 3′ complementarity over hundreds of millions of years of evolution suggests that this is likely to be a functional miR-10 target site. Extensive 5′ and 3′ sequence conservation is also seen for other 3′ compensatory sites, e.g., the two let-7 sites in lin-41 or the miR-2 sites in grim and sickle [ 17 , 26 , 36 ]. The miRNA 3′ End Determines Target Specificity within miRNA Families Several families of miRNAs have been identified whose members have common 5′ sequences but differ in their 3′ ends. In view of the evidence that 5′ ends of miRNA are functionally important [ 26 , 27 , 29 , 42 ], and in some cases sufficient (present study), it can be expected that members of miRNA families may have redundant or partially redundant functions. According to our model, 5′ dominant canonical and seed sites should respond to all members of a given miRNA family, whereas 3′ compensatory sites should differ in their sensitivity to different miRNA family members depending on the degree of 3′ complementarity. We tested this using the wing disc assay with 3′ UTR reporter transgenes and overexpression constructs for various miRNA family members. miR-4 and miR-79 share a common 5′ sequence that is complementary to a single 8mer seed site in the bagpipe 3′ UTR ( Figure 5 A and 5 B). The 3′ ends of the miRNAs differ. miR-4 is predicted to have 3′ pairing at approximately 50% of the maximally possible level (−10.8 kcal/mol), whereas the level of 3′ pairing for miR-79 is approximately 25% maximum (−6.1 kcal/mol), which is below the average level expected for random matches (see below). Both miRNAs repressed expression of the bagpipe 3′ UTR reporter, regardless of the 3′ complementarity ( Figure 5 B). This indicates that both types of site are functional in vivo and suggests that bagpipe is a target for both miRNAs in this family. Figure 5 Target Specificity of miRNA Family Members (A) Diagrams of 3′ UTR conservation in six drosophilid genomes (horizontal black bars) and the location of predicted miRNA target sites. Above is the 3′ UTR of the myogenic transcription factor bagpipe (bap) showing the predicted target site for the Brd box miRNA family, miR-4 and miR-79 (black box below the UTR). Alignment of miR-4 and miR-79 illustrates that they share a similar seed sequence (except that mir-4 has one extra 5′ base) but have little 3′ end similarity. Below are the conserved sequences in the3′ UTRs of the pro-apoptotic genes grim and sickle. Predicted target sites for the K Box miRNAs miR-11, miR-2b, and miR-6 are shown below the UTR. Alignment of miR-11, miR-2b, and miR-6 illustrates that they share the same family motif but have little similarity in their 3′ ends. (B) The bagpipe (bap) 3′ UTR reporter gene is regulated by miR-4 and miR-79. Alignments of the two miRNAs to the predicted target site show good 8mer seed matches (left). Overexpression of miR-4 or miR-79 under ptcGal4 control downregulated the bagpipe 3′ UTR reporter (right). (C) Left: Alignment of K box miRNAs with the single predicted site in the grim 3′ UTR and regulation by overexpression of miR-2 (top), but not by miR-6 (middle) or miR-11 (bottom). Right: Alignment of K box miRNAs with the two predicted sites in the sickle 3′ UTR. Regulation by overexpression of miR-2 was strong (top), regulation by miR-6 was weaker (middle), and miR-11 had little effect (bottom). (D) Effect of clones of cells lacking dicer-1 on expression of UTR reporters for predicted miRNA-regulated genes. Mutant cells were marked by the absence of β-Gal expression (red). EGFP expression is shown in green. Both channels are shown separately below in black and white. Mutant clones are indicated by yellow arrows. Expression of a uniformly transcribed reporter construct lacking miRNA target sites was unaffected in dicer-1 mutant cells (first column). The UTR reporter for the bantam miRNA target hid was upregulated in the mutant cells (second column). The bagpipe (bap) UTR reporter was upregulated in dicer-1 clones (third column). The grim (fourth column) and sickle (fifth column) UTR reporters were upregulated. To test whether miRNA family members can also have non-overlapping targets, we used 3′ UTR reporters of the pro-apoptotic genes grim and sickle, two recently identified miRNA targets [ 26 ]. Both genes contain K boxes in their 3′ UTRs that are complementary to the 5′ ends of the miR-2, miR-6, and miR-11 miRNA family [ 26 , 32 ]. These miRNAs share residues 2–8 but differ considerably in their 3′ regions ( Figure 5 A). The site in the grim 3′ UTR is predicted to form a 6mer seed match with all three miRNAs ( Figure 5 C, left), but only miR-2 shows the extensive 3′ complementarity that we predict would be needed for a 3′ compensatory site with a 6mer seed to function (−19.1 kcal/mol, 63% maximum 3′ pairing, versus −10.9 kcal/mol, 46% maximum, for miR-11 and −8.7 kcal/mol, 37% maximum, for miR-6 ). Indeed, only miR-2 was able to regulate the grim 3′ UTR reporter, whereas miR-6 and miR-11 were non-functional. The sickle 3′ UTR contains two K boxes and provides an opportunity to test whether weak sites can function synergistically. The first site is similar to the grim 3′ UTR in that it contains a 6mer seed for all three miRNAs but extensive 3′ complementarity only to miR-2 . The second site contains a 7mer seed for miR-2 and miR-6 but only a 6mer seed for miR-11 ( Figure 5 C, right). miR-2 strongly downregulated the sickle reporter, miR-6 had moderate activity (presumably via the 7mer seed site), and miR-11 had nearly no activity, even though the miRNAs were overexpressed. The fact that a site is targeted by at least one miRNA argues that it is accessible (e.g., miR-2 is able to regulate both UTR reporters), and that the absence of regulation for other family members is due to the duplex structure. These results are in line with what we would expect based on the predicted functionality of the individual sites, and indicate that our model of target site functionality can be extended to UTRs with multiple sites. Weak sites that do not function alone also do not function when they are combined. To show that endogenous miRNA levels regulate all three 3′ UTR reporters, we compared EGFP expression in wild-type cells and dicer-1 mutant cells, which are unable to produce miRNAs [ 14 ]. dicer-1 clones did not affect a control reporter lacking miRNA binding sites, but showed elevated expression of a reporter containing the 3′ UTR of the previously identified bantam miRNA target hid ( Figure 5 D). Similarly, all 3′ UTR reporters above were upregulated in dicer-1 mutant cells, indicating that bagpipe, sickle, and grim are subject to repression by miRNAs expressed in the wing disc. Taken together, these experiments indicate that transcripts with 5′ dominant canonical and seed sites are likely to be regulated by all members of a miRNA family. However, transcripts with 3′ compensatory sites can discriminate between miRNA family members. Genome-Wide Occurrence of Target Sites Experimental tests such as those presented above and the observed evolutionary conservation suggest that all three types of target sites are likely to be used in vivo. To gain additional evidence we examined the occurrence of each site type in all Drosophila melanogaster 3′ UTRs. We made use of the D. pseudoobscura genome, the second assembled drosophilid genome, to determine the degree of site conservation for the three different site classes in an alignment of orthologous 3′ UTRs. From the 78 known Drosophila miRNAs, we selected a set of 49 miRNAs with non-redundant 5′ sequences. We first investigated whether sequences complementary to the miRNA 5′ ends were better conserved than would be expected for random sequences. For each miRNA, we constructed a cohort of ten randomly shuffled variants. To avoid a bias for the number of possible target matches, the shuffled variants were required to produce a number of sequence matches comparable (±15%) to the original miRNAs for D. melanogaster 3′ UTRs. 7mer and 8mer seeds complementary to real miRNA 5′ ends were significantly better conserved than those complementary to the shuffled variants. This is consistent with the findings of Lewis et al. [ 27 ] but was obtained without the need to use a rank and energy cutoff applied to the full-length miRNA target duplex, as was the case for vertebrate miRNAs. Conserved 8mer seeds for real miRNAs occur on average 2.8 times as often as seeds complementary to the shuffled miRNAs ( Figure 6 A). For 7mer seeds this signal was 2:1, whereas 6mer, 5mer, and 4mer seeds did not show better conservation than expected for random sequences. To assess the validity of these signals and to control for the random shuffling of miRNAs, we repeated this procedure with “mutant” miRNAs in which two residues in the 5′ region were changed. There was no difference between the mutant test miRNAs and their shuffled variants ( Figure 6 A). This indicates that a substantial fraction of the conserved 7mer and 8mer seeds complementary to real miRNAs identify biologically relevant target sites. Figure 6 Computational Analysis of Target Site Occurrence (A) Genome-wide occurrence of conserved 5′ seed matches. Histogram showing the ratio of 5′ seed matches for a set of 49 5′ non-redundant miRNAs and the average of their ten completely shuffled variants for different seed types (black bars). A ratio of one (red line) indicates no difference between the miRNA and its shuffled variants. The same ratio for mutated miRNAs and their shuffled variants shows no signal (white bars). The inset depicts shuffling of the entire miRNA sequence (wavy purple line). (B) Target site conservation between D. melanogaster and D. pseudoobscura . Histogram showing the average conservation of the 3′ UTR sequence (16 nt) upstream of a conserved 8mer seed match that would pair to the miRNA 3′ end. All sites were binned according to their conservation, and the percentage of sites in each bin is shown for sites identified by 49 5′ non-redundant miRNA sequences (grey) and their shuffled control sequences (black, error bars indicate one standard deviation). (C) 3′ pairing preferences for miR-7 target sites. Histogram showing the distribution of 3′ pairing energies for miR-7 (red bars) and the average of 50 3′ shuffled variants (black bars) for all sites identified genome-wide by 6mer 5′ seed matches for miR-7. The inset illustrates shuffling of the 3′ end of miRNA sequence only (wavy purple line). Because the miRNA 5′ end was not altered, the identical set of target sites was compared for pairing to the 3′ end of real and shuffled miRNAs. (D) 3′ pairing preferences for miRNA target sites. Histograms showing the ratio of the top 1% 3′ pairing energies for the set of 58 3′ non-redundant miRNAs and their shuffled variants. The y-axis shows the number of miRNAs for each ratio. Real miRNAs are shown in red; mutant miRNAs are shown in black. Left are shown combined 8- and 7mer seed sites. Right are shown combined 5- and 6mer seed sites. For combined 8- and 7mer seeds, 1% corresponds to approximately ten sites per miRNA; for combined 6- and 5mer, to approximately 25 sites. The difference between the real and mutated miRNAs improves if fewer sites per miRNA are considered. (E) Non-random signal of 3′ pairing. Plot of the ratio of the number of target sites for the set of 58 3′ non-redundant miRNAs and their shuffled miRNA 3′ ends (y-axis) with 3′ pairing energies that exceed a given pairing cutoff (x-axis). 100% is the pairing energy for a sequence perfectly complementary to the 3′ end. As the required level of 3′ pairing energy increases, fewer miRNAs and their sites remain to contribute to the signal. Plots for the real miRNAs extended to considerably higher 3′ pairing energies than the mutants, but as site number decreases we observe anomalous effects on the ratios, so the curves were cut off when the number of remaining miRNAs fell below five. 3′ compensatory and canonical sites depend on substantial pairing to the miRNA 3′ end. For these sites, we expect UTR sequences adjacent to miRNA 5′ seed matches to pair better to the miRNA 3′ end than to random sequences. However, unlike 5′ complementarity, 3′ base-pairing preference was not detected in previous studies looking at sequence complementarity and nucleotide conservation because UTR sequences complementary to the miRNA 3′ end were not better conserved than would be expected at random [ 27 ]. On this basis, we decided to treat the 5′ and 3′ ends of the miRNA separately. For the 5′ end, seed matches were required to be fully conserved in an alignment of orthologous D. melanogaster and D. pseudoobscura 3′ UTRs (we expected one-half to two-thirds of these matches to be real miRNA sites). We first investigated the overall conservation of UTR sequences adjacent to the conserved seed matches and found that overall the sequences are not better conserved than a random control with shuffled miRNAs ( Figure 6 B). For both real and random matches, the number of sites increases with the degree of 3′ conservation (up to the 80% level), reflecting the increased probability that sequences adjacent to conserved seed matches will also lie in blocks of conserved sequence ( Figure 6 B). For real 7mers and 8mers we found a slightly higher percentage of sites between 30% and 80% identity than we did for the shuffled controls. In contrast, the ratio of sites with over 80% sequence identity was smaller for real 7- or 8mers than for random ones, meaning that in highly conserved 3′ UTR blocks (>80% identity) the ratio of random matches exceeds that of real miRNA target sites. This caused us to question whether the degree of conservation for sequences adjacent to seed matches correlates with miRNA 3′ pairing as would be expected if the conservation were due to a biologically relevant miRNA target site. Indeed, we found that the best conserved sites adjacent to seed matches (i.e., those with zero, one, or two mismatches in the 3′ UTR alignment) and the least conserved sites (i.e., those with only three, two, or one matching nucleotides) are not distinguishable in that both pair only randomly to the corresponding miRNA 3′ end (approximately 35% maximal 3′ pairing energy, data not shown). The observation that miRNA target sites do not seem to be fully conserved over their entire length is consistent with the examples shown in Figure 4 in which only the degree of 3′ pairing but not the nucleotide identity is conserved (miR-7/hairy), or at least the unpaired bulge is apparently not under evolutionary pressure (miR-10/Scr). Although this result obviously depends on the evolutionary distance of the species under consideration (see [ 43 ] for a comparison of mammalian sites), it shows that conclusions about the contribution of miRNA 3′ pairing to target site function cannot be drawn solely from the degree of sequence conservation. We therefore chose to evaluate the quality of 3′ pairing by the stability of the predicted RNA–RNA duplex. We assessed predicted pairing energy between the miRNA 3′ end and the adjacent UTR sequence for both Drosophila species and used the lower score. Use of the lower score measures conservation of the overall degree of pairing without requiring sequence identity. Figure 6 C shows the distribution of the 3′ pairing energies for all conserved 3′ compensatory miR-7 sites identified by a 6mer seed match, compared to the distribution of 50 miR-7 sequences shuffled only in the 3′ part, leaving the 5′ unchanged. This means that real and shuffled miRNAs identify the same 5′ seed matches in the 3′ UTRs, which allows us to compare the 3′ pairing characteristics of the adjacent sequences. We also required 3′ shuffled sequences to have similar pairing energies (±15%) to their complementary sequences and to 10,000 randomly selected sites to exclude generally altered pairing characteristics. The distributions for real and shuffled miRNAs were highly similar, with a mean of approximately 35% of maximal 3′ pairing energy and few sites above 55%. However, a small number of sites paired exceptionally well to miR-7 at energies that were far above the shuffled averages and not reached by any of the 50 shuffled controls. This example illustrates that there is a significant difference between real and shuffled miRNAs for the sites with the highest 3′ complementarity, which are likely to be biologically relevant. Sites with weaker 3′ pairing might also be functional, but cannot be distinguished from random matches and can only be validated by experiments (see Figure 5 ). To provide a global analysis of 3′ pairing comprising all miRNAs and to investigate how many miRNAs show significantly non-random 3′ pairing, we considered only the sites within the highest 1% of 3′ pairing energies. The average of the highest 1% of 3′ pairing energies of each of 58 3′ non-redundant miRNAs was divided by that of its 50 3′ shuffled controls. This ratio is one if the averages are the same, and increases if the real miRNA has better 3′ pairing than the shuffled miRNAs. To test whether a signal was specific for real miRNAs, we repeated the same protocol with a mutant version of each miRNA. The altered 5′ sequence in the mutant miRNA selects different seed matches than the real miRNA and permits a comparison of sequences that have not been under selection for complementarity to miRNA 3′ ends with those that may have been. Figure 6 D shows the distribution of the energy ratios for canonical (left) and 3′ compensatory sites (right) for all 58 real and mutated 3′ non-redundant miRNAs. Most real miRNAs had ratios close to one, comparable to the mutants. But several had ratios well above those observed for mutant miRNAs, indicating significant conserved 3′ pairing. A small fraction of sites show exceptionally good 3′ pairing. If we use 3′ pairing energy cutoffs to examine site quality for all miRNAs, we expect sites of this type to be distinguishable from random matches. The ratio of the number of sites above the cutoff for real versus 3′ shuffled miRNAs was plotted as a function of the 3′ pairing cutoff ( Figure 6 E). For low cutoffs the ratio is one, as the number of sites corresponds to the number of seed matches (which is identical for real and 3′ shuffled miRNAs). For increasing cutoffs, the ratios increase once a certain threshold is reached, reflecting overrepresentation of sites that pair favorably to the real miRNA 3′ end but not the 3′ shuffled miRNAs. The maximal ratio obtained for mutated miRNAs never exceeded five, which we used as the threshold level to define where significant overrepresentation begins. For 8mer seed sites overrepresentation began at 55% maximal 3′ pairing; for 7mer seed sites, at 65%; for 6mer seed sites, at 68%; and for 5mer seed sites, at 78%. There was no statistical evidence for sites with 4mer seeds. We also tested whether sequences forming 7mer or 8mer seeds containing G:U base-pairs, mismatches, or bulges were better conserved if complementary to real miRNAs. We did not find any statistical evidence for these seed types. Analysis of 3′ pairing also failed to show any non-random signal for these sites. This suggests that such sites are few in number genome-wide and are not readily distinguished from random matches. Nonetheless, our experiments do show that sites of this type can function in vivo. The let-7 sites in lin-41 provide a natural example. Most Sites Lack Substantial 3′ Pairing The experimental and computational results presented above provide information about 5′ and 3′ pairing that allows us to estimate the number of target sites of each type in Drosophila. The number of 3′ compensatory sites cannot be estimated on the basis of 5′ pairing, because seed matches of four, five, or six bases cannot be distinguished from random matches, reflecting that a large number of randomly conserved and non-functional matches predominate ( Figure 6 A). Significant 3′ pairing can be distinguished from random matches for 6mer sites above 68% maximal 3′ pairing energy, and above 78% for 5mers ( Figure 6 E). Using these pairing levels gives an estimate of one 3′ compensatory site on average per miRNA. The experiments in Figure 5 provide an opportunity to assess the contribution of 3′ pairing to the ability of sites with 6mer seeds to function. The 6mer K box site in the grim 3′ UTR was regulated by miR-2 (63% maximal 3′ pairing energy), but not by miR-11, which has a predicted 3′ pairing energy of 46%. Similarly, the 6mer seed sites for miR-11 in the sickle 3′ UTR had 3′ pairing energies of approximately 35% and were non-functional. We can use the 63% and 46% levels to provide upper and lower estimates of one and 20 3′ compensatory 6mer sites on average per miRNA. For 5mer sites, the examples in Figure 1 show that sites with 76% and 83% maximal 3′ pairing do not function. At the 80% threshold level, we expect less than one additional site on average per miRNA, suggesting that 3′ compensatory sites with 5mer seeds are rare. The predicted miR-10 site in Scr (see Figure 4 ) is one of the few sites with a 5mer seed that reaches this threshold (100% maximum 3′ pairing energy; −20 kcal/mol). It is likely that other sites in this group will also prove to be functionally important. The overrepresentation of conserved 5′ seed matches (see Figure 6 A) suggests that approximately two-thirds of sites with 8mer seeds and approximately one-half of the sites with 7mer seeds are biologically relevant. This corresponds to an average of 28 8mers and 53 7mers, for a total of 81 sites per miRNA. We define canonical sites as those with meaningful contributions from both 5′ and 3′ pairing. Given that 7- and 8mer seed matches can function without significant 3′ pairing, it is difficult to assess at what level 3′ pairing contributes meaningfully to their function. The range of 3′ pairing energies that were minimally sufficient to support a weak seed match was between 46% and 63% of maximum pairing energy (see Figure 5 C). If we take the 46% level as the lower limit for meaningful 3′ pairing, over 95% of sites would be considered seed sites. This changes to 99% for pairing energies that can be statistically distinguished from noise (55% maximal; see Figure 6 E) and remains over 50% even for pairing energies at the average level achieved by random matches (30% maximal). It is clear from this analysis that the majority of miRNA target sites lack substantial pairing in the 3′ end in nearby sequences. Indeed the 3′ pairing level for the three seed sites for miR-4 in Brd are all less than 25% (i.e., below the average for random matches) and Brd was thus not predicted as a miR-4 target previously [ 26 , 28 , 35 ]. Again, we note the caveat that some of sites that we identify as seed could in principle be supported by 3′ pairing to more distant upstream sequences, but also that such sites would be difficult to distinguish from background computationally and that it is unclear whether large loops are functional. If there were statistical evidence for 3′ pairing that is lower than would be expected at random for some sites, this would be one line of argument for a discrete functional class that does not use 3′ pairing and would therefore suggest selection against 3′ pairing. Although the overall distribution of 3′ pairing energies for real miRNA 3′ ends adjacent to 8mer seed matches is very similar to the random control with 3′ shuffled sequences ( Figure 7 ; R 2 = 0.98), we observed a small but significant overrepresentation of real sites on both sides of the random distribution, which leads to a slightly wider distribution of real sites at the expense of the peak values around 30% pairing. Bearing in mind that one-third of 8mer seed matches are false positives (see Figure 6 A), we can account for the noise by subtracting one-third of the random distribution. We then see two peaks at around 20% and 35% maximum pairing energy, separated by a dip. Subtracting more (e.g., one-half or two-thirds) of the random distribution increases the separation of the two peaks, suggesting that the underlying distribution of 3′ pairing for real 8mer seed sites might indeed be bimodal. This effect is still present, though less pronounced, if 7mer seed matches are included. No such effect is seen for the combined 5- and 6mer seed matches. In addition, we see no difference between a random (noise) model that evaluates 3′ pairing of 3′ shuffled miRNAs to UTR sites identified by real miRNA seed matches and a random model that pairs the real (i.e., non-shuffled) miRNA 3′ end to randomly chosen UTR sequences, thus excluding bias due to shuffling. Overall, these results suggest that there might indeed be a bimodal distribution due to an enrichment of sites with both better and worse 3′ pairing than would be expected at random. We take this as evidence that seed sites are a biologically meaningful subgroup within the 5′ dominant site category. Figure 7 Distribution of 3′ Pairing Energies for 8mer Seed Matches Shown is the distribution (number of sites versus 3′ pairing) for 8mer seed matches identified genome-wide for 58 3′ non-redundant miRNAs (black) compared to a random control using 50 3′ shuffled miRNAs per real miRNA (grey). Note that the distribution for real miRNAs is broader at both the high and low end than the random control and has shoulders close to the peak. The red, blue, and green curves show the effect of subtracting background noise (random matches) from the real matches at three different levels, which reveals the real matches underlying these shoulders. Overall, these estimates suggest that there are over 80 5′ dominant sites and 20 or fewer 3′ compensatory sites per miRNA in the Drosophila genome. As estimates of the number of miRNAs in Drosophila range from 96 to 124 [ 44 ], this translates to 8,000–12,000 miRNA target sites genome-wide, which is close to the number of protein-coding genes. Even allowing for the fact that some genes have multiple miRNA target sites, these findings suggest that a large fraction of genes are regulated by miRNAs. Discussion We have provided experimental and computational evidence for different types of miRNA target sites. One key finding is that sites with as little as seven base-pairs of complementarity to the miRNA 5′ end are sufficient to confer regulation in vivo and are used in biologically relevant targets. Genome-wide, 5′ dominant sites occur 2- to 3-fold more often in conserved 3′ UTR sequences than would be expected at random. The majority of these sites have been overlooked by previous miRNA target prediction methods because their limited capacity to base-pair to the miRNA 3′ end cannot be distinguished from random noise. Such sites rank low in search methods designed to optimize overall pairing energy [ 16 , 17 , 26 , 27 , 28 , 30 , 35 ]. Indeed, we find that few seed sites scored high enough to be considered seriously in these earlier predictions, even when 5′ complementarity was given an additional weighting (e.g., [ 28 , 43 ]. We thus suspect that methods with pairing cutoffs would exclude many, if not all, such sites. In a scenario in which protein-coding genes acquire miRNA target sites in the course of evolution [ 4 ], it is likely that seed sites with only seven or eight bases complementary to a miRNA would be the first functional sites to be acquired. Once present, a site would be retained if it conferred an advantage, and sites with extended complementarity could also be selected to confer stronger repression. In this scenario, the number of sites might grow over the course of evolution so that ancient miRNAs would tend to have more targets than those more recently evolved. Likewise, genes that should not be repressed by the miRNA milieu in a given cell type would tend to avoid seed matches to miRNA 5′ ends (“anti-targets” [ 4 ]). Although a 7- to 8mer seed is sufficient for a site to function, additional 3′ pairing increases miRNA functionality. The activity of a single 7mer canonical site is expected to be greater than an equivalent seed site. Likewise, the magnitude of miRNA-induced repression is reduced by introducing 3′ mismatches into a canonical site. Genome-wide, there are many sites that appear to show selection for conserved 3′ pairing and, interestingly, many sites that appear to show selection against 3′ pairing. In vivo, canonical sites might function at lower miRNA concentrations and might repress translation more effectively, particularly when multiple sites are present in one UTR (e.g., [ 42 ]). Efficient repression is likely to be necessary for genes whose expression would be detrimental, as illustrated by the genetically identified miRNAs, which produce clear mutant phenotypes when their targets are not normally repressed (“switch targets” [ 4 ]). Prolonged expression of the lin-14 and lin-41 genes in Caenorhabditis elegans mutant for lin-4 or let-7 causes developmental defects, and their regulation involves multiple sites [ 17 , 36 , 37 ]. Similarly, multiple target sites allow robust regulation of the pro-apoptotic gene hid by bantam miRNA in Drosophila [ 19 ]. More subtle modulation of expression levels could be accomplished by weaker sites, such as those lacking 3′ pairing. Sites that cannot function efficiently alone are in fact a prerequisite for combinatorial regulation by multiple miRNAs. Seed sites might thus be useful for situations in which the combined input of several miRNAs is used to regulate target expression. Depending on the nature of the target sites, any single miRNA might not have a strong effect on its own, while being required in the context of others. 3′ Complementarity Distinguishes miRNA Family Members 3′ compensatory sites have weak 5′ pairing and need substantial 3′ pairing to function. We find genome-wide statistical support for 3′ compensatory sites with 5mer and 6mer seeds and show that they are used in vivo. Furthermore, these sites can be differentially regulated by different miRNA family members depending on the quality of their 3′ pairing (e.g., regulation of the pro-apoptotic genes grim and sickle by miR-2, miR-6, and miR-11 ). Thus, members of a miRNA family may have common targets as well as distinct targets. They may be functionally redundant in regulation of some targets but not others, and so we can expect some overlapping phenotypes as well as differences in their mutant phenotypes. Following this reasoning, it is likely that the let-7 miRNA family members differentially regulate lin-41 in C. elegans [ 17 , 45 ]. The seed matches in lin-41 to let-7 and the related miRNAs miR-48, miR-84, and miR-241 are weak, and only let-7 has strong 3′ pairing. On this basis, it seems likely that lin-41 is regulated only by let-7 . In contrast, hbl-1 has four sites with strong seed matches [ 38 , 39 ], and we expect it to be regulated by all four let-7 family members. As all four let-7 -related miRNAs are expressed similarly during development [ 6 ], their role as regulators of hbl-1 may be redundant. let-7 must also have targets not shared by the other family members, as its function is essential. lin-41 is likely to be one such target. The idea that the 3′ end of miRNAs serves as a specificity factor provides an attractive explanation for the observation that many miRNAs are conserved over their full length across species separated by several hundreds of millions of years of evolution. 3′ compensatory sites may have evolved from canonical sites by mutations that reduce the quality of the seed match. This could confer an advantage by allowing a site to become differentially regulated by miRNA family members. In addition, sites could retain specificity and overall pairing energy, but with reduced activity, perhaps permitting discrimination between high and low levels of miRNA expression. This might also allow a target gene to acquire a dependence on inputs from multiple miRNAs. These scenarios illustrate a few ways in which more complex regulatory roles for miRNAs might arise during evolution. A Large Fraction of the Genome Is Regulated by miRNAs Another intriguing outcome of this study is evidence for a surprisingly large number of miRNA target sites genome-wide. Even our conservative estimate is far above the numbers of sites in recent predictions, e.g., seven or fewer per miRNA [ 27 , 28 , 29 ]. Our estimate of the total number of targets approaches the number of protein-coding genes, suggesting that regulation of gene expression by miRNAs plays a greater role in biology than previously anticipated. Indeed, Bartel and Chen [ 46 ] have suggested in a recent review that the earlier estimates were likely to be low, and a recent study by John et al. [ 43 ], published while this manuscript was under review, predicts that approximately 10% of human genes are regulated by miRNAs. We agree with these authors' suggestion that this is likely an underestimate, because their method identifies an average of only 7.1 target genes per miRNA, with few that we would classify as seed sites lacking substantial 3′ pairing. A large number of target sites per miRNA is also consistent with combinatorial gene regulation by miRNAs, analogous to that by transcription factors, leading to cell-type-specific gene expression [ 47 ]. Sites for multiple miRNAs allow for the possibility of cell-type-specific miRNA combinations to confer robust and specific gene regulation. Our results provide an improved understanding of some of the important parameters that define how miRNAs bind to their target genes. We anticipate that these will be of use in understanding known miRNA–target relationships and in improving methods to predict miRNA targets. We have limited our evaluation to target sites in 3′ UTRs. miRNAs directed at other types of targets or with dramatically different functions (e.g., in regulation of chromatin structure) might well use different rules. Accordingly, there may prove to be more targets than we can currently estimate. Further, there may be additional features, such as overall UTR context, that either enhance or limit the accessibility of predicted sites and hence their ability to function. For example, the rules about target site structure cannot explain the apparent requirement for the linker sequence observed in the let-7/lin-41 regulation [ 48 ]. Further efforts toward experimental target site validation and systematic examination of UTR features can be expected to provide new insight into the function of miRNA target sites. Materials and Methods Fly strains ptcGal4; EP miR278 was provided by Aurelio Teleman. The control, hid, grim, and sickle 3′ UTR reporter transgenes, and UAS- miR-2b are described in [ 19 , 26 ]. For UAS constructs for miRNA overexpression, genomic fragments including miR-4 (together with miR-286 and miR-5 ) and miR-11 were amplified by PCR and cloned into UAS-DSred as described for UAS- miR-7 [ 26 ]. Details are available on request. UAS- miR-79 (also contains miR-9b and miR-9c ) and UAS- miR-6 ( miR-6–1, miR-6–2, and miR-6–3 ) were kindly provided by Eric Lai. dcr-1 Q1147X is described in [ 14 ]. Clonal analysis Clones mutant for dcr-1 Q1147X were induced in HS-Flp;dcr-1 FRT82/armadillo-lacZ FRT82 larvae by heat shock for 1 h at 38 °C at 50–60 h of development. Wandering third-instar larvae were dissected and labeled with rabbit anti-GFP (Torrey Pines Biolabs, Houston, Texas, United States; 1:400) and anti-β-Gal (rat polyclonal, 1:500). Reporter constructs The bagpipe 3′ UTR was PCR amplified from genomic DNA (using the following primers [enzyme sites in lower case]: AAtctaga AGGTTGGGAGTGACCATGTCTC and AActcgag TATTTAGCTCTCGGGTAGATACG) and cloned downstream of the tubulin promoter and EGFP (Clontech, Palo Alto, California, United States) in Casper4 as in [ 26 ]. Single target site constructs Oligonucleotides containing the target site sequences shown in the figures were annealed and cloned downstream of tub>EGFP and upstream of SV40polyA (XbaI/XhoI). Clones were verified by DNA sequencing. Details are available on request. EGFP intensity measurements NIH image 1.63 was used to quantify intensity levels in miRNA-expressing and non-expressing cells from confocal images. Depending on the variation, between three and five individual discs were analyzed. 3′ UTR alignments For each D. melanogaster gene, we identified the D. pseudoobscura ortholog using TBlastn as described in [ 26 ]. We then aligned the D. melanogaster 3′ UTR obtained from the Berkeley Drosophila Genome Project to the D. pseudoobscura 3′ adjacent sequence (Human Genome Sequencing Center at Baylor College of Medicine) using AVID [ 49 ]. For individual examples, we manually mapped the D. melanogaster coding region to genomic sequence traces (National Center for Biotechnology Information trace archive) of D. ananassae, D. virilis, D. simulans, and D. yakuba by TBlastn and extended the sequences by Blastn-walking. These 3′ UTR sequences were then aligned to the D. melanogaster and D. pseudoobscura 3′ UTRs using AVID. miRNA-sequences Drosophila miRNA sequences were from [ 44 , 50 , 51 ] downloaded from Rfam ( http://www.sanger.ac.uk/Software/Rfam/mirna/index.shtml ). The 5′ non-redundant set (49 miRNAs) comprised bantam, let-7, miR-1, miR-10, miR-11, miR-100, miR-124, miR-125, miR-12, miR-133, miR-13a, miR-14, miR-184, miR-210, miR-219, miR-263b, miR-275, miR-276b, miR-277, miR-278, miR-279, miR-281, miR-283, miR-285, miR-287, miR-288, miR-303, miR-304, miR-305, miR-307, miR-309, miR-310, miR-314, miR-315, miR-316, miR-317, miR-31a, miR-33, miR-34, miR-3, miR-4, miR-5, miR-79, miR-7, miR-87, miR-8, miR-92a, miR-9a, and miR-iab-4–5p. Additional miRNAs in the 3′ non-redundant set were miR-2b, miR-286, miR-306, miR-308, miR-311, miR-312, miR-313, miR-318, and miR-6. miRNA shuffles and mutants For the completely shuffled miRNAs, we shuffled the miRNA sequence over the entire length and required all possible 8mer and 7mer seeds within the first nine bases to have an equal frequency (±15%) to the D. melanogaster 3′ UTRs (i.e., same single genome count). For the 3′ shuffled miRNAs, we shuffled the 3′ end starting at base 10 and required the shuffles to have equal (±15%) pairing energy to a perfect complement and to 10,000 randomly chosen sites. For each miRNA we created all possible 2-nt mutants (exchanging A to T or C, C to A or G, G to C or T, and T to A or G) within the seed (nucleotides 3–6) and chose the one with the closest alignment frequencies to the real miRNA in D. melanogaster 3′ UTRs and in the conserved sequences in D. melanogaster and D. pseudoobscura 3′ UTRs. Seed matching and site evaluation For each miRNA and seed type we found the 5′ match in the D. melanogaster 3′ UTRs and required it to be 100% conserved in an alignment to the D. pseudoobscura ortholog allowing for positional alignment errors of ±2 nt. When searching 7mer to 4mer seeds we masked all longer seeds to avoid identifying the same site more than once. For each matching site we extracted the 3′ adjacent sequence for both genomes, aligned it to the miRNA 3′ end starting at nucleotide 10 using RNAhybrid [ 35 ], and took the worse energy. Supporting Information Accession Numbers The miRNA sequences discussed in this paper can be found in the miRNA Registry ( http://www.sanger.ac.uk/Software/Rfam/mirna/index.shtml ). NCBI RefSeq ( http://www.ncbi.nlm.nih.gov/RefSeq/ ) accession numbers: bagpipe (NM_169958), Brd (NM_057541), grim (NM_079413), hairy (NM_079253), hid (NM_079412), lin-14 (NM_077516), lin-41 (NM_060087), and Scr (NM_206443). GenBank ( http://www.ncbi.nlm.nih.gov/Genbank/ ) accession numbers: sickle (AF460844) and D. simulans hairy (AY055843).

D:\keerthana\PMC001xxxxxx\PMC1043860.xml

1043861

Seeds of Destruction: Predicting How microRNAs Choose Their Target

Compare the gene number of fruitfly (13,000) to human (20,000), and it's pretty clear that complexity emerges not just from gene number but from how those genes are regulated. In recent years, it's become increasingly clear that one class of molecules, called microRNAs (miRNAs), exert significant regulatory control over gene expression in most plant and animal species. A mere 22 nucleotides long, miRNAs control a cell's protein composition by preventing the translation of protein-coding messenger RNAs (mRNAs). When a miRNA pairs with an mRNA, through complementary base pairing between the molecules, the mRNA is either destroyed or is not translated. Hundreds of miRNAs have been found in animals, but functions for just a few have been identified, mostly through genetic studies. Many more functions could be assigned if miRNA targets could be predicted. This approach has worked in plants, because miRNAs and their targets pair through the near perfect complementarity of their base pairs. But the molecules follow different rules in animals—duplexes contain just short stretches of complementary sequence interrupted by gaps and mismatches—which makes predicting miRNA targets a challenge. In a new study, Stephen Cohen and his colleagues at the European Molecular Biological Laboratory in Germany establish basic ground rules for miRNA–mRNA pairing using a combination of genetics and computational analyses, and identify different classes of miRNA targets with distinct functional properties. Although the miRNA is only 22 nucleotides long, its 5′ and 3′ ends seem to have distinct roles in binding. Cohen and colleagues show that miRNA functional targets can be divided into two broad categories: those that depend primarily on pairing to the miRNA's 5′ end (called 5′ dominant sites), with varying degrees of 3′ pairing, and those that also need the miRNA's 3′ end (called 3′ compensatory sites). Surprisingly, miRNAs can regulate their targets simply by strong pairing with so-called seed sites that consist of just seven or eight bases complementary to the miRNA 5′ end. Target sites with weaker 5′ complementarity need supplemental pairing with the miRNA's 3′ end to function. The finding that so little sequence complementarity is needed means that there are many more target sites than had been previously recognized. The miRNA 3′ end, while not essential, is expected to confer some function, since it tends to be conserved in animals—miRNA 3′ ends provide an additional measure of regulatory control by permitting the function of target sites that have only limited complementarity to the miRNA 5′ end. The authors speculate that seed sites might be the first functional sites acquired by protein-coding genes that require repression, and that additional sites might be acquired to promote stronger repression. Based on their experimental results, Cohen and colleagues searched the Drosophila genome for biologically relevant targets, and estimate that the fly has about 100 sites for every miRNA in its genome. Since the fruitfly has anywhere from 96 to 124 miRNAs, that means it has 8,000 to 12,000 target sites (in the 11,000 genes sampled). This indicates that miRNAs regulate a large fraction of protein-coding genes. Of the known animal miRNAs, many regulate critical developmental processes. This new approach to predicting targets should help reveal just how much regulatory control actually flows from these tiny bits of RNA.

D:\keerthana\PMC001xxxxxx\PMC1043861.xml

1043862

Cracking the Olfactory Code

For Proust, a taste of cookie was enough to trigger vivid recollections of his childhood, the first of a long string of reveries that he fashioned into his famous memoir Remembrance of Things Past . For many animals, too, tastes and smells are evocative and play a crucial role in finding food, allowing them to build on past successes and to learn how to find their next meal. To locate blooming flowers, for example, honeybees rely heavily on scent. They can associate a whiff of an aldehyde, say, with a nectar-filled orchid. Then later they'll seek out the same or similar scents. To succeed in the wild, they must be able to distinguish relevant scents at varying concentrations, and within complex milieus of other scents. But to find food in varied conditions and adapt to new situations, they also have to generalize from past experience. Through both physiological and behavioral studies, scientists have investigated the response to smell in a wide range of organisms and have suggested that two key properties of scent-inducing chemicals are the functional class, such as alcohol or aldehyde, and the carbon-chain length. Bees trained to associate a particular chemical with a reward, for example, can then generalize to some extent to other chemicals with the same functional groups or similar carbon-chain lengths. In these situations, bees are surprisingly consistent in both in their behavior (extending their proboscis to an odor previously associated with food) and in their brains (brain activity in smell-processing centers). Each set of data, behavioral and neural, can be thought of as a “code” underlying the bee's response: present a scent, and a bee's brain and body will tend to react in a certain way. Linking smellperception and neural activity in the bee (Image: Axel Brockmann) A new study of smell perception in honeybees ( Apis mellifera ) published in PLoS Biology gives a more comprehensive picture of how bees react to a suite of scents and also shows a remarkable correspondence between the codes for the insects' behavior and brain activity. The researchers, led by Martin Giurfa, first trained bees to associate a specific chemical, such as the alcohol 1-nonanol, with a sucrose reward. Then the researchers tested the bees' response to a set of other chemicals, varying in carbon-chain length from six to nine, and with four different functional groups: aldehydes, ketones, and primary and secondary alcohols. By watching how often the bees generalized—that is, how often they responded positively to a particular scent when they'd been trained on another—the researchers could assign perceptual “distances” between pairs of chemicals. Drawing together all these distances, they created a preliminary map of the bees' “perceptual space,” similar to how surveyors measure distances between landmarks to map a landscape. From this comparison they found, for example, that the bees generalized more by functional group than by carbon-chain length. Previously, Giovanni Galizia's group, which works closely with Giurfa's group, had recorded bees' brain responses to the same pairs of scents, assigning distances within centers of activity for each scent. Giurfa's team compared these two sets of data and found that the perceptual and neural distances correlated well, which suggests there's a species-specific code that ties together the insects' brain and behavior. The brain recordings covered only a quarter of the bees' main smell-processing center, the antennal lobe. Future studies with new methods of microscopy that visualize more of the brain and which focus on the olfactory message sent by the antennal lobe to higher-order brain centers should only improve our ability to investigate the correlations between brain and behavior, the authors say. Such studies would go even further toward cracking the codes underlying animals' perception and memory.

D:\keerthana\PMC001xxxxxx\PMC1043862.xml

1044830

Recombination Difference between Sexes: A Role for Haploid Selection

Why the autosomal recombination rate differs between female and male meiosis in most species has been a genetic enigma since the early study of meiosis. Some hypotheses have been put forward to explain this widespread phenomenon and, up to now, only one fact has emerged clearly: In species in which meiosis is achiasmate in one sex, it is the heterogametic one. This pattern, known as the Haldane-Huxley rule, is thought to be a side effect, on autosomes, of the suppression of recombination between the sex chromosomes. However, this rule does not hold for heterochiasmate species (i.e., species in which recombination is present in both sexes but varies quantitatively between sexes) and does not apply to species lacking sex chromosomes, such as hermaphroditic plants. In this paper, we show that in plants, heterochiasmy is due to a male-female difference in gametic selection and is not influenced by the presence of heteromorphic sex chromosomes. This finding provides strong empirical support in favour of a population genetic explanation for the evolution of heterochiasmy and, more broadly, for the evolution of sex and recombination.

Introduction Sex differences in recombination were discovered in the first linkage studies on Drosophila [ 1 , 2 ] and Bombyx (Tanaka [1914] in [ 3 ]) almost one century ago. However, this observation remains today largely unexplained despite several attempts. Based on very limited observations (see Table 1 ), especially of Bombyx, in which the female is heterogametic, Haldane [ 3 ] suggested, as far as “these facts are anything more than a coincidence,” that the lower autosomal recombination rate in the heterogametic sex may reflect a pleiotropic consequence of selection against recombination between the sex chromosomes. Later, Huxley [ 4 ] showed that Gammarus males also recombined less than females. He gave the same evolutionary explanation, although he restricted it to cases of a marked sex difference. Table 1 Data on Which the Haldane-Huxley Rule is Based Listed are the data available to Haldane [ 4 ] when he proposed the Haldane-Huxley rule a r m represents recombination in males b r f represents recombination in females c Plus and minus symbols indicate the direction of heterochiasmy, and zero indicates achiasmy This conjecture has now been confirmed for achiasmate species (i.e., species in which only one sex recombines) and is referred to as Haldane-Huxley rule: Nei [ 5 ] showed theoretically that tight linkage should evolve on Y or W chromosomes, and Bell [ 6 ] compiled a large dataset showing that achiasmy evolved 29–34 times independently, each time with no recombination in the heterogametic sex. However, for heterochiasmate species, three problems with the Haldane-Huxley pleiotropy explanation were discovered [ 7 , 8 ]. The first problem arose when substantial variation in male-female differences in recombination rate was found between pairs of autosomes within mice [ 8 ] and Tribolium [ 9 , 10 ], and between genotypes for the same pair of autosomes [ 11 ]. The second problem was the discovery that hermaphrodite species (the platyhelminth Dendrocoelum [ 12 ] and the plant Allium [ 13 ]) may present strong heterochiasmy between male and female meiosis despite having no sex chromosomes or even sex-determining loci. The third problem was the discovery of species in which the heterogametic sex recombines more than the homogametic one (e.g., in some Triturus species) [ 14 ]. Because of these contradictory observations, variation in heterochiasmy has remained difficult to explain because of the absence of an alternative theory as well as the lack of a clear pattern in the data. In 1969, Nei [ 5 ] worked out the first “modifier” model to study the evolution of sex differences in recombination, and concluded for autosomes that “the evolutionary mechanism of these sex differences is not known at present.” Surveying an updated dataset, Bell [ 6 ] concluded that “female gametes experience more crossing over among hermaphroditic plants (and perhaps animals), but this is not invariably the case among gonochoric animals (…) certainly (this) has never received any explanation.” The idea that heterochiasmy may be explained by a sex rather than by a sex chromosome effect, which was ignored by Haldane because of Bombyx, was reconsidered. This led Trivers [ 15 ] to suggest that, because only males with very good gene combinations reproduce (relative to females, for whom reproduction success is often less variable), they should recombine less to keep intact these combinations. He accounted for exceptions by variation in the regime of sexual selection. The idea was criticized by Burt et al. [ 16 ], who also questioned the correlations—with an updated dataset—between heterochiasmy and either sex or heterogamety. These authors tried to correlate the level of heterochiasmy with the amount of “opportunity for sex-specific selection,” but failed to find an effect. They were tempted to advocate neutrality, but were puzzled by the positive correlation between male and female recombination rate and by evidence showing compensation (e.g., female mice tend to recombine more on the X, as if they were compensating for no recombination in males; similarly, no species is known with achiasmy in both sexes [ 16 ]). In 1994, Korol et al. [ 17 ] insisted on a possible role for gametic selection but did not give evidence in favour of this claim. Recently, Lenormand [ 18 ], using Nei's modifier approach, showed that it is very difficult to explain heterochiasmy by sex-specific diploid selection. Rather, a sex difference in selection during haploid phase, or a sex difference in diploid selection on imprinted genes, is a more likely explanation. He predicted that, as far as haploid selection is concerned, the sex experiencing the more intense haploid selection should recombine less. Indeed, when allelic effects interact to determine fitness (i.e., when there is “epistasis,” either negative or positive), recombining decreases mean fitness in the population of the next generation [ 19 ]. This effect occurs because recombination breaks up combinations of genes that have previously been built up by selection. For a given average recombination rate between sexes and for a given average epistasis between male and female haploids, it is always advantageous for the haploid population (male or female) with the greatest absolute value in epistasis to be produced with the lowest amount of recombination. In this way, the “recombination load” that the haploid population is exposed to is minimized. In this paper, we would like to come up with a more quantitative evaluation of the possible role of haploid selection in shaping heterochiasmy. For that purpose, we first updated the dataset of Burt et al. [ 16 ] on heterochiasmy, focusing on genetic maps that have become available over the last 15 years. We then determined how fast heterochiasmy evolves, in order to measure the amount of phylogenetic inertia on this trait. Finally, we determined whether variables such as gender, heterogamety, or the opportunity for selection in the haploid phase, could explain variation in heterochiasmy. If there is selection with substantial epistasis on some genes during the haploid phase, we expect the sex with the greater opportunity for haploid selection to show less recombination. Alternatively, if selection during the haploid phase is weak or without substantial epistasis, we do not expect it to produce a directional bias in the amount of recombination displayed by either sex. Results/Discussion Sex Chromosomes Heterochiasmy is a fast-evolving trait, and phylogenetic inertia does not satisfactorily explain its distribution. In contrast to achiasmy, we found that heterochiasmy is not influenced by the nature of the sex chromosomes. This is interesting, because it suggests that achiasmy and heterochiasmy are influenced by qualitatively different evolutionary forces, although they seem to differ only quantitatively. It would be useful to determine whether achiasmy evolved to reduce the average recombination rate or to change the relative amount of recombination between the sexes. The two situations may be discriminated by determining whether the homogametic sex in achiasmate species tends to recombine more than in closely related chiasmate species. Evidence for such compensation would indicate that achiasmy did not evolve to reduce the average recombination rate. In the absence of such compensation, however, achiasmy may simply reflect selection for tight linkage. In such a situation, we propose that Haldane-Huxley rule may be caused by the converse argument to the one previously considered: The presence of achiasmy only in the heterogametic sex may reflect selection to maintain nonzero recombination rate on X or Z chromosomes in the homogametic sex. In species in which the average autosomal recombination rate is selected against (i.e., towards a lower equilibrium value), loss-of-function (recombination) mutations with an effect restricted to one sex may spread only if they affect the heterogametic sex, because mutations suppressing recombination in the homogametic sex completely suppress recombination on the X or Z chromosome. The same argument applies to XO species and may explain why achiasmy is associated only with the heterogametic sex. In addition, this hypothesis does not require the existence of genes suppressing recombination between the sex chromosomes with autosomal pleiotropic effects. Under this hypothesis, there is no reason to find an effect of the presence of heteromorphic sex chromosome on the amount of heterochiasmy, as originally envisioned by Haldane and Huxley. Overall, this hypothesis would explain why heterochiasmy and achiasmy differ qualitatively and why we do not observe any effect of sex chromosomes on heterochiasmy. Heterochiasmy in Animals In animals, male-female dimorphism in haploid selection may also contribute to heterochiasmy. In general, there is no female haploid phase in animals, because meiosis is completed only at fertilisation. As far as at least some genes are expressed and under selection during the male haploid phase, this would tend to bias towards tighter linkage in males. Sets of genes responsible for male-specific meiotic drive systems would be good candidates and are often found in tight linkage. Measuring the opportunity for haploid selection in animals may be possible within some groups. Imprinting may, however, act as a confounding effect in many groups of animals while trying to measure the opportunity for “haploid” selection. Within-species comparisons of imprinted regions or of regions with sex-specific recombination using high-resolution maps [ 20 ] may be more fruitful to discriminate among potential causes of heterochiasmy in animals. In particular, there is evidence in humans that the reduction in crossing-over associated with imprinting is in the direction that theory predicts, even if this pattern is consistent with other explanations [ 21 ]. Finally, understanding exceptions within groups (e.g., male marsupials, contrarily to most mammals, recombine more than females of the species [ 22 ]) may also shed light on the different hypotheses. Heterochiasmy in Plants We found that plant heterochiasmy is correlated with the opportunity for male and female haploid selection. Female meiosis tends to exhibit lower recombination rates relative to male meiosis when selection is intense among female gametophytes (e.g., in Pinaceae) or mild among male gametophytes (e.g., in highly selfing species). This pattern is expected if heterochiasmy is determined by the relative magnitude of haploid selection in male and female individuals. Finding a pattern consistent with this general population genetic prediction is, of course, not firm evidence that male-female dimorphism in haploid selection is the evolutionary force generating heterochiasmy. Other correlates of selfing rates might have to be closely examined [ 23 ]. However, we consider this explanation the most parsimonious so far. Our finding provides, therefore, the first empirical evidence for a theory explaining male-female differences in the amount of recombination and contributes to our understanding of contradictory observations that have puzzled geneticists for almost a century. It also indicates that the amount of recombination may be shaped by indirect selection, and, therefore, corroborates theories based on selection and variation for the evolution of sexual reproduction. Materials and Methods An extended dataset We measured heterochiasmy as the log of the male/-to-female ratio ( ρ ) of autosomal recombination rate measured either with chiasma number or map length. We log-transformed the ratio to avoid bias due to measurement error in the denominator. Chiasma-count data for different species were compiled by Burt et al. [ 16 ], and we used their dataset, adding a few recent studies. We compiled genetic map data and linkage studies in animals and plants for which both a male and a female map were available. Only homologous fragments (i.e., between shared markers) in male and female maps were considered (especially in low-resolution maps). Heterochiasmy data were available for 107 species, with 46 sets of data based on genomic maps ( Table 2 ). Table 2 Dataset Pooled by Species with Levels of Phylogenetic Grouping Used in the Analysis Note that references given in Burt et al. [ 17 ] were not repeated here a K, kingdom. Numeric indicators in this column are: 1, Animalia; 2, Plantae b P, phylum. Numeric indicators in this column are: 1, Arthropoda; 2, Chordata; 3, Embryophyta; 4, Platyhelminthes c C, class. Numeric indicators in this column are: 1, Actinopterygii; 2, Amphibia; 3, Magnoliopsidae (subclass asterids); 4, Aves; 5, Coniferopsida; 6, Insecta; 7, Liliopsida; 8, Mammalia; 9, Magnoliopsidae (subclass rosids); 10, Trematoda; 11, Turbellaria d Data refers to linkage map (LM) or chiasma count (CC) e Male and female indicate the value for the chiasma count or map length for each sex f Ratio refers to male/female recombination rate g V sc refers to the presence or absence of sex chromosome (see Materials and Methods , “Sex chromosome effect”) h Data were obtained from maps DBNordic2 and NIAIJapan ( http://www.genome.iastate.edu/pig.html ) [ 54 , 55 ] ND, no data Table 2 Continued Phylogenetic inertia Heterochiasmy may evolve so slowly that there is important phylogenetic inertia. Alternatively, it may be so fast-evolving that the amount of heterochiasmy takes on nearly independent values among related species. In the same way, heterochiasmy may be so variable between genotypes within a species that it may be difficult to measure and irrelevant to analyse species specific effects. In order to get a picture of phylogenetic inertia on heterochiasmy, we estimated the phylogenetic autocorrelation of ρ using Moran's I spatial autocorrelation statistic [ 24 ]. When standardized, values of Moran's I vary from −1 to 1. Positive values indicate that heterochiasmy is more similar than random within a taxonomic level, whereas negative values indicate that it is more different. Because a few species had multiple estimates of heterochiasmy, we also estimated the within-species correlation. The resulting correlogram is shown in Figure 1 . We found that heterochiasmy is a fast-evolving trait: Genotypes tend to be correlated within a species (I/I max = 0.38, p = 7.9%), but this correlation is lower among species within genera (I/I max = 0.18, P-value = 13%), and very low when comparing genera within families (I/I max = 0.039, p = 63%). This pattern is very different from the one observed for highly autocorrelated traits using the same method (for instance, mammalian body size [ 25 ]). This analysis indicates that there is very little phylogenetic inertia overall on heterochiasmy, but that the species level is appropriate for our dataset. However, this low level of inertia may nevertheless inflate type-I error while testing the effect of independent variables on heterochiasmy. In order to avoid this problem, we tested the association between different variables and heterochiasmy using a generalized estimating equations linear model correcting for the full phylogeny (see below) [ 26 ]. Figure 1 Phylogenetic Correlogram of Heterochiasmy and Selfing Rate The y-axis represents Moran's I rescaled to enable comparisons between each taxonomic level for heterochiasmy ( ρ , solid line) and selfing rate ( V m , dashed line). The x-axis represents the taxonomic level: /S is the correlation within species, S/G is the correlation of species within genera, etc. F, family; O, order; C, class; P, phylum; K, kingdom. Filled points indicate significance at p = 0.05. Sex chromosome effect For each species, we reported the presence of sex chromosomes. We defined the variable V sc with the following values: −1 for XY/XX species, −1/2 for XO/XX or XY/XX without pseudoautosomal regions (marsupials), 0 for species without sex-chromosomes, and +1 for ZZ/ZW species. We distinguished the −1 and −1/2 cases to reflect the fact that, in the latter, recombination does not occur between sex chromosomes, so we expect a lower current selection pressure to suppress recombination. Under the Haldane-Huxley hypothesis, the presence of sex chromosomes is supposed to favour reduced recombination rate in the heterogametic sex. We therefore expect a positive effect of the variable V sc on ρ . We did not find such an effect in animals or plants (the linear effect of V sc on ρ is not significantly different from zero [ p = 0.75 in animals and p = 0.52 in plants], assuming species were independent), and this result is unchanged if the −1 and −1/2 cases are not distinguished. Given this negative result, there was no need to do a phylogenetic correction. Gametic selection In animals from our dataset, there is no female haploid phase because the completion of meiosis occurs only at fertilisation (sperm triggers the end of meiosis). In male gametes, very few genes are expressed, and sperm phenotype is determined mostly either by the diploid genotype of the paternal tissue or by its mitochondrial genome. Imprinted genes, which can also affect the evolution of heterochiasmy [ 18 , 21 ], may be as numerous as haploid-expressed genes and act as a confounding factor while evaluating the “opportunity” for male or female gametic selection. As a consequence, we did not attempt to evaluate the opportunity for haploid selection in animals. Rather, we focused on plants, in which there is both a male (pollen) and female (ovule) haploid phase and during which many genes are expressed (e.g., as many as 60% of genes may be expressed in the male gametophyte [ 27 , 28 ]). In order to evaluate the effect of the “opportunity for selection” for male haploid phase on ρ, we used selfing rate as an indirect variable estimating the degree of pollen competition. We assume that with high selfing rates, there is less genetic variation among competing pollen grains and, therefore, less scope for haploid selection. We defined V m (the degree of male gamete competition in plants) using three values depending on the amount of selfing: 0 for dioecious, self-incompatible or largely outcrossing (less than 5% selfing reported) species; 1 for species exhibiting low selfing rates (less than 30% reported); and 2 for other species. We used these three broad categories to reflect the fact that selfing rate is often variable within species and that it is often measured indirectly and with low precision. We therefore expect a positive effect of the variable V m on ρ if the opportunity for male gametic selection favours smaller ρ values, as predicted by the modifier model [ 18 ]. We tested this effect using the 57 species for which we were able to estimate V m ( Table 3 ). We used a linear model in R [ 29 ] assuming that all species are either independent or phylogenetically related. In the latter case, we used a generalized estimating equations linear model [ 26 ] with a plant phylogenetic tree to the family level using data from Davies et al. [ 30 ], and several calibration points, including the Picea / Pinus divergence approximately 140 million years ago [ 31 ], that are not included in the Davies et al. dataset. We found an effect in the right direction with or without correcting for the phylogeny (linear effect of ρ on V m , p < 0.0002 in both cases, Figure 2 ). The fact that selfing plants exhibit higher recombination rates than their outcrossing relatives has been mentioned previously in the literature [ 32 , 33 ]. However, in most cases, recombination was measured only in male meiosis. It would be valuable to reexamine this trend in the light of our results that recombination in male meiosis is typically greater than in female meiosis among selfers. Figure 2 Logarithm of Male-Female Ratio in Recombination Rate in Plants Mean and 95% confidence interval of ρ is shown for different groups of plants, assuming normality and independent data points The number of species in each group is indicated next to the mean. Table 3 Plant Species Used to Test the Effect of Male and Female Opportunity for Selection a Ratio refers to male-to-female recombination rate LM, linkage map; CC, chiasma count; n, haploid number of chromosomes; V m , measure of male opportunity for haploid selection; V f , measure of female opportunity for haploid selection In order to evaluate the effect of the “opportunity for selection” during the female haploid phase on ρ in plants, we contrasted angiosperms with gymnosperms. In angiosperms, ovules do not compete much with each other on a mother plant, because resource accumulation starts after fertilisation (i.e., during fruit development in the diploid phase). In Pinus (three species in our dataset; see Table 2 ), male meiosis, female meiosis, and pollination occur in the year prior to fertilisation, but the pollen tube stops growing until the next spring, while the female gametophytes continue to accumulate resources and compete with each other over the course of the year. The same situation occurs in Picea, although the period between female meiosis and fertilisation is only 2–3 mo [ 34 ]. Perhaps more importantly, the endosperm (which is the organ managing resources for the zygote) is haploid in Pinaceae, in contrast to the double fertilisation that occurs in angiosperms to produce at least a diploid (typically triploid) endosperm [ 35 , 36 ]. We therefore expect that ρ should be greater in Pinaceae, compared to angiosperms. We assigned V f (the degree of female gamete competition in plants) the values 1 for gymnosperms and −1 for angiosperms. We expected a positive effect of the variable V f on ρ according to the modifier model. An effect in the right direction was indeed detected (linear effect of V f on ρ, p = 0.011 and p = 0.0001, with and without correcting for the phylogeny as above, respectively; see Figure 2 ).

D:\keerthana\PMC001xxxxxx\PMC1044830.xml

1044831

Engineering Gene Networks to Emulate Drosophila Embryonic Pattern Formation

Pattern formation is essential in the development of higher eukaryotes. For example, in the Drosophila embryo, maternal morphogen gradients establish gap gene expression domain patterning along the anterior-posterior axis, through linkage with an elaborate gene network. To understand the evolution and behaviour of such systems better, it is important to establish the minimal determinants required for patterning. We have therefore engineered artificial transcription-translation networks that generate simple patterns, crudely analogous to the Drosophila gap gene system. The Drosophila syncytium was modelled using DNA-coated paramagnetic beads fixed by magnets in an artificial chamber, forming a gene expression network. Transient expression domain patterns were generated using various levels of network connectivity. Generally, adding more transcription repression interactions increased the “sharpness” of the pattern while reducing overall expression levels. An accompanying computer model for our system allowed us to search for parameter sets compatible with patterning. While it is clear that the Drosophila embryo is far more complex than our simplified model, several features of interest emerge. For example, the model suggests that simple diffusion may be too rapid for Drosophila -scale patterning, implying that sublocalisation, or “trapping,” is required. Second, we find that for pattern formation to occur under the conditions of our in vitro reaction-diffusion system, the activator molecules must propagate faster than the inhibitors. Third, adding controlled protease degradation to the system stabilizes pattern formation over time. We have reconstituted transcriptional pattern formation from purified substances, including phage RNA polymerases, ribonucleotides, and an eukaryotic translation extract. We anticipate that the system described here will be generally applicable to the study of any biological network with a spatial component.

Introduction Engineering a system to emulate a particular behaviour can be an extremely informative approach to systems biology [ 1 , 2 , 3 , 4 , 5 ]. Even if the natural biochemical interactions are well characterized, it remains a considerable challenge to reconstruct a physical system with the appropriate behaviour. Step-by-step reconstruction allows theoretical assumptions and models to be refined. Not only is complexity reduced by removing the context of the whole organism, but by reconstitution of pattern formation from purified substances (in this case, RNA polymerases, ribonucleotides, and a translation extract) the sufficiency of a proposed mechanism of pattern formation can be demonstrated. In this work, our primary aim was the development of an in vitro system that allows the careful buildup of complex networks under controlled conditions. To demonstrate the usefulness of such a system, we decided to reconstruct a developmental pattern-formation program based on the formation of a gradient of a transcription activator—a “morphogen”—and to link it to a network of transcription repressors. In a sense, we set out to design a patterning system similar to chemical reaction-diffusion systems (see below). However, the use of components such as transcription activators and repressors in an in vitro transcription-translation system made the system significantly closer to a biological system, albeit highly simplified when compared to the complexity of eukaryotic transcription [ 6 ]. Patterning systems can be thought of as belonging to one of two principal types: First, there are systems with homogeneous initial conditions that self-organise after early random symmetry-breaking events ([ 7 , 8 ]; for a biological example, Fucus, see [ 9 ]). Second, there are systems with initial localisation of the components, which can form concentration gradients of activities from their respective sources [ 10 , 11 ]. The first class of patterning system, involving reaction-diffusion from initially homogeneous conditions, was first proposed by Turing in 1952 [ 7 ] and was further developed by Meinhardt and Gierer in the 1970s, in their model of patterning with short-range autoactivator and long-range lateral inhibitor [ 8 ]. Although there are many likely biological candidates for such activator-inhibitor systems (reviewed in [ 12 ]), none has been reengineered from first principles. In contrast, significant progress has been made in reconstituting purely chemical reactions that self-organise, such as the Belousov-Zhabotinski reactions [ 13 , 14 , 15 ] and Turing-type, Meinhardt-Gierer (M-G), and oscillatory reactions [ 16 , 17 , 18 , 19 , 20 ]. In a more biological context, similar spatiotemporal patterns, consisting of propagating concentration waves, have been modelled for a glycolytic enzyme oscillator in yeast [ 21 ]. In the second class of patterning system, where the initial conditions are nonhomogeneous, patterning begins with an asymmetry: a localised source of morphogen (shape-defining molecules that form a concentration gradient and function in a concentration-dependent manner so as to determine positional information in a patterning field [ 22 ]; reviewed in [ 23 , 24 , 25 ]). The question of how morphogen gradients are formed and maintained is still a matter of keen debate and study [ 26 , 27 , 28 ], with many proposed mechanisms (reviewed in [ 29 , 30 ]). In the simplest case, a stable gradient could be formed by passive diffusion [ 31 ] and uniform degradation, although it has been suggested that enhanced morphogen degradation near the source leads to increased robustness against morphogen fluctuations during patterning [ 32 ]. Also, “ligand trapping” by the receptor for a morphogen can have significant effects on the shape of a morphogen gradient [ 33 ], as in the case of Torso diffusing in the extracellular space surrounding the Drosophila oocyte [ 34 ]. In most cases in metazoa, morphogens define patterns over fields of many cells (reviewed in [ 30 , 35 ]), but there is one special case in embryonic development that has been particularly well studied, in which a morphogen gradient operates in a single-celled, multinuclear syncytium: the early Drosophila embryo [ 10 , 11 ]. In this system, early patterning is mediated by maternal morphogen factors, which are thought to diffuse and form gradients to guide patterning within a large multinuclear cell (reviewed in [ 36 , 37 , 38 ]). After egg deposition, an embryo forms a segmentation pattern within 3 h, under the influence of a hierarchical sequence of gene expression interactions involving gap genes, pair-rule genes, and segment polarity genes [ 39 ]. Maternal elements—in particular, the morphogen Bicoid—guide this process, setting distinct initial conditions. Work by Driever and Nüsslein-Volhard [ 10 , 11 ] demonstrated that Bicoid protein possessed three characteristics of a classic morphogen: (i) a localised source of cytoplasmic activity (through bicoid RNA transport to the anterior pole of the egg, involving microtubules and maternal genes [ 40 , 41 , 42 , 43 , 44 ]); (ii) formation of a concentration gradient from the source; and (iii) concentration-dependent activity that determines positional information within the gradient (reviewed in [ 38 ]). Bicoid has at least two functions that contribute to its function as a morphogen: transcription activation and translation inhibition. Acting as a transcription factor, Bicoid can activate a number of downstream gap genes, including hunchback, knirps, giant, and Krüppel, whose products cross-react in a complex and mainly repressive interaction network to modulate each other's expression (reviewed in [ 37 ]; modelled in [ 45 , 46 ]). However, Bicoid does not simply function independently as a morphogen at the top of the gap gene hierarchy. Although Bicoid is responsible for anterior expression of zygotic Hunchback [ 47 ], it actually requires maternally expressed Hunchback as a cofactor to function anteriorly [ 48 ]. Meanwhile, in the posterior, the maternal hunchback mRNA is initially translationally repressed by the posterior determinant Nanos [ 49 ]. Moreover, the terminal gap genes tailless and huckebein are activated independently but serve to repress zygotic gap gene expression at the poles of the embryo, thus influencing patterning [ 50 , 51 , 52 ]. In its other role, as a translation inhibitor of caudal mRNA, Bicoid initially inhibits the uniformly expressed mRNA to form a concentration gradient of the protein Caudal [ 53 , 54 ]. Thus, the Caudal gradient [ 55 ] is essentially the inverse of the Bicoid gradient, and Caudal functions as a transcription activator in the posterior of the egg, further influencing expression of the gap gene network. Some of the important interactions in the gap gene network are shown in Figure 1 A, although it should be noted that this overview is an oversimplification and does not consider differences in maternal and zygotic factor expression over time, nor does it consider all of the factors involved. Figure 1 Gene Circuits and Chambers (A) Principal interactions in the Drosophila gap gene network, modelled after [ 37 ]. Relative levels and distributions of Hunchback (Hb), Giant (Gt), Krüppel (Kr), Knirps (Kni), Bicoid (Bcd), and Caudal (Cad) shown from anterior (left) to posterior (right). Green arrows indicate activation, red T-bars repression. (B) Artificial gene network design, with transcription activators T7 and SP6 polymerases, and zinc finger repressors A, B, and C. Genes are immobilised on paramagnetic beads, and T7 forms a directional concentration gradient. (C) Principal interactions in a simple designed network. (D) Transcription-translation chamber. Genes for repressor A are localised at the “poles,” whereas B and C are ubiquitous. Gel slabs 4–6 have been excised, exposing the magnets below, illustrating gel dissection for Western blot analysis. (E) Normalised Western data for four replicate chambers, showing mean levels of A, B, and C after 20 min (± One standard deviation). (F) Sample Western blot from the four-replicate experiment. From the outset, we chose to model our system around elements of the Drosophila gap gene network ( Figure 1 A), because we could study many elements of morphogenesis, such as gradient formation and the sufficiency of cross-repression for setting pattern boundaries, without the need for considering multiple cells, membrane-bound receptors, and cell-to-cell interactions. Using an in vitro model, we wanted to address the following questions about patterning. First, how easy is it to generate an expression pattern in a gradient, using a diffusing activator from a localised source? Second, one of the outstanding issues in the field is to what extent correct positioning of the gap protein domain boundaries is specified by maternal morphogen gradients and by cross-repression between gap genes; a recent model suggests that repression is crucial for patterning and that threshold-dependent interpretation of the maternal morphogen concentration is not sufficient [ 56 ]. We therefore wanted to test the effect of transcription repression on pattern formation directly, by progressively adding more repression interactions in a designed gene network. Third, uniform degradation of a diffusing morphogen is often assumed to account for steady-state gradient formation, so we set out to test the effects of adding controlled degradation to an in vitro patterning system. Finally, we wanted to use our model to see how the scale and pattern of the system are affected by the relative rates of diffusion of individual components, and whether nonuniform diffusion of activators and inhibitors are required to form a pattern. Results Design of the Network and Development of the In Vitro Experimental Platform We began by designing a simplified gene network to emulate elements of the Drosophila gap gene system ( Figure 1 ). The aim was to develop a fully synthetic approach in which protein analogues completely unrelated to Drosophila would emulate some of the transcription activation and repression interactions thought to be important for patterning in the gap gene system. As activators, two sequence-specific polymerases were employed, T7 and SP6, that have been used successfully by others to engineer gene networks [ 57 ]. These two polymerases bind to their respective consensus DNA recognition sites to initiate transcription, and thus represent an extremely simplified mode of transcription when compared to the multifactor complexes required for eukaryotic transcription (reviewed in [ 6 ]). T7 polymerase was chosen to be the “master activator” of the system and, by crude analogy, was expected to carry out some of the functions of Bicoid, namely transcription activation of downstream members in the gap gene hierarchy, in a concentration-dependent manner, from a localised source [ 10 , 11 ]. In Drosophila , the Bicoid morphogen gradient initially controls the shape of the Caudal protein gradient through translational repression of maternal mRNA [ 53 , 54 ], although later Caudal expression is under zygotic transcriptional control. To simplify this level of complexity, we decided to model Caudal transcription activation by a second gradient of T7 polymerase, from the opposite pole to our primary “Bicoid” gradient (compare Figure 1 A and 1 C). Residual activation between other members of the gap gene members (e.g., Figure 1 A, Hunchback activating Krüppel or Krüppel activating knirps ) was modelled nonexplicitly by having a homogeneous distribution of a second sequence-specific transcription activator, SP6 polymerase ( Figure 1 A and 1 C). Repression interactions between gap gene members were modelled by constructing three site-specific repressors to represent the repressor activities of Hunchback, Giant, and Krüppel ( Figure 1 B, repressors A, B, and C, respectively). The repressors were derived from artificial zinc finger DNA-binding domains that were engineered by phage display [ 58 ]. Variable gene-repression networks were therefore constructed by placing binding sites for the repressors in the appropriate gene expression constructs (see for comparison Figure 2 ). Repressor sites were either overlapping with the polymerase initiation sites (demonstrated to be effective using triplex-forming oligonucleotides [ 59 ]), or immediately downstream of the initiation sites ( Figure 2 ). Therefore, by changing the identity of the repressor sites, the connectivity of the network could readily be modified to add or remove cross-repressive interactions. Figure 2 Map of the Constructs Used in This Study The repressor binding sites overlap with T7 or SP6 promoters and vary between constructs. In this way, it is possible to alter the connectivity of the repressive interactions by the products of genes A, B, and C. Repressive interactions are denoted by T-bars. The start codon of each gene is in Kozak context and is denoted by “GCC ATG G.” Key to the strategy was the development of an experimental platform in which to model the volume of the Drosophila embryo and to carry out artificial gene network reactions. Plastic chambers were therefore developed, constructed over printed templates on petri dishes ( Figure 1 D; see also Materials and Methods and Protocol S1 ). The chambers were filled with a customised transcription-translation mixture, allowing gene network reactions to be carried out in situ. Additionally, small bar magnets were fixed under the chamber to create a spatially defined array, over which paramagnetic beads could be dispensed. By coating such streptavidin-linked beads with biotinylated PCR products, specific gene network constructs were tethered and sublocalised on the array. Furthermore, ultra-low melting point agarose was added to the transcription-translation mixture, both to increase viscosity and to allow the reaction to be “fixed” in a gelling step at 4 °C. Through fixing, gel slices could be excised and assayed by Western blotting against FLAG-epitope tags on the expressed proteins. This design therefore enabled quantification of each output species present in the network (genes A, B, and C) for any given chamber position and time point. The chambers were constructed such that the system components could be pipetted wherever desired, either homogeneously mixed with the transcription-translation mix or pipetted at defined loci, such as at the edges or “poles” of the chamber. As described above, these system components included the two soluble, purified transcription activators (T7 and SP6 polymerases) and three bead-tethered zinc finger transcription-repressor constructs, A, B, and C, which were themselves activated by the polymerases and could cross-repress each other (see Figure 1 ). Positional information was therefore introduced into the artificial system in two ways. First, by injecting purified T7 polymerase at either pole of the chamber, the Bicoid activator distribution could be transiently modelled. Second, beads coated with different gene network constructs (genes A, B, and C), could be fixed at different positions on the magnetic array (see Figure 1 D). For example, repressor A genes were placed solely at the chamber edges (“poles”) to model, loosely, the distribution of embryonic Hunchback activity. This part of the model is a significant oversimplification: Although Hunchback is eventually expressed in two domains, one anterior and one posterior [ 48 ], it is only expressed anteriorly in the early embryo. Furthermore, while maternal hunchback mRNA is evenly distributed, the anterior domain of Hunchback protein forms through zygotic translation and transcription activation (under the control of Bicoid), while maternal RNA is translationally repressed posteriorly, under the influence of Nanos [ 49 , 60 , 61 ]. In the later phase of hunchback regulation, the posterior Hunchback domain forms through a combination of factors, including activation by Tailless [ 62 , 63 ] and hunchback autoactivation [ 64 ]. To complete the model, the genes for repressors B and C were distributed uniformly throughout the chamber on magnetic beads, to represent the ubiquitous distribution of genes in nuclei, throughout the embryo. Therefore, the spatial expression of genes B and C, who represent the downstream gap gene members giant and Krüppel (see Figure 1 A and 1 C), was dependent on differential activation by the T7 polymerase gradient and crossregulation between gene network members. However, it should be noted that the initial expression of these genes in Drosophila may not be achieved by crossregulation, because localised mRNA is seen before any protein is detectable (Krüppel and Giant are only detected unambiguously in early cycle 13 [ 46 ]). Pattern Generation In Vitro from a Transcription Network In our first experiments, we constructed a simple, minimal network with sequential transcription activation and repression (see Figure 1 B and 1 C). Although this basic system is far less complex than the Drosophila gap gene system, it was indeed sufficient to generate a crude target behaviour (see Figure 1 E and 1 F). Qualitatively, the pattern can be explained as follows, Gene A is activated by T7 polymerase from its source at either end of the chamber, and so is expressed most highly at these poles. Gene B is similarly activated, and so it is also less expressed in the middle of the cell. However, since gene B is repressed by protein A, its levels are also reduced at either pole. Finally, Gene C is activated by a ubiquitous SP6 polymerase, but is repressed by proteins A and B, and is consequently centrally distributed. Progressing from the minimal network, we explored systems with a variety of connectivities ( Figure 3 ), including a control network without repression interactions ( Figure 3 A), and one with extensive mutual or feedback interactions ( Figure 3 C). These were compared with the original network ( Figure 3 B) in a series of time-course experiments. Generally, we observed that the more repression interactions in a system, the lower the overall protein production but the “sharper” the pattern. Figure 3 Alternative Gene Networks At five set time points (15, 25, 35, 60, and 90 min), transcription-translation chambers were dissected into nine slabs for Western blot analysis. (A) Control network with no repression sites between genes A, B, and C. (B) Minimally repressed network (compare Figure 1 ). (C) Mutual repression network with extensive negative interactions between species. Adding protease (“+ Degradation”) creates weak but time-stable patterns for both the “Repressed” and “Mutual” networks (35 versus 90 min). Quantitated graphs for the above data are available in Protocol S1 . All patterns degenerated to a significant degree by 60 min, indicating the transience of the system ( Figure 3 , 60 min). However, by adding Factor Xa protease, we were able approximately to match levels of production and degradation. Thus, the outputs became sharper, weaker, and more dynamically stable, hardly varying between 35 and 90 min ( Figure 3 B and 3 C, “+ Degradation”). Computer Exploration of Parameter Space To study parameter sensitivity in our system more comprehensively, we constructed a computer model of the chamber and networks ( Protocol S1 ). A series of coupled differential equations were simulated, yielding expression levels of gene products A, B, and C, for the three different levels of network connectivity coded by our gene network designs ( Figure 4 ). The modelling was carried out at two scales, that of our experimental system (18 mm long) and that of a Drosophila embryo (500 μm long). As in the experimental system ( Figures 3 and 4 A), the simulations revealed a large difference of pattern between the unrepressed and repressed systems. The patterns are more similar, however, between the simple and mutually repressed networks ( Figure 4 B) but, as in our in vitro experiments (e.g., see Figure 3 B and 3 C, 15 min), adding feedback repression makes the peaks better resolved. Figure 4 Comparison of Experimental Data and Computer Simulations Data are shown for the three gene networks described in Figure 3 , showing outputs for proteins A (cyan), B (magenta) and C (dark blue). (A) Quantitated Western blot data from Figure 3 , after 25 min. (B) Simulation data plotted as percentage of total output protein against chamber length, at the chamber (18-mm) or Drosophila (0.5-mm) scale. The model is described in full in Protocol S1 . Next, we explored the sensitivity of the simple repression network to diffusion parameters ( Figure 5 ). Generally, we found that the A- and C-peaks were least sensitive to parameter variation, as there are no antagonistic forces against their formation ( Figure 5 A and 5 C). By contrast, gene B is more sensitive: Twin-peak formation correlates with the relative diffusion ratios of activator (T7) and other mRNA/protein components ( Figure 5 B). To generate “target behaviour,” the activator must diffuse more rapidly than other species, within certain limits (approximately 5- to 50-fold faster for Figure 5 B). However, the absolute values (and ratios) for diffusion merely alter the timing of the transient B-peak formation in a system of a given scale. For simplicity, only the 0.5-mm system is illustrated in Figure 5 ; similar conclusions were drawn from the 18-mm scale model. Figure 5 Varying Diffusion and Degradation Parameters Computer model of gene network, scaled to Drosophila length (0.5 mm). Diffusion parameters are varied for mRNA (Dm), protein (Dp), and T7 activator (DX). Data are plotted as percentage of total output protein (y-axes) against chamber position (x-axes), for 10-min simulations. (A) Outputs for protein A. (B) Output for protein B. Graphs with “target behaviour” are shaded grey, and the four asterisks mark the parameter sets used to generate outputs for proteins A and C. (C) Outputs for protein C. (D) Effect of adding protease degradation to B-output, shown at 15-min intervals, over a 2.5-h time course (parameters: DX = 0.43 μm 2 s −1 ; Dm = Dp = 0.02 μm 2 s −1 ; t 1/2 = 770 s). As in our chamber experiments, the computer model output became more “time-stable” by adding a degradation element ( Figure 5 D). Drosophila may exploit such mechanisms to some extent, since Bicoid protein degrades in vivo (t 1/2 ≤ 1800 s [ 11 ]), although bicoid mRNA is unusually stable [ 65 ]. Discussion To develop a fully synthetic approach that will emulate elements of gap gene expression domain pattern formation, we created an in vitro transcription-translation system that allows flexible spatial gene network construction. The system is widely applicable, allowing control over factors such as localisation or diffusion, and the ability to add or remove components at will. Repressive Interactions and Pattern Formation A basic aim of our system was to see whether we could engineer a gradient of protein expression, using a diffusing activator from a localised source. We found this task straightforward in the transcription-translation chambers, using injected T7 polymerase, and this led us to try more complex expression-repression interactions. We constructed three types of gene network—unrepressed, simple repressed, and mutually repressed (see Figure 3 )—representing different levels of network connectivity. Generally, in both our in vitro and computer models, we found that adding more connections resulted in better-resolved patterning, although the absolute levels of gene expression were reduced. Our in vitro results are essentially qualitative at this stage, but appear to agree with the observations of others—that crossrepression is crucial for the control of patterning boundaries [ 56 ]. It will be interesting to learn whether more sophisticated elements can be engineered into the system to begin to emulate the more complex features of gap gene expression domain patterning. For example, dynamic anterior shifts are seen in domain expression over time because of asymmetric gap-gap crossrepression [ 56 ]. Asymmetric repression and other circuits could, in the future, be engineered into our chambers by altering the repressor binding sites in the appropriate constructs. Such a system would require component turnover to achieve steady-state patterning. We have begun to tackle this project through our experiments with controlled protease degradation, but a further requirement would be to have autocatalytic production of T7 polymerase from a localised source, rather than the injected pulse of purified polymerase in our current model. Interestingly, our experimental data showed a reproducible degree of patterning even in the unrepressed system ( Figure 3 A, 15 min, C output). Because gene C is activated by a separate polymerase (SP6), this patterning cannot result from competition for activator. Therefore, competition for other resources (such as ribosomes, nucleotides, and tRNAs) may allow A and B to “inhibit” C. Indeed, supplying extra components (particularly wheat germ extract and SP6 polymerase together) increases protein production under these conditions, including that of C (unpublished data). If competition can generate patterns, albeit less well defined ones than repression-connected networks, this could perhaps represent an evolutionary “network precursor” state: Weak patterns could be generated by localisation and competition between factors, and these could later be consolidated by evolution of a “true” negative network connection. However, this hypothesis may not be relevant to the situation inside an insect egg, as this has probably evolved to deliver nutrients very efficiently to the embryo, even at a very primitive evolutionary stage. It therefore remains to be seen whether competition effects would be as significant in vivo. Since our models use minimal components to achieve spatial pattern formation, they demonstrate the ease with which very simple networks might evolve. Patterning may be achieved with only localisation, diffusion, and some kind of functional network connection, such as transcription activation, competition, or repression. The addition of extra layers of network properties, such as controlled degradation, could then fix and stabilize such patterns. In fact, since sublocalisation—followed by stepwise addition of network components—is sufficient to generate crude patterns, it might provide a plausible mechanism for early spatial network evolution inside a single cell. However, it should be noted that gap gene expression domain patterning probably evolved by a different mechanism, from an earlier multicellular state, where segments were added sequentially through polar growth. In fact, bicoid is absent in most other insects, and it has been proposed that Drosophila evolved bicoid by duplication of the homeodomain-encoding gene zerknüllt, found in lower Diptera [ 66 ]. Diffusion Rates and Patterning We were intrigued by our observations, both in vitro and in silico, that patterning required the activator molecule to diffuse or propagate more rapidly than the inhibitors. This is interesting because it is the opposite of the M-G system (described in the Introduction and reviewed in [ 12 ]). Long-range activation is not unknown in chemical patterning systems [ 67 ], although many biological models appear to require the M-G criterion for long-range inhibition (e.g., [ 68 ]). The other obvious differences between our system and the M-G model are the initial localisation of components and the lack of autocatalysis of the activator. It will be interesting to determine whether such M-G patterning systems can be recreated in our chambers, once further factors are considered, such as the avoidance of “autocatalytic explosions” or “global inhibitions.” The computer model that we developed allowed us to test a broad range of parameters, such as diffusion and degradation rates ( Figure 5 ), revealing differences between the requirements for patterning among the different species in the gene network. First, the more-connected member of the network (gene B) was much more sensitive to parameter variation than the less-connected members (genes A and C). This is perhaps to be expected, since protein B has two separate boundaries of expression (defined as a function of T7 distribution and both A and C expression), whereas proteins A and C have only single “edges” to be defined. Another important feature of the system emerged when scaling the parameters for the model patterns to Drosophila scale ( Figure 5 , 0.5 mm “embryo,” 2.5 h). We found that, assuming simple diffusion, B-peak formation was compatible only with unphysiologically slow diffusion values (diffusion constants for mRNA [Dm] and protein [Dp] = 0.02 μm 2 s −1 ; for T7 activator [DX] = 0.43 μm 2 s −1 ). Since cellular proteins are expected to diffuse more rapidly (approximately 1–100 μm 2 s −1 ), this could be an artefact, reflecting the simplicity of our model. Nonetheless, simple diffusion still appears too rapid to account for Drosophila -scale patterning. It should be noted that a potential barrier to free diffusion is the active nuclear import of Bicoid and Hunchback [ 69 ]. In a separate example, diffusion of pair-rule transcripts is overridden by microtubule transport [ 70 ]. Controlled sublocalisation may therefore be crucial to limit apparent diffusion in vivo, allowing more precise patterning. Perspectives The understanding of how precision of patterning is achieved in Drosophila is still far from complete. In a recent study, it was shown that the Bicoid profile is far more variable between embryos than that of Hunchback, but the mechanism by which this noise is filtered remains unknown [ 71 ]. As more and more detailed experimental data are collected [ 72 ], and new mechanisms are proposed to account for patterning, it will be important to test the sufficiency of these mechanisms through experimental reconstitution. For such purposes, the chambers described here may be easily adapted to test different hypotheses. In vitro systems are a useful first step towards testing the sufficiency of a network—which might then be reengineered in the original target organism. Combining simple reconstruction with theoretical modelling is a useful tool to discover and test general design principles in gene networks [ 73 , 74 , 75 , 76 ]. Until now, however, the spatial component essential in many biological processes has been ignored in these approaches. We anticipate that other networks, such as signalling cascades or metabolic networks, might also be studied using our system and that the spatial element, introduced through the beads, might provide new insights into complex systems. Materials and Methods Magnetic chamber construction A detailed, step-by-step description of the construction of the chamber can be found in Protocol S1 . Briefly, nine stirring-bar magnets (1.5 mm × 8 mm; VWR International, Vienna, Austria; #4429025) were inserted vertically into a plasticine-filled standard petri dish, creating a magnetic array (see Figure 1 D). Construction was guided with a grid template, laser-printed on a transparent acetate sheet, and fixed over the magnets and plasticine. The template was a 3 mm × 18 rectangle with nine subdivisions (“slabs”). A sterile cell culture dish (Nalge Nunc, Rochester, New York, United States; #150350) was fixed immediately above the magnetic array. Chamber borders (1 mm deep) were constructed on the base of this second dish, following the template, using strips cut from adhesive Hybriwell chambers (Sigma-Aldrich, St. Louis, Missouri, United States; #H1159–100EA). Gene network constructs Maps of the constructs are illustrated in Figure 2 . Repressors A, B, and C were derived from previously engineered zinc fingers [ 58 ]. Repressor A contained six zinc fingers, recognising the sequence 5′- AGGGAGGCGGACTGGGGA-3′, fused to the residues 11–55 of the Kox-1 repression domain [ 77 ] and a six-repeat FLAG epitope tag [ 78 ]. Repressor B contained six zinc fingers, recognising the sequence 5′- AGGGAGGCGGGAGCTTTC-3′ and fused to a three-repeat FLAG-tag. Repressor C contained three zinc fingers, recognising the sequence 5′- GGAGCTTTC-3′, fused to the Kox domain and a three-repeat FLAG-tag. The following polymerase consensus promoter regions were used: T7, 5′- TAATACGACTCACTATA G GGAG-3′; SP6, 5′- ATTTAGGTGACACTATA G AAGGG-3′. The gene network promoters were linked with neutral or repressor sites to the polymerase promoters. In the following nucleotide sequences, zinc finger binding sites are indicated in lowercase, initiation nucleotides in bold, and promoter overlaps underlined. Unrepressed T7, 5′- TAATACGACTCACTATA G GGAGAAACACCATAG-3′ (see Figure 3 A, constructs A and B, and Figure 3 B, construct A). Unrepressed SP6, 5′- ATTTAGGTGACACTATA G AAGGGAAACACCATAG-3′ (see Figure 3 A, construct C). T7 repressed by A (and weakly by B), 5′- TAATACGACTCACTATa g ggaggcggactgggga -3′ (see Figure 3 B, construct B). SP6 repressed by A (and weakly by B), 5′- ATTTAGGTGACACTATA G Aagggaggcggactgggga-3′ (see Figure 3 B and 3 C, construct C). T7 repressed by A and C (and weakly by B), 5′- TAATACGACTCACTATa g ggaggcggactggggaTggagctttc-3′ (see Figure 3 C, construct B). T7 repressed by C (and weakly by B), 5′- TAATACGACTCACTATA G ggagctttc-3′ (see Figure 3 C, construct A). Constructs were cloned in pCaSpeR4, sequenced, and used to generate PCR DNA for in vitro transcription-translation. Gene network reactions Paramagnetic beads were coated with PCR DNA (with one primer biotinylated) using a Dynabeads Kilobase Binder Kit (Dynal, Oslo, Norway; #601.01). Typically, gene A was used at 800 fmol per 10 μl of beads, resuspended in 8 μl of water; 200 fmol of gene B and 140 fmol of gene C were combined with 20 μl of beads, and resuspended in 20 μl of water. Transcription-translation mixture was prepared that included 2.5 μl of water; 28 μl of ultra-low melting point agarose (Sigma; #A2576) solution (prepared as 1.5% [w/v] in boiling water and cooled to 30 °C); and TNT Coupled Wheat Germ Extract System (Promega, Madison, Wisconsin, United States; #L4130 and #L4140), which comprised 20 μl of TNT wheat germ extract, 1.2 μl of TNT reaction buffer, 0.6 μl of amino acid mixture (1 mM), 1.2 μl of RNasin (not included in TNT kit), and 0.5 μl of SP6 polymerase. 54 μl of this mixture was dispensed per chamber. For degradation experiments, 2.25 units of Factor Xa (Amersham Biosciences, Little Chalfont, United Kingdom) were added per chamber. Coated Dynabeads were injected at appropriate positions over the magnetic array: typically, 100 fmol of gene A (1 μl), 5 fmol of gene B, and 3.5 fmol of gene C (0.5 μl). T7 polymerase (0.5 μl; from Promega TNT kit) was immediately injected at the chamber edges. After timed incubations at 25 °C, chambers were transferred to 4 °C for 35 min, to form a gel. Gel slices were cut with a razor blade (guided by the printed template) and aspirated with a P10 Gilson pipette. Samples were mixed with 10 μl of SDS-loading buffer and analysed by SDS-PAGE, Western blotting, and ECL, with anti-M2 FLAG antibody (Sigma; #F3165). Further details on this step can be found in Protocol S1 . Computer modelling A Perl script was written to simulate the diffusion-coupled expression of genes A, B, and C, by T7 and SP6 phage polymerases, in a translation extract. The program parameters and script are fully described in Protocols S1–S3 . 18 mm-scale chamber model: Parameters included separate diffusion (and degradation) rates for RNA and protein; a separate apparent diffusion for injected T7, modelled from experimental observations (rapid initial diffusion with exponential decay; Section 5 of Protocol S1 ); estimated binding constants for all interacting species (zinc finger dissociation constants were estimated from previous work on related three- and six-finger constructs [ 58 , 79 , 80 ]); and estimated transcription-translation rates. For adapting the model to the 0.5-mm Drosophila scale, chamber size was scaled down, and only simple diffusion was allowed for all components; for simplicity, transcription-translation rates were not varied (Section 3 of Protocol S1 ) . Supporting Information Protocol S1 Detailed Description of Model (1.2 MB PDF). Click here for additional data file. Protocol S2 Parameter File for Simulations This file contains the default parameters for the computer model in a format that can be read by the Perl script. (5 KB DOC). Click here for additional data file. Protocol S3 Computer Program Script for Simulations This text file is a Perl script to run the computer simulations described in the manuscript. (22 KB DOC). Click here for additional data file. Accession Numbers The Locuslink ( http://www.ncbi.nlm.nih.gov/LocusLink/ , or GeneID ( www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene , accession numbers of the genes and proteins discussed in this paper are Bicoid (40830), caudal (35341), giant (31227), huckebein (40549), hunchback (41032), knirps (40287), Krüppel (38012) , Nanos (42297), tailless (43656), Torso (35717), and zerknüllt (40828).

D:\keerthana\PMC001xxxxxx\PMC1044831.xml

1044832

Recent Origin and Cultural Reversion of a Hunter–Gatherer Group

Contemporary hunter–gatherer groups are often thought to serve as models of an ancient lifestyle that was typical of human populations prior to the development of agriculture. Patterns of genetic variation in hunter–gatherer groups such as the !Kung and African Pygmies are consistent with this view, as they exhibit low genetic diversity coupled with high frequencies of divergent mtDNA types not found in surrounding agricultural groups, suggesting long-term isolation and small population sizes. We report here genetic evidence concerning the origins of the Mlabri, an enigmatic hunter–gatherer group from northern Thailand. The Mlabri have no mtDNA diversity, and the genetic diversity at Y-chromosome and autosomal loci are also extraordinarily reduced in the Mlabri. Genetic, linguistic, and cultural data all suggest that the Mlabri were recently founded, 500–800 y ago, from a very small number of individuals. Moreover, the Mlabri appear to have originated from an agricultural group and then adopted a hunting–gathering subsistence mode. This example of cultural reversion from agriculture to a hunting–gathering lifestyle indicates that contemporary hunter–gatherer groups do not necessarily reflect a pre-agricultural lifestyle.

Introduction The Mlabri are an enigmatic group of about 300 people who nowadays range across the Nan, Phrae, and Phayao provinces of north and northeastern Thailand and the Sayaburi province of western Laos [ 1 , 2 ]. Their traditional lifestyle is to move frequently through the dense forests of the high mountains, building temporary structures of bamboo sticks thatched with banana leaves, which they occupy for a few days, until the leaves turn yellow (thus accounting for their traditional Thai name, Phi Tong Luang, which means “spirit of the yellow leaves”). First contacted by Europeans in 1936 [ 3 ], they are unique among the hill tribes of northern Thailand in that, until recently, they subsisted by hunting and gathering combined with occasional barter trade with villagers. The origins of the Mlabri are controversial. Some investigators have assumed that there is a direct connection between the Mlabri and the ancient Hoabinhian hunting–gathering culture of Southeast Asia [ 1 ]. However, a limited investigation of blood group variation [ 4 ] raised the possibility that the Mlabri originated via a founder event from an agricultural group, and preliminary linguistic analyses support this idea. The Mlabri language seems lexically most closely related to Khmu and Tin, two languages of the Khmuic branch of the Mon-Khmer sub-family of Austro-Asiatic languages, both of which are spoken in agricultural highland villages [ 5 ]. The cluster of dialects jointly referred to as Tin, or Mal/Prai, [ 6 ] is spoken in the Thailand–Laos border region that the Mlabri also occupy, whereas Khmu is spoken over a much wider area [ 7 ]. The grammar of Mlabri additionally has features that deviate markedly from typical Mon-Khmer, suggesting that Mlabri developed as a result of contact between speakers of a Khmuic language and speakers of a quite different language of unknown affiliation [ 2 , 8 ]. We report here the results of an investigation of genetic diversity in the Mlabri, to see whether patterns of genetic variation might provide further insights into the question of an agricultural versus hunting–gathering origin for the Mlabri. The rationale for using genetic analyses to investigate this question is that previous work has shown that hunter–gatherer groups typically differ from their agricultural neighbors in having reduced genetic diversity and high frequencies of unique mtDNA types [ 9 , 10 , 11 , 12 , 13 , 14 ], so we might expect a similar pattern if the Mlabri have always been hunter–gatherers. The genetic results, combined with linguistic and cultural evidence, suggest that the most probable explanation for the origin of the Mlabri is an extreme founder event from an agricultural group, followed by adoption of a hunting–gathering lifestyle. Results/Discussion Genetic Analyses: mtDNA Diversity We analyzed 360 bp of the first hypervariable segment (HV1) of the mtDNA control region in 58 Mlabri; surprisingly, all of the sequences were identical, with the following differences from the reference sequence [ 15 ]: 16140C, 16189C, and 16266A, as well as the common Asian 9-bp deletion in the intergenic region between the cytochrome oxidase subunit II and lysine tRNA genes [ 16 ]. No other human population has been found to lack mtDNA HV1 variation, and mtDNA HV1 variation in six other hill tribes (all agricultural groups) from the same region of Thailand was significantly higher ( Figure 1 ; Table 1 ). Figure 1 Genetic Diversity in the Mlabri and Other Hill Tribes Genetic diversity based on mtDNA HV1 sequences, Y-STR haplotypes, and autosomal STR (A-STR) genotypes in the Mlabri, compared to the average genetic diversity for six other hill tribes. The haplotype diversity is indicated for the mtDNA and Y-STR data, while the average heterozygosity is indicated for the autosomal STR loci. Table 1 Genetic Diversity Parameters Based on mtDNA HV1 Sequences, Y-STR Haplotypes, and Autosomal STR Genotypes for the Mlabri and the Six Other Hill Tribes Diversity in the Mlabri is significantly lower than the average for the other groups for all three genetic systems, based on t -tests (not shown) a Probability of the observed heterozygosity excess under the stepwise mutation model, Wilcoxon one-tailed test Y-Chromosome Diversity We analyzed nine short tandem repeat (STR) loci on the Y chromosome in 54 Mlabri, and again found significantly reduced variation in the Mlabri compared to the other six hill tribes ( Figure 1 ; Table 1 ). The Mlabri had just four Y-chromosome STR (Y-STR) haplotypes, two of which differed by a single repeat at a single locus from one each of the other two haplotypes ( Table 2 ). The Y-STR haplotype diversity in the Mlabri is again lower than that reported for any other human population [ 17 , 18 ]; the Akha, one of the six other hill tribes, also exhibited very low Y-STR diversity ( Table 1 ). The average variance in the allele size distribution at the nine Y-STR loci shows an even greater contrast between the Mlabri and the other hill tribes: the average variance was 0.11 for the Mlabri, versus an average of 1.45 for the other six hill tribes. Table 2 Y-STR Haplotypes in the Mlabri The number of repeats for the allele at each locus in the four haplotypes is given Autosomal DNA Diversity We analyzed nine autosomal STR loci in the Mlabri and the other six hill tribes, and again found significantly reduced variation in the Mlabri ( Figure 1 ; Table 1 ). The genotype frequencies did not deviate significantly from Hardy–Weinberg expectations for any locus in the Mlabri; however, even though these nine STR loci are on different chromosomes and hence unlinked, eight pairs of loci exhibited significant linkage disequilibrium (LD) ( p < 0.05; Figure 2 ), as measured by a likelihood ratio test [ 19 ]. This is significantly more ( p < 0.01) than the 1.8 pairs expected by chance (out of 36 pairwise comparisons) to exhibit this level of LD. For each of the six agricultural hill tribes, the number of pairs of loci exhibiting significant LD was within expectations ( Figure 2 ). Moreover, the p -value of the likelihood ratio test is a measure of the strength of the association between two loci [ 19 ]; the average p -value was 0.20 for the Mlabri, versus 0.31–0.55 for the other six hill tribes, indicating that overall associations between these unlinked loci were stronger in the Mlabri than in the other hill tribes. However, the sample size for the Mlabri for the autosomal STR analyses was larger than the sample size for the other hill tribes ( n = 35 for the Mlabri, versus n = 29–30 for the others), so it is possible that the lower average p -value for the Mlabri reflects more statistical power due to a larger sample size and not more LD. To test this, we sampled 30 Mlabri at random and redid the LD analysis; the conclusions did not change, indicating that the lower average p- value for the Mlabri does reflect more LD in the Mlabri. Figure 2 Associations amongst Unlinked Autosomal STR Loci in the Mlabri and the Other Hill Tribes Probability values of the likelihood ratio test of association versus no association for nine unlinked autosomal STR loci in the Mlabri and six other hill tribes (average probability over the 36 pairs of loci in parentheses). One explanation for the reduced diversity at mtDNA, Y-STR loci, and autosomal STR loci, and the significant number of pairs of unlinked autosomal STR loci in LD, is a severe reduction in population size in the Mlabri. Following such an event, the number of alleles is reduced more than the heterozygosity, leading to an excess of observed heterozygosity compared to that expected for the observed number of alleles under mutation–drift equilibrium [ 20 ]. We therefore compared the observed and expected heterozygosity (at mutation–drift equilibrium, conditioned on the observed number of alleles) for the autosomal STR loci in the Mlabri and the six other hill tribes, under a stepwise mutation model. Only the Mlabri exhibited a significant excess of observed heterozygosity ( Table 1 ). Although more complicated scenarios are possible, the simplest explanation is that the Mlabri (but not the other hill tribes) have undergone a severe reduction in population size, as also indicated by the mtDNA and Y-STR haplotype data, and as also suggested by a previous study of blood group variation [ 4 ]. Population Size Reduction in the Mlabri Assuming that there was a reduction in population size in the Mlabri that set the mtDNA and Y-chromosome diversity near or equal to zero, the coalescence times for the Mlabri mtDNA and Y-STR haplotypes provide an upper estimate as to when the population reduction occurred. We therefore applied Bayesian-based coalescence analysis [ 21 ] to the mtDNA sequences and the Y-STR haplotypes from the Mlabri and the other six hill tribes. For the six agricultural hill tribes, the resulting estimates of coalescence time are broadly distributed ( Figure 3 ), indicating little information in the data (except for the Akha, who do show a pronounced peak in the posterior probability distribution for the Y-STR data, in accordance with their lower Y-STR haplotype diversity). By contrast, the estimates of coalescence time for the Mlabri show a sharp peak ( Figure 3 ), with a median time of 770 y (approximate 95% credible interval 250–4,270 y) for the mtDNA sequences and 490 y (approximate 95% credible interval 170–1,290 y) for the Y-STR haplotypes. Figure 3 Time to the Most Recent Common Ancestor for mtDNA and Y-STR Types for the Mlabri and the Other Hill Tribes Posterior probability distribution of the time back to the most recent common ancestor for the mtDNA (A) and Y-STR haplotype (B) data for the Mlabri and six other hill tribes. Both the mtDNA and the Y-STR data therefore indicate that the Mlabri underwent a substantial reduction in population size about 500–800 y ago (and not more than about 1,300 y ago, if the mtDNA and Y-chromosome data reflect the same event). There are two possible scenarios: (1) a bottleneck, in which the Mlabri were reduced from a formerly large population to a much smaller population size, which then increased to the current level of about 300 individuals; or (2) a founder event, in which the Mlabri began as a very small number of individuals, became isolated, and then increased over time to their present size. Similar reductions in genetic diversity are predicted under either scenario, so the genetic data cannot distinguish between these. But some information can be obtained by considering the magnitude of the reduction in population size needed to completely eliminate mtDNA diversity in the Mlabri. The amount of population size reduction needed to completely eliminate mtDNA diversity in the Mlabri depends on how much mtDNA diversity was present prior to the size reduction. We assumed that the ancestral Mlabri population would have the same mtDNA diversity as one of the other hill tribes and then estimated the amount of population size reduction needed to completely eliminate mtDNA diversity by resampling with replacement various numbers of mtDNA types from the ancestral (pre-bottleneck) population. For example, we started with an ancestral population with the same distribution of mtDNA types as the Akha. We then sampled two mtDNA types (with replacement) from this ancestral population, repeated this procedure 1,000 times, and found that 243 out of the 1,000 resamples of size two had no mtDNA diversity; thus, the probability is 0.243 that a reduction to just two individuals would eliminate mtDNA diversity in an ancestral population that started with the same mtDNA diversity as the Akha. We then repeated this procedure, sampling three mtDNA types (with replacement), and obtained a probability of 0.007 that there would be no mtDNA diversity following a reduction to three individuals. Therefore, if the Mlabri were derived from a population with the same mtDNA diversity as the Akha, the population had to be reduced to not more than two unrelated females, in order to completely eliminate mtDNA diversity. This resampling analysis was carried out six times, with the putative ancestral mtDNA diversity corresponding to each of the six hill tribes. The results of this analysis were that for five of the ancestral populations, resampling three (or more) individuals gave a probability of no mtDNA diversity of less than 0.05; for the remaining ancestral population (which had the same starting mtDNA diversity as the Red Karen), resampling four (or more) individuals gave a probability of no mtDNA diversity of less than 0.05. We also carried out a similar analysis for the Y-STR types in the Mlabri. Here we again assumed an ancestral population with the same Y-STR haplotype diversity as one of the other hill tribes, then determined the maximum number of individuals that could be sampled at random that would have not more than two Y-STR types (since the four Y-STR types in the Mlabri consist of two pairs that differ by a single-step mutation at a single locus). The results of this analysis were that at most 3–6 individuals (depending on which hill tribe the ancestral population resembled most in terms of Y-STR diversity) could have been present after the size reduction, otherwise, with greater than 95% probability, more than two Y-STR types would have been retained. A critical assumption is the amount of genetic diversity present in the ancestral Mlabri population prior to the size reduction. The estimates used in the above analysis are based on agricultural populations, which in general have more mtDNA diversity than hunter–gatherer populations. We therefore also constructed putative ancestral populations with frequency distributions of mtDNA types identical to those found in the !Kung and in African Pygmies [ 22 ]; the results of the resampling analysis were the same. Another assumption of this analysis is that the event that led to the population size reduction completely eliminated the mtDNA diversity. Alternatively, some mtDNA diversity may have been present after the population size reduction, but was subsequently lost because of drift. Loss of mtDNA diversity due to subsequent drift is not likely if there was a single event reducing the Mlabri population size that was followed by population growth, since mtDNA diversity is retained in growing populations [ 23 ]. However, if the reduction in size occurred over several generations, then it may not have been as dramatic a bottleneck as the resampling analysis implies. To investigate this further, we employed a Bayesian approach, following the procedure previously used to estimate the number of founders for the Maoris [ 24 ] but allowing for new mutations, to estimate the number of founders for the Mlabri, assuming various time periods since the founding event. The results ( Figure 4 ) indicate that the most probable number of founders is one over all time periods; however, for longer time periods since the founding event, there is decreasing information on the number of founders from the observation of no mtDNA diversity in the Mlabri. As expected, the longer the time since the founding event (i.e., the slower the population growth rate), the greater the influence of drift in eliminating diversity that might have been present in the founding population. Nevertheless, given that the coalescent analyses indicate an upper date for the origin of the Mlabri of about 1,000 y ago, the lack of mtDNA diversity in the Mlabri is most consistent with a very small founding population size, perhaps even only one female lineage. Figure 4 Number of Founding Individuals in the Mlabri, Given No mtDNA Diversity Posterior probability distribution for the number of founding individuals (k), conditioned on the observation of no diversity in a sample of 58 mtDNA sequences and the time since the founding event. The prior probability is indicated by the dashed black line. Origin of the Mlabri The group that gave rise to the founder event that established the Mlabri could have been either a hunter–gatherer group, in which case the Mlabri maintained their hunting–gathering lifestyle from before, or an agricultural group, in which case the Mlabri subsequently adopted their current hunting–gathering lifestyle. While the genetic data cannot unequivocally distinguish between these two possibilities, they do suggest the latter. Other hunter–gatherer groups typically share few, if any, mtDNA types with neighboring agricultural groups, consistent with long-term isolation of the hunter–gatherer groups. For example, !Kung, African Pygmies, Andamanese Islanders, and south Indian hunter–gatherer groups can readily be distinguished from nearby agricultural groups on the basis of their mtDNA sequences [ 9 , 13 , 14 , 25 ]. By contrast, the Mlabri mtDNA sequence has been reported in other, agricultural hill tribes [ 26 , 27 ], and identical or closely related sequences have also been reported from Southeast Asia and China [ 9 , 28 , 29 ]. Similarly, the Mlabri Y-STR haplotypes are identical or closely related (differing by a single-step mutation at one locus) to Y-STR haplotypes found in Southeast Asia and Oceania [ 30 , 31 ]. Also, the Mlabri do not exhibit any alleles at the nine autosomal STR loci that are not found in the agricultural hill tribes. The widespread sharing of mtDNA, Y-STR, and autosomal STR alleles between the Mlabri and agricultural groups in Southeast Asia are not expected if the Mlabri have always been hunter–gatherers. Instead, the genetic data suggest that the Mlabri are derived from an agricultural group. Moreover, the Mlabri vocabulary and folklore also give some evidence of ancient familiarity with agriculture coexisting with hunting and gathering (J. Rischel, personal communication). While preliminary in nature, the available linguistic evidence suggests that the present-day Mlabri language arose after some speakers of a Khmuic language, most likely Tin, became isolated and subsequently experienced intensive contact with speakers of some other, presently unknown language [ 2 , 8 ]. Just how long ago the Mlabri and Tin languages diverged cannot be determined, but it has been suggested that Tin branched from Khmu about 600 y ago, and that Tin then branched into two varieties (Mal and Prai) some 200–300 y ago [ 6 , 32 ]. These time estimates are based on a calibration of the chronology of sound changes in Tin against reasonably secure datings of sound changes in neighboring languages; the actual time depth may be underestimated, but most likely by not more than a few centuries. Thus, the linguistic evidence would date the origin of the Mlabri at less than 1,000 y ago, in excellent agreement with the genetic evidence. Other data that may shed light on the origins of the Mlabri, such as historical information, are scarce, since the Mlabri do not have a written language and the first recorded contact was only in 1936. However, the Tin Prai have an oral tradition concerning the origin of the Mlabri (J. Rischel, personal communication), in which several hundred years ago, Tin Prai villagers expelled two children and sent them downriver on a raft. They survived and escaped into the forest, turning to a foraging lifestyle and thus becoming the Mlabri. Although it is difficult to know how to evaluate such oral traditions, this story nevertheless intriguingly parallels the genetic and linguistic evidence concerning the origins of the Mlabri. In sum, genetic, linguistic, and cultural data all suggest a founding event in the Mlabri, involving a single maternal lineage and 1–4 paternal lineages some 500–1,000 y ago, from an ancestral agricultural population. The Mlabri then subsequently adopted their present hunting and gathering lifestyle, possibly because the group size at the time of founding was too small to support an agricultural lifestyle. Other examples of such cultural reversion are rare; probably the best known involves Polynesian hunter–gatherers on the Chatham Islands and the South Island of New Zealand [ 33 ], who abandoned agriculture and adopted a maritime-based foraging subsistence because of the rich marine resources and the inability of these islands to support cultivation of tropical crops. Other hypothesized examples of cultural reversion, such as the Punan of Borneo [ 34 ], the Guajá and other lowland Amazonian groups [ 35 ], and the Sirionó of Bolivia [ 36 ], are controversial, as it is not clear whether these groups are descended directly from earlier hunter–gatherer groups or whether they indeed have undergone cultural reversion. Detailed genetic analyses, as carried out here for the Mlabri, may shed further light in these cases. In any event, our conclusion that the Mlabri, a present-day group of hunters and gatherers, was founded recently and in all probability from an agricultural group further supports the contention that contemporary hunter–gatherer groups cannot be automatically assumed to represent the pre-agricultural lifestyle of human populations, descended unchanged from the Stone Age [ 37 ]. Indeed, even if they have not reverted from an agricultural lifestyle, most (if not all) contemporary hunter–gatherer groups interact with, and have evolved and changed along with, agricultural groups [ 38 ]. The Mlabri provide a unique opportunity to investigate the circumstances and consequences of a reversion from an agricultural to a hunting–gathering lifestyle that apparently was not dictated by purely ecological reasons (as in the case of Polynesian hunter–gatherers). Materials and Methods Samples There are three linguistically distinct subgroups of Mlabri [ 39 ], designated A, B, and C. Subgroup A (also known erroneously as “Mrabri”) is the only group that has been studied in detail [ 1 ]; subgroup B (minor Mlabri) is practically extinct [ 2 ], and subgroup C (formerly “Yumbri”) comprises less than 30 people [ 39 ]. Blood samples and genealogies of 91 Mlabri (all from subgroup A) were obtained with informed consent in 1999, and cell lines were prepared and DNA was extracted from the cell lines. The genealogical data were used to identify and exclude known relatives from the genetic analyses. Data on mtDNA and Y-STR variation from six agricultural hill tribes in the same geographic region (Akha, Lahu, White Karen, Red Karen, CR Lisu [from near Chiang Rai], and MHS Lisu [from near Mae Hong Son]), all of whom speak Sino-Tibetan languages, were published previously [ 26 ]. Genetic analyses The first hypervariable segment (HV1) of the mtDNA control region (nucleotide positions 16,024–16,385) was amplified and sequenced directly, as described previously [ 29 ], from 58 Mlabri. PCR analysis of the intergenic region between the cytochrome oxidase subunit II and lysine tRNA genes, which harbors an informative 9-bp deletion, was carried out as described previously [ 28 ]. Nine Y-STR loci (DYS385a, DYS385b, DYS389I, DYS389II, DYS390, DYS391, DYS392, DYS393, and DYS394) were amplified and genotypes determined, using previously described methods [ 30 ], for 54 Mlabri. Nine autosomal STR loci (D3S1358, vWA, FGA, D8S1179, D21S11, D18S51, D5S818, D13S317, and D7S820) plus the amelogenin locus were amplified with the AmpFLSTR Profiler Plus PCR Amplification Kit (Applied Biosystems, Foster City, California, United States), using 2–4 ng of DNA in a 15-μl reaction volume. Genotypes were determined by fragment analysis on an ABI377 (Applied Biosystems) for 35 Mlabri, 29 Lahu, and 30 individuals from each of the other hill tribes. Statistical analyses Genetic diversity, heterozygosity, and tests for goodness of fit to Hardy–Weinberg expectations were calculated with Arlequin 2.000 [ 40 ]. LD was estimated as the probability of the likelihood of the data assuming linkage equilibrium versus the likelihood of the data assuming association [ 19 ]; Arlequin 2.000 was used to obtain maximum-likelihood estimates of the haplotype frequencies for each pair of loci with the EM algorithm [ 41 ], and the null distribution of the p -value of the likelihood ratio test was generated by 10,000 random permutations. The program Bottleneck ( http://www.ensam.inra.fr/URLB/bottleneck/bottleneck.html ) was used to compare the observed heterozygosity at each autosomal STR locus to that expected at mutation–drift equilibrium for the observed number of alleles, assuming a stepwise mutation model. Bayesian-based coalescence analyses of Y-STR haplotypes [ 42 ] were performed using the software Batwing ( http://www.maths.abdn.ac.uk/~ijw/downloads/batwing/batguide/node6.html ) and previously described prior distributions for the initial effective population size, population growth rate, and Y-STR mutation rates [ 30 ]. The coalescence time for mtDNA HV1 sequences was also estimated by a Bayesian procedure [ 21 ] as described previously for Xq13.3 sequences [ 43 ], using the same initial effective population size and population growth rate priors as for the Y-STR analysis, and a γ-distribution with parameters α = 14.74 and β = 0.0005 (corresponding mean = 0.00737) as a prior for the mutation rate [ 44 ]. Resampling of mtDNA and Y-STR types, in order to estimate the magnitude of population size reduction needed to eliminate mtDNA and reduce Y-STR diversity, was performed with the software Resample ( http://www.pbs.port.ac.uk/~woodm/resample.htm ). Bayesian analysis of the number of founders for the Mlabri was performed by pooling the mtDNA types in the other hill tribes to obtain a starting population, from which a certain number of founding mtDNA types were selected at random, assuming a uniform prior distribution between one and 20 founders. The sample was then allowed to grow from the number of founders to size 300 (the current size of the Mlabri population) over various time intervals, such that the shorter the time interval, the faster the growth rate. Simulations were performed both under the assumption of no new mutations, and with a mutation rate of one mutation/sequence/10,000 y. For each combination of parameters, 1,000,000 simulations were carried out. The simulation results were converted via Bayes's theorem into a posterior probability for the number of founding individuals, conditioned on the observation of no diversity in a random sample of size 58 (the sample size in this study). In practice, the posterior probability distributions were independent of the mutation rate (analyses not shown).

D:\keerthana\PMC001xxxxxx\PMC1044832.xml

1044833

Modeling the Mutualistic Interactions between Tubeworms and Microbial Consortia

The deep-sea vestimentiferan tubeworm Lamellibrachia luymesi forms large aggregations at hydrocarbon seeps in the Gulf of Mexico that may persist for over 250 y. Here, we present the results of a diagenetic model in which tubeworm aggregation persistence is achieved through augmentation of the supply of sulfate to hydrocarbon seep sediments. In the model, L. luymesi releases the sulfate generated by its internal, chemoautotrophic, sulfide-oxidizing symbionts through posterior root-like extensions of its body. The sulfate fuels sulfate reduction, commonly coupled to anaerobic methane oxidation and hydrocarbon degradation by bacterial–archaeal consortia. If sulfate is released by the tubeworms, sulfide generation mainly by hydrocarbon degradation is sufficient to support moderate-sized aggregations of L. luymesi for hundreds of years. The results of this model expand our concept of the potential benefits derived from complex interspecific relationships, in this case involving members of all three domains of life.

Introduction Complex positive species interactions have been shown to expand the ecological niche and increase the persistence of the organisms involved in a variety of systems. In terrestrial systems, increased diversity of mycorrhizal symbionts is correlated with increased biodiversity of plant communities, resulting in greater stability and longer persistence at the community level [ 1 ]. In marine ecosystems, the coral Oculina arbuscula harbors a majid crab, Mithrax forceps, that prevents overgrowth of macroalgae and shading of the corals [ 2 ]. This allows O. arbuscula to maintain its facultative mutualism with photosynthetic zooxanthellae in well-lit habitats off the Atlantic coast of North Carolina, increasing the amount of energy available to the coral for growth and reproduction. At cold seeps in the Cascadia [ 3 , 4 ] and Aleutian [ 5 ] subduction zones, bioirrigation through burrow formation and bioturbation by clams ( Calyptogena spp.) has been shown to significantly affect the distribution of microbial anaerobic methane oxidation. Lamellibrachia luymesi inhabits areas associated with advection of hydrocarbons and other reduced chemicals to the seafloor (hydrocarbon or brine seeps) on the upper Louisiana slope (ULS) of the Gulf of Mexico from 400 to 1,000 m depth. L. luymesi does not posses a digestive system; rather, it acquires energy via internal sulfide-oxidizing bacterial symbionts [ 6 ]. L. luymesi differs from other vestimentiferan tubeworms by its ability to use a posterior extension of its body, the “root,” to acquire sulfide from interstitial pools in sediments [ 7 , 8 ]. Near the anterior plumes of tubeworms, sulfide concentrations typically decline below 0.1 μM as the tubeworms approach 1 m in length [ 9 ]. By using its roots, L. luymesi is able to delve into deeper sediment layers, providing access to more persistent sulfide sources. In the apparent absence of lethal predation [ 10 , 11 ], the most significant hazard that this vestimentiferan tubeworm faces is sulfide limitation. Its high uptake rate of sulfide from hydrocarbon seep sediments, estimated at over 30 μmol · h −1 for a moderate-sized individual [ 12 ], suggests that sulfide flux may be limiting in L. luymesi 's habitat. A diverse chemosynthetic community relies on the sulfide generated as a by-product of anaerobic degradative processes in the Gulf of Mexico [ 10 , 11 ]. Reduction of seawater sulfate utilizing methane or other hydrocarbons as electron donors produces the majority of sulfide available at ULS seeps [ 13 , 14 ]. Anaerobic methane oxidation is most commonly carried out by microbial consortia consisting of sulfate-reducing bacteria along with methanogenic archaea executing reverse methanogenesis [ 15 , 16 ]. Methane oxidation linked to sulfate reduction and subsequent authigenic carbonate precipitation constrain ocean–atmosphere carbon fluxes [ 3 , 4 ], accounting for up to 20% of the global methane flux to the atmosphere [ 17 ]. Oxidation of other hydrocarbons and organic material, carried out by sulfate-reducing bacteria in monoculture and in consortia with other microbes [ 18 ], may account for a larger proportion of sulfate depletion in ULS sediments [ 14 ]. These processes can result in a decoupling of sulfate reduction and methane oxidation rates [ 14 ], and form carbonates consisting mainly of non-methane-derived carbon [ 19 ]. L. luymesi may influence these anaerobic processes by utilizing its roots to release the sulfate generated by its symbionts during sulfide oxidation [ 7 , 8 , 12 ]. This hypothetical mechanism would provide sulfate for anaerobic methane oxidation and hydrocarbon degradation at sediment depths normally devoid of energetically favorable oxidants, thereby augmenting exogenous sulfide production. In this study, we address the question of whether known biogeochemical processes could supply sulfide at rates sufficient to match the requirements of long-lived L. luymesi aggregations. In the diagenetic model presented here, the hypothesized release of sulfate in sediments with sufficient electron donors results in sulfide generation at rates matching the sulfide uptake rate of L. luymesi aggregations for over 250 y. We speculate that the mutual benefits derived from the syntrophy among symbiotic tubeworms and microbial consortia implicit in the model would expand our current concept for the potential complexity of positive interspecific interactions and the benefits they confer. Results/Discussion L. luymesi Sulfate Release Allows Persistence of Aggregations The model predicts that inputs from known sources, including diffusion and advection of deep sulfide along with reduced seawater sulfate, will support a moderately-sized aggregation of 1,000 individuals for an average of 39 y (range, 22 to 78 y) ( Figure 1 ). A smaller aggregation of 200 individuals could be maintained with these sources for an average of 64.1 y (standard deviation, 10.6 y). In this model configuration, the duration of adequate sulfide flux is not congruent with the known sizes of aggregations and existing age estimates of L. luymesi individuals and aggregations. Adding sulfate release by tubeworm roots to the model results in sulfide generation and flux at rates that match the demands of large aggregations, allowing the tubeworms to survive for over 250 y ( Figure 1 ). This additional source of sulfate results in a two-orders-of-magnitude increase in sulfate flux in older (>100 y) aggregations, accounting for over 90% of sulfate available after only 24 y. The sulfate released by the tubeworms would be used for anaerobic methane oxidation and hydrocarbon degradation. The nature of the relationship between symbiotic tubeworms and microbial consortia that we are proposing is a coupling of the sulfur cycle only, and not carbon. Light dissolved inorganic carbon (DIC) resulting from the oxidation of hydrocarbons is apparently not taken up by tubeworms as the carbon stable isotope signatures of L. luymesi are heavier than would be expected from a methane-derived DIC source [ 20 , 21 ]. In addition, the well-studied hydrothermal vent tubeworm, Riftia pachyptila, obtains carbon in the form of CO 2 across its plume [ 22 ]. However, this does not necessarily exclude the passive diffusion of DIC across the root surface, which could account for some of the variability observed in L. luymesi carbon stable isotope signatures [ 20 , 21 ]. By augmenting the sulfate supply to microbial consortia for sulfate reduction, large aggregations of tubeworms may survive for hundreds of years in the model, mirroring the population sizes and individual lengths regularly observed and collected at seeps on the ULS [ 23 ]. Figure 1 Ratio of Sulfide Supply to Sulfide Uptake Rate of L. luymesi Aggregations Equilibrium line (1:1 ratio) and average, maximum, and minimum values for 1,000 iterations presented. Supply rate based on known sources without sulfate release by tubeworm roots shown in blue. Sulfide supply declines below demand after approximately 40 y. Supply rate including sulfate release from tubeworm roots shown in red, with sulfate release constrained by tubeworm symbionts' sulfide oxidation rate. Sulfide supply exceeds demand for the duration of the model. Model Results Are Robust to Parameter Variation An alternate hypothesis to explain the discordance between estimated sulfide supply and uptake rates is the presence of locally elevated seepage rates. Sensitivity analyses were carried out to determine the potential effects of uncertainty in seepage rate on supply estimated for aggregations without root sulfate release. A 10% increase in seepage rate resulted in a 5.6% increase in sulfide supply to aggregations 200 y old and older. This corresponds to only 16.4% of the sulfide required, which does not serve to extend aggregation survivorship (average, 39 y; range, 21 to 79 y) beyond that determined for lower flow rates. To supply the sulfide flux required by older aggregations, seepage rate would have to be at least 363 mm · y −1 . This is over ten times greater than the rate used in the model (32 mm · y −1 ), which is the highest region-wide estimate for the Gulf of Mexico [ 24 ]. A rate of over 300 mm · y −1 approaches rates reported for active venting of fluids ( Table 1 ). Active venting would result in the visual manifestation of seepage in the form of methane bubbles and oil droplets, which are generally restricted to mussel ( Bathymodiolus childressi ) beds at these sites [ 25 ]. In addition, larger, and therefore older [ 26 ], aggregations have lower epibenthic sulfide concentrations [ 8 , 9 , 25 ] suggesting that seepage becomes less vigorous over time and is not in the form of active venting in larger tubeworm aggregations. While difficult to obtain, in situ measures of advection rate of fluids at Gulf of Mexico seeps could be used to test these assumptions and may lend insight into the relationship between variability in tubeworm growth rate and sulfide availability. Table 1 Reported Seepage Rates for Hydrocarbon and Methane Seeps The high degree of variability in growth rate and recruitment rate could also affect the ratio of supply and demand in the model. In an aggregation exhibiting anomalously low recruitment, the size of the rhizosphere would increase more rapidly than the biomass of the aggregation. This would lead to high rates of sulfide delivery and generation and low rates of sulfide uptake by tubeworm roots. When initial recruitment rate ( a in equations 1 and 2 ) is decreased by 10%, the length of time that supply exceeds demand increases by 3.7%. This effect appears to be linear, with a 20% decrease in initial recruitment rate resulting in a 7.4% increase in persistence. If growth rate is increased, thereby increasing the rate of rhizosphere growth in terms of surface area for diffusion and advection, there appears to be little effect of the ratio of supply to demand (20% increase in growth—0% change in persistence time). In fact, increasing growth to the upper limits of the error term ( equation 5 ) lowers the amount of time that the aggregation can be supported since biomass and sulfide demand increase more rapidly than increases in supply resulting from additional surface area. By decreasing growth rate, aggregations may be supported for longer periods of time, with a 20% decrease leading to a 6.3% increase in persistence time and a decrease of 88% leading to persistence for over 250 y. While an 88% lower growth rate lies outside of the range of existing growth data, this could be accomplished by ceasing growth for extended periods of time in a quiescent stage. This possibility remains to be investigated in L. luymesi . By utilizing a variable recruitment rate in the model, both between realized aggregations and between years within a model run, along with a growth error term encompassing the full range of observed growth data, the model is capable of generating aggregations within the range of the 10%–20% variability tested in this analysis. Even these outlying aggregations (presented as maxima and minima in Figure 1 ) support the qualitative conclusions drawn from model results. While the model was based on empirical data to the greatest degree possible, estimates of many of the parameters necessary to resolve the model were not available or are extremely difficult to measure in deep water with existing technology. Uptake rates were measured in the laboratory [ 8 ] for relatively small individuals (<50 cm). While we attempted to approximate metabolic scaling by covarying uptake and growth rates, it is possible that large individuals require even lower sulfide flux. Model predictions are not overly sensitive to variability in this parameter. A reduction by 10% of the overall sulfide uptake rate results in a 5.2% increase in persistence time. To maintain an aggregation for over 250 y, mass-specific uptake rate would have to be reduced 6-fold. While this could also be accomplished by entering a period of quiescence as mentioned before, there is no existing evidence for this ability in vestimentiferans. The second version of the model is based on the assumption that L. luymesi is capable of releasing sulfate through its roots. It should be noted that in the model, sulfate release is constrained by the rate of sulfate generation by the tubeworm's sulfide-oxidizing symbionts, resulting in the near 1:1 ratio of supply and demand in Figure 1 . Though modeled sulfate flux across the roots into the rhizosphere may exceed 20 mmol · h −1 in older aggregations, the roots provide an ample respiratory surface such that rates of sulfate flux per unit root surface area do not exceed 0.4 μmol · h −1 · cm −2 in the model. It remains possible that a proportion of the sulfate could be released through the plume of the tubeworms, though the energy required to pump sulfate against a concentration gradient (seawater [SO 4 ] = 29 mM) [ 13 ] suggests that it would be more energetically favorable for the sulfate to passively diffuse out of the roots. It is also possible that sulfate flux could be increased by active bioirrigation delivering seawater to deeper sediment layers through the tubeworm tubes. This could allow the sulfide-oxidizing symbionts to store some of the oxidized sulfide as elemental sulfur rather than releasing it as sulfate, while maintaining sufficient sulfate flux to deeper sediment layers for sulfide generation. These mechanisms remain hypothetical and require further experimental investigations to evaluate their potential role in this system. Tubeworms Impact Seep Biogeochemistry Tubeworm sulfate release, in conjunction with high sulfide uptake rates, could contribute to the observation of declining advection rate in older aggregations. By increasing sulfate flux to deeper sediments, L. luymesi increases integrated rates of anaerobic methane oxidation and hydrocarbon degradation, which would enhance authigenic calcium carbonate precipitation within the rhizosphere. Under the conditions of root sulfate release in the model, calcium carbonate precipitation is rapid (0.109 to 0.316 μmol · l −1 · sec −1 ) in the first 53 y, with rates declining exponentially thereafter. By creating a barrier to fluid advection [ 4 ], this could result in the observed decrease in epibenthic sulfide concentration in older aggregations [ 8 , 9 ] and the predicted cessation of tubeworm recruitment around this time [ 12 , 23 ]. In order to prevent the precipitation of carbonate directly on the root surface, L. luymesi individuals may release hydrogen ions as well as sulfate through their roots. While hydrogen ion flux through the roots has not yet been empirically demonstrated, none of the nearly 5,000 tubeworms examined as part of this study were observed to have carbonate formed directly on their roots, suggesting that this form of precipitation is inhibited in some manner. In the model, diffusion of hydrogen ions across the root surface (the only form of release explicitly modeled) accounts for less than 40% of ion generation when carbonate precipitation is most vigorous. We speculate that L. luymesi may utilize the excess hydrogen ions generated by their sulfide-oxidizing symbionts to periodically raise the rate of hydrogen ion flux from their roots. This would not only supply additional hydrogen ions to sulfate-reducing bacteria, but could inhibit carbonate precipitation on the tubes and subsequent reduction of the root area utilizable as a respiratory surface. Further pH reduction could dissolve existing carbonate in sediments beneath the rhizosphere, thereby opening seepage pathways and allowing further root growth. This possibility is corroborated by the observation of young tubeworms that had apparently bored through bivalve shells in an experimental system (R. Carney, personal communication). Empirical measurements of hydrogen ion flux across the root tissue of L. luymesi are required to test these hypothetical mechanisms. The release of sulfate by tubeworm roots potentially explains the frequent observation of highly degraded hydrocarbons in the vicinity of large tubeworm aggregations [ 27 ]. The majority of sulfate supplied by tubeworm roots is utilized for microbial hydrocarbon degradation in the model ( Figure2 ). This process alone accounts for over 60% of the sulfide available to aggregations after approximately 80 y. In the absence of liquid and solid phase hydrocarbons, methane flux would have to be approximately four times the rate in the model in order to fuel sufficient sulfate reduction to support an aggregation for over 200 y. This could occur in sediments overlying rapidly sublimating gas hydrates, and hydrate abundance has been previously suggested as a potential factor influencing the distribution of chemosynthetic communities in the Gulf of Mexico [ 10 ]. However, model results indicate that large chain hydrocarbons are the most significant energy source for sulfate reduction in tubeworm-dominated sediments. Increased integrated rates of hydrocarbon degradation would lead to highly biologically altered hydrocarbon pools among the roots of tubeworm aggregations. Hydrocarbon oxidation has been implicated as one of the dominant processes in the carbon cycle at ULS seeps, accounting for over 90% of the carbon in carbonates collected in the vicinity of tubeworm aggregations [ 19 ]. Model analysis indicates that the minimum amount of organic carbon (including hydrocarbons as well as buried organic material) in sediments required to supply sulfide at rates matching aggregation demand (1:1 supply:uptake ratio) is 1.03% by weight, remarkably close to the lowest value found in any of the seep sediment core samples (1.2%) [ 13 , 28 ], and greater than that found in ULS sediments away from seeps (0.71%) [ 29 ]. Determination of organic carbon concentration in sediments beneath tubeworm aggregations is necessary to test the prediction that elevated carbon content at seeps, primarily resulting from oil seepage, provides the energy source required to generate sufficient sulfide for tubeworm aggregations. Figure 2 Sources of Sulfide Available to Tubeworm Aggregations over Time in the Model Sources of sulfide include advection and diffusion of sulfide from deep sources (yellow) or sulfate reduction using methane (blue), buried organic carbon (green), or C 6+ hydrocarbons (dark grey) as electron donors. Sulfate is provided by diffusion from sediments surrounding the rhizosphere, diffusion at the sediment–water interface, and release from tubeworm roots. Additional sulfate flux from tubeworm roots could also explain the high apparent sulfate diffusion coefficients determined for tubeworm-impacted sediments [ 13 ]. Anomalous sulfate fluxes have been proposed to be a result of bioturbation and bioirrigation by macrofauna [ 3 , 5 ], and recycling by microbial mats [ 13 ]. The results of the model presented here provide evidence for macrofaunal sulfur recycling, an additional component to be considered in future investigations of cold seep biogeochemistry. The hypothesized release of sulfate by tubeworm roots potentially explains numerous, apparently disparate observations, hinting at the great impact that L. luymesi aggregations may have on their abiotic environment. While the proposed interactions between symbiotic tubeworms and sulfate-reducing bacteria are essential for the persistence of L. luymesi aggregations in the model, we suggest that there are significant effects on the microbial community as well. This syntrophy will increase the abundance of sulfate-reducing bacteria and therefore increase the rates of anaerobic methane oxidation and hydrocarbon degradation carried out by microbial consortia that rely on sulfate as an oxidant. Tubeworm-generated sulfate supplies a more energetically favorable electron acceptor below the normal depth of sulfate penetration at seeps, relaxing the limitation on anaerobic oxidative processes at these sediment depths. Deeper sediment layers then become habitable to sulfate reducers, significantly altering the microbial community structure within the rhizosphere. Model configurations neglect the potential role of bioirrigation of seawater sulfate through L. luymesi tubes, which could further increase sulfate supply to deeper sediment layers. The possible role of tubeworm roots as substrata for the growth of microbial consortia, analogous to the habitat afforded mycorrhizal symbionts of higher plants, remains another possible benefit for the microbes. These predictions may be tested by determination of the relative abundance of microbial consortia at different depths of sediments both impacted by and isolated from tubeworms. Localization of the microbes on the root surface would provide evidence for a more intricate relationship. It is our hope that the results of this model may provide the impetus for future rigorous experimental tests of these ideas. Summary The model results presented here are consistent with the hypothesis that L. luymesi releases sulfate into hydrocarbon-rich sediments to fuel sulfide generation, allowing for the persistence of the longest-lived animal known. The importance of this process to sulfide generation in the modeled rhizosphere implies a complex relationship between an animal with bacterial endosymbionts and external sulfate-reducing bacteria, often in consortia with methane-oxidizing or hydrocarbon-degrading microbes. This positive interspecific relationship, including members of all three domains, would benefit both the tubeworms and the microbial consortia involved. This expands our existing concept of the potential for complexity in mutualisms and the benefits they may confer. Further complex relationships are likely to be discovered through continued research into the role of positive species interactions at the individual and community levels. Materials and Methods This study couples an individual-based population growth and sulfide uptake model [ 12 ] to a diagenetic diffusion/advection model to compare the relative magnitude of sulfide supply and uptake for long-lived tubeworm aggregations. A series of 1,000 iterations of the model under three different initial conditions (known sources of sulfate, known sources plus root sulfate supply, and known sources with elevated seepage rates) were carried out. The rhizosphere (volume of sediment encompassed by the root system of an aggregation) is modeled as an inverted dome beneath the sediment with a radius equal to the average root length of the population ( Figure 3 ). The rhizosphere was approximated by a series of two-dimensional discs at 2-cm intervals in order to reduce the complexity of a three-dimensional solution for a sphere of changing size. Sulfate (SO 4 2− ), methane (CH 4 ), sulfide (HS − ), bicarbonate (HCO 3 − ), and hydrogen ion (H + ) fluxes across the rhizosphere boundary are determined. Sulfate reduction rates using methane, larger chain hydrocarbons, and buried organic matter as electron donors are modeled in order to estimate the sulfide available to tubeworm aggregations as they change in size over the course of 250 y. Figure 3 Model Construction Population model includes individual size-specific growth and mortality rates, and population size-specific recruitment rate. Growth rate was determined by in situ staining of tubeworm aggregations using a blue chitin stain (in situ photograph of stained aggregation demonstrating annual growth shown here) and collection after 12–14 mo. Diagenetic model included advection and diffusion of sulfate, sulfide, methane, bicarbonate, and hydrogen ions as well as organic carbon content of sediments. Fluxes across the rhizosphere (root system) boundary were compared to sulfide uptake rates for simulated aggregations to determine whether sulfide supply could match the required uptake rates of aggregations (for specific methodology see methods). HC, C 6+ hydrocarbons; orgC, organic carbon; ox, oxidation reaction; red, reduction reaction. Population growth model The population growth model follows the methodology presented in [ 12 ] and includes population growth, mortality rate, individual growth rate, and sulfide uptake rate. The parameters underlying the population growth model were refined using growth data from an additional 615 individuals and population data from an additional 11 aggregations comprising 4,908 individuals. The model presented here includes data from a total of 23 tubeworm aggregations from three nearby sites (Green Canyon oil lease blocks 184, 232, and 234) collected over a period of 7 y on the ULS to arrive at generalized population growth parameters. L. luymesi individuals are dioecious, with males releasing sperm into the water column. Fertilization is believed to be external [ 30 ], though sperm has been found within the oviducts of females of the hydrothermal vent tubeworm R. pachyptila [ 31 ]. Eggs and embryos are positively buoyant and develop into a swimming trochophore-like larval stage within 3 d of fertilization [ 32 ]. Larvae are lecithotrophic and may remain in the water column for several weeks [ 32 ]. They require hard substrata for settlement, and acquire symbionts from their environment after metamorphosis [ 33 , 34 ]. Settlement is initially rapid, and continues until the available substrate is occupied [ 12 , 23 , 35 ]. Population sizes of aggregations collected with existing sampling devices typically vary between 100 and 1,500 individuals ([ 12 , 23 ]; this study), though far larger aggregations covering tens to hundreds of square meters are common at the sites sampled. Previous studies have shown that L. luymesi has an average longevity of 135 y [ 12 ], and requires an average of 210 y to reach 2 m in length [ 26 ], a size not uncommon among collected animals. Mortality events are exceedingly rare, dropping below 1% annual mortality probability for animals over 30 cm [ 12 ]. The expanded datasets of growth and mortality rates included here extend the longevity estimate for L. luymesi to an average of 176 y and the estimated age of a 2-m-long animal to 216 y. At the beginning of each iteration, population growth parameters are chosen for the following population growth model: where N is population size, t is time (in years), K is carrying capacity (set to 1,000 individuals for all simulations presented here), a describes the initial slope of the line, b defines the degree of density dependence, and c is a shape parameter. The first parameter (a) was generated using the following function: where ɛ [N(0,1)] is a normally distributed random deviate with an average of zero and a standard deviation of one. This allows the initial recruitment rate to vary within the range of all recruitment trajectories that have been observed [ 12 ]. The other parameters were not normally distributed; therefore, the log-transformed distributions were used to define the distribution of the random numbers generated. As the three parameters in the model were significantly correlated (ln( a ) and ln( b ), r = −0.853, p < 0.001; ln( a ) and ln ( c ), r = −0.461, p = 0.036), values of b and c were chosen from their relationship with a : The value of a was allowed to vary each year according to the pooled standard error associated with the estimates of a from the empirical data (standard error, 0.105). Once population size equaled or exceeded carrying capacity, recruitment was ceased, representing the lack of additional substrate or sulfide available in the water column. Once recruitment was determined for that year, the individual-based portion of the model began. Each individual was traced through time with respect to its length, root length, mass, mortality probability, mass-specific sulfide uptake rate, sulfate excretion rate, and hydrogen ion elimination rate. Growth rates of tubeworms were determined by staining tubes in situ ( Figure 3 ) and collection 12 to 14 mo later ([ 26 ]; this study). Individual growth rate was determined from the following function ( Figure 4 ): Figure 4 L. luymesi Growth Rate Size-specific growth of L. luymesi determined from stained tubeworms. Different colors indicate growth data from different aggregations. Blue points labeled “2000” are all from Bergquist et al. [ 26 ]. Other colored points refer to submersible dive numbers from 2003 when stained aggregations were collected. (A) Growth function and 95% confidence interval for size-specific growth. (B) Error function fitted to the residuals of the model. Length (l) is defined here as the distance from the anterior end of the tube to an outer tube diameter of 2 mm following the methodology of [ 26 ]. All growth rates were standardized to 365 d. The error term is an additional function fitted to the residuals of the first regression function ( Figure 4 B), resulting in a variable growth rate. This error term was used rather than varying growth within the 95% confidence interval of the regression of length and growth rate because of the high degree of variability in growth among individuals. It should also be noted that there is a certain degree of variability in growth rate between aggregations ( Figure 4 ). This may be attributable to spatial or temporal variability in seepage rate or sulfide concentration between aggregations. Aggregations may be subject to persistently differing conditions on a small (meter) scale, or may encounter periodic fluctuations in habitat characteristics. Because we are uncertain whether this variation is persistent on the temporal scales that we are simulating, between-aggregation variability is not explicitly modeled, though by chance certain realized aggregations deviated from mean growth rate. The ratio of root length to tube length was determined from individual length using the following function: Annual mortality rate was approximated as the size-specific frequency of empty tubes in collected aggregations [ 12 ] with an overall annual mortality rate of 0.569%. This approximation is conservative and likely overestimates yearly mortality, as available data indicate that empty tubes should persist longer than 1 y [ 12 , 36 ]. Mortality probability was determined for each 10-cm size class using the following function: where m is mortality probability and l is length. Individuals were considered dead if their probability of mortality exceeded a uniform random number between zero and one. By using generalized population growth parameters in the model presented here, we attempt to encompass the range of empirical data from sampled aggregations in our examination of sulfide uptake and supply rates. Taken together, the population growth model including recruitment, growth, and mortality provides a good qualitative if not quantitative fit for any individual aggregation, reflecting the size frequency of tubeworms within sampled aggregations [ 12 ]. It should be noted that the modeling of specific aggregations was not the aim of this study; rather, an attempt has been made to encompass the variability observed in the various populations of tubeworms that have been sampled. To examine the effect of uncertainty in the population growth parameters, sensitivity analyses were carried out. The initial slope of the recruitment rate ( a in equation 1 ) was varied while individual size-specific growth rate was held constant (no error term in equation 5 ). Growth rate was then varied while holding the initial rate of population growth constant (no error term in equation 2 ). The effect of a 10% change in each parameter was determined, and then changes of greater magnitude were examined to determine the fastest rate of population or individual growth that could be supported by the sulfide available to the aggregation in the absence of sulfate release. Individual sulfide uptake was allowed to vary within the range of laboratory-determined sulfide uptake rates according to that individual's growth rate for that year: where u is uptake rate (in micromoles per gram per hour), m is mass (in grams), and g is growth rate (in centimeters per year). Growth rate was divided by the maximum growth rate (10 cm · y −1 ) such that highest growth rates resulted in highest uptake rates. By scaling uptake rate with growth, we approximate metabolic scaling, resulting in a decline in uptake rate by a factor of 3.7 over the range of tubeworm sizes in this study [ 12 ]. The amount of sulfate that could be excreted by each individual was determined from the amount generated by sulfide oxidation carried out by the internal chemoautotrophic symbionts assuming constant internal sulfate concentration, thereby accounting for changes in body volume. We do not account for the binding of sulfur by free amino acids, as this is believed to relatively minor compared the flux rates of sulfate and sulfide, and is reversible [ 37 ]. Hydrogen ions are also generated in the oxidation of sulfide by the tubeworm symbionts. Hydrogen ion elimination rate was determined in the model in the same fashion as sulfide uptake, with growth rate determining the variability in this metabolic flux according to laboratory-measured ion fluxes (mean, 10.96 μmol · g −1 · h −1 ; standard deviation, 1.88 μmol · g −1 · h −1 ) [ 38 ]. Simple diffusion of hydrogen ions across the root surface was included in the model, though the exact mode of proton flux has not yet been determined experimentally for L. luymesi [ 38 ]. As diffusion across the roots accounts for a relatively small proportion of total proton flux (less than 10% in large individuals), additional pathways are likely and require further investigation. Geochemical setting Known sources of sulfide available to L. luymesi aggregations are sulfide transported with seeping fluids [ 10 ] and sulfide generated via reduction of seawater sulfate [ 39 , 40 ]. The majority of the sulfide present at ULS sites is believed to be related to sulfate reduction coupled to anaerobic hydrocarbon oxidation [ 14 , 39 ]. Other potential sources of sulfide associated with seepage include anaerobic oxidation of deeply buried organic material [ 10 ], “sour” hydrocarbons containing a proportion of sulfur [ 41 ], and hydrocarbon interactions with sulfur-bearing minerals such as gypsum and anhydrite found in the salt dome cap rocks of the ULS [ 8 , 42 , 43 ]. Concentrations of all chemical species in the sediments surrounding the rhizosphere were derived from the dataset included in Arvidson et al. [ 13 ] and Morse et al. [ 28 ]. Only those sediment cores taken around the “drip line” of tubeworm aggregations that contained detectable sulfide concentrations were used. Due to the vagaries of sampling with a submersible in sediments heavily impacted by carbonate and roots, those cores with detectable sulfide are believed to more accurately represent conditions around the periphery of the rhizosphere. Dissolved organic carbon (DOC) concentration was used as an estimate of methane concentration. While other forms of DOC make up this total concentration, methane accounts for 90%–95% of the hydrocarbon gasses dissolved in pore waters [ 28 ]. In seep sediments, the majority of DOC is likely to be in the form of hydrocarbon gasses. Because estimates of organic acid concentrations were not available, they could not be explicitly modeled. This would not affect the overall concentration of electron donors in the model, but could affect the sulfate reduction rate. Since sulfate reduction rate estimates for methane seeps in the Gulf of Mexico are among the highest recorded [ 14 , 39 ], any differences in DOC composition (e.g., higher relative concentrations of dissolved organic acids) would serve to lower the overall sulfate reduction rate and sulfide availability. Sulfide supply estimates presented are likely overestimated most by the model without root sulfate release owing to the greater reliance on anaerobic methane oxidation in this form of the model. Simulations including sulfate release by tubeworms are affected to a lesser extent as the concentration of electron donors is not limiting in this model configuration. Solid and liquid phase organic carbon was separated into hydrocarbons and buried organic material according to their relative concentrations in hydrocarbon seep and surrounding Gulf of Mexico sediments. Background sediments on the ULS contain 0.71% organic carbon by weight [ 29 ]. At hydrocarbon seeps on the ULS, organic carbon accounts for 4.47% of total weight. This was assumed to be the sum of background organic input plus carbon in the form of C 6+ hydrocarbons. It is possible that the higher biomass located at ULS seeps in the form of non-living macrofaunal and microbial materials may also contribute to the increased organic carbon concentration, but without empirical estimates, this could not be accounted for in the model. Hydrocarbons may consist of between 50% and 95% labile materials [ 44 , 45 , 46 ]. Based on existing data on degradation rates and residual hydrocarbons subjected to degradation [ 42 , 47 ], a value of 50% labile material was used here. These assumptions of hydrocarbon concentration and degradation potential are therefore believed to be conservative. The following functions were fitted to the sulfide, sulfate, and methane concentration profiles ( Figure 5 ) to determine the boundary conditions at any given depth: Figure 5 Concentration Profiles of Sulfate, Sulfide, and DOC Points represent average concentration at a given depth from 13 sediment cores taken around the periphery of tubeworm aggregations (see Materials and Methods and original data in [ 13 , 28 ]). Best-fitted line based on least squares fit of equation 9 . where C 0 is initial concentration, C ∞ is concentration at infinite distance, and C i is concentration at depth d . As there were no existing data for sediments below 30 cm, concentrations at infinite depth ( C ∞ ) were used (SO 4 2− = 0 mmol · l −1 , HS − = 12 mmol · l −1 , DOC = 11 mmol · l −1 , DIC = 20 mmol · l −1 , pH = 7.78). The first derivatives of the sulfide and methane profiles were used for the calculation of advective flux from depth. The first derivative of the sulfate profile was used for diffusive flux across the water–sediment interface of the rhizosphere, with advection rate subtracted from diffusive flux of sulfate across this surface. Advection (seepage) rate varied with time according to the following function: where t is simulation time in years and sed is sedimentation rate (6 cm · 1,000 y −1 ) [ 29 ]. Early seepage rate approximated the highest flux rates measured or estimated for methane seeps and declined with time in the model to the highest estimates for persistent, region-wide seepage in the Gulf of Mexico ( Table 1 ). This follows a pattern of hydrocarbon seep development, with the highest seepage rates early in the evolution of the local seepage source followed by occlusion of fluid migration pathways by carbonate precipitation, hydrate formation, and possibly tubeworm root growth. By using the highest rate estimated (32 mm · y −1 = 0.000365 cm · h −1 in equation 10 ) as the basal seepage rate, we are testing the possibility that tubeworm aggregations could survive under the most favorable conditions possible in the absence of tubeworm sulfate supply. For sediments encompassed by the rhizosphere, sulfide, sulfate, methane, DOC, and hydrogen ion concentration profiles were determined iteratively prior to model implementation using a central difference scheme: where C i ( t ) is concentration in cell i at time t, D is the diffusion coefficient, k is the maximum reaction rate, and K s is the half-saturation constant for the reaction ( Table 2 ). Reactions included anaerobic methane oxidation ( equation 17 ), tubeworm sulfide uptake rate ( equation 8 ), and carbonate precipitation rate ( equation 22 ). The concentration in each 2 × 2 cm cell was calculated at 1 h time steps. At the end of each year, diffusion distance increased. The number of cells (total diffusion distance) was determined by the average root length of L. luymesi populations as realized in independent runs of the population growth model described above, and included here as model input only. A separate function was fitted to each of the concentration profiles: Table 2 Parameters Involved in Diagenetic Model a Diffusion coefficients all corrected for temperature, pressure, and salinity according to Stumm and Morgan [ 51 ] and Pilson [ 52 ] b All disassociation constants corrected for temperature, salinity, and pressure according to Stumm and Morgan [ 51 ] and Pilson [ 52 ] except: CaOH, no correction; CaHCO 3 , CaSO 4 , CaSO 4 H 2 O, MgHCO 3 , temperature only; H 2 CO 3 , temperature and salinity only; and HSO 4 , temperature and pressure only where d is radial distance. The relationship between the parameter a and distance was used to generate concentration profiles for each disc comprising the rhizosphere. Because of the tight linear relationship between diffusion distance and the shape of the curve, concentration profiles could be generated for a disc of any size using the following function: where α is 1.7344 and β is 1.0104 for HS − , α is 0.2111 and β is 0.3363 for SO 4 2− , and α is 0.1626 and β is 0.2518 for CH 4 . Diffusional fluxes of sulfide, sulfate, and methane were calculated according to the first and second derivatives of the concentration profiles as determined by the diameter of each disc. Model implementation The model estimates sulfide availability to the aggregation as a whole by summing the fluxes separately determined for each 2-cm disc composing the rhizosphere. Depth-dependant boundary conditions were set for each disc separately based on the sediment profiles ( Figure 5 ). Diffusional fluxes into each disc were calculated from the shape of the concentration profiles according to the following function [ 48 ]: where C is concentration, r is disc radius, and D s is the diffusion coefficient corrected for porosity by: where D o is the diffusion coefficient corrected for temperature and pressure, n is the chemical species-specific constant, and φ is porosity. The value of n was set to 2.75 as this was found to be a reasonable fit for all chemical species examined [ 49 ]. The ionic states of each species at the average pH value of tubeworm-dominated sediments (7.78) were used for the determination of diffusion coefficients. Porosity was determined from the following function: where φ z is porosity at depth z, φ 0 is porosity at the sediment–water interface, and φ ∞ is porosity at infinite depth; φ 0 was set at 0.841, φ ∞ at 0.765, and a at 0.210, as determined from the best fit with the porosity data ( Figure 6 ) from Morse et al. [ 28 ]. Figure 6 Sediment Porosity Values Points represent average porosity at a given depth from 13 sediment cores taken around the periphery of tubeworm aggregations (see Materials and Methods and original data in [ 13 , 28 ]). Best-fitted line based on least squares fit of equation 9 . Diffusion across the sediment–water interface of the rhizosphere was also considered as an additional input of sulfate and hydrogen ions. This was included as one-dimensional diffusion across a circular surface (subtracting the area encompassed by the tubeworm tubes) with diffusion distance equal to rhizosphere diameter, and concentration differential from seawater concentration to the average concentration within the rhizosphere. Sulfate and hydrogen ion diffusion across the root surface was then added (if included in the set of model realizations) as simple Fickian diffusion. Concentration differential was the difference between internal concentration and average concentration for each disc of the rhizosphere assuming roots were evenly proportioned according to the volume encompassed by each disc. Internal sulfate concentration and pH ( Table 2 ) represented an average of the values determined for R. pachyptila [ 22 ], a hydrothermal vent tubeworm. Internal sulfate concentrations and pH of L. luymesi have not been reported, but these values are generally consistent within taxa [ 50 ]. Uptake of sulfide and release of sulfate were summed across the entire tubeworm population, again assuming an even distribution of roots within the rhizosphere. The paucity of empirical data on the location of any individual tubeworm's roots within an aggregation precluded modeling space explicitly; therefore, it is assumed that each individual has equal access to the resources available within the rhizosphere. Within the rhizosphere, sulfide generation may be limited by sulfate supply, electron donor availability, or sulfate reduction rate. Sulfate supply was determined as the sum of flux across the series of discs approximating the rhizosphere dome, across the sediment–water interface, and from root sulfate (if available). Available sulfate is utilized for anaerobic methane oxidation first (the more energetically favorable process), then hydrocarbon and organic matter degradation. Electron donors included methane, complex hydrocarbons, and buried organic material. Solid and liquid phase hydrocarbons and organic material were assumed to be homogenous within the rhizosphere. Methane supply was determined as the sum of flux across each rhizosphere disc boundary. Hydrocarbon and organic material concentrations were determined as the amounts encompassed within the rhizosphere volume minus that oxidized in previous years. Sulfate reduction rate was determined from the relative amounts of the various electron donors with higher rates (0.71 μmol · ml −1 · h −1 ) for methane oxidation and lower rates (0.083 μmol · ml −1 · h −1 ) for organic matter or hydrocarbon degradation [ 39 ]. Microbes carrying out these processes are assumed to be evenly distributed within the rhizosphere. Total hydrogen sulfide availability to the aggregation was determined as the sum of sulfide diffusion and advection across each rhizosphere disc and sulfide generated within the rhizosphere from sulfate reduction according to the following reactions: SO 4 2− + CH 4 → HS − + HCO 3 − + H 2 O SO 4 2− + 2CH 2 O → HS − + 2HCO 3 − + H 2 O SO 4 2− + 1.47C n H 2n+2 → HS − + 1.47HCO 3 − + H 2 O Bicarbonate (HCO 3 − ) is generated at a 1:1 stoichiometry during anaerobic methane oxidation and a 2:1 stoichiometry in the degradation of organic material. As hydrocarbons are degraded forming smaller chain hydrocarbons and organic acids, bicarbonate is generated at different stoichiometries. Because different-sized hydrocarbons and organic acids were not accounted for in the model, a rough average of these stoichiometries (1.47:1) based on toluene, ethylbenzene, xylene, and hexadecane degradation [ 18 ] was used to determine the amount of bicarbonate generated per mole of carbon. Hydrogen ions are also used up in a 1:1 stoichiometry with sulfate in the sulfate reduction half reaction as included in reaction 17. In order to account for carbonate precipitation, the model traced DIC concentration, calcium concentration, hydrogen ion concentration, buffer capacity, carbonate saturation, and carbonate precipitation rate. The buffer state of the rhizosphere was calculated to determine changes in pH resulting from hydrogen ion flux. Buffer capacity ( β ) was calculated using the following function [ 51 ]: where A and B represent the concentrations of the various acids and bases in the buffer system. In addition to hydrogen and hydroxyl ions, the buffer system included carbonate (CO 2 , H 2 CO 3 , HCO 3 − , and CO 3 2− ), sulfide (H 2 S and HS − ), sulfate (HSO 4 − and SO 4 2− ), and borate (B[OH] 4 − and B[OH] 3 ) speciation. Current pH was used to determine the ionic state of each species according to temperature-, pressure-, and salinity-corrected disassociation constants when available [ 51 , 52 ] ( Table 2 ). Change in pH was determined from hydrogen ion flux and buffer capacity as follows: Saturation state is highly dependent on the degree to which calcium and bicarbonate form complexes with other ions. The “free” calcium was determined as the proportion of calcium that is not associated with complexed bicarbonate (HCO 3 − ), carbonate (CO 3 2− ), hydroxyl (OH − ), or sulfate (SO 4 2− ) ions. Free carbonate was determined as the amount not forming complexes with calcium (Ca + ) or magnesium (Mg + ) ions in solution. Saturation state was then calculated from the product of the concentrations of free calcium and carbonate divided by the solubility product constant. If the saturation state was above one, then carbonate precipitation occurred at a rate determined by: where k 1 is 0.00597 l · mol −1 · sec −1 and k 3 = 0.456 l · mol −1 · sec −1 [ 51 ]. Because there is no empirical relationship between weight percent of carbonate and sediment porosity in tubeworm-dominated sediments [ 28 ], precipitation did not directly affect porosity. Precipitation was accounted for in the model by subtracting the volume of carbonate precipitate from the total volume encompassed by the rhizosphere. At the end of each annual time step, model output included average length of individuals, population size, sulfide uptake rate, sulfide supply rate, root sulfate flux (if included), root hydrogen ion flux, amount of sulfide supply accounted for by each process (sulfide seepage, anaerobic methane oxidation, organic matter degradation, and hydrocarbon degradation), number of individuals that could be supported by sulfide supply, carbonate precipitation rate, volume of carbonate precipitate, and pH.

D:\keerthana\PMC001xxxxxx\PMC1044833.xml

1044834

Location Coding by Opponent Neural Populations in the Auditory Cortex

Although the auditory cortex plays a necessary role in sound localization, physiological investigations in the cortex reveal inhomogeneous sampling of auditory space that is difficult to reconcile with localization behavior under the assumption of local spatial coding. Most neurons respond maximally to sounds located far to the left or right side, with few neurons tuned to the frontal midline. Paradoxically, psychophysical studies show optimal spatial acuity across the frontal midline. In this paper, we revisit the problem of inhomogeneous spatial sampling in three fields of cat auditory cortex. In each field, we confirm that neural responses tend to be greatest for lateral positions, but show the greatest modulation for near-midline source locations. Moreover, identification of source locations based on cortical responses shows sharp discrimination of left from right but relatively inaccurate discrimination of locations within each half of space. Motivated by these findings, we explore an opponent-process theory in which sound-source locations are represented by differences in the activity of two broadly tuned channels formed by contra- and ipsilaterally preferring neurons. Finally, we demonstrate a simple model, based on spike-count differences across cortical populations, that provides bias-free, level-invariant localization—and thus also a solution to the “binding problem” of associating spatial information with other nonspatial attributes of sounds.

Introduction Topographic representation is a hallmark of cortical organization: primary somatosensory cortex contains a somatotopic map of the body surface, primary visual cortex contains a retinotopic map of visual (retinal) space, and primary auditory cortex contains a cochleotopic map of sound frequency. The necessity of auditory cortex for normal sound localization (which is disrupted by cortical lesions [ 1 , 2 , 3 ]) strongly implies a cortical representation of auditory space. That representation has been reasonably expected to consist of a spatiotopic map, based on the existence of such maps in other sensory systems and on the view, proposed by Jeffress [ 4 ], that spatial processing in the auditory brainstem and midbrain might involve a “local code” consisting of topographic maps of interaural spatial cues. A local code, or “place code,” is one in which particular locations in space, or the spatial cues that correspond to those locations, are represented by neural activity at restricted locations in the brain. Evidence for local coding of auditory space has been demonstrated in mammalian superior colliculus [ 5 , 6 ] and in avian inferior colliculus (IC) [ 7 , 8 ] and optic tectum (homologous to mammalian superior colliculus) [ 9 ]. Nevertheless, local spatial coding has not thus far been demonstrated in the mammalian ascending auditory pathway. If the Jeffress model is correct and a local code for spatial cues exists subcortically, one might anticipate local coding to be maintained in the cortex, where the various cues might finally be integrated into a coherent map of auditory space. Numerous studies, however, have failed to provide evidence for such a map. The spatial tuning of neurons is often characterized using rate–azimuth functions (RAFs), which specify the average response rate (spikes per trial or per second) as a function of stimulus location in the horizontal dimension. Throughout the auditory cortex, such functions typically exhibit broad peaks (up to 180° wide) that cover the contralateral hemifield and broaden further with increasing sound level [ 10 , 11 , 12 , 13 , 14 ]. Similar functions have been reported for cortical sensitivity to interaural cues [ 15 , 16 ], and for spatial and interaural sensitivity in the auditory brainstem and midbrain [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 ], thus questioning Jeffress's view of binaural processing in mammals. The emerging alternative view replaces the local code with a “distributed code,” in which sound-source locations are represented by patterns of activity across populations of broadly tuned neurons [ 12 , 24 , 25 ]. In the past, we argued for a distributed spatial code in the auditory cortex in part because the broad spatial tuning of cortical neurons would seem to preclude the existence of a local code and also because individual neurons are able to transmit spatial information throughout much, if not all, of auditory space [ 25 , 26 ]. At least implicitly, we have advocated a uniform distributed code, assuming that uniform sampling of space by RAF peaks is required for maximally accurate spatial coding. Spatial centroids of neurons in the posterior auditory field (PAF), for example, sample space more uniformly than neurons in the primary auditory field (A1), and we have suggested that this feature partially underlies the increased ability of ensembles of PAF neurons to accurately signal sound-source locations [ 14 ]. A number of observations demonstrate, however, that the auditory cortex samples space nonuniformly. RAFs are plotted for a selection of neurons in the dorsal zone (DZ) of auditory cortex in Figure 1 , to illustrate a common observation of location-sensitive auditory cortical neurons: the majority favor contralateral stimulation, and typically exhibit either “hemifield” or “axial” tuning [ 11 ], responding to stimuli located throughout contralateral space or near the acoustic axis of the contralateral pinna, respectively. A smaller number of ipsilaterally tuned units are also observed, the majority of which exhibit hemifield or axial tuning characteristics similar to those of contralateral units. In A1 and DZ, ipsilateral- and/or midline-tuned neurons may be arranged in bands—parallel to the tonotopic axis—that interdigitate with bands of contralaterally tuned cells [ 25 , 27 , 28 , 29 ]. The overall preponderance of contralateral tuning among cortical units seems to justify the view that each hemisphere represents the contralateral spatial hemifield, a view that is also supported by the contralateral sound-localization deficits that follow auditory cortical lesions [ 2 , 30 , 31 ]. Even within a single hemifield, however, no strong evidence for a topographic representation has been reported, and the observation that many units share similar hemifield RAFs demonstrates a profound inhomogeneity in the way cortical populations sample auditory space. Figure 1 Example RAFs Plotted are normalized mean spike counts (y-axis) elicited by broadband stimuli (20 dB above unit threshold) varying in azimuth (x-axis). Lines represent units recorded in cortical area DZ. Left: contralaterally responsive units. Right: ipsilaterally responsive units. Additional evidence that the cortical representation of auditory space is inhomogeneous comes from studies of the ability of cortical responses to classify stimulus locations. Stecker et al. [ 14 ] found that the responses of most spatially sensitive units in cat cortical areas A1 and PAF could accurately discriminate the lateral hemifield (left versus right) of a stimulus, but often confused locations within the hemifield. This is shown for six PAF neurons represented by confusion matrices in Figure 2 . Similarly, Middlebrooks et al. [ 12 ] measured median localization errors—based on neural-network analyses of responses in the second auditory field (A2) and the field of the anterior ectosylvian sulcus (AES)—between 37.5° ± 8.9° and 43.7° ± 10.2 ° , just under the theoretical limit of 45 ° attainable through perfect left/right discrimination and within-hemifield confusion. Taken together, these results suggest that auditory space is represented within the cortex by a population of broadly tuned neurons, each of which is able to indicate the lateral hemifield from which a sound originated, but generally little more. Figure 2 Classification Performance of Accurate PAF Units from [ 14 ] Neural spike patterns were classified according to the stimulus location most likely to have elicited them. In each panel, a confusion matrix plots the relative proportion of classifications of each target azimuth (x-axis) to each possible response azimuth (y-axis). Proportions are indicated by the area of a circle located at the intersection of target and response locations. Example units were selected from among those transmitting the most spatial information in their responses. In each case, discrimination of contralateral azimuths (negative values) from ipsilateral azimuths (positive values) is apparent, accompanied by significant within-hemifield confusion. As such, neural responses are sufficient for left/right discrimination only, and the spatial information transmitted by the most accurate units tends not to be much greater than one bit per stimulus. Results/Discussion Preferred Locations Oversample Lateral Regions of Contralateral Space—Steepest RAF Slopes Straddle the Midline The idea that sound locations are signaled by the peaks of RAFs, which tend to be centered deep within the lateral hemifields, is at odds with localization behavior, which shows greatest resolution near the interaural midline [ 32 , 33 ]. An alternative view, however, has emerged for the processing of interaural time and level differences by cortical and subcortical neurons. In that view, locations are coded by the slopes, rather than the peaks, of rate–interaural-time-difference or rate–interaural-level-difference functions [ 22 , 24 , 34 ]. Moreover, these slopes appear aligned with the interaural midline and provide maximum spatial information in that region [ 20 ]. If a similar arrangement can explain the inhomogeneity of spatial sampling in the auditory cortex, then we would expect to find cortical RAF slopes to be steepest near the interaural midline as well. In this report, we compare the responses of neurons in primary auditory cortex (A1) and two higher-order auditory cortical fields (PAF and DZ) in the cat. Compared to A1, areas PAF and DZ exhibit spectrotemporally complex responses that are significantly more sensitive to variations in sound-source location [ 14 , 25 ]. Therefore, these areas are the most likely candidate regions of cat auditory cortex for spatial specialization. PAF, in particular, appears necessary for sound localization by behaving cats [ 30 ]. Figure 3 depicts the distribution of preferred locations (“azimuth centroids”; see [ 14 ]) along with locations of peak RAF slopes in all three fields. As we have reported previously, centroid distributions in Figure 3 reveal a preponderance of contralateral sensitivity regardless of cortical area or stimulus level [ 14 , 25 ]. Distributions of peak-slope location, however, are tightly clustered around the frontal midline (median ± standard error [see Materials and Methods ] in A1, +15° ± 2.5°; in DZ, +5° ± 2.1°; in PAF, −5° ± 2.6°). Values in A1 fall significantly farther into the ipsilateral field than do those in DZ ( p < 0.0004) or PAF ( p < 0.0002), consistent with both the broader spatial tuning and less extreme azimuth centroids of A1 compared to PAF or DZ units [ 14 , 25 ]. Overall, the positioning of RAF slopes near the interaural midline suggests that auditory space is sampled inhomogeneously by the cortical population; the midline represents a transition region between locations eliciting responses from populations of contralateral- and ipsilateral-preferring units. Figure 3 RAF Slopes Are Steepest near the Interaural Midline Plotted are summaries of preferred locations (centroids) and points of maximum RAF slope for 254 units recorded in A1 (left), 411 in PAF (middle), and 298 in DZ (right) for levels 20 and 40 dB above threshold (thr) (bottom and top rows, respectively). In each panel, units are sorted by centroid (blue crosses) on the y-axis. Thin red lines denote the region of azimuth (x-axis) containing the centroid and bounded by the points of steepest slope. For units with centroids lateralized more than 10° from the midline, we marked either the steepest positive slope (for ipsilaterally tuned units) or negative slope (for contralateral units) with a black circle. These points represent the location of most rapid response change that occurs toward the front of the animal (relative to the centroid; for units that respond throughout the frontal hemifield, this point can occur toward the rear). Distributions of centroid (blue line) and peak slope (black line), calculated using kernel density estimation with 20 ° rectangular bins, are plotted below each panel. These indicate that while preferred locations (centroids) are strongly biased toward contralateral azimuths, peak slopes are tightly packed about the interaural midline, consistent with the opponent-channel hypothesis. Neural Response Patterns Discriminate Best across Midline Modulation of spike count is generally the most salient location-sensitive feature of neural responses, especially when data are averaged over many trials. However, temporal features of the neural response—such as first-spike latency, the temporal dispersion of spikes, and specific temporal features such as prototyped bursts of spikes or periods of inhibition—could also play an important role in stimulus coding by cortical neurons, and we have studied this role using pattern-recognition analyses applied to spike patterns [ 12 , 14 , 35 ]. Here, we assess the ability of neural spike patterns to subserve pairwise discrimination of stimulus locations by adapting the pattern-recognition approach of Stecker et al. [ 14 ] to a discrimination paradigm. This approach is similar to the receiver-operating-characteristic analysis used to estimate interaural and/or spatial thresholds from neural spike counts [ 36 , 37 , 38 ], with the addition of spike-timing information. Given a spike pattern—a smoothed, bootstrap-averaged peristimulus time histogram (2-ms bins) that approximates the instantaneous probability of spike firing over the course of 200 ms following stimulus onset—elicited by stimulation from an unknown location in space, the algorithm estimates the relative likelihood that the pattern was evoked by a sound from each of the 18 tested locations. From these relative likelihoods, we compute the index of discriminability, d′ [ 39 ], for each pair of stimulus locations. In Figure 4 (right), pairwise d′ is plotted as a function of the midpoint and separation between paired stimuli for a single PAF unit; the contour d′ = 1 (dashed line) indicates the spatial discrimination threshold. Note that in this example suprathreshold discrimination is possible at much narrower stimulus separations when the stimuli span the interaural midline (left/right discrimination) than in cases of front/back discrimination spanning +/− 90 ° , about which point many features of the neural response (e.g., spike rate and latency) are symmetrical. As a result, the minimum discriminable angle (MDA, defined as the minimum separation along the d′ = 1 contour) of 25 ° is found at a best azimuth (BA, the midpoint location of the most discriminable pair) of −5 ° , near the frontal midline. Figure 4 Discrimination Analysis Based on Responses of One PAF Unit Left: raster plot of spike times (x-axis) recorded in response to broadband noise stimuli varying in azimuth (y-axis). Note the strong modulation of spike count, response latency, and temporal features of the response between contralateral and ipsilateral locations. Right: pairwise spatial discrimination. Colors indicate d′ values for pairs of stimulus locations varying in separation (y-axis) and overall azimuth (x-axis, midpoint of two azimuths). The dashed line indicates threshold discrimination ( d′ = 1), and the red circle marks the unit's MDA (y-axis) and BA (x-axis). In Figure 5 , MDA is plotted as a function of BA for the entire population of A1, PAF, and DZ units in which discrimination thresholds could be calculated. Overall, two main features of the results should be noted. First, despite the broad azimuth tuning of cortical neurons, the majority can discriminate between stimuli separated by less than 40 ° . A number of neurons successfully discriminate even smaller separations—especially in DZ, in which the median MDA (30.5 ° ± 2.5 ° ) is significantly smaller than in A1 (40 ° ± 2.6°; p < 0.007) or PAF (43 ° ± 3.8 ° ; p < 0.0002). Note that MDAs of even the most sensitive units exceed behavioral estimates of 5°–6 ° minimum audible angles in cats [ 40 ], but likely underestimate the true neuronal performance because loudspeaker separations were tested in minimum steps of 20 ° , and thus discrimination at smaller separations can only be assessed through extrapolation. Second, the distribution of BAs is tightly clustered around the interaural midline, with 50% of BA values falling within 18.5 ° (PAF), 25 ° (A1), or 26 ° (DZ) of the 0 ° or 180 ° azimuth. Note that this does not mean that units cannot discriminate off-midline locations. It does indicate, however, that the majority of units capable of discriminating between stimulus azimuths do so best for location pairs near the interaural midline. Very few units exhibit BAs located far within either lateral hemifield, although A1 units exhibit significantly more ipsilateral BA values (median, +14.5 ° ± 3.9 ° ) than those in PAF (0 ° ± 2.9 ° ; p < 0.004) or DZ (5 ° ± 4.3 ° ; p < 0.05). As with the analysis of RAF slopes, the pairwise discrimination data reveal an inhomogeneous arrangement of spatial sampling by neurons in the cortical population. Accurate discrimination is found where RAF slopes are steepest (the midline), rather than where units respond most strongly (the lateral poles). Figure 5 MDA by BA MDA (y-axis) is plotted against BA (x-axis) for each unit exhibiting suprathreshold spatial discrimination (see Materials and Methods ). Symbols indicate the cortical area of each unit. Left and lower panels plot distributions of MDA and BA (in numbers of units per rectangular 20 ° bin), respectively. An Opponent-Channel Code for Auditory Space? We have demonstrated quantitatively that the representation of auditory space in the cortex is inhomogeneous, consisting mainly of broadly tuned neurons whose responses change abruptly across the interaural midline. The population of auditory cortical neurons, then, appears to contain at least two subpopulations broadly responsive to contralateral and ipsilateral space. Neurons within each population exhibit similar spatial tuning and thus appear redundant with respect to spatial coding. The similarity of spatial tuning of units in these populations stands in contrast to their more profound differences in frequency tuning, for example. Each subpopulation, or “spatial channel,” is capable of representing locations on the slopes of their response areas (i.e., across the interaural midline) by graded changes in response—a “rate code” for azimuth, generalized to incorporate spatially informative temporal features of the cortical response [ 14 , 35 , 41 ]. In other words, each spatial channel encodes space more or less panoramically, as we have argued previously for individual cortical neurons [ 12 ], although it now seems clear that some regions of space are represented with greater precision than others. In the past, we have argued that auditory space is encoded by patterns of activation across populations of such panoramic neurons. Here, we amend that view—which remains tenable—to reflect the observed inhomogeneity of spatial sampling in the cortex and account for differences in coding accuracy of midline and other locations. Following the proposals of von Békésy [ 42 ] and van Bergeijk [ 43 ] regarding interaural coding in the brainstem, we propose that auditory space is encoded specifically by differences in the activity of two broad spatial channels corresponding to subpopulations of contralateral and ipsilateral units within each hemisphere (i.e., by a left/right opponent process). We will refer to this proposal as the opponent-channel theory of spatial coding in the auditory cortex. An important consequence of the opponent-channel theory is that spatial coding may be robust in the face of changes in stimulus level. As is evident from past work, an important constraint on spatial coding in the cortex is the level dependence of many neurons' tuning widths, such that sharp tuning is seen predominantly for low-level stimulation. For example, a number of narrowly tuned units in Figure 3 exhibit locations of peak slope that closely track their centroids at 20 dB above threshold, and one could argue that such units form the basis of a local (e.g., topographic) spatial code when stimulus levels are low. Such a code, however, would be significantly impaired by increases in stimulus level—predicting that sound localization should be most accurate at low levels. That prediction is not borne out in psychophysical tests [ 44 , 45 ], and we have argued that spatial coding in the auditory cortex must employ relatively level-invariant features of the neural response [ 12 ]. Rather than relying on such features as they naturally occur, the opponent-channel mechanism constructs level-invariant features by comparing the activity of neurons that respond similarly to changes in level but differentially to changes in location, similarly to the coding of color by opponent-process cells in the visual system [ 46 ]. To illustrate the level invariance achieved by opponent-process coding, we analyzed the ability of cortical population responses to signal sound-source locations in the frontal hemifield under different stimulus-level conditions. The analysis (see Materials and Methods ) is simplified in a number of ways—for example, it utilizes a simple linear decision rule that weights contralateral and ipsilateral input equally, sums across multiple neurons within each subpopulation (ignoring any complexity of neural circuitry), combines data across different cortical areas known to exhibit different spatial sensitivities, and reduces each neural response pattern to a single overall spike count—but serves as a “proof of concept” that differences between the responses of neural subpopulations with quasi-independent spatial tuning can be used to estimate sound-source locations in an unbiased manner when stimulus levels vary, whereas the individual population responses cannot. Population responses (means of normalized spike rate across neurons in a population) to stimuli varying in location and level were computed separately for subpopulations composed of contralateral units or ipsilateral units in our sample of recordings in A1, PAF, and DZ. These subpopulations correspond to hypothetical “left” and “right” channels of a spatial coding mechanism. Classification of stimulus locations was based on either one of the subpopulation responses or the difference between the two, and involved linear matching to templates computed from a separate training set [ 14 ]. Population and difference RAFs are plotted in Figure 6 (left), along with confusion matrices (similar to Figure 2 ) for sound-source classification based on each (right). Relatively accurate classification is exhibited by both subpopulation responses and by their difference when test and training sets reflect the same stimulus level. When training and test sets differ, however, responses are systematically biased. After training with 20-dB stimuli, localization of 40-dB stimuli by the contralateral subpopulation is biased toward the contralateral hemifield, because 40-dB ipsilateral stimuli and 20-dB contralateral stimuli elicit similar responses. Similarly, when trained with 40-dB stimuli, localization of 20-dB stimuli is biased toward the ipsilateral hemifield. This pattern is clear in the responses of both the contralateral-preferring and ipsilateral-preferring subpopulations. Classification based on their difference, however, is relatively unbiased. Significant undershoot (central responses for peripheral stimulus locations) results from compression of population RAFs by intense sound. While we know of no behavioral data relating to the effects of stimulus level on sound localization by cats, undershoot has been reported in numerous studies of their localization behavior [ 47 , 48 , 49 ]. Such undershoots, however, need not be assumed to reflect a limitation of the underlying neural representation of auditory space. Figure 6 Difference between Channel Responses Is Less Sensitive to Changes in Level Than Are Channel Responses Themselves Left: population responses (y-axis; see Materials and Methods ) are plotted as a function of azimuth (x-axis) for stimuli presented 20 dB (red) and 40 dB (blue) above unit thresholds. Population responses were computed separately for subpopulations composed of contralateral units (top) or ipsilateral units (middle) corresponding to hypothetical “left” and “right” channels of an opponent-channel spatial coding mechanism. The difference (bottom) between responses of the two subpopulations is more consistent across stimulus level than is either subpopulation response alone. Error bars indicate the standard deviation of responses across 120 simulated trials. Right: stimulus–response matrices (confusion matrices; see Figure 2 ) showing the proportion (area of black circle) of responses to a given (unknown) stimulus azimuth (x-axis) classified at each response azimuth (y-axis). Classification assigned each neural population response in the “test” set to the stimulus azimuth whose mean population response in an independently selected set of “training” trials was most similar. In some conditions, test and training trials were drawn from the same set of (matching level) trials: 20 dB (first column) or 40 dB (far right column). In others, test and training trials reflected different-level stimuli: 40-dB test stimuli classified based on a 20-dB training set (second column), or 20-dB test stimuli classified based on a 40-dB training set (third column). The contralateral and ipsilateral subpopulation responses (top and middle rows) accurately localize fixed-level stimuli, but are strongly biased when tested at non-trained stimulus levels. In contrast, the difference between responses (bottom row) remains relatively unbiased in all conditions, although responses to stimuli at untrained levels do exhibit compressed range and increased variability of classification. Based on its manner of level-invariant spatial coding, it seems clear that an opponent-channel mechanism should behave similarly in the presence of any stimulus change (e.g., in frequency, modulation, or bandwidth) that acts to increase or decrease the response of both channels. This suggests an efficient means for combining spatial information with information about other stimulus dimensions. This general principle of opponent-process coding should hold in any case where both channels exhibit similar sensitivity to the nuisance dimension (level, frequency, etc.) but dissimilar sensitivity to space, and illustrates one strength of the opponent-channel coding strategy: the ability to recover spatial information from the responses of neurons that are strongly modulated by other stimulus dimensions. As long as some of the cortical neurons involved in coding a particular acoustic feature are contralaterally driven and others are ipsilaterally driven, the spatial location of that feature can be computed without imposing additional distortion of its neural representation. Note that the opponent-channel theory as presented here involves contralateral and ipsilateral channels within each hemisphere. This feature is based on the observation of both types of neurons in a single hemisphere, and on the results of unilateral cortical lesions, which produce localization deficits mainly in contralesional space [ 2 , 30 , 31 , 50 ]. The lesion data prevent us from considering opponent-channel mechanisms that place each channel in a separate hemisphere (e.g., left-hemisphere contralateral units versus right-hemisphere contralateral units) because in that case unilateral lesions should abolish localization throughout the entire acoustic field. As proposed here, however, the opponent-channel mechanism in either hemisphere should be capable of coding locations throughout space, not just in the contralateral hemifield. This would suggest that only bilateral lesions could produce localization deficits, which is also not the case. At this point, we can merely speculate that auditory cortical structures in each hemisphere provide input only to those multimodal spatial or sensorimotor structures that subserve localization behavior in contralateral space and, furthermore, that these inputs cannot be modified in adulthood following cortical lesions. General Discussion In summary, the available data suggest that space is sampled nonuniformly in all fields of auditory cortex, with the majority of neurons responding broadly within one hemifield and modulating their responses abruptly across the interaural midline. Consistent with this view, we found cortical responses to be most sensitive to changes in stimulus azimuth at midline locations. Cortical neurons' RAFs tend to be steepest near the midline even though their preferred locations are found distributed throughout the contralateral hemifield. Spatial discrimination by neural responses is also best at or near the interaural midline. Results of both analyses are compatible with the existence of a limited number of spatial channels in the cortex, and incompatible with either a uniform distributed representation or a local representation (e.g., a topographic map). The relative paucity of units with sharp tuning peaking near the midline strongly suggests that behavioral sound-localization acuity is mediated by the slopes and not the peaks of spatial receptive fields. In this report, we consider a model of spatial coding based on differences in the response rates of two broad spatial channels in the auditory cortex. It is similar to the mechanism proposed by Boehnke and Phillips [ 51 ] to account for differences in human gap detection when gaps are bounded by auditory stimuli occurring in the same or opposite hemifields. In each proposal, neural response rates are compared across channels, but each is also consistent with information encoded in the relative response timing of cortical neurons [ 25 , 52 ]. Although the psychophysical and physiological data seem to agree on a two-channel mechanism, it is important to note that in this study, we treat units that respond more strongly to forward than rearward locations (“axial” units) as equivalent to units that respond equally to both quadrants (“hemifield” units). Similarly, we do not specifically examine the small number of units that respond best to midline locations. Distributed coding of interaural intensity by neural populations differing in binaural facilitation has been suggested previously [ 24 ]; similarly, populations of midline and/or axial units could be treated separately in a three-, four-, or five-channel opponent model of spatial coding. Such a model would follow the general principles of opponent-channel coding described here, but might differ in its ability to accurately code locations over wide regions of azimuth (see [ 24 ]). That the representation of space appears inhomogeneous in both primary and higher-order auditory cortical fields argues against the existence of a topographic “space map” within sensory cortex, pushing the emergence of any such map further into central structures than previously expected. The processing of interaural cues begins at the level of the superior olivary complex, but the integration of such cues into a complete topographic map of auditory space is presumed to begin with processing at the level of the IC or cortex. The suggestion that interaural cues are represented by a limited number of binaural channels in the IC [ 22 ] seems to imply that the space map must emerge at the level of auditory cortex or beyond, and the results of this study, along with others [ 15 , 16 ], suggest that a “limited channel” code is maintained throughout primary and non-primary fields of the auditory cortex as well. PAF, in particular, appears to sit at the top of the auditory cortical processing hierarchy [ 53 ] but is similar to primary auditory cortex (A1) in this regard. We should note that spatial coding must subserve at least two distinct behavioral tasks, namely, the discrimination of sound-source locations and the localization of individual sources (e.g., orientation, or pointing). Much of the current discussion has focused on aspects of spatial coding relevant to discrimination, and on the observation that the RAF slopes of cortical neurons are better suited to the discrimination of nearby locations than are their broad RAF peaks. Nevertheless, we are interested in general mechanisms of spatial representation, and argue that cortical neurons' broad spatial tuning suggests that neither aspect of sound localization is likely mediated by RAF peaks in cat cortex. This stands in contrast to the neural mechanism for sound localization in the IC of the owl, where sharp circumscribed spatial receptive fields form a place code for localization [ 7 ]. Owls' behavioral discrimination of spatial locations, however, is sharper than these neural receptive fields, and appears—as in mammals—to be mediated by receptive-field slopes [ 38 ]. Thus, the owl makes use of place and rate codes for different behavioral tasks. The cat's auditory cortex, on the other hand, lacks the sharp spatial tuning necessary for map-based localization, so one coding strategy underlies both types of behaviors. It seems clear that these different coding strategies in owls and cats necessitate different mechanisms for generating motor responses and orienting to sound sources. The owl's space map exhibits a straightforward correspondence between restricted neural activity and locations in space, which might be ideal for computing audiovisual correspondence but requires further translation into motor coordinate systems before action can take place. It is possible that the opponent-channel code is transformed into a similar auditory space map within multisensory or sensorimotor areas, that is, not within auditory cortex itself. Alternatively, opponent-channel population codes in the auditory domain might be directly transformed into population codes in the motor domain without an intervening map-like representation. In either case, we could argue that the fundamental mode of spatial coding within the auditory system per se is non-topographic. In fact, it might be that auditory spatial topography is an emergent property of widespread neural populations and is evident only in perception and behavior, not in the physiology of single neurons. In considering the relative advantages of opponent-channel spatial coding within the cortex, one might wonder whether the formation of a spatiotopic map would be necessary or desirable. As described above, the opponent-channel mechanism could subserve behavior without an intervening map, and it provides an efficient means of combining information about space with information about other stimulus features. In this regard, at least, the opponent-channel mechanism solves—or simply avoids—the so-called binding problem [ 54 ] of how multiple stimulus features can be associated to create a unified neural representation. It does so without recourse to specialized mechanisms for binding [ 55 ] and without an explosion in the number of neurons necessary for a complete combinatorial code [ 56 ]. So long as feature maps (e.g., of frequency) contain neurons of each class (i.e., contralateral and ipsilateral), the spatial position of any particular feature can be reconstructed without the difficulty of binding activity in one feature map (frequency) with that in another (location). Finally, the three cortical fields studied in this report exhibited similar evidence for an opponent-channel mechanism, despite previously reported differences in their spatial sensitivity [ 14 ]. Although such differences appear modest when assessed physiologically, studies indicate that some fields are more critical for localization behavior than others [ 30 ]. An intriguing question for future research involves cortical fields—such as the anterior auditory field—that are not necessary for accurate localization. Are spatial channels maintained in such fields, or are they combined to produce space-invariant representations of other stimulus features? Materials and Methods Data analyzed for this report were collected from extracellular recordings of 254, 411, and 298 units in areas A1, PAF, and DZ (respectively) of the cortex of chloralose-anesthetized cats [ 14 , 25 ]. Methods of animal preparation, stimulus delivery, unit recording, and basic analysis have been described previously [ 14 ], and were approved by the University of Michigan Committee on Use and Care of Animals. Stimuli were delivered from loudspeakers placed in the free field, and consisted of 80-ms broadband noise bursts presented at levels 20–40 dB above unit threshold. Stimulus locations spanned 360 ° of azimuth in 20 ° steps, and are identified by angular distance from the frontal midline (0 ° ). Positive azimuths increase to to the right (ipsilateral to the recording site), whereas negative values correspond to contralateral locations on the cat's left side. Unit activity was recorded extracellularly from the right cerebral hemisphere using 16-channel electrode arrays (“Michigan probes”), and spikes were sorted off-line based on principal-components analysis of their waveshapes. Locations of peak slope and centroids Each unit's preferred location was characterized by the azimuth centroid of response (dark blue crosses in Figure 3 ; see [ 14 ]); this is the spike-count-weighted average of contiguous stimulus locations eliciting a normalized response at or above 75% of maximum spike count per stimulus presentation. We additionally determined the locations of peak slope for each unit by smoothing its RAF (circular convolution with a 40 ° boxcar) and calculating the first spatial derivative of the result. Maximum and minimum values of the derivative indicate two peak-slope azimuths for each unit (black circles and endpoints of red horizontal lines in Figure 3 ). Spatial discrimination by neural response patterns Analyses of pairwise spatial discrimination (see Figures 4 and 5 ) employed a statistical pattern-recognition algorithm [ 14 ] to estimate the relative likelihood of each stimulus location, given the temporal pattern of neural response to a single (unknown) stimulus. We computed, for each pair of locations θ 1 and θ 2 in the loudspeaker array, the index of pairwise discriminability d′ [ 39 ] based on the estimated relative likelihoods: where z ( P ) represents scaling to the standard normal distribution and the probability P of responding “1” is given by the (estimated) relative likelihood l of location θ 1 (versus θ 2 ), conditional on the actual stimulus location θ i . The analysis produces a map of d′ between each pair of stimulus locations, plotted in coordinates of stimulus separation and overall location in Figure 4 . The map was interpolated to find a contour of d′ = 1, which we define as threshold discrimination. The smallest stimulus separation along the threshold contour defines the MDA, and the overall location of that stimulus pair defines the unit's BA. Symbols in Figures 4 and 5 indicate values of MDA and BA for individual units. Evaluation of a simple population code for space To assess the level invariance of opponent-channel coding, we analyzed a simplified model of population spatial coding in the cortex. For each neural unit in a channel (e.g., a subpopulation of contralateral-preferring units), we accumulated a list of responses (spike counts normalized to the maximum response across all trials) on each trial with a given combination of stimulus azimuth and level. Azimuths were confined to the frontal hemifield (−80° to +80°) to avoid front–back confusions, which obscure but do not alter the appearance of bias in classification responses, and levels were either 20 or 40 dB above individual unit thresholds. We then computed population responses by randomly selecting one trial (with matching stimulus azimuth and level) from each unit and computing the mean of individual responses. We repeated the selection process 120 times for each combination of azimuth and level to simulate a set of 120 population “trials.” The mean of these population responses for each stimulus is plotted on the left in Figure 6 . Separate “training” and “test” sets of population responses were computed by this method and used to assess the ability of subpopulations to classify stimulus locations. Individual population responses in the test set were classified to the azimuth with the most-similar mean population response across the training set. Confusion matrices in Figure 6 plot the proportion of test-set responses assigned to each stimulus azimuth. In some conditions, test and training sets were drawn from the same trials (matching level); in other conditions, training and test sets differed in stimulus level. We tested classification based on responses of a contralateral subpopulation, an ipsilateral subpopulation, and on the difference between subpopulation responses. Contralateral and ipsilateral subpopulations were composed of all units with centroids falling farther than 30° into the corresponding hemifield in our sample of A1, PAF, and DZ units. Differences were computed from the two subpopulation responses on a trial-by-trial basis, and classification was tested in the same manner as for the population responses themselves. Statistical procedures Tests of statistical significance in this study were conducted using a 5,000-permutation bootstrap test (see [ 14 ] for details), reported to one significant digit. Standard error of the median, where reported, was obtained using a 2,000-permutation bootstrap, drawing N (the total number of data points) samples from the data with replacement on each permutation and recomputing the median. Distributions in Figures 3 and 5 were computed by kernel density estimation (convolution) with a 20 ° rectangular window to obtain a continuous function of units per 20 ° bin.

D:\keerthana\PMC001xxxxxx\PMC1044834.xml

1044835

Grasping the Intentions of Others with One's Own Mirror Neuron System

Understanding the intentions of others while watching their actions is a fundamental building block of social behavior. The neural and functional mechanisms underlying this ability are still poorly understood. To investigate these mechanisms we used functional magnetic resonance imaging. Twenty-three subjects watched three kinds of stimuli: grasping hand actions without a context, context only (scenes containing objects), and grasping hand actions performed in two different contexts. In the latter condition the context suggested the intention associated with the grasping action (either drinking or cleaning). Actions embedded in contexts, compared with the other two conditions, yielded a significant signal increase in the posterior part of the inferior frontal gyrus and the adjacent sector of the ventral premotor cortex where hand actions are represented. Thus, premotor mirror neuron areas—areas active during the execution and the observation of an action—previously thought to be involved only in action recognition are actually also involved in understanding the intentions of others. To ascribe an intention is to infer a forthcoming new goal, and this is an operation that the motor system does automatically.

Introduction The ability to understand the intentions associated with the actions of others is a fundamental component of social behavior, and its deficit is typically associated with socially isolating mental diseases such as autism [ 1 , 2 ]. The neural mechanisms underlying this ability are poorly understood. Recently, the discovery of a special class of neurons in the primate premotor cortex has provided some clues with respect to such mechanisms. Mirror neurons are premotor neurons that fire when the monkey performs object-directed actions such as grasping, tearing, manipulating, holding, but also when the animal observes somebody else, either a conspecific or a human experimenter, performing the same class of actions [ 3 , 4 , 5 ]. In fact, even the sound of an action in the dark activates these neurons [ 6 , 7 ]. In the macaque, two major areas containing mirror neurons have been identified so far, area F5 in the inferior frontal cortex and area PF/PFG in the inferior parietal cortex [ 8 ]. Inferior frontal and posterior parietal human areas with mirror properties have also been described with different techniques in several labs [ 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 ]. It was proposed early on that mirror neurons may provide a neural mechanism for understanding the intentions of other people [ 22 ]. The basic properties of mirror neurons, however, could be interpreted more parsimoniously, such as that mirror neurons provide a mechanism for recognizing the observed motor acts (e.g., grasping, holding, bringing to the mouth). The mirror neuron mechanism is, in fact, reminiscent of categorical perception [ 23 , 24 ]. For example, some mirror neurons do not discriminate between stimuli of the same category (i.e., the sight of different kinds of grasping actions can activate the same neuron), but discriminate well between actions belonging to different categories, even when the observed actions share several visual features. These properties seem to indicate an action recognition mechanism (“that's a grasp”) rather than an intention-coding mechanism. Action recognition, however, has a special status with respect to recognition, for instance, of objects or sounds. Action implies a goal and an agent. Consequently, action recognition implies the recognition of a goal, and, from another perspective, the understanding of the agent's intentions. John sees Mary grasping an apple. By seeing her hand moving toward the apple, he recognizes what she is doing (“that's a grasp”), but also that she wants to grasp the apple, that is, her immediate, stimulus-linked “intention,” or goal. More complex and interesting, however, is the problem of whether the mirror neuron system also plays a role in coding the global intention of the actor performing a given motor act. Mary is grasping an apple. Why is she grasping it? Does she want to eat it, or give it to her brother, or maybe throw it away? The aim of the present study is to investigate the neural basis of intention understanding in this sense and, more specifically, the role played by the human mirror neuron system in this type of intention understanding. The term “intention” will be always used in this specific sense, to indicate the “why” of an action. An important clue for clarifying the intentions behind the actions of others is given by the context in which these actions are performed. The same action done in two different contexts acquires different meanings and may reflect two different intentions. Thus, what we aimed to investigate was whether the observation of the same grasping action, either embedded in contexts that cued the intention associated with the action or in the absence of a context cueing the observer, elicited the same or differential activity in mirror neuron areas for grasping in the human brain. If the mirror neuron system simply codes the type of observed action and its immediate goal, then the activity in mirror neuron areas should not be influenced by the presence or the absence of context. If, in contrast, the mirror neuron system codes the global intention associated with the observed action, then the presence of a context that cues the observer should modulate activity in mirror neuron areas. To test these competing hypotheses, we studied normal volunteers using functional magnetic resonance imaging, which allows in vivo monitoring of brain activity. We found that observing grasping actions embedded in contexts yielded greater activity in mirror neuron areas in the inferior frontal cortex than observing grasping actions in the absence of contexts or while observing contexts only. This suggests that the human mirror neuron system does not simply provide an action recognition mechanism, but also constitutes a neural system for coding the intentions of others. Results Subjects watched three different types of movie clips (see Figure 1 ): Context, Action, and Intention, interspersed with periods of blank screen (rest condition). The Context condition consisted of two scenes with three-dimensional objects (a teapot, a mug, cookies, a jar, etc). The objects were arranged either as just before having tea (the “drinking” context) or as just after having tea (the “cleaning” context). The Action condition consisted of a hand grasping a cup in the absence of a context on an objectless background. Two types of grasping actions were shown in the same block an equal number of times: a precision grip (the fingers grasping the cup handle) and a whole-hand prehension (the hand grasping the cup body). In the Intention condition, the grasping actions (also precision grip and whole-hand prehension shown for an equal number of times) were embedded in the two scenes used in the Context condition, the “drinking” context and the “cleaning” context ( Figure 1 ). Here, the context cued the intention behind the action. The “drinking” context suggested that the hand was grasping the cup to drink. The “cleaning” context suggested that the hand was grasping the cup to clean up. Thus, the Intention condition contained information that allowed the understanding of intention, whereas the Action and Context conditions did not (i.e., the Action condition was ambiguous, and the Context condition did not contain any action). Figure 1 Six Images Taken from the Context, Action, and Intention Clips The images are organized in three columns and two rows. Each column corresponds to one of the experimental conditions. From left to right: Context, Action, and Intention. In the Context condition there were two types of clips, a “before tea” context (upper row) and an “after tea” context (lower row). In the Action condition two types of grips were displayed an equal number of times, a whole-hand prehension (upper row) and a precision grip (lower row). In the Intention condition there were two types of contexts surrounding a grasping action. The “before tea” context suggested the intention of drinking (upper row), and the “after tea” context suggested the intention of cleaning (lower row). Whole-hand prehension (displayed in the upper row of the Intention column) and precision grip (displayed in the lower row of the Intention column) were presented an equal number of times in the “drinking” Intention clip and the “cleaning” Intention clip. Figure 2 displays the brain areas showing significant signal increase, indexing increased neural activity, for Action, Context, and Intention, compared to rest. As expected, given the complexity of the stimuli, large increases in neural activity were observed in occipital, posterior temporal, parietal, and frontal areas (especially robust in the premotor cortex) for observation of the Action and Intention conditions. Figure 2 Areas of Increased Signal for the Three Experimental Conditions Threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05. Notably, the observation of the Intention and of the Action clips compared to rest yielded significant signal increase in the parieto-frontal cortical circuit for grasping. This circuit is known to be active during the observation, imitation, and execution of finger movements (“mirror neuron system”) [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 25 , 26 , 27 ]. The observation of the Context clip compared to rest yielded signal increases in largely similar cortical areas, with the notable exceptions of the superior temporal sulcus (STS) region and inferior parietal lobule. The STS region is known to respond to biological motion [ 28 , 29 ], and the absence of the grasping action in the Context condition explains the lack of increased signal in the STS. The lack of increased signal in the inferior parietal lobule is also explained by the absence of an action in the Context condition. Note that, in monkeys, inferior parietal area PF/PFG contains mirror neurons for grasping [ 8 ]. Thus, it is likely that the human homologue of PF/PFG is activated by the sight of the grasping action in the Action and Intention conditions, but not in the Context condition, where the action is not presented. The Context condition activates the inferior frontal areas for grasping, even though only graspable objects—but no grasping actions—are shown. In the monkey brain, ventral premotor area F5 contains, in addition to mirror neurons, a population of cells called canonical neurons [ 4 ]. These neurons fire during the execution of grasping actions as well as during the passive observation of graspable objects, but not during the observation of an action directed at the graspable object. Neurons with these properties mediate the visuo-motor transformations required by object-directed actions [ 30 , 31 ] and are likely activated by the sight of the Context clips. The critical question for this study was whether there are significant differences between the Intention condition and the Action and Context conditions in areas known to have mirror properties in the human brain. Figure 3 displays these differences. The Intention condition yielded significant signal increases—compared to the Action condition—in visual areas and in the right inferior frontal cortex, in the dorsal part of the pars opercularis of the inferior frontal gyrus ( Figure 3 , upper row). The increased activity in visual areas is expected, given the presence of objects in the Intention condition, but not in the Action condition. The increased right inferior frontal activity is located in a frontal area known to have mirror neuron properties, thus suggesting that this cortical area does not simply provide an action recognition mechanism (“that's a grasp”) but rather it is critical for understanding the intentions behind others' actions. Figure 3 Signal Increases for Intention minus Action and Intention minus Context Threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05. The black arrow indicates the only area showing signal increase in both comparisons. The area is located in the dorsal sector of pars opercularis, where mirror activity has been repeatedly observed [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 27 ]. See Tables S1 and S2 for coordinates of local maxima. To further test the functional properties of the signal increase in inferior frontal cortex, we looked at signal changes in the Intention condition minus the Context condition ( Figure 3 , lower row). These signal increases were most likely due to grasping neurons located in the inferior parietal lobule, to neurons responding to biological motion in the posterior part of the STS region, and to motion-responsive neurons present in the MT/V5 complex. Most importantly, signal increase was also found in right frontal areas, including the same voxels—as confirmed by masking procedures—in inferior frontal cortex previously seen activated in the comparison of the Intention condition versus Action condition. Thus, the differential activation in inferior frontal cortex observed in the Intention condition versus Action condition, cannot be simply due to the presence of objects in the Intention clips, given that the Context clips also contain objects. From the contrasts Intention–Action and Intention–Context it is clear that the strongest activity in right inferior frontal cortex is present in the Intention condition. This could be due to two factors, not mutually exclusive: (1) a summation of canonical and mirror neurons activity, and (2) additional activation of mirror neurons of the inferior frontal cortex that code the action the agent will most likely make next. Because in the Intention clips the same action was shown in two contexts (“drinking” and “cleaning”), one can test the intention-coding hypothesis by analyzing the signal increase during observation of the Intention clips. A differential signal increase for the “drinking” Intention clip compared to the “cleaning” Intention clip would indicate neural activity specifically coding the intention of the agent. This logic would hold only if there is no differential signal increase in the “drinking” and “cleaning” Context conditions, when no action is displayed. To test this hypothesis, we compared the signal change in the inferior frontal area in the two Intention clips and the two Context clips. The “drinking” Intention clip yielded a much stronger response than the “cleaning” Intention clip ( p < 0.003; Figure 4 ). In contrast, no reliable difference was observed between the “drinking” Context clip and the “cleaning” Context clip ( p > 0.19). These findings clearly show that coding intention activates a specific set of inferior frontal cortex neurons and that this activation cannot be attributed either to the grasping action (identical in both “drinking” and “cleaning” Intention clips) or to the surrounding objects, given that these objects produced identical signal increase in the “drinking” and “cleaning” Context clips, when no action was displayed. Figure 4 Time Series of the Inferior Frontal Area Showing Increased Signal in the Comparisons Intention minus Action and Intention minus Context The drinking Intention condition yielded a much stronger response than the cleaning Intention condition ( p < 0.003), whereas no reliable difference was observed between the drinking and cleaning Context conditions ( p > 0.19). The time series represents the average activity for all subjects in all voxels reaching statistical threshold in the right inferior frontal cortex. Automaticity of the Human Mirror Neuron System We also tested whether a top-down modulation of cognitive strategy may affect the neural systems critical to intention understanding. The 23 volunteers recruited for the experiment received two different kinds of instructions. Eleven participants were told to simply watch the movie clips (Implicit task). Twelve participants were told to attend to the displayed objects while watching the Context clips and to attend to the type of grip while watching the Action clips. These participants were also told to infer the intention of the grasping action according to the context in which the action occurred in the Intention clips (Explicit task). After the imaging experiment, participants were debriefed. All participants had clearly attended to the stimuli and could answer appropriately to questions regarding the movie clips. In particular, all participants associated the intention of drinking to the grasping action in the “during tea” Intention clip, and the intention of cleaning up to the grasping action in the “after tea” Intention clip, regardless of the type of instruction received. The two groups of participants that received the two types of instructions had similar patterns of increased signal versus rest for Action, Context, and Intention (see Figures S1 and S2 ). The effect of task instructions is displayed in Figure 5 . In all conditions, participants that received the Explicit instructions had signal increases in the left frontal lobe, and, in particular, in the mesial frontal and cingulate areas. This signal increase is likely due to the greater effort required by the Explicit instructions [ 32 , 33 ], rather than to understanding the intentions behind the observed actions. In fact, participants receiving either type of instructions understood the intentions associated with the grasping action equally well. Critically, the right inferior frontal cortex—the grasping mirror neuron area that showed increased signal for Intention compared to Action and Context—showed no differences between participants receiving Explicit instructions and those receiving Implicit instructions. This suggests that top-down influences are unlikely to modulate the activity of mirror neuron areas. This lack of top-down influences is a feature typical of automatic processing. Figure 5 Significant Signal Changes in Subjects Receiving Explicit Instructions Compared to Subjects Receiving Implicit Instructions in the Three Tasks Versus Rest Threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05. The two black arrows indicate two foci of activity in dorsal premotor cortex that are located deep in the sulci and thus not easily visible on the three-dimensional surface rendering. See Tables S3–S5 for coordinates of local maxima. Discussion The data of the present study suggest that the role of the mirror neuron system in coding actions is more complex than previously shown and extends from action recognition to the coding of intentions. Experiments in monkeys demonstrated that frontal and parietal mirror neurons code the “what” of the observed action (e.g., “the hand grasps the cup”) [ 4 , 6 , 8 , 34 ]. They did not address, however, the issue of whether these neurons, or a subset of them, also code the “why” of an action (e.g., “the hand grasps the cup in order to drink ”). The findings of the present study showing increased activity of the right inferior frontal cortex for the Intention condition strongly suggest that this mirror neuron area actively participates in understanding the intentions behind the observed actions. If this area were only involved in action understanding (the “what” of an action), a similar response should have been observed in the inferior frontal cortex while observing grasping actions, regardless of whether a context surrounding the observed grasping action was present or not. Before accepting this conclusion, however, there are some points that must be clarified. First, one might argue that the signal increase observed in the inferior frontal cortex was simply due to detecting an action in any context. That is, it is the complexity of observing an action embedded in a scene, and not the coding of the intention behind actions, that determined the signal increase. A second issue, closely related to the first one, is the issue of canonical neurons. These neurons fire at the sight of graspable objects. Because they are also located in the inferior frontal cortex, one might be led to conclude that the increased activity we observed in the Intention clips was due to the presence of objects. Note, however, that canonical neurons do not fire at the sight of an action directed to a graspable object, even though the object is visible [ 35 ]. A strong argument against both these objections is that the activity in inferior frontal cortex is reliably different between “drinking” Intention clips and “cleaning” Intention clips, even though graspable objects were present in both conditions. In contrast, no differences in activity in the inferior frontal region were observed when “drinking” and “cleaning” clips of the Context condition were compared. Thus, the simple presence of an action embedded in a scene is not sufficient to explain the findings. Similarly, the sum of canonical and mirror neurons cannot account for the observed signal increase in the Intention condition, because this increase should be identical for both “drinking” and “cleaning.” Because “drinking” and “cleaning” contexts determined different activations in the Intention condition, it appears that there are sets of neurons in human inferior frontal cortex that specifically code the “why” of the action and respond differently to different intentions. An important issue to consider in interpreting these data is the relationship between the present results and the activity of single neurons in the activated area. On the basis of our current knowledge of physiological properties of the inferior frontal cortex, the most parsimonious explanation of the findings reported here is that mirror neurons are the likely neurons driving the signal changes in our study. This proposal needs, however, a clarification. The characteristic property of most mirror neurons is the congruence between their visual and motor properties. A neuron discharging during the execution of grasping also fires during observation of grasping done by another individual. This property cannot account for the present findings, specifically, the differences in response observed between the drinking and cleaning Intention clips. Our results suggest that a subset of mirror neurons in the inferior frontal cortex discharge in response to the motor acts that are most likely to follow the observed one. In other words, in the Intention condition, there is activation of classical mirror neurons, plus activation of another set of neurons coding other potential actions sequentially related to the observed one. This interpretation of our findings implies that, in addition to the classically described mirror neurons that fire during the execution and observation of the same motor act (e.g., observed and executed grasping), there are neurons that are visually triggered by a given motor act (e.g., grasping observation), but discharge during the execution not of the same motor act, but of another act, functionally related to the observed act (e.g., bringing to the mouth). Neurons of this type have indeed been previously reported in F5 and referred to as “logically related” neurons [ 34 ]. In that previous study, however, the role of these “logically related” mirror neurons was never theoretically discussed and their functions remained unclear. The present findings not only allow one to attribute a functional role to these “logically related” mirror neurons, but also suggest that they may be part of a chain of neurons coding the intentions of other people's actions. What are the possible factors that selectively trigger these “logically related” mirror neurons? The most straightforward interpretation of our results is that the selection of these neurons is due to the observation of an action, also coded by classical mirror neurons, in a context in which that action is typically followed by a subsequent specific motor act. In other words, observing an action carried out in a specific context recalls the chain of motor acts that typically is carried out in that context to actively achieve a goal. Another possible explanation of how mirror neurons are triggered can be related not only to the context, but also to the way in which the action is performed. It is more common to grasp the handle of the cup with a precision grip while drinking, and to use a whole-hand prehension while cleaning up. Thus, the grasp itself may convey information about the intention behind the grasping action. Although this consideration is very plausible, in general, there are reasons to believe that it is unlikely that this mechanism played a role in our study. First, in all presented grasping actions, when the handle was on the same side of the approaching hand, the grasp was always a precision grip, but when the handle was on the opposite side of the approaching hand, the grasp was always a whole-hand prehension. Thus, the hand always adopted the type of grasp afforded by the orientation of the cup, minimizing the impression that the type of grip would reflect the intentional state of the agent. Second, this hypothesis cannot explain the empirical data. In fact, in both drinking and cleaning Intention clips there was always the same number of precision grips and whole-hand prehensions. However, as Figure 4 shows, the drinking Intention entailed a much larger signal increase than the cleaning Intention. Thus, the differential brain responses in the two Intention clips cannot be explained by a possible meaning conveyed by the grasp type, and cannot even be explained by a possible “compatibility effect” between grasp type and context type (for instance, a whole-hand prehension in a context suggesting cleaning). The stronger activation of the inferior frontal cortex in the “drinking” as compared to the “cleaning” Intention condition is consistent with our interpretation that a specific chain of neurons coding a probable sequence of motor acts underlies the coding of intention. There is no doubt that, of these two actions, drinking is not only more common and practiced, but also belongs to a more basic motor repertoire, while cleaning is culturally acquired. It is not surprising, therefore, that the chain of neurons coding the intention of drinking is more easily recruited and more widely represented in the inferior frontal cortex than the chain of neurons coding the intention of cleaning. The conventional view on intention understanding is that the description of an action and the interpretation of the reason why that action is executed rely on largely different mechanisms. In contrast, the present data show that the intentions behind the actions of others can be recognized by the motor system using a mirror mechanism. Mirror neurons are thought to recognize the actions of others, by matching the observed action onto its motor counterpart coded by the same neurons. The present findings strongly suggest that coding the intention associated with the actions of others is based on the activation of a neuronal chain formed by mirror neurons coding the observed motor act and by “logically related” mirror neurons coding the motor acts that are most likely to follow the observed one, in a given context. To ascribe an intention is to infer a forthcoming new goal, and this is an operation that the motor system does automatically. Materials and Methods Participants Through newspaper advertisements we recruited 23 right-handed participants, with a mean age of 26.3 ± 6.3. Eleven participants (six females) received Implicit instructions while 12 participants (nine females) received Explicit instructions. Participants gave informed consent following the guidelines of the UCLA Institutional Review Board. Handedness was determined by a questionnaire adapted from the Edinburgh Handedness Inventory [ 36 ]. All participants were screened to rule out medication use, a history of neurological or psychiatric disorders, head trauma, substance abuse, and other serious medical conditions. Image acquisition Images were acquired using a GE 3.0T MRI scanner with an upgrade for echo-planar imaging (EPI) (Advanced NMR Systems, Woburn, Massachusetts, United States). A two-dimensional spin-echo image (TR = 4,000 ms, TE = 40 ms, 256 by 256, 4-mm thick, 1-mm spacing) was acquired in the sagittal plane to allow prescription of the slices to be obtained in the remaining sequences. This sequence also ensured the absence of structural abnormalities in the brain of the enrolled participants. For each participant, a high-resolution structural T2-weighted EPI volume (spin-echo, TR = 4,000 ms, TE 54 ms, 128 by 128, 26 slices, 4-mm thick, 1-mm spacing) was acquired coplanar with the functional scans. Four functional EPI scans (gradient-echo, TR = 4,000 ms, TE = 25 ms, flip angle = 90, 64 by 64, 26 slices, 4-mm thick, 1-mm spacing) were acquired, each for a duration of 4 min and 36 s. Each functional scan covered the whole brain and was composed of 69 brain volumes. The first three volumes were not included in the analyses owing to expected initial signal instability in the functional scans. The remaining 66 volumes corresponded to six 24-s rest periods (blank screen) and five 24-s task periods (video clips). In each scan there were two Context clips (during tea; after tea), one Action clip, and two Intention clips (drinking; cleaning) (see next section). The order of presentation of the clips was counterbalanced across scans and participants. Stimuli and instructions There were three different types of 24-s video clips (Context, Action, and Intention). There were two types of Context video clips. They both showed a scene with a series of three-dimensional objects (a teapot, a mug, cookies, a jar, etc). The objects were displayed either as just before having tea (“drinking” context) or as just after having had tea (“cleaning” context). In the Action video clip, a hand was shown grasping a cup in absence of a context on an objectless background. The grasping action was either a precision grip (the hand grasping the cup handle) or a whole-hand prehension (the hand grasping the cup body). The two grips were intermixed in the Action clip. There were two types of Intention video clips. They presented the grasping action in the two Context conditions, the “drinking” and the “cleaning” contexts. Precision grip and whole-hand prehension were intermixed in both “drinking” and “cleaning” Intention clips. A total of eight grasping actions were shown during each Action clip and each Intention clip. The participants receiving Implicit instructions were simply instructed to watch the clips. The participants receiving Explicit instructions were told to pay attention to the various objects displayed in the Context clips, to pay attention to the type of grip in the Action clip, and to try to figure out the intention motivating the grasping action in the Context clips. All participants were debriefed after the imaging session. Data processing GE image files were converted in Analyze files and processed with FSL ( http://www.fmrib.ox.ac.uk/fsl ). Brain volumes within each fMRI run were motion corrected with Motion Correction using the Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB) Linear Image Registration Tool (MCFLIRT) [ 37 ]. Spatial smoothing was applied using a Gaussian-weighted kernel of 5 mm at full-width half-maximum, and data were high-pass filtered with sigma = 15.0 s and intensity normalized. Functional images were first registered to the co-planar high-resolution structural T2-weighted EPI volume after non-brain structures had been removed with FMRIB's Brain Extraction Tool (BET) from the co-planar high-resolution T2-weighted EPI volume [ 38 ]. The co-planar high-resolution structural T2-weighted EPI volume was subsequently registered to the Montreal Neurological Institute Talairach-compatible MR atlas averaging 152 normal subjects using FMRIB's Linear Image Registration Tool (FLIRT) [ 37 ]. Statistical analyses Data analyses were performed by modeling the three conditions (Context, Action, and Intention) as stimulus functions, applying the general linear model as implemented in FSL ( http://www.fmrib.ox.ac.uk/fsl ). Statistical analyses were carried out at three levels: an individual-run level; a higher-order, multiple-runs individual-subject level; and a further higher-order intra- and inter-group comparison level. Time-series statistical analyses were carried out using FMRIB's Improved Linear model (FILM) with local autocorrelation correction [ 39 ]. Higher-level intra- and inter-group statistics were carried out using mixed effect (random effects) implemented in FLAME (FMRIB's Local Analysis of Mixed Effects) [ 40 ]. Z image statistics were performed with a threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05 [ 41 , 42 ]. The signal change displayed in Figure 4 was statistically analyzed with repeated measures ANOVA and subsequent planned contrasts. Supporting Information Figure S1 Significant Signal Changes in Subjects Receiving Implicit Instructions for Each Task Versus Rest With a threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05. (1 MB JPG). Click here for additional data file. Figure S2 Significant Signal Changes in Subjects Receiving Explicit Instructions for Each Task Versus Rest With a threshold of Z = 2.3 at voxel level and a cluster level corrected for the whole brain at p < 0.05. (1.1 MB JPG). Click here for additional data file. Table S1 Intention minus Action Local maxima in Talairach coordinates. Only the six local maxima with highest Z score in each cluster are provided in the table. Cluster size is in voxels. When the cluster encompasses more than one anatomical location, the localization given corresponds to the local maxima with the highest Z score. (82 KB PDF). Click here for additional data file. Table S2 Intention minus Context Local maxima in Talairach coordinates. Only the six local maxima with highest Z score in each cluster are provided in the table. Cluster size is in voxels. When the cluster encompasses more than one anatomical location, the localization given corresponds to the local maxima with the highest Z score. LH, left hemisphere; RH, right hemisphere; TPO, temporo-parieto-occipital. (82 KB PDF). Click here for additional data file. Table S3 Effect of Task Instructions: Explicit minus Implicit Instruction, Action Versus Rest Local maxima in Talairach coordinates. Only the six local maxima with highest Z score in each cluster are provided in the table. Cluster size is in voxels. When the cluster encompasses more than one anatomical location, the localization given corresponds to the local maxima with the highest Z score. SMA, supplementary motor area. (82 KB PDF). Click here for additional data file. Table S4 Effect of Task Instructions: Explicit minus Implicit Instruction, Context Versus Rest Local maxima in Talairach coordinates. Only the six local maxima with highest Z score in each cluster are provided in the table. Cluster size is in voxels. When the cluster encompasses more than one anatomical location, the localization given corresponds to the local maxima with the highest Z score. (82 KB PDF). Click here for additional data file. Table S5 Effect of Task Instructions: Explicit minus Implicit Instruction, Intention Versus Rest Local maxima in Talairach coordinates. Only the six local maxima with highest Z score in each cluster are provided in the table. Cluster size is in voxels. When the cluster encompasses more than one anatomical location, the localization given corresponds to the local maxima with the highest Z score. ACC, anterior cingulate cortex; VLPFC, ventrolateral prefrontal cortex. (82 KB PDF). Click here for additional data file.

D:\keerthana\PMC001xxxxxx\PMC1044835.xml

1044836

Selection on Sex Cells Favors a Recombination Gender Gap

Males and females of the same species can be strikingly different. Peacocks strut around with flashy feathers to attract mates, while peahens blend into their surroundings with more subdued colors. But differences are not always as obvious or easily explainable as in this classic example. Even the amount of genetic reshuffling that goes on during egg and sperm production differs between males and females in most species. An evolutionary reason for this has eluded researchers since the phenomenon was originally discovered in fruitflies, Chinese silk worms, and amphipods almost 100 years ago. Genetic diversity among organisms is promoted when genetic information is rearranged during meiosis, the cell division process that yields sperm and eggs (generically called gametes). During this genetic reshuffling, chromosome pairs overlap, forming structures called chiasmata (“crosses” in Greek), and physically recombine. This process does not just create diversity, it is also an example of diversity—recombination rates vary across chromosomes, sexes, and species. Male and female recombination rates differ An early 20th century hypothesis to explain the sex difference in recombination proposed that recombination is restrained within a pair of unlike sex chromosomes (X and Y, for example) and that the suppression spills over to the rest of the chromosomes. Under this idea, the sex with dissimilar sex chromosomes (XY instead of XX, for example) should be the one with the least amount of recombination in all chromosomes. But that is not always the case. Some hermaphroditic species of flatworms, for example, lack sex chromosomes altogether but still display marked differences in male and female recombination rates. In one salamander genus, more reshuffling unexpectedly occurs in the sex with two different sex chromosomes. In a new study analyzing an updated dataset of 107 plants and animals, Thomas Lenormand and Julien Dutheil bolster the argument against the recombination suppression hypothesis by showing that in species with sex chromosomes, the sex with two dissimilar sex chromosomes doesn't necessarily have a reduced recombination rate. Additionally, they found that, as a trait, the sex difference in recombination rate is not a lot more similar between two species in the same genus than between two species in different genera, suggesting that the difference evolves quickly. An alternative hypothesis suggests that sexual selection might play a role in recombination differences. Reproductive success among males is often highly influenced by selection, so mixing up successful genetic combinations in males could be evolutionarily counterproductive. But in past studies, sexual selection was not related to variation in recombination rates. Putting a new twist on this hypothesis, Lenormand and Dutheil realized that selection was not necessarily limited to the adult stage and that differences in selection among eggs or sperm might help account for recombination differences between the sexes. The authors reasoned that more opportunity for selection on sperm than egg should correspond to less recombination during sperm than egg production (and vice versa), consistent with the idea that genetic combinations surviving selection should remain more intact in the sex experiencing the strongest selection at the gametic stage. Though male gametes might be expected to be under stronger selection in many species, in true pines it seems to be the female gametes. The ovules compete with each other for resources over an entire year before being fertilized, and, indeed, from the dataset analysis, ovule production involves low recombination rates compared with male pollen in this group. In males, the opportunity for pollen competition was indirectly estimated using self-fertilization rates. The authors assumed that pollen grains competing for ovules of a self-fertilizing plant would be genetically similar and therefore experience less selection. Again, in the analysis, low selection correlated with less recombination in female gamete production, as predicted. Is selection among eggs and sperm the evolutionary force generating sex-based variation in genetic shuffling? By demonstrating that differences may be influenced by gamete selection in plants, this work has added clarity to otherwise contradictory observations.

D:\keerthana\PMC001xxxxxx\PMC1044836.xml

1044837

How the Brain Signals a Sound Source

Navigating one's environment requires sensory filters to distinguish friend from foe, zero in on prey, and sense impending danger. For a barn owl, this boils down mostly to homing in on a field mouse scurrying in the night. For a human—no longer faced with the reputedly fearsome saber-toothed Megantereon —it might mean deciding whether to fear rapidly approaching footsteps from behind on a dark, desolate street. How does the brain encode auditory space? The long-standing model, based on the work of Lloyd Jeffress, proposes that the brain creates a topographic map of sounds in space and that individual neurons are tuned to particular interaural time differences (difference in the time it takes for a sound to reach both ears). Another key aspect of this model is that the location of a sound source is encoded by the identity of responding neurons. Discriminating sound locations from neural data Evidence for local coding of auditory space has been shown in the brains of owls and in a subcortical region of small mammals, but no such map has been found in the higher centers of the mammalian auditory cortex. What's more, electrophysiological recordings in mammals indicate that most neurons show the highest response to sounds emanating from the far left or right and that few neurons show that kind of response to sounds approaching head-on—even though subjects are best at localizing sounds originating in front of them. Faced with such contrary evidence, other investigators have suggested that sound localization may rely on a different kind of code—one based on the activity distributed over large populations of neurons. In a new study, Christopher Stecker, Ian Harrington, and John Middlebrooks find evidence to support such a population code. In their alternative model, groups of neurons that are broadly responsive to sounds from the left or right can still provide accurate information about sounds coming from a central location. Although such broadly tuned neurons, by definition, cannot individually encode locations with high precision, it is clear from the authors' model that the most accurate aural discrimination occurs where neuron activity changes abruptly, that is, at the midpoint between both ears—a transition zone between neurons tuned to sounds coming from the left and those tuned to sounds coming from the right. These patterns of neuronal activity were found in the three areas of the cat auditory cortex that the authors studied. These findings suggest that the auditory cortex has two spatial channels (the neuron subpopulations) tuned to different sound emanations and that their differential responses effect localization. Neurons within each subpopulation are found on each side of the brain. That sound localization emerges from this opponent-channel mechanism, Stecker et al. argue, allows the brain to identify where a sound is coming from even if the sound's level increases, because it is not the absolute response of a neuron (which also changes with loudness) that matters, but the difference of activity across neurons. How this opponent-channel code allows an animal to orient itself to sound sources is unclear. However auditory cues translate to physical response, the authors argue that the fundamental encoding of auditory space in the cortex does not follow the topographic map model. How neurons contribute to solving other sound-related tasks also remains to be seen.

D:\keerthana\PMC001xxxxxx\PMC1044837.xml

1044838

Engineering Gene Networks to Probe Embryonic Pattern Formation in Flies

When it comes to making the perfect body, it's all about expressing the right genes in the right place at the right time. This process begins even before the sperm and egg combine to form a zygote, because maternal factors are laid down in the egg that help establish the key axes of the body. After fertilization, precisely coordinated interactions between proteins called morphogens and a network of gene regulators establish a fly's anterior–posterior axis and its pattern of segments in just three hours. In a new study, Mark Isalan, Caroline Lemerle, and Luis Serrano simulated segmentation patterning by creating a synthetic embryo and engineering an artificial version of the gap gene network, the first patterning genes expressed in the zygote. This simple system, combined with computer simulations to test network parameters, identifies significant features of the complex embryo and could do the same for other complex biological systems. An artificial network to study patterning in development One of the first molecules to act is the Bicoid protein. This morphogen is present in a concentration gradient—highest at the future head end. Different gap genes (so-called because their mutations create gaps in the segmentation pattern) respond to different levels of Bicoid, and are therefore switched on in different parts of the embryo. Expressed gap genes in turn modulate each other's activity. In the fruitfly, all of this action takes place while the embryo is a syncytium—having many nuclei but no cell membranes to separate them. Isalan et al. created a model of segmentation patterning by using a tiny plastic chamber containing various purified genes, proteins, metabolites, and cell extracts to mimic the gap gene network. Some of the genes were attached to magnetic microbeads, so that their location could be controlled by magnets anchored to the bottom of the chamber. The authors investigated a number of open questions about pattern formation, including how a morphogen diffusing from a local source generates an expression pattern along a gradient and how transcriptional repression sets pattern boundaries. After testing the system to mimic a simple network of sequential gene transcription and repression, the authors increased the components and connectivity of the network, starting with systems that had no repression interactions and moving on to systems that had different levels of cross-repression. Patterns generated by networks involving repression were much different from those generated by networks lacking repression, fitting with observations that patterning boundaries in living flies require cross-repression. But even the unrepressed system generated reproducible patterns, possibly caused by simple competition between the proteins. While such a situation likely bears little resemblance to that inside a fly egg, the authors suggest that any such competition effects would have to be tested in flies. In any case, this simplified approach can test hypotheses of how simple networks might evolve inside a cell. And since many aspects of Drosophila embryonic patterning remain obscure, these synthetic chambers will provide a powerful resource for testing different hypotheses.

D:\keerthana\PMC001xxxxxx\PMC1044838.xml

1044839

Tubeworm May Live Longer by Cycling Its Sulfur Downward

Any organism may be limited by some essential nutrient in short supply—nitrogen for a plant on poor soil, for instance, or iron for phytoplankton in the open ocean. For the long-lived tubeworm Lamellibrachia luymesi , what it needs most and has least is sulfide. L. luymesi lives clustered around hydrocarbon-releasing ocean floor seeps in the Gulf of Mexico, and it—or rather, its menagerie of internal bacterial symbionts—uses the high-energy sulfide the way plants use sunlight, extracting energy and releasing the waste products, in this case sulfate. With a lifespan of up to 250 years, L. luymesi is among the longest-lived of all animals, but how it obtains sufficient sulfide to keep going for this long has been a mystery. In this issue, Erik Cordes and colleagues propose a model in which, by releasing its waste sulfate not up into the ocean but down into the sediments, L. luymesi stimulates the growth of sulfide-producing microbes, and ensures its own long-term survival. An aggregation of Lamellibrachia luymesi in the Gulf of Mexico (Photo: Ian MacDonald) The sulfide L. luymesi needs is created by a consortium of bacteria and archaea that live in the sediments surrounding the vent. These chemoautotrophs use energy from hydrocarbons to reduce sulfate to sulfide, which L. luymesi absorbs through its unique “roots,” extensions of its body that it tunnels into the sediments. Measurements of sulfide and sulfate fluxes in the water near the vents are inconsistent with the observed tubeworm colony size and individual longevity, leading Cordes et al. to propose that L. luymesi also uses its roots to release sulfate back to the microbial consortia from which it draws its sulfide. Without this return of sulfate, the model predicts an average lifespan of only 39 years in a colony of 1,000 individuals; with it, survival increases to over 250 years, matching the longevity of actual living tubeworms. The model, which was based largely on empirical data, is relatively unperturbed by changes in hydrocarbon seep rate, or in the growth and recruitment rates for the colony. The authors note that the proposed return of sulfate into deep sediments would, in theory, increase the local rate of carbonate rock formation, creating a barrier to fluid circulation into the sediments. Their model predicts this to occur after about 50 years, in line with observed reductions in tubeworm recruitment in colonies of this age. They propose that carbonate precipitation may be inhibited if roots can also release hydrogen ions, a possibility open to further testing. Their model also explains several biogeochemical anomalies observed near tubeworm colonies, including elevated levels of highly degraded hydrocarbons and higher than predicted rates of sulfur cycling. To date, the proposed return of sulfate to the sediments through the roots is only a hypothesis—albeit one with much to support it—that still awaits direct confirmation. By providing a model in which this hypothetical interaction provides real benefits and explains real observations, the authors hope to stimulate further research into the biology of L. luymesi . For more on tubeworms, see “Microfauna–Macrofauna Interaction in the Seafloor: Lessons from the Tubeworm” (DOI: 10.1371/journal.pbio.0030102 ), also in this issue.

D:\keerthana\PMC001xxxxxx\PMC1044839.xml

1044840

Predicting the Future: Mirror Neurons Reflect the Intentions of Others

One of the more intriguing recent discoveries in brain science is the existence of “mirror neurons,” a set of neurons in the premotor area of the brain that are activated not only when performing an action oneself, but also while observing someone else perform that action. It is believed mirror neurons increase an individual's ability to understand the behaviors of others, an important skill in social species such as humans. A critical aspect of understanding the behavior of another person is recognizing the intent of his actions—is he coming to praise me or to bury me? In this issue, Marco Iacoboni and colleagues use functional magnetic resonance imaging (fMRI) to show that the mirror neuron system tracks not only the actions, but also the intentions, of others. The researchers presented subjects with one of three types of movie clips, “context,” “action,” and “intention.” The “context” clip came in two versions. In the first, a mug of tea, a teapot, a pitcher of cream, and a plate of cookies sit neatly on a nondescript surface. In the second, the mug is empty, the pitcher is on its side, and a napkin lies crumpled beside scattered cookie crumbs—the dregs of an apparently well-enjoyed snack. The “action” clip shows only an empty mug, being grasped either by the whole hand around the rim (called whole-hand prehension), or with the fingers on the handle (precision grip). The “intention” clip puts it all together, providing the context needed to understand the intent of the action. In the first context scene, the mug is grasped in the context of a well-laid spread, signaling intent to drink the tea. In the second, the mug is grasped against the backdrop of the rest of the tea-time dregs, signaling the intent to clean up. In both cases, the type of grasp used is alternated, to avoid implying intent merely based on grip type. The essential interpretive technique in fMRI is to subtract the activation patterns from two different stimuli, thereby highlighting brain regions that are activated differentially in response to the difference in stimulus. In this study, “intention minus context” showed areas involved in recognizing both action and intention, while “intention minus action” showed areas for both context and intention. Comparing these two results, Iacoboni and colleagues found that an area in the right inferior frontal cortex, an area known to be involved in the mirror system, was activated only by scenes in which intention could be inferred. In the authors' words, “it appears that there are sets of neurons in human inferior frontal cortex that specifically code the ‘why’ of the action and respond differently to different intentions.” They note that these neurons differ in function from previously defined mirror neurons in that they apparently code not current actions, but some aspect of future ones. In this interpretation, an action observed within a familiar context activates mirror neurons for “logically related” actions, those that most likely will follow the observed one. This suggests the mirror neuron system is intimately involved not only with understanding the behavior of others, but predicting it as well.

D:\keerthana\PMC001xxxxxx\PMC1044840.xml

1044841

An Evolutionary Road Less Traveled: From Farming to Hunting and Gathering

Invested with the arguably unique capacity for self-reflection, humans may well have asked the question, “Where did we come from?” ever since the dawn of self-awareness. From this universal question come origin stories as diverse as the cultures who tell them. In some cases, little is known about a population's evolutionary history aside from these stories—such is the case for the Mlabri people of Southeast Asia. Until expanding agricultural development and modernization encroached on their forest homelands, the Mlabri lived mostly as nomadic hunter–gatherers in the forests of northeastern Thailand and western Laos. This lifestyle is unique among the other so-called hill tribes of Thailand—who all farm—raising the possibility that the Mlabri descended from the ancient Hoabinhian hunting–gathering culture of Southeast Asia and practice a way of life that predates agriculture. The Mlabri of Thailand: Their history holds a lesson for anthropologists (Photo: Takafumi Ishida) Scant historical information exists on Mlabri language, culture, and origin, but Mlabri traditions speak to a long history as hunter–gatherers. The oral traditions of a neighboring hill tribe, the Tin Prai, paint a slightly different picture: several hundred years ago, legend has it, Tin Prai villagers sent two banished children downriver on a raft; the children, who survived by foraging in the forest, became the first Mlabri. In a new study, Mark Stoneking and colleagues use the tools of molecular anthropology to investigate the agricultural versus hunting–gathering origin of the Mlabri and reveal a scenario remarkably similar to the traditional origin stories. The notion that genetic analyses can shed light on this question, the authors explain, comes from a body of research indicating that hunting–gathering groups have a lower level of genetic diversity and a higher frequency of unique mitochondrial (mtDNA) sequence types than neighboring agricultural groups. In this study, Stoneking and colleagues compared the genetic diversity of the Mlabri with that of six other agriculture-based hill tribes by analyzing specific regions of each population's mtDNA, Y chromosomes, and autosomes (non-sex chromosomes). mtDNA and Y chromosomes can help uncover clues to evolutionary origins because both are in effect haploid systems (i.e., there is only one copy of the Y chromosome and a lot of identical copies of mtDNA present in each cell), and so do not undergo recombination. This in turn means that observed genetic variations likely result from random mutation—which is assumed to occur at a predictable rate—allowing scientists to estimate the age of the genetic variation found in a population. The mtDNA analysis revealed something remarkable: all the Mlabri mtDNA sequences were identical. Not only did all of the other hill tribes show “significantly higher” variation, but this lack of variation hasn't been found in any other human population. The Y-chromosome and autosome analyses revealed the same reduced diversity, indicating a “severe reduction in population size” for the Mlabri. This reduction likely happened 500 to 800 years ago, Stoneking and colleagues conclude, and at most 1,000 years ago. But how? Since genetic analyses can't distinguish between a population bottleneck and a founding event, the authors used simulations to calculate the amount of population reduction required to completely eliminate mtDNA diversity, arriving at “not more than two unrelated females” and “perhaps even only one.” But were the first Mlabri farmers or hunter–gatherers? Unlike other hunting–gathering groups, the Mlabri share closely related mtDNA, autosomal, and Y-chromosome sequences with both the agriculture-practicing hill tribes and other agricultural groups in Southeast Asia. Linguistic studies suggest that the Mlabri language arose after speakers of a related language, probably Tin, split off and came into contact with another, as yet unknown language, an event that likely happened less than 1,000 years ago. The genetic and linguistic evidence indicates that the Mlabri were “founded” between 500 to 1,000 years ago by a single maternal lineage and one to four paternal lineages from an agricultural culture. With too few hands to farm, this tiny group likely turned to hunting and gathering. Altogether, Stoneking and colleagues conclude, these findings caution against automatically assuming that contemporary hunter–gatherer groups “represent the pre-agricultural lifestyle of human populations, descended unchanged from the Stone Age.” Interestingly, the authors' scenario of Mlabri origins is not so different from the story told by the Tin.

D:\keerthana\PMC001xxxxxx\PMC1044841.xml

1054879

Cell-by-Cell Dissection of Gene Expression and Chromosomal Interactions Reveals Consequences of Nuclear Reorganization

The functional consequences of long-range nuclear reorganization were studied in a cell-by-cell analysis of gene expression and long-range chromosomal interactions in the Drosophila eye and eye imaginal disk. Position-effect variegation was used to stochastically perturb gene expression and probe nuclear reorganization. Variegating genes on rearrangements of Chromosomes X, 2, and 3 were probed for long-range interactions with heterochromatin. Studies were conducted only in tissues known to express the variegating genes. Nuclear structure was revealed by fluorescence in situ hybridization with probes to the variegating gene and heterochromatin. Gene expression was determined alternately by immunofluorescence against specific proteins and by eye pigment autofluorescence. This allowed cell-by-cell comparisons of nuclear architecture between cells in which the variegating gene was either expressed or silenced. Very strong correlations between heterochromatic association and silencing were found. Expressing cells showed a broad distribution of distances between variegating genes and their own centromeric heterochromatin, while silenced cells showed a very tight distribution centered around very short distances, consistent with interaction between the silenced genes and heterochromatin. Spatial and temporal analysis of interactions with heterochromatin indicated that variegating genes primarily associate with heterochromatin in cells that have exited the cell cycle. Differentiation was not a requirement for association, and no differences in association were observed between cell types. Thus, long-range interactions between distal chromosome regions and their own heterochromatin have functional consequences for the organism.

Introduction From the broad level of the whole chromosome down to the individual gene, interphase chromosomes in every organism studied adhere to common organizational principles (reviewed in [ 1 ]). An aspect of chromosome structure important for organism function is long-range chromosomal interactions (LRCIs) between distant loci. LRCIs have been linked with gene silencing by insulators in Drosophila [ 2 , 3 ] and with Polycomb silencing of homeotic genes [ 4 , 5 ]. LRCIs between euchromatic loci and heterochromatin can silence genes (reviewed in [ 6 , 7 ]). LRCIs are not static, for example, the polar organization of Drosophila embryonic chromosomes changes as homologous loci pair and LRCIs within the nucleus form [ 8 , 9 , 10 , 11 ]. Mouse immune cells adopt unique contacts between silenced genes and heterochromatin during differentiation and cell fate specification [ 12 , 13 , 14 ]. These changes appear to be functional rather than merely structural, such that altering LRCIs appears to have profound biological consequences. Live studies of green-fluorescent-protein-tagged chromosomal loci reveal how LRCIs can change. Individual loci exhibit Brownian motion constrained to a defined volume, as observed in yeast [ 15 , 16 , 17 ], mammalian cells [ 15 , 18 ], and Drosophila [ 17 , 19 ]. Constraints are under developmental and cell cycle control, as evidenced by the observation that individual loci in male Drosophila pre-meiotic spermatocyte nuclei are more tightly confined in late G2 than in early G2 [ 19 ]. Relaxing constraints to allow considerable motion permits new LRCIs to form, while constraining loci more tightly can stabilize them. Developmental control of locus confinement could reconfigure a basic polar chromosomal organization into relatively stable developmental and cell-fate-specific architectures. Drosophila position-effect variegation (PEV) is an ideal system to study the functional consequences of altered LRCIs (reviewed in [ 20 , 21 ]). PEV occurs when chromosome rearrangements juxtapose euchromatic genes and heterochromatin, producing a variegated expression pattern such that the gene is silenced in some but not all cells. These rearrangements also cause the affected genes to form long-range interactions with heterochromatin in a subset of cells [ 9 , 22 ]. Genetic evidence suggests that PEV may utilize these long-range interactions to silence genes. PEV can skip over one gene to silence another [ 23 ] or silence a wild-type locus on a homologous chromosome [ 9 , 22 , 24 ]. In the case of bw D variegation, chromosome rearrangements that alleviate PEV move the affected gene farther away from heterochromatin, while rearrangements that move the locus closer to heterochromatin enhance PEV [ 25 , 26 ]. This suggests that juxtaposition between a gene and heterochromatin allows for gene-to-heterochromatin interactions that can cause silencing. Once formed, these contacts may cause a gene to be silenced either by repackaging the gene into heterochromatin or by a specific silencing activity sequestered within heterochromatin itself. Interaction with heterochromatin does seem to correlate with the silencing of specific genes, but the connection between association with heterochromatin and silencing has not been directly verified. Fluorescence in situ hybridization (FISH) techniques to identify chromosomal and heterochromatic loci are not generally compatible with the detection of gene expression. Furthermore, studies that have examined the connection between LRCIs and silencing by modulating the amount of PEV-induced repression have given conflicting results: one study found a correlation between relaxed silencing and relaxed association [ 22 ], whereas another did not [ 24 ]. Because the affected gene's expression was not compared to its association with heterochromatin on a cell-by-cell basis, it remains unclear whether, in a given cell, a heterochromatin-associated locus was silenced or expressed. For the first time, to our knowledge, we present an experimental system that compares the expression of a variegating gene and its association with heterochromatin on a cell-by-cell basis for three different variegating genes in Drosophila whole-mount tissues. Multiple lines were chosen to ensure that results could be generally applied to PEV, rather than being limited to a specific rearrangement. The positions of variegating chromosomal loci and regions of heterochromatin were probed by FISH while fluorescent detection of eye pigments or variegating gene proteins marked gene expression. The affected gene in each line is quite far (>10 MB) from the centromere, with a block of heterochromatin placed nearby either through insertion or inversion. This provided an easy assay for long-range interactions with centromeric heterochromatin, as an interacting gene would relocate a significant distance across the nucleus to interact with centromeric regions. The variegating genes used are expressed in the Drosophila eye or eye imaginal disk; therefore, our examinations were limited to these tissues. Cell types, developmental stages, and cell cycle states were identified to determine what effects these variables might have on LRCIs with heterochromatin. Our results show that interactions between variegating loci and heterochromatin are tightly correlated with gene silencing. Cells in which the variegating gene has been silenced exhibit a tight distribution of distances, suggesting direct interaction between heterochromatin and the distally located gene. Expressing cells, however, show a much broader distribution, with far fewer interactions between variegating loci and heterochromatin. Furthermore, these interactions with heterochromatin are primarily found in non-dividing cells in the eye disk. Differentiation is not required for LRCIs to occur, and interactions did not differ between cell types. Results LRCIs Are Strongly Correlated with Gene Silencing Using gene expression as a functional assay, the consequences of LRCIs between chromosomal loci and heterochromatin were probed on a cell-by-cell basis. The first variegating gene examined was the brown gene on the bw D rearrangement. Expression of brown RNA has only been detected in recently eclosed adult heads; therefore, gene expression experiments were confined to this tissue [ 27 ]. Our experiments were greatly simplified by working in a scarlet background as in [ 26 ] as this eliminated the brown pigments from the eye. Therefore brown -expressing cells contained red pteridine pigment while non-expressors contained none ( Figure 1 ). Figure 1 Gene Expression and LRCIs Color images show combined FISH–gene expression results for each of the three lines studied. (A and B) bw D adult eye tissue. brown -expressing cells with pigment are blue, DAPI is red, single-copy FISH marking the brown locus is white, and heterochromatic single is green. (A) Wider view of eye. (B) Close-up view showing two expressing cells at right and lower left, and one silenced cell at the lower right. (C and D) In(3L)BL1 eye disk tissue. Blue is anti-beta-galactosidase staining, green is anti-lamin staining, white is single copy FISH, red is heterochromatic FISH signal. (C) Wide-angle view of disk. (D) Close-up view showing silenced cells and expressing cells (blue in nucleus). (E and F) In(1) rst 3 eye disk tissue. Blue is anti-White staining, green is anti-lamin, red marks heterochromatic FISH signals, and white is single-copy FISH to the white gene. (E) Wide-angle view of disk. (F) Close-up showing expressing cell ringed by White protein and silenced cell nearby with no staining. Gene-to-heterochromatin distance distributions from brown -expressing cells were quite different from those of silenced cells ( Figure 2 ). The distribution of distances from expressing cells was broad, centered around 1 μm and extending to about 2 μm ( Figure 2 A). Looking at the cumulative percentage plots, few nuclei (10%) had variegating-gene-to-heterochromatin distances shorter than 0.5μm, and barely 25% had distances shorter than 0.75 μm ( Figure 2 G, blue line). Silenced cells, however, had a tight distribution of distances centered around 0.5 μm, with virtually all variegating-gene-to-heterochromatin distances being under 1 μm ( Figure 2 D). Cumulative percentage plots show nearly 40% of distances were shorter than 0.5 μm, and nearly 80% of nuclei had distances shorter than 0.75 μm ( Figure 2 G, red line). The Mann–Whitney U test revealed that these differences were highly significant. ( p < 0.0001; Table 1 ). Figure 2 Silenced Genes Associate Tightly with Heterochromatin Physical distances in microns between heterochromatic sequences near each chromosome's centromere and the variegating gene were measured and sorted based on expressing or non-expressing cells. (A–C) Expressing cells. (D–F) Silenced cells. (G–I) percentile plots of (A–F), in which blue represents expressing cells, red represents silenced cells, and green represents wild-type nuclei. bw D nuclei are Su( bw D )208 cells in adult eyes; In(1) rst 3 and In(3L)BL1 data are from imaginal eye disks. Table 1 Summary of Reported Results The second variegating gene examined was the white gene in the In(1) rst 3 . Antisera developed against the unique N-terminus of the White protein (see Figure S1 and part 1 of Protocol S1 ) stained white -expressing cells in In(1) rst 3 disks as expected from their eye pigmentation ( Figure 1 E and 1 F). LRCIs between white and the 1.688 satellite correlated strongly with PEV-mediated silencing of white ( Figure 2 B, 2 E, and 2H). As before, cells expressing the variegating gene had a broad distribution of distances, in this case centered at 1.5 μm and continuing to 4 μm ( Figure 2 B). Cumulative percentage plots reveal that only 20% of nuclei expressing white had variegating-gene-to-heterochromatin distances under 1 μm ( Figure 2 H, blue line). The distance distribution for silenced cells was much narrower than that for expressing cells, with distances centered around 0.8 μm and extending to only 2 μm in total distance ( Figure 2 E). Cumulative percentage plots show that over 60% of nuclei from silenced cells had variegating-gene-to-heterochromatin distances of under 1 μm ( Figure 2 H, red line) compared to 20% for expressing cells. Differences between expressing and silenced populations of cells were highly significant ( p < 0.0001; Table 1 ). Despite the marked differences in association levels between expressing and silenced cells, the distributions of distances overlapped considerably between expressing and non-expressing cells ( Figure 2 B and 2 E). The third variegating gene examined was the heat-shock inducible beta-galacosidase transgene located on the In(3L)BL1 rearrangement. The expression of beta-galactosidase in In(3L)BL1 showed no preference for cell type, and was only found behind the morphogenic furrow ( Figure 1 C and 1 D) [ 28 , 29 ]. In(3L)BL1 nuclei showed patterns of interaction and expression similar to those seen in the brown and white variegating lines. Expressing cells showed a broad distance distribution centered around 1.5 μm and extending out to nearly 4 μm ( Figure 2 C). Cumulative percentage plots show barely 15% of expressing cells had variegating-gene-to-heterochromatin distances 1 μm or less ( Figure 2 I, blue line). In(3L)BL1 silenced nuclei displayed a rather narrow distribution, with a median distance of under 1 μm ( Figure 2 F). As in In(1) rst 3 disks, cumulative percentage plots also showed more than 60% of silenced nuclei had distances between the silenced gene and heterochromatin under 1 μm ( Figure 2 I, red line). The narrow distribution of distances meant that silenced nuclei rarely had variegating-gene-to-heterochromatin distances greater than 2 μm, while expressing cells had greater distances in at least 50% of cells. Also similar to In(1) rst 3 , variegating genes in silenced cells were not 100% associated with heterochromatin ( Figure 2 C and 2 F), with 40% of silenced nuclei having variegating-gene-to-heterochromatin distances greater than 1 μm. Additionally, some expressing cells had variegating-gene-to-heterochromatin distances less than 1 μm, in this case roughly 10%. Interaction between Variegating Genes and Heterochromatin Primarily Occurs in Non-Dividing Cells While it was clear that silencing of a variegating gene was tightly correlated with its interaction with heterochromatin, it was not known what other factors might affect this interaction. A study of how cell cycle progression and differentiation might affect interactions between a variegating gene and heterochromatin was undertaken. The cell cycle had previously been implicated as a force that periodically disrupts LRCIs [ 9 , 30 ]. The third-instar eye imaginal disk is an excellent tissue to explore this possibility because it contains well-separated populations of dividing and non-dividing cells. The anterior portion of the eye imaginal disk contains dividing cells ( Figure 3 A [left of dotted line] and 3 B). The morphogenic furrow contains cells arrested in G1. Posterior to the morphogenic furrow, many cells cease dividing and differentiate ( Figure 3 A [right of dotted line] and 3 C). Because of our sample preservation methods (see Materials and Methods ), it was also possible to distinguish each individual photoreceptor cell and cone cell by their unique three-dimensional position and nuclear shape ( Figure 3 D; [ 31 ]). Our results therefore contain three groups of nuclei: nuclei from dividing cells anterior to the morphogenic furrow and differentiated cone cell nuclei and photoreceptor cell nuclei from behind the morphogenic furrow. Figure 3 Preservation of Tissue Architecture and Cell-by-Cell Identification in the Drosophila Eye Disk bw D eye disks were processed for FISH and stained for nuclear lamin as described in Materials and Methods . Red, fluorescein lamin stain; blue, AACAC-cy5 heterochromatic probe; green, rhodamine-labeled P1 probe covering brown . (A) Low-magnification view of eye disk showing anterior (left) and posterior (right) portions of the disk. Morphogenic furrow is marked with a dotted line. (B) View of cells anterior to morphogenic furrow. These cells are still dividing and have not yet undergone differentiation. (C) Cells posterior to morphogenic furrow. These cells have ceased dividing and are differentiating into adult eye structures. (D) Four cone cell clusters from the two boxes in (C), with each photoreceptor and cone cell identified. Apical cells are at the top, basal at the bottom. bw D LRCIs were dramatically different between dividing and non-dividing cells in bw D disks ( Figure 4 ). Cells anterior to the morphogenic furrow ( Figure 4 A) had a roughly Gaussian distribution of distances. The cumulative percentage plots show that few nuclei had less than 1 μm distance between the variegating gene and heterochromatin (<10%) ( Figure 4 J, red line). Figure 4 Association Primarily Occurs in Non-Dividing Cells Physical distances in microns between heterochromatic probe signals and P1 probe signals were measured for each line and sorted based upon their position relative to the morphogenic furrow and cell fate. (A–C) Cells anterior to morphogenic furrow. (D–F) Differentiated cone cells behind the morphogenic furrow. (G–I) Differentiated photoreceptor cells behind the morphogenic furrow. (J–L) Percentile plots for histograms in (A–I). Blue, wild-type cells; red, anterior cells; green, cone cells; pink, photoreceptor cells. Entirely different results were found for cone and photoreceptor cells behind the morphogenic furrow. The distributions in both cases were substantially skewed, with a long tail of distances representing cells in which brown was greater than 1 μm from heterochromatin ( Figure 4 D and 4 G). Cumulative percentage plots reveal that interactions were far more common in differentiated cells behind the morphogenic furrow than in the dividing anterior cells ( Figure 4 J, purple and green lines). The majority of nuclei in both cone cells and photoreceptor cells had their variegating gene closer than 1 μm to the centromeric heterochromatin on 2R. Statistical tests showed that the distributions of cone cells and photoreceptor cells were significantly different from dividing anterior cells ( p < 0.0001; Table 1 ) but were not significantly different from one another ( p = 0.15). This argues that differences between cell types do not have a significant effect upon brown -to-centeromeric-heterochromatin LRCIs. Cone cells and photoreceptor cells are also different in shape and nuclear volume. The fact that no difference in heterochromatic association was found suggests that in these cell types nuclear shape and volume do not play a meaningful role in LRCIs with heterochromatin. In(1) rst 3 The inverted X chromosome In(1) rst 3 displayed similar results to that of bw D . Gene-to-heterochromatin distance distributions in anterior cells were nearly Gaussian ( Figure 4 B), such that few nuclei had variegating genes closer than 1 μm to the centromeric heterochromatin (<10%; Figure 4 K, red line). Nuclei posterior to the morphogenic furrow displayed a skewed distribution similar to bw D ( Figure 4 E and 4 H). Cone cells and photoreceptor cells displayed nearly identical distributions and were not statistically distinct ( p = 0.0319). These distributions are markedly different from those of anterior cells ( Figure 4 B), with most nuclei exhibiting a shorter variegating-gene-to-heterochromatin distance ( Figure 4 E and 4 H). The cumulative percentage plots confirm this ( Figure 4 K, purple and green lines), in that posterior cells routinely showed association levels 40% higher than anterior cells. In(3L)BL1 The line In(3L)BL1 was a notable exception to previous results as it had an almost bimodal distribution of distances in cells anterior to the morphogenic furrow ( Figure 4 C). The cumulative percentage plots show dividing nuclei with variegating-gene-to-heterochromatin distances less than 1 μm (25%; Figure 4 L, red line). This may be due to confounding effects of the multiply inverted balancer TM3. Gene-to-heterochromatin distance distributions from differentiated cells were not substantially skewed towards shorter distances relative to anterior cells ( Figure 4 F and 4 I). There were few noticeable differences, as shown in cumulative percentage plots ( Figure 4 L, purple and green lines relative to blue). These differences were not statistically significant between cone and anterior cells ( p = 0.0567; see Table 1 ) nor between photoreceptor and anterior cells ( p = 0.6600; Table 1 ). Distributions from cone and photoreceptor cells were not significantly different ( p = 0.0184; Table 1 ; Figure 4 F and 4 I), nor were there any noticeable differences between the cell types in the cumulative percentage plots ( Figure 4 L, purple and green lines). Data from each variegating locus were compared with the behavior of loci on wild-type chromosomes; variegating loci behaved quite differently from those on wild-type chromosomes (see Table S1 and part 2 of Protocol S1 ). Differentiation Is Not Necessary for Association Because LRCIs were found in differentiated nuclei, this suggested that differentiation was important for interaction with heterochromatin. Alternatively, cessation of the cell cycle may allow loci to interact with heterochromatin because the periodic anaphases are eliminated [ 9 ]. The unique cell cycle profile of the eye imaginal disk presented an elegant way to distinguish between these two possibilities. Eye disk cells anterior to the morphogenic furrow divide in a band (I in Figure 5 A and 5 B) and arrest in the furrow at G1, as cartooned in Figure 5 A and 5 B. Posterior to the furrow, cells either differentiate into ommatidia or replicate their DNA. Some cells in this undifferentiated pool divide in the second mitotic wave (II in Figure 5 A and 5 B) and differentiate immediately. The remaining cells pause in G2 and await a signal to divide [ 32 ]. These G2-arrested cells behind the morphogenic furrow have been in interphase since the first mitotic wave ([ 32 ]; Figure 5 A, shaded nuclei). If interaction requires differentiation, these G2-arrested cells should show less association than differentiated cells. If these G2 cells show equivalent levels of association, then a sufficiently long cell cycle permits variegating genes to interact with heterochromatin. G2 cells are easy to identify in third-instar eye disks as their nuclei react to anti–cyclin A antisera and are basal to differentiated and dividing nuclei ( Figure 5 D and 5 E; [ 33 , 34 ]). G1-arrested cells are also clearly identified as a band of unstained nuclei between the anterior and posterior portions of the eye disk ( Figure 5 C). Figure 5 Identification of Cell Cycle Stages in Drosophila Eye Disks A combination of immunofluorescence to cyclins and tissue architecture was used to identify cells in specific cell cycle states. (A) Cartoon of eye disk that shows populations of dividing and cell-cycle-arrested cells. I, first mitotic wave; II, second mitotic wave. G2-arrested cells are filled nuclei located basally posterior to the furrow. (B) Cartoon of same disk viewed from above apical side of disk. Anterior is to the left. (C–E) Immunofluorescence against G2 cyclins identifies cells in G1 and G2 arrest. Red is anti–cyclin A, green is single-copy FISH signal, and blue is heterochromatic FISH signal. (C) G1-arrested cells can be identified as those near and in the morphogenic furrow that do not show G2 cyclin staining. (D) G2-arrested cells posterior to the morphogenic furrow show dense anti–cyclin A staining. (E) Close-up of cells shown in (D). G2-arrested cells in the bw D line showed variegating-gene-to-heterochromatin distances similar to those of differentiated cells ( Figure 6 ). As in differentiated cells, the majority of G2-arrested nuclei had distances between the variegating gene and heterochromatin less than 1 μm. As shown in the cumulative percentage plots, G2-arrested nuclei were nearly indistinguishable from differentiated nuclei ( Figure 6 M, yellow and purple lines). These two distributions were also similar by statistical tests ( p > 0.2; Table 1 ). Similar results were seen with In(1) rst 3 and In(3L)BL1 G2-arrested nuclei, such that the gene-to-heterochromatin distance distributions in arrested nuclei were indistinguishable from those of differentiated cells ( Figure 4 H and 4 I compared to Figure 6 K and 6 L). Cumulative percentage plots likewise do not show any differences between arrested nuclei and differentiated ones ( Figure 6 N and 6 O), and these distributions were not statistically distinct ( p > 0.1; Table 1 ). Figure 6 Differentiation Is Not Required for Association Physical distances in microns between heterochromatic probe signals and P1 probe signals were measured for each line and sorted based upon staining for G2 cyclins. (A–C) Cells anterior to morphogenic furrow. (D–F) G1-arrested cells (in furrow, no cyclin staining). (G–I) G2-arrested cells, basal and stained for cyclins. (J–L) Differentiated cells behind the morphogenic furrow. (M–O) Percentile plots for (A–L). Blue, wild-type cells; red, anterior cells; green, G1-arrested cells; pink, G2-arrested cells; yellow, differentiated cells. bw -to-heterochromatin distances in bw D G1-arrested nuclei displayed a bimodal distribution ( Figure 6 D), unlike the unimodal distributions seen for anterior cells ( Figure 6 A) and differentiated cells ( Figure 4 J). One peak shows a distance distribution similar to that seen for G2-arrested and differentiated cells ( Figure 6 G and 6 J), while a second is intermediate between that of differentiated and anterior nuclei. Cumulative percentage plots show a marked difference between G1-arrested nuclei and all other populations, including an inflection in the curve between the two peaks in the G1 distribution ( Figure 6 M, red line). It appears that the morphogenic furrow is a transition state, whereby variegating genes that were once quite far from heterochromatin ( Figure 6 A) begin to associate with centromeric regions on the same chromosome. Similar behavior is also seen in In(1) rst 3 and In(3L)BL1 nuclei ( Figure 6 E and 6 F). Distributions in these two lines are not bimodal, but they are intermediate between anterior and differentiated nuclear distributions. Cumulative percentage plots also show distinctions between these nuclei and the other populations ( Figure 6 N and 6 O, red lines). Statistical tests showed that all three distributions of G1 nuclei were distinct from anterior cells ( p < 0.0001; Table 1 ), but that only bw D G1 nuclei were different from differentiated cells ( p < 0.0001; Table 1 ). Variegating Genes Associate with Heterochromatin at Their Own and Other Centromeres An appreciable percentage of loci were found far from centromeric cis- heterochromatin, even in PEV-silenced nuclei. These silenced yet far loci may form LRCIs with heterochromatin on other chromosomes ( trans -heterochromatin). While other authors found that variegating rearrangements involving bw preferentially associated with cis- heterochromatin [ 9 , 35 ], we felt this issue merited reexamination. In two sets of experiments per line, probes marking variegating loci were compared with two different heterochromatic probes: one marked cis- heterochromatin while another identified trans -heterochromatin. Each variegating gene was examined for promiscuous interaction with centromeric heterochromatin on each large chromosome. Data were sorted based on whether the variegating gene was within 1 μm of heterochromatin on its own chromosome ( Figure 7 A), on a different chromosome ( Figure 7 B), or on both chromosomes ( Figure 7 C). Figure 7 Interaction of Variegating Genes with Heterochromatin on Other Chromosomes All three lines were examined to see how often variegating genes interacted with heterochromatin on their own and other chromosomes. (A) Close-up of In(1) rst 3 nucleus showing variegating gene interaction with heterochromatin on its own chromosome. (B) Close-up of In(1) rst 3 nucleus showing variegating gene interaction with heterochromatin on a different chromosome. (C) In(1) rst 3 nucleus showing variegating gene interaction with heterochromatin on both its own and another chromosome. (D–F) Bar plots of interaction frequencies of variegating genes with their own heterochromatin and that found on other chromosomes. A distance of 1.0 μm was chosen as the maximum distance for interaction to occur. As in Dernburg et al. [ 9 ], the brown locus on the bw D chromosome interacted primarily with heterochromatin on the same chromosome ( Figure 7 D). Interactions with trans -heterochromatin on Chromosomes X and 3 where the brown gene was not associating with cis -heterochromatin were generally limited to 5%–10% of examined nuclei. Interactions where the brown gene was close to heterochromatin on multiple chromosomes were also infrequent, between 2%–5% of examined nuclei. The In(3L)BL1 chromosome differed from the bw D chromosome in that comparable levels of association were observed between cis- and trans -heterochromatin ( Figure 7 E). The In(3L)BL1 chromosome's transgene interacted similarly with heterochromatin on Chromosomes 2 and 3 (20% of nuclei examined), but only 10% with the X chromosome. Interactions of the variegating gene with heterochromatin on multiple chromosomes were less common, at around 5%–10% of nuclei examined. The variegating locus on In(3L)BL1 seemed to show a preference for associating either with heterochromatin on its own chromosome or that of Chromosome 2. The In(1) rst 3 line showed similar levels of association with heterochromatic blocks on all three chromosomes, with around 10% of nuclei exhibiting distances less than 1 μm ( Figure 7 F). Unlike in the other lines examined, the variegating gene was rarely close to heterochromatin on multiple chromosomes, suggesting that the white locus interacts with only one block of heterochromatin at a time. While the bw D chromosome exhibited a marked preference for associating with heterochromatin on its own chromosome, the other two lines did not. This suggests that the tails of the distances seen in the silenced cells of In(1) rst 3 and In(3L)BL1 could well be explained by interaction with heterochromatin on other chromosomes. bw D ' s infrequent association with blocks of heterochromatin on other chromosomes may explain the smaller tails seen in the distributions of bw D disk nuclei. Discussion The variegating rearrangements studied here cause a nuclear reorganization with clear consequences for the organism. The rearrangements alter the behavior of affected loci relative to wild-type (see Table S1 and part 2 of Protocol S1 ), changing their nuclear position. This in turn increases the likelihood that nearby genes will contact and form persistent interactions with heterochromatin. Those loci that do interact with heterochromatin may be silenced. Supporting this conclusion is the observation that variegating-gene-to-heterochromatin distance distributions in silenced cells are vastly different from those of expressing cells ( Figure 2 D– 2 F). Silenced nuclei show a tight unimodal distribution with a peak centered around short distances, 0.5 μm in the case of bw D , and 0.8 μm in the case of In(3L)BL1 and In(1) rst 3 . We posit that the lower variability of these silenced cell distributions means that variegating genes are forming persistent contacts with heterochromatin. Because the same level of association was seen between recently differentiated nuclei and those that had differentiated as many as 12–18 h before (see Figure S2 and part 5 of Protocol S1 ), this suggests that the interactions are quite persistent and do not appear and disappear periodically. Distances of nearly a micron may seem inordinately high. However, since the analysis scheme measured between FISH signal centers, distances will never be zero even when FISH signals overlap considerably. This is compounded by the relatively large volume occupied by heterochromatic FISH signals. However, FISH signals that we classified as interacting based on distance were often touching or partially overlapping ( Figure 1 ). Unlike silenced cells, expressing cells have a broader, nearly Gaussian distribution centered at 1.0–1.5 μm and extending to nearly 4 μm ( Figure 2 A– 2 C). We interpret the more variable distribution to mean that expressing loci are not interacting with heterochromatin and are therefore less restricted. Despite clear differences between expressing and silenced loci behavior, we do not see absolute distinctions between them. Appreciable numbers of silenced loci are far from cis- heterochromatin, while some expressing loci are quite close. Some of these results may be explained by limitations of our experimental procedures. Not all centromeric heterochromatin is labeled by our FISH methods. It remains possible that a “far silenced” gene may be interacting with heterochromatin not labeled by our oligonucleotide probes in either cis- or trans -heterochromatin. Two lines exhibited significant association with trans -heterochromatin, suggesting that this could explain cases where silenced genes are not associated with labeled cis- heterochromatin. Some of the expressing yet heterochromatin-close cases may have expressed the gene for some time before contact with heterochromatin silenced the variegating locus. Our use of accumulated gene product to mark gene expression would mask these examples. While FISH methods do not drastically perturb nuclear structure, these methods may impact interloci distances somewhat [ 36 , 37 ]. Heat shock utilized for the In(3L)BL1 line may explain its unique behavior relative to other chromosomes, as heat shock has been shown to affect nuclear structure [ 2 ]. A live chromosomal imaging system [ 15 ] that tracks the long-range interactions of a variegating gene with heterochromatin should provide the best control for sample manipulations discussed here. Specific biological interactions may also explain the “far silenced” and “close expressing” loci. Interaction with heterochromatin may be required for the establishment but not the maintenance of silencing. Once a gene is silenced, chromosomal motion may pull it away from centromeric heterochromatin without changing the affected gene's expression. A second possibility is that the proximal block of heterochromatin near the variegating locus can occasionally silence the gene by itself. For example, the block of heterochromatin near the white gene in In(1) rst 3 is known to be a weak silencer [ 38 ]. The observation that variegating loci in some expressing cells seem to interact with heterochromatin (upper limit of 20%) may mean that interaction does not guarantee silencing. Early transcriptional activation allows a gene to escape PEV, arguing that once a gene is expressed, future interactions with heterochromatin might not silence it [ 39 ]. The timing of interaction relative to the normal expression pattern of the gene may therefore be critical. Some genes vary in their sensitivity, such that interaction with heterochromatin may not necessarily result in silencing [ 40 , 41 ]. Based on Figure 2 and the results in Table 1 , we can determine what the differences between expressing and silenced cells mean in terms of chromosomal and nuclear dimensions. In silenced cells, the median distance between the variegating locus and heterochromatin is nearly halved relative to expressing cells. For example in In(1) rst 3 the distance falls from around 1.5 μm in expressing cells to 0.85 μm in silenced ones. Based upon calculations of linear base pair distance, this is what we would expect if the silenced locus is located 9 MB closer to heterochromatin than in the expressing case, or about half of a chromosome arm. Given the complex folding of an interphase chromosome arm, this reduction is likely more pronounced than a linear calculation would suggest. Additionally, the behavior of a silenced locus is different in ways far more dramatic than merely “closeness” to heterochromatin. The distribution of distances for silenced loci are also less variable than for expressing loci, roughly half that of an expressing locus. This becomes particularly clear when we imagine the variegating-gene-to-heterochromatin distance in statistical terms, using two standard deviations as a 95% confidence interval. Using the In(1) rst 3 case as an example, the expressing locus might be anywhere from 0 to 3 μm from heterochromatin, while the silenced locus will not be found farther than 1.7 μm away. Thinking about this in terms of three-dimensional nuclear volume, this means than an expressing locus will inhabit a 95% confidence volume five times larger than that of a silenced locus. We posit that the shorter median distance and reduced variability in position relative to heterochromatin suggests a persistent interaction. It is not known to what extent the association of a variegating gene with heterochromatin is a cause or an effect of gene silencing. Even though our results argue that the two events are connected, a gene could be silenced by local factors and associate with heterochromatin independently. Future examinations of this problem would require a live study to pinpoint precisely when interactions occur. If interaction occurs before the gene is silenced, that suggests that the interaction could be a cause of silencing. If interactions occur after silencing, however, this would suggest that the interaction is merely a side effect. Studying the persistence of such associations will reveal whether interaction with heterochromatin causes loci to be more constrained than non-interacting loci, which could explain why gene-to-heterochromatin distance distributions from expressing cells are so much broader than those of silenced cells. Now that specific interactions with heterochromatin have been shown to be closely correlated with silencing, one can ask how such interactions form. Time-lapse studies have shown that individual loci can move substantial distances within the nucleus, eventually bringing the variegating locus into contact with heterochromatin [ 19 ]. Additional live studies found that chromosomal loci can be held in place after contacting certain structures such as the nuclear envelope or the nucleolus [ 42 ]. Once contact is made, the proximal and centromeric blocks of heterochromatin may mediate a persistent interaction through multimerization of proteins such as HP1. Loci on wild-type chromosomes will not form persistent interactions because they lack a proximal block of heterochromatin. Interestingly, studies have found that as interphase progresses, nuclear motion decreases such that contacts between euchromatic loci and heterochromatin may stabilize [ 19 , 43 , 44 ]. Recent insights into the molecular behavior of heterochromatin proteins suggest how persistent interaction with heterochromatin may silence genes (reviewed in [ 45 ]). We hypothesize that the associated loci, because of intimate (possibly molecular) contact, acquire the molecular features of heterochromatin. The chromatin of PEV-silenced genes acquire heterochromatic features such as a regular nucleosome array, insensitivity to nucleases [ 46 ], and binding of heterochromatin proteins [ 47 , 48 ]. This altered chromatin structure seems to occlude the affected gene's promoter, preventing the loading of RNA polymerase and transcriptional activators, thereby preventing gene expression [ 49 ]. A model of how LRCIs with heterochromatin occur in the eye disk is outlined in Figure 8 . In dividing cells anterior to the morphogenic furrow, few loci interact with heterochromatin. As cells pass into the morphogenic furrow, the G1 cell cycle arrest allows loci to explore their chromosome territory and make contacts with regions of heterochromatin on their own and possibly other chromosomes. Loci that contact heterochromatin may form persistent interactions that cause silencing. The normal expression pattern of the white gene, for example, is now altered by association with heterochromatin. Instead of being expressed in every r8 cell behind the morphogenic furrow, as in a wild-type chromosome, its expression pattern is periodically interrupted by silenced cells. Loci that contact but do not form persistent interactions remain relatively unconstrained. Figure 8 Model of Variegating Gene Association with Heterochromatin and Its Consequences Cartoon of eye disk showing when and where association occurs and its effects on gene silencing. Anterior is to the left. Long-range interactions are not observed anterior to the morphogenic furrow in regularly dividing cells. In the morphogenic furrow, G1 arrest allows variegating genes the time to begin to find and associate with blocks of heterochromatin. Once behind the morphogenic furrow, variegating loci in more cells associate with heterochromatin, and once formed these interactions persist. When a developmental signal is sent for a gene to be expressed, its association with heterochromatin greatly affects its chances for being expressed, such that a gene interacting with heterochromatin is far less likely to be activated than one that is not. This is the first study to examine the expression state of a chromosomal locus and its interaction with heterochromatin on a cell-by-cell basis within intact tissue. In each of the three cases examined, relocalization of a gene to heterochromatin was tightly correlated with gene silencing. These results strongly suggest that spatial organization of the Drosophila genome is an integral part of organism function. Materials and Methods Drosophila strains and chromosomes Three classes of rearrangements were used, each representing one of Drosophila's three largest chromosomes: X, 2, and 3. Each variegates for a different gene ( Table 2 ) and has a proximal block of heterochromatin placed into or nearby the variegating locus. It was expected that the small block of heterochromatin near the variegating gene would mediate interaction between the variegating gene and centromeric heterochromatin without silencing the variegating gene directly. Table 2 List of Rearranged Chromosomes Studied Satellite sequences are taken from [ 66 ] Bloomington, Bloomington Stock Center, University of Indiana, Bloomington, Indiana, United States; Eissenberg, Joel Eissenberg, St. Louis University, St. Louis, Missouri, United States; Hawley, Scott Hawley, Stower's Institute, Kansas City, Missouri, United States; Henikoff, Stephen Henikoff, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States; ND, not determined; NR, not relevant The X chromosome was represented by the line In(1) rst 3 y 1 car 1 . Variegation for white was produced by mating In(1) rst 3 males with virgin females of the genotype C(1)RM v, v; C(4) ci ev R × X^Y, In(1)EN, v f B; C(4) ci ev R to produce normal phenotype XX/Y females and variegating X/O males [ 50 ]. X/O males were distinguished from XX/Y females by the expression pattern of white in third-instar eye disks (see Results). The second chromosome was represented by two different lines containing the bw D insertion [ 51 ]. Experiments studying the bw D rearrangement in imaginal disk tissue were performed with larvae homozygous for the bw D chromosome. Control experiments with bw D / bw + tissues produced identical results (data not shown). Experiments studying the long-range interactions with heterochromatin and silencing in adult eyes used the line Su( bw D )208/SM1. The In(2LR)SM1, Cy balancer carries a wild-type brown gene to allow bw variegation to be observed. The Su( bw D )208 rearrangement increased the number of expressing cells per eye from the normal bw D / bw + expression level (<5%) to about 20%–30% [ 26 ]. Experiments were carried out in a scarlet (st) background to eliminate brown ommochrome pigment; brown -expressing cells appeared red against a field of white non-expressors, as opposed to red-brown expressors against a field of brown non-expressors. All gene expression experiments in adult eyes were in flies with the genotype In(2LR)SM1 Cy /Su( bw D )208; st / st . The third chromosome was represented by the beta-galactosidase variegating line In(3L)BL1 [ 28 ]. The precise genotype for the In(3L)BL1 line is Df(1) y w, In(3L)BL1/Tm3 hb-lacZ . Diagrams of all variegating chromosomes including FISH staining of polytene chromosomes are shown in Figure 9 . Figure 9 Variegating Lines and Probes Used in This Study Variegating lines were used for three different chromosomes, each variegating for a different gene. For full details see Materials and Methods . (A) Diagram of each line showing approximate breakpoints and locations of variegating genes. FISH probes were made from P1 clones covering each gene and from heterochromatic sequences unique to each chromosome (see Materials and Methods ). (B–D) FISH probes for each chromosome. FISH probes were hybridized to polytene squashes to show cytological location of each probe. Regions of proximal heterochromatin from inversion and insertion are marked with an arrow. Each chromosome spread is taken from individual experiments. (B) Probe used in experiments for the line In(1) rst 3 . (C) Probe used in experiments for the line In(3L)BL1. (D) Probe used in experiments for the line bw D . FISH probes FISH probes marked the three-dimensional locations of chromosomal loci. Repetitive heterochromatin sequences were hybridized with hapten-conjugated DNA oligonucleotide probes. Each probe's location is presented in Figure 9 A. Chromosome 3 heterochromatin was hybridized to a 46-bp DNA oligonucleotide containing the sequence ( CCCGTACTGGT) 4 corresponding to the dodeca satellite sequence. Chromosome 2 heterochromatin was hybridized to a 35-bp DNA oligonucleotide containing the sequence ( AACAC) 7 . X heterochromatin was hybridized to a PCR product of 359 bp comprising the bulk of the 1.688 satellite sequence. Satellite sequences and probe construction are discussed in more detail in [ 52 ]. DNA oligonucleotides were synthesized by Invitrogen (Carlsbad, California, United States) using sequences as published in [ 52 ]. Heterochromatic probes made from these oligonucleotides were labeled with digoxygenin-conjugated nucleotides by terminal transferase. Digoxygenin-labeled probes were detected by sheep anti-dig F(ab)2 fragments coupled to Cy5 (Amersham Biosciences, Piscataway, New Jersey, United States). Euchromatic probes were made from P1 phage clones provided by the Berkeley Drosophila Genome Project and Genome Systems (St. Louis, Missouri, United States) [ 53 , 54 ]. Experiments with bw D lines used P1 clone DS003480, which hybridizes to 59E ( Figure 9 B) [ 9 ]. Experiments with the In(1) rst 3 line used P1 clone DS006812, corresponding to 3C2–3. For the line In(3L)BL1, the P1 clone DS000374 was utilized, which hybridizes to 65D2–3. P1 DNA was amplified in bacteria and purified via alkaline lysis and silica gel chromatography (Qiagen Midi Tips, Qiagen, Valencia, California, United States). The purified DNA was digested with restriction endonucleases and labeled with Rhodamine 4-dUTP (FluoroRed, Amersham Biosciences) by terminal transferase as described previously [ 52 ]. Sample preparation Eye imaginal disks were dissected from climbing third-instar larvae raised at 16 °C and fed on a yeast paste/glucose/agar/instant Drosophila food (Carolina Biological Supply, Burlington, North Carolina, United States) recipe. Eye disks were dissected into modified M3 medium [ 55 ] or Grace's and fixed for 15 min in 3.7% formaldehyde and Buffer A+ [ 56 ]. After fixation, disks were permeablized for 30 min in Buffer A+/0.1% Triton X-100 (Pierce Chemical, Rockford, Illinois, United States) and transferred to PBS (pH 7.4) with 0.1% Tween 20, unless specified otherwise. Adult eyes from recently eclosed (0–3 h) adult heads were dissected into a drop of Grace's medium and fixed for 15 min in Buffer A+/3.7% formaldehyde. Heads were transferred back into Grace's medium and the eyes cut away with a 15° microsurgical knife (Surgical Specialties Corporation, Reading, Pennsylvania, United States). These eyes were fixed for an additional 15 min in A+/3.7% formaldehyde and then pried away from the compound lens with forceps and a scalpel modified into a scoop (made by Daron Brown, Fine Science Tools, Foster City, California, United States). Bovine serum albumin (5%) prevented eyes from sticking together. Detection of gene expression Detection of gene expression in In(3L)BL1 relied on immunofluorescence of the beta-galactosidase gene product after heat shock at 37 °C. Eppendorf tubes containing a 1-ml agarose plug were preheated for one hour at 37 °C in a circulating water bath. The plug was removed, larvae were placed inside the eppendorf tube, and the pre-warmed plug was loosely placed over the larvae and the tube sealed. Larvae were incubated in the water bath for 5 min, followed by immersion in a room temperature water bath and a 1 h recovery period. Eye disks were dissected, followed by blocking for 4 h in PBS +0.1% Tween 20 (PBT) plus 5% dry milk (Nestle, Solon, Ohio, United States). Disks were incubated with a mouse monoclonal antibody in PBT at a dilution of 1:1,000 (Promega, Madison, Wisconsin, United States) for 2 h. After three 30-min washes in PBT, disks were incubated with horseradish-peroxidase-conjugated anti-mouse secondary antibodies (Jackson ImmunoResearch, West Grove, Pennsylvania, United States) at a 1:500 dilution in PBT for 1 h. After three more washes in PBT, the horseradish-peroxidase-coupled antibody was treated with fluorescent coumarin-coupled tyramide signal amplification reagents (TSA-direct, PerkinElmer Life Sciences, Wellesly, Massachusetts, United States) to develop a stable fluorescent stain that would survive FISH. Detection of gene expression in the line In(1) rst 3 relied on immunofluorescence to the white gene product. Details of anti-White antibodies are presented in Figure S1 and part 1 of Protocol S1 . Anti-White antibody staining proceeded as anti-beta-galactosidase staining, except that anti-White primary antibody was used at a concentration of 0.5 μg/ml in PBT and stained overnight. The primary antibody was detected by anti-rabbit horseradish-peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch) at dilution of 1:500, followed by treatment with tyramide as described above. For bw D , adult eyes were embedded in polyacrylamide immediately after dissection (see below), stained with 0.5 μg/ml DAPI in PBS (pH 7.4), and imaged at low magnification with an Olympus (Tokyo, Japan) 20 × 0.8 NA lens to identify gross structural features for later realignment. Select regions of each eye were imaged under a 1.4 NA 100x lens to acquire high-resolution three-dimensional images of each nucleus. The precise positions of both high and low magnification datasets were recorded for realignment after FISH. Expressing cells were identified by their bright autofluoresence. Detection of cell cycle arrest Experiments to determine the role of differentiation and the cell cycle were carried out in third-instar eye disks. A rabbit anti–cyclin A antibody detected the presence of G2 cyclins [ 34 ]. G1-arrested cells were identified by their location within the morphogenic furrow and the absence of cyclin A. FISH Whole-mount disks were processed for in situ hybridization as described elsewhere [ 52 ]. After FISH, tissues were stained in 2XSSCT with anti-Lamin antibodies [ 57 ] and an FITC-labeled goat anti-mouse secondary antibody to mark the nuclear periphery (Jackson ImmunoResearch). Before imaging, all tissues were washed three times in 50 mM Tris (pH 8.0) and embedded in polyacrylamide pads as in [ 58 ] to support three-dimensional structure. Adult eyes were processed for FISH similar to [ 58 ] with modifications described below. Adult eyes were placed within a nail polish ring on a #1.5 20 × 30 mm coverslip in 13.5 ml of Tris (pH 8.0). Then 6.5 ml of 3X activated acrylamide buffer (Buffer A+, 15% acrylamide, and 0.333% bis-acrylamide) was added. Twenty-five microliters of 20% ammonium persulfate and 25 μl of 20% sodium sulfite were added to 500 μl of acrylamide buffer immediately before addition to tissue. The acrylamide buffer drop was mixed by repeated pipetting, and sealed with a clean 20 × 30 mm coverslip. After 30 min of polymerization, the top coverslip was removed and pads washed four times in Buffer A+, followed by DAPI staining and pigment imaging. After pigment imaging, pads were stepped into 2XSSCT/50% formamide and incubated with probe and hybridization solution overnight at 37 °C before denaturation. Pre- and post-FISH washes proceeded as described for eye disks in PCR tubes, except that washes were extended to 1 h apiece in humid chambers with agitation. Eyes were denatured as described elsewhere [ 52 , 58 ], except that denaturation was extended to 10 min. After FISH, eyes were washed five times with 50% formamide/2xSSCT, stepped into PBT, and stained with DAPI. Three-dimensional imaging Before mounting, all tissues were washed three times in 50 mM Tris (pH 8.0) to remove any detergent. Disks were mounted in Vectashield (Vector Labs, Burlingame, California, United States) while adult eyes were mounted with ProLong (Molecular Probes, Eugene, Oregon, United States). Three-dimensional imaging was carried out on a computer-controlled epifluorescence microscope [ 59 , 60 ]. All images were collected on a Peltier-cooled CCD camera (Photometrics, Tucson, Arizona) connected to an SGI Indigo2 workstation (SGI, Mountain View, California, United States). Imaginal disks were imaged with a 60x Olympus 1.4 NA oil lens while adult eyes were imaged on an Olympus 20 × 0.8 NA oil lens followed by a Nikon (Tokyo, Japan) 100 × 1.4 NA oil lens. For adult eyes, the same tissues examined for pigmentation were re-imaged and carefully aligned with pre-FISH images. Data were processed after collection by a constrained iterative deconvolution software program [ 61 ]. Image analysis Image analysis was performed with the Image Visualization Environment software package PRIISM ([ 62 ]; D. Diggs and E. Branlund, unpublished data). The Water algorithm automatically identified nuclei [ 63 ] and segmented them into three-dimensional objects. After segmentation the center point of each object was the starting point to find the nuclear periphery. The nuclear lamin signal was found by searching for pixels above an intensity threshold a specified radial distance from the center point. Pixels above the threshold were marked and refined by a second- and third-order surface harmonic expansion [ 10 ]. Individual FISH signals within each modeled nucleus were identified based on their intensity peak above background (M. Lowenstein and D. Diggs, unpublished data). FISH signal distances were measured and sorted by nucleus membership. Distances were measured from the intensity-weighted center of mass rather than from the edge of each FISH signal. This means that even overlapping FISH signals will return a measured distance. Differences between different groups of nuclear distances were tested for statistical significance by the Mann–Whitney U test (see Table 1 ), which is a two-sample rank-sum test for position that makes no assumptions regarding the distributions of sample data [ 64 ]. Plots and statistical comparisons were performed by the Statview 4.5 (Abacus Concepts, Berkeley, California, United States) analysis program. Results are summarized in Table 2 . Data presentation Groups of distances mentioned previously were tested for statistical significance relative to one another (see Table 1 ). Variegating chromosomal loci close to heterochromatin were considered to be interacting, while loci farther away were considered to be not interacting. Data were presented in two different ways. First, standard histograms presented the full distribution of the data including any subpopulations. Second, cumulative percentile plots with percent association between variegating genes and their own centromeric heterochromatin plotted as a function of gene-to-heterochromatin distance more clearly distinguished varying interaction levels between different cell populations. This plotting method also revealed that the exact placement of a cutoff for interaction between loci was unimportant, as clear differences could be seen at distances less, equal to, or greater than 1 μm. As a general rule 1 μm in distance was used as a cutoff for interaction. This cutoff was not intended to be mechanistically meaningful, but to serve as a rule of thumb for thinking about the results. Monte Carlo analysis and correlation between nuclear distances and nuclear dimensions Distance distributions in eye disk nuclei were compared to pairwise distances measured from 50 points randomly placed within the nucleus. Also, distances between chromosomal loci were examined for their sensitivity to nuclear shape and volume. Data and results are presented in Table S2 and parts 3 and 4 of Protocol S1 . Supporting Information Protocol S1 Anti-White Antibody Generation, and the Effects of Gene Expression, Nuclear Dimensions, and Time on Gene to Heterochromatin Distances (31 KB DOC). Click here for additional data file. Figure S1 Production and Characterization of Anti-White Antisera Full details available in Materials and Methods . (A) Western blots showing specificity of antibody for a single polypeptide of approximately 70 kDa in larval (L) and adult (A) extracts. Preimmune sera show no reactivity for that polypeptide. (B) Immunofluorescence of eye disks with affinity-purified anti-White antisera. Anti-White staining surrounded each nucleus, possibly because the protein is enriched in the endoplasmic reticulum. Expression was seen only in four cone cell ommatidial clusters several cell rows behind the morphogenic furrow. (C) Immunofluoresence in eye disks of the white variegating line In(1) w m51b . Gaps in the white expression pattern are marked with dotted polygons. (D) Vertical series of images through a single four cone cell cluster in the eye disk stained for White protein (cyan) and the nuclear lamin (green). In every cell cluster examined, expression was only seen in the eighth photoreceptor. Each panel represents a step of about 2.5 μm from the most apical to basal parts of the cluster. Total distance from basal edge of cell cluster to cone cells is about 10 μm. (2.9 MB TIF). Click here for additional data file. Figure S2 Comparison of Different Cell Populations Based upon Time after Differentiation or Cell Cycle Arrest Individual rows of differentiated nuclei were compared to one another to see whether differences in interaction levels existed between younger and older cells. (A–C) Row-by-row analysis method for lines bw D , In(1) rst 3 , and In(3L)BL1 respectively. Each color represents a different row, rows 4–8 respectively. (D–F) Scattergrams for each row based on pooling of data from three different disks per line. Y-axis is distance in microns; X-axis is row number. (G–I) Percentile plots of each row. (2.6 MB TIF). Click here for additional data file. Table S1 Variegating Locus-to-Heterochromatin Distances in Wild-Type Cells and Random Distributions (53 KB DOC). Click here for additional data file. Table S2 Effects of Nuclear Dimensions on Variegating Gene-to-Heterochromatin Distances (40 KB DOC). Click here for additional data file.

D:\keerthana\PMC001xxxxxx\PMC1054879.xml

1054880

Highly Nonrandom Features of Synaptic Connectivity in Local Cortical Circuits

How different is local cortical circuitry from a random network? To answer this question, we probed synaptic connections with several hundred simultaneous quadruple whole-cell recordings from layer 5 pyramidal neurons in the rat visual cortex. Analysis of this dataset revealed several nonrandom features in synaptic connectivity. We confirmed previous reports that bidirectional connections are more common than expected in a random network. We found that several highly clustered three-neuron connectivity patterns are overrepresented, suggesting that connections tend to cluster together. We also analyzed synaptic connection strength as defined by the peak excitatory postsynaptic potential amplitude. We found that the distribution of synaptic connection strength differs significantly from the Poisson distribution and can be fitted by a lognormal distribution. Such a distribution has a heavier tail and implies that synaptic weight is concentrated among few synaptic connections. In addition, the strengths of synaptic connections sharing pre- or postsynaptic neurons are correlated, implying that strong connections are even more clustered than the weak ones. Therefore, the local cortical network structure can be viewed as a skeleton of stronger connections in a sea of weaker ones. Such a skeleton is likely to play an important role in network dynamics and should be investigated further.

Introduction Understanding cortical function requires unraveling synaptic connectivity in cortical circuits, that is, establishing the wiring diagrams. Although smaller wiring diagrams can be reconstructed using electron microscopy [ 1 ], reconstruction on the scale of a cortical column is impossible with current technology (but see [ 2 ] for a promising approach). Even if such a possibility were within reach, synaptic connectivity likely varies from animal to animal and within one animal over time. Therefore, a reasonable approach is to describe synaptic connectivity statistically, or probabilistically. Such statistical description may be based on random sampling of connections with multineuron recordings [ 3 , 4 , 5 ]. For example, electrophysiological recordings from neuronal pairs showed that the probability of connection is often low [ 3 , 5 , 6 , 7 , 8 , 9 , 10 , 11 ], suggesting that the network is sparse. Such statistical approaches may create the impression that synaptic connectivity in local cortical circuits is random. This view is consistent with previous suggestions [ 12 , 13 , 14 ], but hard to reconcile with cortical functionality, which must rely on specificity of connections [ 15 , 16 , 17 , 18 ]. In general, statistical sampling of connections does not imply that the underlying network has random connectivity. Indeed, statistical sampling has already revealed several nonrandom features in cortical connectivity. In particular, specific connectivity patterns exist between different classes in the local circuit [ 3 , 19 , 20 ]. Within one putatively homogeneous class, the number of reciprocally connected pairs is greater than expected in a random network [ 5 , 6 , 11 ]. Distribution of the number of synapses per connection is non-Poisson and has low variance [ 6 , 21 ]. At the same time, the distribution of the connection strength is broad [ 5 , 6 , 10 , 11 ]. But many questions remain unanswered: Are there nonrandom features in patterns involving more than two neurons? What is the distribution of synaptic connection strength? Are synaptic connection strengths correlated? Recently, new approaches for network connectivity analysis have been developed and various nonrandom features were discovered in natural and man-made networks. In particular, many networks are scale-free, that is, the number of connections per node (node degree) often follows a power-law distribution [ 22 ]. Also, many networks exhibit the small-world property, that is, high local clustering of connections in combination with a short path between any two nodes [ 23 , 24 ]. In addition, probability of connection between nodes depends on how many connections they have [ 25 ]. Although local cortical networks may possess these properties, existing connectivity data are not sufficient for such analyses. These data are obtained by random sampling of connections and call for other approaches. One such approach is to explore local structures in network connectivity by studying the distribution of few-node connectivity patterns, or motifs [ 26 , 27 ]. Another such approach is analyzing the utilization (or, in this case, the strength) of connections [ 28 , 29 , 30 ]. In this paper, we apply a combination of statistical methods to a large dataset from hundreds of simultaneous quadruple whole-cell recordings from visual cortex in developing rats. Our results confirm previous indications of nonrandomness and point out several new ones. In particular, we show that the distribution of connection strengths between pyramidal neurons is non-Poisson and find correlations in the strength of the connections sharing pre- or postsynaptic neurons. Also, we find several overrepresented three-neuron connectivity patterns, or motifs. Surprisingly, we find that some few-neuron motifs can play an important role in the dynamics of layer 5 local cortical networks because they are composed of exceptionally strong connections. This suggests a novel view of the local cortical network, in which a skeleton of stronger connections is immersed in a sea of weaker ones. Results We studied connectivity among thick tufted layer 5 neurons in rat visual cortex with quadruple whole-cell recordings ( Figure 1 A and 1 B). Thick tufted layer 5 pyramidal neurons are important projection neurons from the cerebral cortex to subcortical areas [ 9 , 31 , 32 ]. Synaptic connection strengths were assessed by evoking action potentials in each of the four cells and measuring the averaged peak excitatory postsynaptic potential (EPSP) amplitudes in the other three cells (see Figure 1 C and Materials and Methods ). Results of these measurements for a sample quadruple recording are shown in Figure 1 D. Each arrow indicates a detected connection with the corresponding connection strengths. The dataset contained a total of 816 quadruple recording attempts (some of these attempts contained data for only two or three neurons, if whole-cell configuration was not successfully established with all four cells). As previously reported [ 5 ], the rate of connectivity was p = 11.6% (931 connections out of 8,050 possible connections), which is similar to that reported for rat somatosensory cortex layers 5 [ 6 , 9 ] and 2/3 [ 11 ], as well as those previously reported for rat visual cortical layers 5 [ 3 , 10 ] and 2/3 [ 11 ]. Figure 1 Illustration of a Quadruple Whole-Cell Recording (A) Dodt contrast image showing four thick-tufted L5 neurons before patching on. (B) Fluorescent image of the same four cells in whole-cell configuration. (C) Average EPSP waveform measured in the postsynaptic neuron (bottom) while evoking action potentials in the presynaptic neuron (top). (D) Diagram of detected synaptic connections and their strengths for this quadruple recording. Two-Neuron Patterns We started by assessing how well a randomly connected network [ 33 ] describes our dataset. In this model, the existence of a connection between any two neurons is independently chosen with a uniform probability p ( Figure 2 A). We test the predictions of this model by classifying all simultaneously recorded pairs of neurons into three classes: unconnected, unidirectionally connected, and bidirectionally connected. Given connection probability p and total number of pairs N, the expected number of unconnected pairs should be N(1 − p) 2 . The expected number of unidirectionally connected pairs should be 2Np(1 − p), and the expected number of bidirectionally connected pairs should be Np 2 . We find that the actual number of bidirectionally connected pairs is four times that of the expected numbers ( p < 0.0001) ( Figure 2 B). The observed overrepresentation of reciprocally connected layer 5 neurons confirms previous reports [ 5 , 6 ]. Such overrepresentation has also been observed in layer 2/3 of the rat visual cortex [ 11 ]. However, whether projections between layers observe this pattern remains an open question. Figure 2 Two-Neuron Connectivity Patterns Are Nonrandom (A) Null hypothesis is generated by assuming independent probabilities of connection. (B) Reciprocal connections are four times more likely than predicted by the null hypothesis ( p < 0.0001, Monte Carlo simulation to test for overrepresentation). Numbers on top of bars are actual counts. Error bars are standard deviations estimated by bootstrap method. Can the overrepresentation of reciprocal connections reflect an experimental artifact? Indeed, such overrepresentation could arise from a nonuniform probability of connection for different neurons. For example, connection probability may depend on interneuron distance. To control for this artifact, we measured distances between neurons in recordings where labeling was performed successfully. We found that the probability of connection does not depend systematically on the interneuron distance ( p = 0.21, chi square test) ( Figures 3 , S1 , and S2 ). This is not surprising because most neurons were located closer than the span of their dendritic (and especially axonal) arbors. Our result is consistent with Holmgren et al.'s study [ 11 ], which found only a small (22%) decrease in connection probability up to 50 μm, with a more significant drop (44%) up to 100 μm for layer 2/3 pyramidal neurons. Figure 3 Probability of Connection among Adjacent Neurons Does Not Depend Strongly on the Interneuron Distance (A) Relative location of labeled neurons in the plane of the section. Positive direction of y-axis is aligned with apical dendrite. Potentially presynaptic neuron is located at the origin. Red—bidirectionally connected pairs; blue—unidirectionally connected pairs; green—unconnected pairs. (B) Histogram showing the numbers of pairs in the three classes as a function of distance between neurons (Euclidian distance was calculated from relative X, Y, Z coordinates). (C) Probability of connection versus interneuron distance. Error bars are 95% confidence intervals estimated from binomial distribution. Another source of nonuniformity in connection probability may be axonal arbors being cut off differentially, depending on the depth of the recorded neuron from the slice surface. (The recording depths were from 10 to 100 μm.) To explore such a possibility we measured the dependence of the connection probability on the recording depth. Neither connection probability, nor strength of connection was found to depend systematically on the recording depth (see Figure S3 ). In addition, for every successfully labeled tissue we measured the distance from the average cell position to the nearest axonal cut point (see Figure S3 ). Again no strong trends in connection probability or connection strength were found. These results show that the cutting artifact is unlikely to explain observed nonrandom features. We also considered the possible artifact of connection probability varying with age. We found a weak decline in connection probability and EPSP amplitude (consistent with Reyes and Sakmann [ 34 ]) within the range used in experiments (P12–P20; see Figure S4 ). Yet, such a decline is insufficient to account for the observed nonrandomness. To demonstrate this, we repeated most of the analysis on a subset of the data from 14 to 16-d-old animals when the majority of measurements were performed (see Figure S5 ). We found that bidirectional connections are also overrepresented in this subset of data. Results of other analyses that will be described later in the paper are also confirmed on this subset ( Figure S5 ). Finally, it is possible that some extreme degree of inhomogeneity in connections probability is able to explain the observed overrepresentation of reciprocal pairs, but this would probably reflect large local inhomogeneity in cortical connectivity patterns—possibly differences between subclasses [ 6 , 35 ], rather than experimental artifacts—and is in line with the main conclusions of this paper. Three-Neuron Patterns We extended our analysis by comparing the statistics of three-neuron patterns to those expected by chance [ 26 , 27 ]. We classify all triplets into 16 classes and count the number of triplets in each class. In order to avoid reporting overrepresented three-neuron patterns just because they contain popular two-neuron patterns, we have revised the null hypothesis[ 26 , 27 ]. The control distribution was obtained numerically by constructing random triplets where constituent pairs are chosen independently, but with the same probability of bidirectional and unidirectional connections as in data ( Figure 4 A). For example, the actual probability of a unidirectional connection is (according to Figure 2 B) 495/(3312 + 495 + 218) = 0.123. Then the probability of unidirectional connection from A to B is 0.123/2 = 0.0615, the same as from B to A (see Figure 4 A). The probability of bidirectional connection is (according to Figure 2 B) 218/(3312 + 495 + 218) = 0.0542. The probability of finding the particular triplet class in Figure 4 A by chance is the product of the probabilities of finding the three constituent pairs and a factor to account for permutations of the three neurons. The ratio of the observed counts and the expected counts for each class are plotted in Figure 4 B. The actual counts are given as numbers on top of the bars. Although triplets from several of these patterns have been reported previously [ 9 , 10 ], small numbers of recordings have precluded statistical analysis. Figure 4 Several Three-Neuron Patterns Are Overrepresented as Compared to the Random Network (A) Null hypothesis for three-neuron patterns assumes independent combinations of connection probabilities of two kinds of two-neuron patterns. (B) Ratio of actual counts (numbers above bars) to that predicted by the null hypothesis. Error bars are standard deviations estimated by bootstrap method. (C) Raw (open bars) and multiple-hypothesis testing corrected (filled bars) p -values. p -values above 0.5 are not shown. According to Figure 4 B, several triplet patterns are overrepresented. Is this overrepresentation statistically significant? Because we look for overrepresentation in many pattern classes at the same time, the raw p -values ( Figure 4 C) overestimate the true significance (underestimate the true p -value). To correct the raw p -values, one has to apply a multiple-hypothesis testing procedure. We chose to correct the p -values by applying a step-down min-P-based multiple-hypothesis testing correction procedure (see Materials and Methods ). The corrected p -values ( Figure 4 C) give the probability of mistakenly reporting at least one of the patterns as overrepresented when it is not. Two-neuron correlations are summarized by saying that if neuron A synapses onto neuron B, then the probability of B synapsing onto A is several times greater than chance. Three-neuron correlations are summarized roughly by saying that if A connects with B and B connects with C (regardless of direction), the probability of connection between A and C is several times greater than chance. Interestingly, similar results have been obtained in the analysis of the Caenorhabditis elegans wiring diagram [ 36 ], which was reconstructed from serial section electron microscopy [ 1 ]. Because different techniques have different biases, the similarity of results suggests that correlations in synaptic connectivity represent a general property of neuronal circuits. Such property may represent evolutionary conservation from invertebrates to mammals or convergence driven by similar computational constraints. Although individual connectivity patterns containing more than three neurons could not be analyzed statistically for the existing dataset ( Table S1 ), we found a 70% overrepresentation of “chain” quadruplets (patterns number 21 23 24 26 28 29 31 32 33 34 35 38 39 41 43 as defined in Figure S6 , p = 0.004) relative to the null hypothesis requiring that a random matrix has the actual proportion of triplet classes. Distribution of Synaptic Connection Strengths Next, we turned our attention to the distribution of synaptic connection strengths as characterized by EPSP amplitude ( Figure 5 A). We estimated the probability density function by binning connection strengths and dividing the number of occurrences in each bin by the bin size. Since there are many more weak connections than strong ones, we chose bins whose sizes increase linearly with the connection strength at the bin center. In other words, bin sizes are uniform on the log scale. The estimated density function is independent of the chosen bin size since the bin size is divided out. Figure 5 Distribution of Synaptic Connection Strength Has a Heavy Tail (A) Estimated probability density function in log–log space, with both lognormal fit ( p [ w ] = 0.426exp[−(ln[ w ] + 0.702) 2 /(2 × 0.9355) 2 ]/ w ) and exponential fit ( p [ w ] = 1.82exp[−1.683 w ]). Notice that the lognormal fit has a heavier tail than the exponential distribution. Error bars are standard deviations estimated by bootstrap method (not shown when narrower than the dot). The numbers on top on the dots are the actual counts (not shown when more than 50). (B) Estimated probability density distribution in semilog space, with the lognormal fit. The lognormal function shows up as a normal function in the semilog space. (C) Empirical cumulative density function for both the probability distribution of synaptic strengths and the synaptic contribution (normalized product of probability and connection strength). They are generated directly from the data rather than the fits. The vertical line illustrates the fact that 17% of the synaptic connections contribute to half of the total synaptic strengths. (D) Probability density function of synaptic connection strengths p(w) fitted by a lognormal function and the synaptic contribution defined as the product of the strength, w, and p(w). The total areas under both curves are normalized to 1. The obtained distribution has a mean of 0.77 mV and a heavy tail, that is, a greater number of strong synaptic connections than expected for either the exponential distribution ( Figure 5 A) or the normal distribution (not shown). There are significantly more connections with strengths above 1 mV than expected by best exponential or normal fit ( p < 0.0001; see Materials and Methods ). We find that the dataset is best fit by a lognormal distribution, which has a bell shape when plotted on a semilog scale ( Figure 5 B). Although our measurements were performed in developing animals, experiments in mature animals have also revealed large single EPSPs (>5 mV) [ 37 ]. The overrepresentation of strong synaptic connections is likely to have important implications for the cortical network dynamics. This is because strong connections are few but powerful. For example, although synaptic connections with strength above 1.2 mV constitute only 17% of all connections, they contribute about half of the total synaptic weight ( Figure 5 C and 5 D). This estimate was obtained by multiplying the number of synaptic connections by the connection strengths (assuming equal presynaptic firing rates). Correlation of Connection Strengths in Two-Neuron Patterns Next, we analyzed the correlations between the strengths of the synaptic connections in two-neuron patterns. We find that the synaptic strengths of the bidirectional connections are on average stronger than the unidirectional synaptic connections (mean 0.95 mV versus 0.61 mV, p = 3.1 × 10 −7 , Student's t -test) in agreement with [ 6 ]. The distribution of connection strengths for the bidirectional connections is expanded toward stronger connections compared to that of unidirectional connections ( Figure 6 A; note the semilog scale). Furthermore, the strengths of the two connections in a bidirectional pair are moderately but significantly correlated with each other ( Figure 6 B). To control for possible systematic variations between different quadruplets, we looked at correlations in the strength of synaptic connections that shared no pre- and postsynaptic neurons and found no significant correlation ( Figure 6 C). Could the correlation in connection strength result from nearby neurons having stronger connections? We do not think so because the strengths of bidirectional connections do not depend strongly on the distance between neurons ( Figure 6 D). Figure 6 Bidirectionally Connected Pairs Contain Connections That Are Stronger and Correlated (A) Synaptic connections in bidirectionally connected pairs are on average stronger than those in unidirectionally connected pairs. The probability density distribution for both the reciprocal (red solid, p(w) = 0.41exp(−(ln w + 0.60) 2 /(2 × 0.976 2 )/w) and nonreciprocal (blue dashed, p(w) = 0.47exp(−(ln w + 0.81) 2 /(2 × 0.834 2 )/w) connections are shown. (B) In bidirectionally connected pairs synaptic connection strengths are moderately but significantly correlated ( R = 0.36, p < 0.0001). (C) Scatter plot of the strength of synaptic connections that shared no pre- and postsynaptic neurons in the same quadruple recording. There might be other connections in the quadruplet besides these two connections. No significant correlation is observed ( R = 0.068, p = 0.48). All correlations calculated using Pearson's R method in log space. (D) Average connection strength for bidirectional connections does not vary systemically with interneuron distance (one-way ANOVA, p = 0.068). Numbers on top of data points are the number of connections. Error bars are standard errors of the mean. Another way to characterize correlations in connection strengths is by analyzing the overrepresentation of the bidirectional motif for different synaptic-strength thresholds. For every threshold value, we modify the dataset by keeping only those synaptic connections that exceed this threshold. In a unidirectionally connected pair, connection is kept only if its strength exceeds the threshold. In a bidirectionally connected pair, if both connections exceed the threshold, they are both kept. If only one of the connections exceeds the threshold, the pair becomes unidirectionally connected. Then we predict the numbers of bidirectional synaptic connections that exceed threshold by using the null model assuming independent probability, as was done for two-neuron patterns. The actual number of bidirectional connections exceeding the threshold is compared with the predicted. We find that, as the threshold is raised, the ratio of actual to expected number of bidirectional connections monotonically increases ( Figure 7 ). This shows that reciprocity of connections is greater for stronger connections. Figure 7 Stronger Connections Are More Likely Reciprocal than Weaker Ones Overrepresentation of bidirectionally connected motifs gets more dramatic for higher threshold of connection strength (counts differ from random with p < 0.001 for all thresholds, Monte Carlo simulation). Significance of monotonicity is assessed by applying the Kolmogorov-Smirnov test ( p < 3.5 × 10 −10 for all successive pairs). Numbers on top of dots show the counts of actual pairs. Three-Neuron Patterns with Strong Connections We also analyzed the overrepresentation of three-neuron patterns as a function of threshold. Because of small numbers of patterns in some classes, we have grouped highly connected patterns (boxed patterns in Figure 8 ) together and calculated the measured counts relative to random for different thresholds. Similar to the two-neuron motifs, overrepresentation of the highly connected motifs gets more dramatic as the threshold is raised ( Figure 8 ). Although the numbers of overrepresented three-neuron patterns are small, they may contribute to the neuronal dynamics in nontrivial ways, for example, by supporting recurrent activity. Furthermore, the contribution of three-neuron patterns depends on the chosen connection strength threshold. The relative fraction of overrepresented patterns in the network of stronger connections is much greater than that in the network of weaker connections. Figure 8 Stronger Connections Are More Clustered than Weaker Ones Relative overrepresentation of highly connected three-neuron motifs monotonically increases as the threshold is raised (counts differ from random p < 0.001 for all thresholds, Monte Carlo simulation). Significance of monotonicity is assessed by applying the Kolmogorov-Smirnov test ( p <3.5 × 10 −10 for all successive pairs). Numbers show the actual triplet counts. For the second to highest threshold, two instances of pattern 12 and one instance of pattern 16 survive. For the highest threshold, one instance each of pattern 12 and 15 survive (one of the connections in pattern 16 drops out and it becomes pattern 15). In addition, we analyzed the strengths of synaptic connections made onto the same neuron, synaptic connections coming out of the same neuron, and synaptic connections onto and out of the same neuron ( Figure S7 ). These strengths are weakly correlated. Correlations in the strength of incoming or outgoing connections may suggest, although not conclusively prove, the presence of neurons with particularly strong connections. Such neurons may be analogous to “network hubs,” or nodes with particularly large numbers of connections (degrees), which are known to exist in other networks [ 22 , 38 ]. Discussion We showed that synaptic connectivity in the local network of layer 5 pyramidal neurons is highly nonrandom. The network consists of sparse synaptic connections that tend to cluster together in the form of overrepresented patterns, or motifs. The distribution of connection strengths has a significant tail; strong connections are few but powerful and even more clustered than the weak ones. These results suggest that the network may be viewed as a skeleton of stronger connections in a sea of weaker ones ( Figure 9 ). Interestingly, the existence of few but powerful synaptic connections makes analyzing the network with few-neuron connectivity patterns a reasonable first step. Indeed one could have thought that, since each neuron receives inputs from thousands of others collectively determining its dynamics, analysis of few-neuron motifs is akin to “searching under the street light.” Yet, the finding of a heavy tail in the connection strength distribution suggests that a lot of power is due to a few connections. Therefore, our analysis has illuminated a significant part of the local cortical architecture, especially if the stronger connections are distributed uniformly among neurons. Naturally, this description is not complete, and future studies should investigate whether stronger synaptic connections are distributed among neurons uniformly or belong preferentially to “hub” neurons. Also, studies involving larger networks of neurons will be needed to provide a more complete understanding of the network structure and function. Figure 9 Statistically Reconstructed Network of 50 Layer 5 Pyramidal Neurons Illustrates That Stronger Connections form a Skeleton Immersed in a Sea of the Weaker Ones Details of statistical reconstruction are given in Materials and Methods . For illustrative purposes, neurons are arranged so that strongly interconnected nodes are close by. Dotted arrows are weak (<1 mV) unidirectional connections; solid arrows are weak bidirectional connections. Red arrows are strong (>1 mV) unidirectional connections with arrow size indicating the strength. Red arrows with double lines are strong bidirectional connections. Although broad distribution of synaptic connections strength has been seen in the cortex [ 6 , 11 ] and in the cerebellum [ 39 ], heavy-tailed distributions have not been suggested as suitable fits previously. For example, in the feed-forward projection from granule to Purkinje cells in the cerebellum, the distribution was fitted by a truncated Gaussian distribution, argued to be optimal for information storage [ 40 ]. It would be interesting to see if analogous theory could be developed to explain the lognormal distribution seen among the layer 5 pyramid recurrent connections. Another relevant observation is that of mini-EPSC amplitudes [ 41 ], which were fitted by a Poisson distribution based on a binomial model of the data. In this case, however, we are looking at direct unitary connections between pairs of neurons rather than individual synapses, and such direct connections between nearby cortical neurons are typically comprised of multiple individual synapses [ 6 , 21 , 34 , 42 ]. Evoked and spontaneous release may also produce different synaptic strength distributions because the underlying molecular mechanisms are different. Alternatively, the lognormal distribution could depend on network activity patterns not present in dissociated cultures. To illustrate the possible impact of the skeleton of strong connections on the network dynamics, let us consider a local network of layer 5 neurons occupying the 300 μm × 300 μm area. According to Peters et al. [ 43 ], there are 4,480 thick tufted layer 5 neurons under 1 mm 2 of cortex, and therefore 400 thick tufted neurons in the considered network. With a connection probability of 0.11, each neuron receives inputs from 44 others. If distribution of connection strength is uniform among neurons, then each neuron has only about 2–3 connections in the greater than 2-mV range. If the corresponding 2–3 presynaptic neurons fire simultaneously, they may drive the postsynaptic neuron to fire. This suggests that a sparse skeleton of strong connections may drive the dynamics of the network. An exceptionally strong connection (>10 mV ) may alone drive a postsynaptic neuron to fire. Suprathreshold EPSPs have been observed previously with paired recordings [ 37 , 44 , 45 ] and with calcium imaging [ 46 ]. However, such connections occur with a very low probability (about 1/1000, estimated from lognormal distribution), meaning that there are only about 20 of such connections in the considered network and that therefore most neurons do not have them. Finally, inhibitory neurons may make it more difficult to drive a postsynaptic neuron to fire and need to be investigated. Because the highly influential, strong, and reliable ( Figure S8 ) synaptic connections in the network are few in number, their exact connectivity pattern and properties might therefore be important and make firing patterns of the involved cortical neurons highly reproducible. This may be manifested in the simultaneous activation of several neurons in organized patterns during spontaneous and evoked activity that has been observed in cortical slices [ 47 , 48 , 49 , 50 ] and elsewhere [ 51 , 52 , 53 , 54 ]. Unfortunately, most current experimental studies rely on random sampling of neurons appropriate for studying the properties of average connections rather than the particularly strong connections. It might be important in the future to devise methods to selectively study particularly strong connections, because of their anticipated large influence on network dynamics [ 35 , 49 ]. Although stronger connections are likely to be important for network dynamics, weaker connections need to be considered as well. Collectively, they could affect the dynamics of the network significantly and might carry out computations with a population code. The weaker connections may be a potent driving force if firing is correlated between neurons. In addition, weaker connections may serve as a potential reserve for cortical plasticity. Indeed, weaker connections could be strengthened easily through a variety of activity-dependent learning rules (see below for an example). In neurobiological literature, synaptic connections have been classified previously by their impact into “drivers” and “modulators” [ 55 ]. Drivers are less numerous and produce stronger impact than modulators. We stopped short of calling stronger connections among layer 5 pyramidal neurons drivers, and weaker connections modulators, for two reasons. First, previously drivers and modulators have been used to describe inputs arising from a priori different subsets of neurons, such as different pathways. Second, we do not find a clearly bimodal distribution of connections strength, suggesting that the distinction between stronger and weaker connections is not clearly defined enough to warrant two separate classes. Next, we consider how observed distributions of synaptic strength and correlations between them might have arisen. Although it is possible that the neurons bound by stronger connections form a distinct subclass defined by perhaps distinct long-range projection patterns, or different channel densities and/or gene expression patterns, it is also possible that the distributions arise as a natural consequence of activity-dependent plasticity rules. The lognormal distribution of synaptic connection strength may be explained by a random multiplicative process, which has been extensively studied previously [ 56 , 57 , 58 , 59 ]. The idea behind this is demonstrated below. Suppose a synaptic connection changes its strength multiplicatively after i th plasticity episode. This can be expressed as w i = F i w i-1 , where w i is the synaptic connection strength after i th plasticity episode, and F i is the fractional change in synaptic strength induced by that episode. Then it is easy to see that where w n is the current synaptic connection strength, w 0 is the initial synaptic connection strength, and F i s are the fractional changes in synaptic connection strength. If we assume all F i s to be independent and identically distributed with finite mean and variance, then by applying central limit theorem, log w n should obey a Gaussian distribution, which implies that w n obeys a lognormal distribution. For a more rigorous treatment, a decay term has to be added to make the distribution stationary, which is analogous to the Gompertz stochastic growth model in ecology [ 56 , 59 ] (also see Appendix S1 ). In a network, the fractional change in synaptic connection strengths due to long-term potentiation (LTP) would have complex dependencies both on current synaptic connection strength and on correlated activity in the network. In a previous study of layer 5 pyramidal cells, we found only a weak—although statistically significant—dependence of the percentage amount of LTP on the pre-LTP EPSP amplitude so that fractional synaptic change due to LTP was in effect approximately constant for most synaptic strengths [ 5 ]. Studies in other brain areas have found a more marked negative dependency [ 60 , 61 , 62 ]. However, this negative dependency could be counterbalanced by the stronger correlation between presynaptic and postsynaptic firing patterns introduced by a stronger synaptic connection. Regardless, it is curious that a simple independency assumption, together with synaptic decay, reproduces the observed distribution, despite the complex interactions in the network. How this is achieved warrants further investigation. Can the overrepresentation of bidirectional connections and the correlation in the reciprocal connection strength arise from known learning rules? For example, the synaptic connections studied in this paper are known to obey a temporally asymmetric spike-timing-dependent plasticity rule [ 5 , 8 ], in which the strength of a synaptic connection changes according to the timing of pre- and postsynaptic spikes. If a presynaptic spike shortly precedes a postsynaptic spike, the synaptic connection is strengthened. Conversely, if a presynaptic spike follows a postsynaptic spike, the synaptic connection is weakened. Simulations have shown that in a recurrent network, if effects of spike pairs are assumed to sum linearly, this rule leads to underrepresentation of bidirectional motifs, instead of the overrepresentation observed here [ 63 ], because the firing statistics are exactly reversed for the reciprocal synaptic connections. However, for highly correlated firing of pre- and postsynaptic cells, depending on the relative durations and amplitudes of the long-term potentiation and long-term depression temporal windows, more potentiation than depression may be triggered for connections going both ways [ 64 , 65 ]. Furthermore, nonlinear spike interactions are known to operate at those synapses. In particular, the spike-timing-dependent plasticity rule becomes temporally symmetric if the pre- and postsynaptic neurons fire at higher than 50 Hz [ 5 ]. Whether these factors can explain this discrepancy, or additional factors need to be considered, remains to be studied. In a wide range of networks, there is a power-law relationship between the numbers of connections a particular node has (its degree) and the abundance of such nodes [ 66 ]. These networks have been termed scale-free networks [ 22 ]. In particular, such a power-law distribution of the number of connections a neuron makes has been reported in C. elegans [ 22 ]. Here, we have not studied the degree distribution because of the lack of adequate data (such as, for example, the full connectivity diagram for the cortical network). We instead analyzed the strengths of the connections and found a lognormal distribution of synaptic connection strengths, which has a heavy tail, similar to the power-law distribution. Similar distributions have been observed in many nonbiological networks [ 67 , 68 ]. In the biological setting, using an in silico model of metabolic flow in yeast, Almaas et al. [ 28 ] found that network use is highly uneven and dominated by several “hot links” that represent high-activity interactions that are embedded into a web of less active interactions. Such heavy-tailed distribution for connection strengths has also been suggested based on experimental data for metabolic flow and gene regulation networks [ 29 , 30 ]. Therefore, a heavy-tailed distribution for connection strengths along with clustering of stronger connections into a backbone might represent a novel universal feature of many networks, in addition to the power-law distribution of number of connections commonly discussed. Such an arrangement would give the stronger links a larger role in the network and might represent a hierarchal organizational scheme of the network structure [ 38 ]. In conclusion, the statistics of connectivity in a local network of layer 5 tufted pyramidal neurons are highly nonrandom and bear similarities to other biological networks. The cortical network is best visualized as a skeleton of stronger connections in a sea of weaker connections. These findings are likely to have important implications for cortical dynamics. Materials and Methods Electrophysiology The dataset used for this study was originally used for the study of long-term plasticity, and the methods were previously described in detail [ 5 , 69 ]. Briefly, acute visual cortical slices were cut from rats aged P12–P20. Rats were anesthetized with isoflurane, decapitated, and the brain was rapidly removed to ice-cold artificial cerebrospinal fluid (in mM: NaCl, 126; KCl, 3; MgCl 2 , 1; NaH 2 PO4, 1; CaCl 2 , 2.5; NaHCO 3 , 25; dextrose, 25; osmolality 320 mOsm, bubbled with 95% O 2 /5% CO 2 [pH 7.4]). Slices were used after at least 1 h of incubation, and up to 11 h after slicing. Recordings were done at 32–34 °C. Whole-cell recording pipettes (5–10 MΩ, 1–2 μm diameter) were filled with (in mM): KCl, 20; (K)Gluconate, 100; (K)HEPES, 10; (Mg)ATP, 4; (Na)GTP, 0.3; (Na)Phosphocreatine, 10; and 0.1% w/v biocytin, adjusted with KOH to pH 7.4, and with sucrose to 290–300 mOsm. Thick tufted L5 neurons were identified at 400X magnification using IR-DIC optics (Olympus BX-50; Olympus, Melville, New York, United States). To ensure that arborizations of recorded L5 neurons were minimally damaged during dissection, slices were used only if L5 apical dendrites were approximately parallel with the slice surface and could be traced most or all of the way to the pial surface. Gigaohm seals were then established on four neurons, after which breakthroughs were performed in quick succession. In some cases, one or two breakthroughs failed, thus yielding triple or double recordings; connections found in these cases were included in the dataset. Signals were amplified with AxoPatch 200B, AxoPatch-1B, and AxoClamp 2B amplifiers (Axon Instruments, Foster City, California, United States) and filtered at 5 kHz. Acquisition was done at 10 kHz using MIO-16E boards (National Instruments, Austin, Texas, United States) and custom software running on Igor Pro (WaveMetrics, Lake Oswego, Oregon, United States) on Macintosh computers (Apple Computer, Cupertino, California, United States). Recordings were terminated if membrane potential changed more than 8 mV or input resistance (measured from 250-ms-long 25 pA hyperpolarizing pulses preceding each trace) changed more than 30% from the baseline. Measurement of synaptic connection strengths We assessed connectivity by averaging ten or more traces. Synaptic connection strength was calculated by averaging the peak EPSP amplitudes (measured using a 1-ms-long window centered on the peak of the averaged EPSP trace) from 45 to 60 responses obtained during a 10- to 15-min-long baseline period just after breakthrough. In some cases (less than 5%), the EPSP amplitude was determined from fewer than 45 responses (although never fewer than 10 responses), typically because the recordings failed. The standard deviation of EPSP amplitude is within 0.04–1.4 mV and depends weakly on the mean EPSP amplitude (see Figure S8 ). As the averaged EPSP waveforms were time-locked to the presynaptic spike, the signal-to-noise ratio was good enough to allow for detection of synaptic connections with strengths as low as 0.01 mV. However, as only ten traces were averaged to determine connectivity, we might have missed connections with very low release probability. Analysis and statistics To evaluate correlations in synaptic connection strength, Pearson's R is calculated using the following standard formula: where X and Y are vectors of paired samples and N is the total number of pairs. The p- value score of significance is calculated based on Fisher's z -score calculated from R. For synaptic connection strengths of reciprocally connected pairs, assignment of connection strengths as X and Y would be arbitrary. Therefore, the R value calculated should not depend on the assignment. We use each pair of X, Y values twice when calculating the R score. For each pair X i , Y i , the pair constructed by flipping the order also entered in the formula; therefore, N is twice the total number of pairs. When calculating the p -value, the number N is taken to be the total number of pairs instead of twice the total number, so as not to overestimate the significance. The correlation scores and p -values calculated this way agree with those calculated from a reshuffling procedure by randomly assigning each neuron as either X or Y and using each pair only once. To analyze the distribution of synaptic connection strength, we generated fits to the data, by using a mean-square error-based procedure from the MATLAB curve-fitting toolbox in the log–log space. To test whether the experimental distribution has a longer tail than the exponential or Gaussian distribution, we chose a threshold T, and counted the number of experimental observations with higher value than T, and denoted it by n, out of a total of N observations. We then calculated the p -value as the probability of generating more than n observations with values larger than T out of N observations from the null distributions. To assess the monotonicity in Figures 7 and 8 , we used the Kolmogorov-Smirnov test. For each threshold of synaptic connection strength, we generated an ensemble of 1,000 random matrix sets with matched connection statistics as described in the Motif finding section below. We then computed the distribution of ratios between occurrence counts in the random ensemble and the observed occurrence counts for motif(s) of interest. These ratios are the inverses of the ratios plotted in Figures 2 and 4 in order to avoid division by zero. We then tested for monotonicity between successive pairs of distributions with the Kolmogorov-Smirnov test. To generate bootstrap distributions for a dataset with N observations, we drew an ensemble of 1,000 trials of N samples each from the dataset with replacement and computed the appropriate statistic on each trial. The statistics from these trials formed the bootstrap distribution. Mean and standard deviations were then computed on the bootstrap distribution of the chosen statistic. Motif finding To find overrepresented motifs, we used a statistic based on how the observed counts compare with the expected counts from the null hypothesis. For the null hypothesis, we generated B = 1,000 sets of random connection matrices. Each set contained as many matrices as the number of quadruplets in experimental data. The randomization procedure is as follows: In the two-neuron case, the probability that neuron A is connected with neuron B is the same as experimentally measured, and the connection from B to A is treated independently. In the three-neuron case, each neuron pair is treated as one unit, and the probabilities of having one-way and bidirectional connections within the pair are the same as measured. But how the three pairs form a triplet is random. In the four-neuron case, a 90 × 90 matrix of connections was generated. The fractional counts for each triplet motif to total triplet counts from this big matrix were matched to experiment data using a simulated annealing procedure (see [ 36 ]). To generate each random connection matrix in each of the B = 1,000 sets of matrices, we randomly picked four neurons from this 90 × 90 random connection matrix to form a 4 × 4 random connection matrix. This procedure matches the probability of observing a triplet motif to experimental data while randomizing how triplets combine to form quadruplets. For each motif, we counted the number of its occurrences in measured data and in each set of random matrices. The p -value for this motif is the fraction of random matrices with occurrence counts above or below the observed occurrence counts. This tests for significant deviation from random, including both overrepresentation and underrepresentation. Since we are testing for many motifs simultaneously, we applied the step-down min-P - based algorithm for multiple-hypothesis correction [ 70 , 71 , 72 ]. This procedure ensures weak control for the familywise error rate, which is defined as the probability of at least one type I error (stating that a pattern is overrepresented when it is not) among the family of hypotheses (all motifs). Weak control refers to the fact that type I error is controlled under the complete null hypothesis when all the null hypotheses are assumed to be false. Strong control, which is not used here, would control type I error rate under any combination of true and false null hypotheses, but is harder to achieve. The idea behind the step-down procedures is to order hypotheses according to the raw p -values in ascending order. Then for a chosen cutoff p -value, the hypotheses are considered successively. For each hypothesis, we test for the possibility of committing at least one type I error for the subset of hypotheses with lower or equal raw p -values. Further tests depend on the outcomes of earlier ones. As soon as one fails to reject a null hypothesis, no further hypotheses are rejected. The real procedure combines the testing for all cutoff p -values into one procedure, as described in more detail below. First, we test for M motifs with an ensemble R of B random matrix sets, R = { R b ,b ∈ {1,…, B }} generated as described above. For each motif i ∈ {1,…, M }, we calculate the mean occurrence counts over the ensemble and denote it with c¯ R i (step 1). Second, we calculate the raw p -values p * i for each motif i for the two-sided statistic T, defined as the absolute value of the difference between the observed counts c and the mean ensemble counts, T = | c − c¯ R i | . We calculate the proportion of sets of random matrices in the ensemble with a larger or equal value for statistic T than observed. for i = 1,…, M (step 2). Third, we then order the raw p -values such that p * k 1 > p * k 2 > … > p * k M (step 3). Fourth, for each R b ,b ∈ {1,…, B } and each motif k i ,i ∈ {1,…, M }, we repeat step 4: count its number of occurrences c k i ,b in R b , calculate T = | c k i ,b − c¯ R i | and the p -value, p k i ,b, as in step 2, and then compute q k i ,B = min l=k 1 ,…,k i p l,b , , the successive minima of the raw p -values (step4). Fifth, the corrected p -values p˜ k 1 are estimated by calculating the proportion of sets of random matrices in the ensemble in which q k i ,b is smaller than or equal to the observed p -value p * k 1 . for i = 1,…, M (step 5). Finally, we enforce the monotonicity constraints by successively setting p˜ k i to max( p˜ k i−1 , p˜ k i for i = 2,…, M (step 6). Statistical reconstruction of the network To generate Figure 9 , links were assigned randomly among 50 nodes with the experimentally measured probability of unidirectional and bidirectional connections. Strengths of connections were drawn from the experimentally measured distribution. Then we manually adjusted the connections to have roughly similar probability of occurrence of three-neuron motifs. In constructing this diagram, we assumed that each individual cell has the same distribution of strong and weak synaptic connections. This assumption could be violated if some cells have many stronger synaptic connections while others have few or none. Whether this is the case should be investigated in future studies. This figure is for illustration purposes only. Positions of recorded neurons To investigate the dependence of connectivity on pairwise distances, we measured the relative coordinates of the recorded cells from slices prepared by biocytin histochemistry after recordings. Distances were not corrected for tissue shrinkage. Since we were most interested in the relationship between connectivity and distance, an equal amount of shrinkage for all slices would not affect our results. Some inhomogeneities of shrinkage were likely, but we did not expect the shrinkage factor to vary greatly across slices. During recording sessions, the approximate relative positions of cells and the positions of recorded quadruplets in the slice were kept in notes. In most cases, these drawings allowed unambiguous identification of recorded cells, and cell positions were then measured on those cases after identification of the recorded cells. If a quadruplet was totally unconnected, drawings were not provided. However, totally unconnected cells did not have to be identified, and the assignment was made randomly. In some cases, some of the cells in the quadruplet were not well stained. If positions of at least three cells out of the quadruplet could be recovered, the positions of those cells were recorded. We defined the position of each cell as the three-dimensional coordinate of the axonal initial segment and measured it using the Neurolucida system (MicroBrightField, Williston, Vermont, United States). We estimate the measurement error to be less than 2 μm in X and Y positions and less than 3 μm in Z positions, based on repeated measurements of the same quadruplets. In about 10% of the cases, the initial segment of the axon was obscured by other cells and could not be positively identified, and the cell position was instead measured from the middle of the base of the cell body. In these cases, the measurement error could be as large as 3 μm in X and Y and 5 μm in Z. For each slice, we also estimated the average position of the main apical dendrites of the quadruplet around 300 μm away. The positive direction for the vector from the mean positions of the cells to the estimated average position of the main apical dendrites was defined to be the pia direction in the slice. We rotated the original relative coordinates of pairs in the X, Y plane so this vector pointed in the positive Y direction. We normalized the vector and defined it as the original coordinates of the new unit Y vector. The new unit X vector is the normal direction to the unit Y vector. To calculate the relative X, Y coordinates of two cells in the new coordinates, we took the dot product of the relative vector calculated in the original coordinates and the unit X and Y vectors, also defined in the original coordinates. The relative Z coordinates were not subject to rotation. Notice that rotation was done on the relative positions of any two cells and not on the positions of each individual cell. A total of 817 cells in 83 triplets and 142 quadruplets were measured, resulting in a total of 2,202 possible connections. For each possible connection, the relative position of the target neuron to the originating neuron was plotted in Figure 3 A. If a connection was present and was involved in a bidirectional connection, the position was indicated with red. If a connection was present, and the reciprocal connection was not present, the position was indicated with blue. If a connection was not present, regardless of the status of the reciprocal connection, the position was indicated with green. Most of the cells included in the dataset came from nearby positions (<50 μm, 82% of pairs), with the remaining 18% of the connections in the 50 μm–110 μm range ( Figure 3 B). The densities of red, green, and blue connections are proportional to each other regardless of the distance from the soma. The connection probability is mostly uniform within the range of three-dimensional distances recorded in the experiments ( Figure 3 C). However, the connection probability for all distances (0.013) is slightly higher than that calculated for the entire dataset (0.0116). This is likely due to less efficient recovery of unconnected quadruplets, as less care was taken to preserve them. The connection strengths do not vary with distance systematically (data not shown), although the distance does seem to have an effect ( p = 0.02, one-way analysis of variance). To control for cutting artifacts, we have measured the closest cut ending of the main axons out of the four cells in the quadruplet to the mean position of the cell bodies. The main axons go toward white matter to innervate subcortical structures. When the distance is small, then we might have cut off more portions of the axonal arbor, and cutting artifact might be a concern. However, since the main axons start branching approximately 100 μm from the cell bodies [ 73 ], the local connectivity might not be greatly affected (see Figure S3 ). We also measured the Z coordinate of the slice surface. From this coordinate and the coordinate of the cells, we can deduce the depth of each cell from the slice surface. Cells closer to the surface might have had larger portions of their axonal arbors cut off, which would reduce their connectivity. However, the connectivity seems to be fairly uniform regardless of the depth of either the originating cell or the target cell (see Figure S3 ). The caveat is that the measurements we have taken from the fixed slices are uncorrected for shrinkage, and differential shrinkage in the Z direction might have randomized a trend that might otherwise be present. Supporting Information Appendix S1 Properties of the Lognormal Distribution (39 KB DOC). Click here for additional data file. Figure S1 Connectivity and Mean Synaptic Strengths of the Connections Are Uniform for All Distances between a Pair of Neurons For the x-axis, positive means the receiving cell is to the right of the sending cell in the slice. For the y-axis, positive means the receiving cell is above the sending cell. For the z-axis, positive means the receiving cell is on top of the sending cell. (A, D, G, and J) Connection probability for X, Y, and Z and angle between the vector connecting two neurons and the x-axis separately (no significant variation; all chi square tests, p > 0.05). Error bars are 95% confidence intervals estimated from binomial distributions. (B, E ,H , and K) Mean synaptic strengths for X, Y, and Z and angle separately (no significant variation; all one-way ANOVA tests, p > 0.05). Error bars are standard errors of the mean. (C, F, I, and L) Histogram of connections for X, Y, and Z and angle separately. (363 KB DOC). Click here for additional data file. Figure S2 The Overrepresentation of Bidirectionally Connected Pairs Is Not Due to Inhomogeneous Connection Probabilities for Neurons of Different Distances Counts relative to random are shown for neurons of different distances. The red line indicates the value calculated for pooled data for neurons of all distances. Notice that the values calculated for certain distances are very similar to that calculated for pooled data. However, the ratio calculated for all distances (3.0) is a bit lower than that calculated for the entire dataset (4.0), probably owing to increased connection probability (0.013 versus 0.0116) caused possibly by inefficient recovery of totally unconnected quadruplets. Error bars are standard deviations from bootstrap. (119 KB DOC). Click here for additional data file. Figure S3 Connection Probability and Mean Synaptic Connection Strengths Are Not Greatly Modified by Cutting (A) Connection probability is uniform with regard to the distance to the closest main axon cut ending ( p = 0.077, chi square test). Notable exception is distances more than 600 μm away, where the connection probability seems to be slightly increased. However, since there are relatively few neurons with axon cut distance of more than 600 μm and the increase in connection probability is not statistically significant, we do not expect this to fully explain our results. (B) Mean synaptic connection strength does not vary systematically with regard to the distance to the closest main axon cut ending (however, mean strength depends on distance; p = 0.02 by one-way ANOVA). (C) Histogram of neurons with certain axon cut distances. (D) Connection probability is uniform with regard to the depth of both the neuron sending the connection and the neuron receiving the connection ( p = 0.99, chi square test). (E) Mean synaptic connection is uniform with regard to the depth of both the neuron sending the connection and the neuron receiving the connection ( p = 0.2, one-way ANOVA). (F) Histogram of recorded neurons with certain depth. Error bars in (A) and (D) are 95% confidence intervals estimated from binomial distribution. Error bars in (B) and (E) are standard errors of the mean. (244 KB DOC). Click here for additional data file. Figure S4 EPSP Size and Rate of Connectivity Does Not Significantly Depend on Animal Age (A and B) We found no statistically significant difference among EPSP amplitudes for animals of different ages (one-way ANOVA in log space, p = 0.36). We note, however, that there is a weak downward trend, in agreement with the observation of [ 34 ] that L5-to-L5 synaptic strength is significantly weaker in P28 animals than in P14 animals. Error bars are standard errors of the mean. (C) The connectivity rate does not depend on animal age (chi square test, p = 0.92). Error bars are 95% confidence intervals calculated from binomial distribution. (218 KB DOC). Click here for additional data file. Figure S5 Main Results of This Paper Are Still Valid for the Subset of Data from P14–P16 Animals (A and B) Overrepresentation of bidirectional connections and highly connected triplets. Numbers on top of bars are actual counts. (C) Synaptic connection strengths are well fit by the lognormal distribution. Number on top of dots are actual counts (not shown when greater than 50). (D) Probability of significant deviation from random for a given triplet motif. (E and F) Increase in overrepresentation of bidirectional connections and highly connected triplets for increasingly higher connection strength thresholds. (847 KB DOC). Click here for additional data file. Figure S6 Quadruplet Catalogue (1.2 MB DOC). Click here for additional data file. Figure S7 Synaptic Strengths of Incoming and Outgoing Connections Are Weakly Correlated (A) Scatter plot of incoming synaptic connection strength. A weak correlation of 0.2 is observed ( p = 0.029). (B) Scatter plot of outgoing synaptic connection strength. A weak correlation of 0.17 is observed ( p = 0.054). (C) Scatter plot of outgoing and incoming synaptic connection strength. A weak correlation of 0.13 is observed ( p = 0.039). All correlations calculated using Pearson's R method in log space. (231 KB DOC). Click here for additional data file. Figure S8 EPSP Standard Deviation Depends Weakly on EPSP Amplitude (A) EPSP standard deviation depends weakly on EPSP amplitude. (B) Coefficient of variation is inversely proportional to the EPSP amplitude. Note the log–log scale. (128 KB DOC). Click here for additional data file. Table S1 Quadruplet Counts Quadruplets are numbered according to the catalogue in Figure S1 . (124 KB DOC). Click here for additional data file.

D:\keerthana\PMC001xxxxxx\PMC1054880.xml

1054881

Dissociation of Cohesin from Chromosome Arms and Loss of Arm Cohesion during Early Mitosis Depends on Phosphorylation of SA2

Cohesin is a protein complex that is required to hold sister chromatids together. Cleavage of the Scc1 subunit of cohesin by the protease separase releases the complex from chromosomes and thereby enables the separation of sister chromatids in anaphase. In vertebrate cells, the bulk of cohesin dissociates from chromosome arms already during prophase and prometaphase without cleavage of Scc1. Polo-like kinase 1 (Plk1) and Aurora-B are required for this dissociation process, and Plk1 can phosphorylate the cohesin subunits Scc1 and SA2 in vitro, consistent with the possibility that cohesin phosphorylation by Plk1 triggers the dissociation of cohesin from chromosome arms. However, this hypothesis has not been tested yet, and in budding yeast it has been found that phosphorylation of Scc1 by the Polo-like kinase Cdc5 enhances the cleavability of cohesin, but does not lead to separase-independent dissociation of cohesin from chromosomes. To address the functional significance of cohesin phosphorylation in human cells, we have searched for phosphorylation sites on all four subunits of cohesin by mass spectrometry. We have identified numerous mitosis-specific sites on Scc1 and SA2, mutated them, and expressed nonphosphorylatable forms of both proteins stably at physiological levels in human cells. The analysis of these cells lines, in conjunction with biochemical experiments in vitro, indicate that Scc1 phosphorylation is dispensable for cohesin dissociation from chromosomes in early mitosis but enhances the cleavability of Scc1 by separase. In contrast, our data reveal that phosphorylation of SA2 is essential for cohesin dissociation during prophase and prometaphase, but is not required for cohesin cleavage by separase. The similarity of the phenotype obtained after expression of nonphosphorylatable SA2 in human cells to that seen after the depletion of Plk1 suggests that SA2 is the critical target of Plk1 in the cohesin dissociation pathway.

Introduction Faithful inheritance of the genome depends on its accurate replication and correct distribution to the two daughter cells. In eukaryotes, the two copies of a chromosome that are generated in S-phase (sister chromatids) remain connected until they are separated in anaphase of mitosis. This physical association (cohesion) allows the mitotic segregation machinery to handle sister chromatids as entities that have to be distributed to opposite poles. Sister chromatid cohesion depends on cohesin, a protein complex that is highly conserved in evolution and consists of at least four subunits: two “structural maintenance of chromosomes” proteins, Smc1 and Smc3, the so-called “kleisin” subunit Scc1 (also called Rad21 or Mcd1), and Scc3 (reviewed in [ 1 ]). Cells of humans, Xenopus, and other higher eukaryotes contain two mitotic orthologs of Scc3, called SA1 and SA2. Cohesin complexes in these cells contain either SA1 or SA2, but not both [ 2 , 3 ]. In order to segregate sister chromatids to opposite poles in anaphase, cohesin has to be removed from chromosomes. In budding yeast, the prevalent mode of cohesin removal is by proteolytic cleavage of the Scc1 subunit at the onset of anaphase by the endopeptidase separase [ 4 , 5 ]. Prior to anaphase, separase is kept inactive by its inhibitor securin [ 5 , 6 , 7 , 8 , 9 , 10 ], and in vertebrate cells also by inhibitory phosphorylation mediated by Cdk1 [ 11 ]. Both securin and Cdk1's activating subunit cyclin B are ubiquitinated at the onset of anaphase by the anaphase-promoting complex/cyclosome, leading to their proteasome-dependent degradation and to separase activation (reviewed in [ 12 ]). In higher eukaryotes, removal of cohesin from chromosomes occurs in at least two steps. During prophase and prometaphase, the bulk of cohesin dissociates from chromosomes without Scc1 cleavage [ 3 , 13 ]. Only minor amounts of cohesin (an estimated 10%) remain on chromosomes up to metaphase, preferentially at centromeres [ 10 , 14 ]. A similarly minor amount of cohesin, presumably the chromosome-bound fraction, is cleaved by separase at the onset of anaphase [ 10 ]. As in budding yeast, the cleavage of Scc1 is essential for anaphase to occur [ 15 ]. Two mitosis-specific protein kinases are required for the cleavage-independent removal of cohesin from chromosome arms: Plk1, called Plx1 in Xenopus, and Aurora-B [ 16 , 17 , 18 ]. Plk1/Plx1, but not Aurora-B, can phosphorylate the cohesin subunits Scc1 and SA2 in vitro [ 16 , 17 ], and is required for their phosphorylation in Xenopus egg extracts [ 16 ]. In these extracts, the ability of cohesin to bind to chromatin correlates inversely with its phosphorylation state [ 16 ]. This observation, and the finding that Plk1 is required for cohesin dissociation from chromosomes, raise the possibility that phosphorylation of cohesin by Plk1 leads to its cleavage-independent dissociation from chromosomes. However, it is unknown whether Plk1's critical target in the cohesin dissociation process is cohesin itself, and whether cohesin phosphorylation is required for dissociation of the complex from chromosomes. So far, the functional relevance of cohesin phosphorylation has been studied only in budding yeast. In this organism, Scc1 is also phosphorylated by a Polo-like kinase, called Cdc5, but this modification does not seem to result in cohesin's dissociation from chromatin; rather, it renders Scc1 more susceptible to cleavage by separase [ 19 , 20 ]. To test whether cohesin phosphorylation is required for its dissociation from chromosome arms during prophase and prometaphase, and to address whether this modification could explain the requirement for Plk1 in this process, we have searched for phosphorylation sites on all four subunits of the human cohesin complex by mass spectrometry. We have identified numerous mitosis-specific sites on Scc1 and SA2, mutated them, and expressed wild-type and nonphosphorylatable forms of both proteins stably at physiological levels in cultured human cells. The analysis of these cell lines, in conjunction with biochemical experiments in vitro, imply that Scc1 phosphorylation is dispensable for cohesin dissociation from chromosomes in early mitosis, but enhances the cleavability of Scc1 by separase. In contrast, our data reveal that phosphorylation of SA2 is essential for cohesin dissociation during prophase and prometaphase, but is not required for cohesin cleavage by separase. The similarity of the phenotype obtained after expression of nonphosphorylatable SA2 (this study) to that seen after the depletion of Plk1 [ 18 ] strongly suggests that SA2 is the critical target of Plk1 in the cohesin dissociation pathway. Results Identification of Mitosis-Specific Phosphorylation Sites on Human Cohesin The SA1, SA2, and Scc1 subunits of vertebrate cohesin complexes are phosphorylated specifically in mitosis [ 2 , 14 , 16 ]. To be able to analyze the functional relevance of these modifications, we mapped mitosis-specific phosphorylation sites on human cohesin by mass spectrometry. We prepared lysates from HeLa cells that had been arrested either at the G1/S transition (interphase) by hydroxyurea or in mitosis by the spindle poison nocodazole, and immunoprecipitated SA1- and SA2-containing cohesin complexes with antibody 447, which recognizes the C termini of both SA1 and SA2 [ 3 ]. Separation of the isolated proteins by SDS-PAGE and staining with silver ( Figure 1 A) or cohesin-specific antibodies ( Figure 1 B) demonstrated reduced electrophoretic mobility of both SA1 and SA2 when the complexes had been isolated from mitotic cells, indicating that the mitosis-specific phosphorylation of these subunits was preserved during the isolation procedure. This notion was confirmed by immunoblotting with phosphothreonine-specific antibodies. This assay yielded a pattern that was consistent with the presence of phosphothreonine residues on SA1, SA2, and Scc1 in mitosis, whereas no phosphothreonine signal was detected on cohesin subunits isolated from interphase cells ( Figure 1 B). Figure 1 Identification of Mitosis-Specific Phosphorylation Sites on Human Cohesin (A and B) Cohesin was immunoprecipitated by antibody 447 (which recognizes SA1 and SA2) from extracts prepared from HeLa cells that were either arrested in S-phase by hydroxyurea (HU) or in mitosis by nocodazole (Noc). Cohesin was eluted by buffer of low pH and analyzed by (A) silver staining and (B) immunoblotting with antibodies to cohesin subunits and phosphorylated threonine (P-Thr). (C) Schematic representation of the phosphorylation sites on Scc1 and SA2 that were identified by mass spectrometry, and of the mutant versions of the proteins that have been generated. The star indicates a phosphorylation site that was found in both interphase and mitotic Scc1. All SA2 constructs used for in vitro experiments lack the 69 N-terminal amino acids. SA2-WT-myc and SA2–12xA-myc cell lines contain the entire open reading frame of 1,231 amino acids. To identify phosphorylation sites, purified cohesin complexes were digested with various proteases in solution and analyzed by high performance liquid chromatography-electrospray ionization-tandem mass spectrometry (HPLC-ESI-MS/MS). In total, 28 phosphorylated serine or threonine residues were identified. Of these, 18 could be assigned unequivocally to specific residues, whereas in ten cases it could not be determined which of several serine and threonine residues in a given peptide was phosphorylated ( Figure 1 C and Table 1 ). Two sites each were found in Smc1 and Smc3, ten in Scc1, and 14 in SA2. We were unable to identify phosphorylated peptides derived from SA1, presumably because SA1-containing cohesin complexes are much less abundant than SA2-containing complexes [ 2 , 3 ]. By analyzing cohesin peptides obtained by digestion with various proteases, we were able to obtain a sequence coverage of approximately 80% for each of the subunits ( Figure S1 and unpublished data). This indicates that our analysis was theoretically able to identify the majority of in vivo phosphorylation sites on cohesin subunits, although we cannot exclude the possibility that some sites have gone undetected. Table 1 Phosphorylation Sites Identified in Cohesin Subunits Where two or more residues are marked, the phosphorylation site could not be unequivocally assigned to any one of the adjacent residues +, phosphorylated; ++, highly phosphorylated DOI: 10.1371/journal.pbio.0030069.t001 The phosphorylation sites we identified on Smc1 and Smc3 were present in complexes from both interphase and mitotic cells, and we therefore did not analyze them any further. Two of these sites (Ser 966 and Ser 957 in Smc1) have been shown to be phosphorylated by the ATM kinase in response to DNA damage [ 21 , 22 ]. Only one of the sites in Scc1 (Ser 153 ), and none in SA2, was found phosphorylated in interphase. All other sites were only identified in cohesin samples from mitotic cells. These results confirm that Scc1 and SA1/2 are specifically phosphorylated in mitosis. Since Plk1/Plx1 can phosphorylate Scc1 and SA1/SA2 in vitro and is required for the phosphorylation of these subunits in Xenopus egg extracts [ 16 ], we compared the sequence surrounding the identified sites to the consensus sequences that have been reported as binding and phosphorylation sites for Plk1. The C-terminal Polo box domain of Plk1 binds to phosphorylation sites for which the consensus S-[pT/pS]-[P/x] has been proposed [ 23 ]. These phosphorylation sites are thought to be generated by proline-directed kinases, such as Cdk1, that “prime” the substrate for subsequent recognition by Plk1. Once bound to the substrate, Plk1 is thought to phosphorylate sites that are distinct from the one that is recognized by the Polo box. Based on in vitro phosphorylation experiments using peptides derived from the phosphatase Cdc25C, Nakajima et al. [ 24 ] have proposed the consensus (E/D)-x-(S/T)Φ-x-(E/D) for Plk1 phosphorylation-sites, where Φ signifies a hydrophobic amino acid, whereas Barr et al. [ 25 ] have suggested the consensus (E/D/Q)-x-(S/T)Φ. One putative phosphorylation site in Scc1 (Thr 186 ) and three putative sites in SA2 (Ser 1065 , Thr 1124 , and Ser 1178 ) match the Polo box-binding consensus sequence, although none of these sites contains a proline residue at the +1 position (these are putative sites because they are present in peptides in which the exact identity of the residue carrying the phosphomoiety could not be determined; Table 1 ). We also note that, although Cdk1 can phosphorylate SA1/2 in vitro [ 2 ], it does not appear to be essential for cohesin dissociation in human cells and Xenopus egg extracts [ 3 , 16 ]. Two phosphorylation sites in Scc1 (Thr 144 and Thr 312 ) match the consensus proposed by Nakajima et al. [ 24 ]. These two sites, in addition to one in Scc1 (Ser 454 ) and three in SA2 (Thr 1109 , Ser 1137 , and Ser 1224 ) conform with the consensus proposed by Barr et al. [ 25 ]. These findings are consistent with the possibility that at least some of the sites in Scc1 and SA2 are directly phosphorylated by Plk1. Alignment of multiple sequences showed that most of the amino acid residues that we found to be phosphorylated in mitotic human Scc1 and SA2 were conserved in orthologs from other vertebrates, but not in more distantly related eukaryotes such as Drosophila, Caenorhabditis elegans, or yeast, where the homology to human cohesin subunits is in overall much lower (unpublished data). When we considered the distribution of the sites, we found that Scc1 phosphorylation sites strikingly clustered around the two separase cleavage sites, whereas SA2 phosphorylation sites were concentrated in the C terminus of the protein ( Figure 1 C). Clustering of Scc1 phosphorylation sites in the vicinity of separase cleavage sites has also been observed in budding yeast [ 19 ]. This indicates that, although individual sites on Scc1 are not conserved from yeast to human, the overall pattern of phosphorylation is conserved. Evolutionary conservation of the overall mitotic phosphorylation pattern, but not of individual sites, has also been observed in the case of the anaphase-promoting complex/cyclosome [ 26 ]. Phosphorylation of Scc1 Is Required for Efficient Cleavage by Separase In Vitro In budding yeast, phosphorylation of Scc1 by the Polo-like kinase Cdc5 enhances the rate of cleavage by separase both in vitro and in vivo [ 19 , 20 ]. We therefore analyzed whether treatment of human Scc1 with Plk1 similarly increases its cleavability by separase ( Figure 2 ). We activated purified human separase-securin complexes by securin destruction in mitotic Xenopus egg extracts and incubated the activated separase with 35 S-labeled recombinant Scc1 as a substrate. Under these conditions, separase cleaves Scc1 at the same sites that are cleaved at anaphase onset in vivo, Arg 172 and Arg 450 ([ 15 ]; Figure 2 A). When untreated Scc1 was incubated with separase, efficient cleavage could only be detected at the more N-terminal site, resulting in the formation of one N-terminal and one C-terminal cleavage product which could be seen by Phosphorimager analysis ( Figure 2 A and 2 B). The C-terminal cleavage product could also be detected by immunoblotting using antibodies to a myc-epitope tag on the C-terminus of Scc1 ( Figure 2 A). When the cleavage reaction was carried out in the presence of active recombinant Plk1, cleavage at the first Scc1 site was slightly enhanced, and cleavage at the second site now became apparent ( Figure 2 A and 2 B), consistent with the possibility that phosphorylation of Scc1 by Plk1 increases the cleavability of Scc1. A cleavage product that corresponds to the fragment in between the two cleavage sites could not be detected under these conditions ( Figure 2 A), indicating that separase cleaves either one or the other site in one Scc1 molecule, but not or only rarely both. Figure 2 Plk1 Facilitates Cleavage of Human Scc1 by Separase In Vitro (A) Recombinant, 35 S-labeled, wild-type and mutant Scc1 (see Figure 1 C) tagged with 9xmyc at the C terminus were incubated with human separase. Recombinant human GST-Plk1 was added to the reaction mixtures where indicated. Samples were withdrawn from the reactions at the indicated time points and analyzed by SDS-PAGE followed by immunoblotting (anti-myc) and Phosphorimager analysis ( 35 S exposure). Arrows indicate full length Scc1-myc (fl), C- and N-terminal fragments resulting from cleavage at Arg 172 (Ct #1, Nt #1, respectively), and C- and N-terminal fragments resulting from cleavage at Arg 450 (Ct #2, Nt #2, respectively). The lower parts of the membrane or gels were exposed longer than the upper parts. The enhancement of cleavage at Arg 172 by Plk1 can be seen particularly well by comparing the intensities of the N-terminal fragments (Nt #1). Note that in the autoradiographs a band can be seen (particularly clearly in the lanes representing the zero time points) that has almost the same electrophoretic mobility as cleavage product Ct #1. This band is distinct from Ct #1 because it migrates a slightly shorter distance and because it is also present in the absence of separase. This band was therefore not included in the quantification in (B). (B) Quantification of the abundance of Scc1-myc and the Scc1-myc cleavage fragments in the assay shown in the left autoradiograph of (A). For the quantification, autoradiographs of identical exposure were used. The sum of the intensities of full-length and all cleavage fragments was set to 100%. Signal intensities for N- and C-terminal fragments resulting from cleavage at the same site were summed. (C) Chromatin fractions were prepared from HeLa cells stably expressing either wild-type Scc1-myc or the mutant Scc1-S 454 A-myc, and were incubated in either interphase or mitotic Xenopus egg extract. Mitosis-specific cleavage of Scc1 was detected by immunoblotting with myc antibodies. We suspected that the observed enhancement of Scc1's cleavability in the presence of Plk1 might be due to phosphorylation at two sites that are directly adjacent to the cleavage sites, Ser 175 and Ser 454 , which we had found to be phosphorylated in mitosis in vivo ( Table 1 ). We therefore mutated these two residues to alanine, thereby creating mutant Scc1-S 175 A/S 454 A (see Figure 1 C), and tested the cleavability of this mutant in the absence or presence of Plk1 in vitro. Scc1-S 175 A/S 454 A could still be cleaved at the first site, and this reaction was still slightly enhanced in the presence of Plk1, but cleavage at the second site was completely abolished even in the presence of Plk1 ( Figure 2 A and 2 B). This result suggested that cleavage at Arg 450 of Scc1 by separase depends on phosphorylation of Ser 454 . The analysis of a Scc1 mutant in which only Ser 454 was changed to alanine (Scc1-S 454 A) confirmed this notion (unpublished data; see Figure 2 C for a related result). Next, we asked whether the enhancement of cleavage at Arg 172 upon incubation with Plk1 is due to phosphorylation of any of the other residues we had identified. We therefore created a Scc1 mutant in which nine of the ten identified phosphorylation sites were mutated to alanine (Scc1–9xA; see Figure 1 C). Mutation of the tenth site was not possible for technical reasons. When Scc1–9xA was incubated with active separase, we found, as expected, that cleavage at Arg 450 was abolished, but cleavage at Arg 172 did occur and was still enhanced by the addition of Plk1 (unpublished data). At present, it is therefore unclear how Plk1 enhances cleavage of Scc1 at Arg 172 . The effect could be due to phosphorylation of one or more of the serine and threonine residues at positions 185–189 ( Table 1 ), which were not included in our mutational analysis; alternatively, it might be due to phosphorylation on one or more residues that our mass spectrometry analysis failed to identify. It also remains a formal possibility that Plk1 enhances cleavage at Arg 172 by phosphorylating separase rather than cohesin. Scc1 Phosphorylation Is Not Essential for the Dissociation of Cohesin from Chromosome Arms and for Progression through Mitosis To address the physiological significance of Scc1 phosphorylation and the resulting enhanced cleavability of Scc1, we generated cell lines that stably express Scc1-S 454 A or Scc1–9xA ( Figure 3 ; see also Figure 1 C). To be able to distinguish the ectopically expressed Scc1 proteins from endogenous Scc1, we tagged the Scc1 mutants with nine myc epitopes at their C termini. We have shown that such a tag does not detectably compromise the ability of Scc1 to assemble into cohesin complexes [ 10 ] that can establish cohesion [ 15 ]. Since expression of mutant Scc1 could have deleterious effects on cells, we furthermore used a doxycycline-regulatable promoter to be able to control the level of expression. To avoid potential overexpression artifacts, we screened by immunoblotting for cell lines in which the ectopically expressed Scc1 is present in amounts that are similar to the amounts of endogenous Scc1 when expression is fully induced by doxycycline ( Figure 3 A). Immunoprecipitation of the ectopically expressed Scc1 with myc antibodies, followed by SDS-PAGE and silver staining, revealed that the mutated forms of Scc1 could associate with Smc1, Smc3, and SA1/2 into cohesin complexes ( Figure 3 B; note that endogenous untagged Scc1 does not coimmunoprecipitate with Scc1-myc, indicating that only one Scc1 molecule is present per cohesin complex). Sucrose density gradient centrifugation experiments in conjunction with immunoblotting showed that most of the ectopically expressed Scc1 was incorporated into cohesin complexes ( Figure S2 A). These observations indicate that any phenotypes (but also the absence of phenotypes) that are observed after ectopic expression of Scc1 are not simply caused by overexpression or by the inability of Scc1 to assemble into cohesin complexes. Figure 3 Characterization of HeLa Cell Lines Stably Expressing Wild-Type or Mutant Forms of Human Scc1 and SA2 (A) Wild-type Scc1 or SA2, or the indicated mutant proteins (see Figure 1 C), all tagged with 9xmyc at the C terminus, were stably and inducibly expressed in HeLa tet-on cells. After induction by treatment with 2 μg/ml doxycycline for 1–3 d, cell extracts were prepared from either logarithmically proliferating cells (i, interphase) or from cells arrested in mitosis by nocodazole (m, mitosis), then immunoblotted. In the case of Scc1 cell lines (upper blots), only data from interphase extracts are shown. Exogenous protein was detected by immunoblotting with myc antibodies (lower blots). Since the 9xmyc-tag caused a reduced mobility in SDS-PAGE compared to the endogenous protein, Scc1- and SA2-immunoblots (upper blots) revealed the relative amounts of exogenous and endogenous protein in the different cell lines. The position of molecular weight markers is indicated on the right side. (B) Extracts were prepared from the different cell lines as indicated. Immunoprecipitation was performed using myc antibodies, followed by SDS-PAGE and silver staining. As a control, the cohesin complex was immunoprecipitated from untransfected HeLa tet-on cells using antibodies to SA2. (C) Extracts were prepared from SA2-WT-myc or SA2–12xA-myc expressing cells, and fractionated by sucrose density gradient centrifugation (5%–30% sucrose), followed by immunoblotting with antibodies recognizing the proteins indicated on the right (inp. = input/unfractionated sample of the extract). To analyze whether mutation of Ser 454 abolishes cleavage of Scc1 at Arg 450 also when Scc1 is part of cohesin complexes that have been loaded onto chromatin in vivo, we incubated chromatin from HeLa cells expressing wild-type Scc1, Scc1-S 454 A, or Scc1–9xA in Xenopus egg extracts. In this assay, Scc1 is cleaved by separase at Arg 172 and Arg 450 if the Xenopus egg extract is in a mitotic state [ 10 ], presumably because separase and Plx1 are active under these conditions. Whereas wild-type Scc1 was efficiently cleaved at both sites in mitotic egg extract, we did not observe any fragment resulting from cleavage at Arg 450 when either Scc1-S 454 A or Scc1–9xA was analyzed ( Figure 2 C and unpublished data), further supporting the conclusion that phosphorylation at Ser 454 of Scc1 is essential for cleavage at Arg 450 . Immunofluorescence analysis demonstrated that the localization of Scc1-S 454 A and Scc1–9xA was very similar to the one of wild-type Scc1 ( Figure S2 B and unpublished data). Both mutants were present in nuclei from telophase through interphase until the next mitosis. In prometaphase, the bulk of mutant cohesin complexes had dissociated from chromosome arms, but small amounts remained at centromeres, even if prometaphase was prolonged by treatment of the cells with nocodazole ( Figure S2 B). Scc1 phosphorylation at the nine mutated sites is therefore essential neither to load cohesin onto chromatin nor to remove cohesin from chromosome arms in early mitosis. Furthermore, we were unable to observe obvious abnormalities at later stages of mitosis and in the overall ability of cells expressing nonphosphorylatable Scc1 to proliferate (unpublished data), indicating that Scc1 phosphorylation at the mutated sites is not essential for the ability of separase to cleave cohesin complexes and to initiate anaphase. The finding that cells expressing Scc1 in which residue 454 cannot be phosphorylated and in which cleavage at Arg 450 is therefore compromised (see Figure 2 C) do not show anaphase defects is in agreement with our previous observation that Scc1 cleavage at Arg 172 is sufficient for the viability of HeLa cells [ 15 ]. Phosphorylation of SA2 Is Essential for the Dissociation of Cohesin from Chromosomes during Prophase and Prometaphase The observation that nonphosphorylatable Scc1 mutants bind chromatin in interphase and dissociate from chromosome arms normally in early mitosis implied that the requirement for Plk1 in cohesin dissociation and the inhibitory effect of cohesin phosphorylation on chromatin binding [ 16 , 18 ] cannot be explained by Scc1 phosphorylation. We therefore asked whether phosphorylation of SA2 might control the association of cohesin with chromosomes. We generated a mutant of SA2 in which serine/threonine residues at 12 sites found to be phosphorylated in mitosis in vivo were mutated to alanine; this mutant was called SA2–12xA (see Figure 1 C). We C-terminally tagged this mutant and wild-type SA2 with myc epitopes and expressed both in HeLa cells in a stable and inducible manner, employing the same strategy that we had used for Scc1. Also in this case we isolated cell lines in which the levels of ectopically expressed SA2 are similar to the levels of endogenous SA2 ( Figure 3 A). Immunoprecipitation experiments with myc antibodies and sucrose density gradient centrifugation showed that both the tagged SA2 wild-type and SA2–12xA proteins were incorporated into cohesin complexes ( Figure 3 B and 3 C). Since SA2 phosphorylation can be catalyzed by purified Plk1 in vitro and depends on Plx1 in Xenopus egg extracts [ 16 ] we asked whether SA2–12xA had lost the ability to be phosphorylated by Plk1. When purified cohesin complexes were incubated with Plk1 and 32 P-γ-ATP, approximately 50% less radiolabel was incorporated into SA2–12xA than into wild-type SA2, whereas phosphorylation of Scc1 in the same complexes was not affected ( Figure 4 A). Because both budding yeast and human Polo-like kinases can phosphorylate many sites in vitro that are not phosphorylated in vivo [ 26 , 27 ], it is possible that the residual phosphorylation of SA2–12xA by Plk1 in vitro occurs on sites that are not phosphorylated in vivo. We therefore analyzed the phosphorylation of wild-type and mutant forms of SA2 under more physiological conditions by incubating 35 S-labeled recombinant SA2 in mitotic Xenopus egg extracts. SA2 phosphorylation in these extracts depends on Plx1 [ 16 ] and causes an electrophoretic mobility shift ( Figure 4 B). This shift was partially abolished when three C-terminal phosphorylation sites were mutated (mutant SA2–6/7/11; see Figure 1 C), and no shift could be seen with SA2–12xA ( Figure 4 B). These results indicate that most mitosis-specific phosphorylation sites have been removed from SA2–12xA. The finding that mitosis-specific phosphorylation sites in SA2 are clustered in the C terminus was also confirmed by generating deletion mutants in which SA2 was progressively shortened from the C terminus (see Figure 1 C), because SA2's mobility shift was completely abolished when at least 124 C-terminal residues were removed ( Figure 4 C). Figure 4 Mutations of the C-Terminal Phosphorylation Sites and C-Terminal Deletions Decrease the Mitotic Phosphorylation of SA2 (A) Cohesin complexes containing wild-type SA2-myc or SA2–12xA-myc were immunopurified by myc antibodies from the respective cell lines. Similar amounts of cohesin were incubated with recombinant GST-Plk1, and the amount of phosphorylation was quantified by 32 P incorporation followed by SDS-PAGE, silver staining (top) and Phosphorimager analysis, and quantification using the software program ImageJ. (B and C) In vitro-translated, 35 S-labeled SA2 tagged at the N terminus with 9xmyc was incubated in interphase Xenopus egg extracts, which were induced to enter mitosis at time point 0 min by addition of nondegradable cyclin B Δ90 and okadaic acid. Samples were collected at the indicated time points and analyzed by SDS-PAGE followed by Phosphorimager analysis. The autoradiographs in (B) show phosphorylation site mutants and in (C) they show C-terminal deletion mutants (see Figure 1 C). The slower-migrating band represents myc-SA2, whereas the faster-migrating band is presumably generated by translation initiation at an internal start codon. To be able to address the physiological relevance of SA2 phosphorylation, we first had to determine whether cohesin complexes containing tagged SA2 behave normally in human cells. Immunofluorescence imaging of wild-type SA2-myc showed a localization pattern that is very similar to the pattern found for Scc1 [ 10 , 14 , 15 , 18 ]. SA2-myc was nuclear in interphase, and mainly present in a soluble form in the cytoplasm during mitosis (unpublished data). When we extracted mitotic cells to remove the soluble pool, we found a minor fraction bound to chromatin, and in prometaphase and metaphase, SA2-myc was enriched at centromeres as compared to chromosome arms ( Figure 5 A). Therefore, the tag in SA2 does not seem to interfere with the behavior of cohesin complexes containing SA2-myc. Figure 5 Phosphorylation of SA2 Is Required for Cohesin Dissociation from Chromosome Arms during Prometaphase (A) Logarithmically proliferating HeLa cells expressing SA2-WT-myc or SA2–12xA-myc were extracted prior to fixation, and stained with myc antibodies. In the upper set of images, kinetochores were labeled with human CREST serum, and DNA was counterstained with DAPI. In the lower set of images, only SA2-myc staining is shown. (B) SA2-myc expression was induced by different amounts of doxycycline (Dox.), and cells were arrested in prometaphase by nocodazole (Noc.) treatment for 3 or 10 h. Cells were spun on glass slides, extracted by detergent, fixed, and processed for immunostaining as in (A). Scale bars 10 μm. When we analyzed the intracellular distribution of the SA2–12xA mutant by immunofluorescence microscopy, we found that this protein was also nuclear throughout interphase and that the nuclear signal could only partially be reduced by removing soluble cohesin complexes by preextraction (unpublished data), indicating that complexes containing SA2–12xA can associate with chromatin like wild-type cohesin. In stark contrast to wild-type SA2, however, SA2–12xA was not strongly enriched at centromeres of prometaphase and metaphase chromosomes, but instead was almost equally abundant on chromosome arms and on centromeres ( Figure 5 A). A very similar distribution of cohesin on chromosome arms and centromeres has been observed after depletion of Plk1 by RNA interference [ 18 ]. These observations indicate that phosphorylation of SA2 is not required for the loading of cohesin onto chromatin, but is essential for its dissociation from chromosome arms during early mitosis. However, it also remained possible that cohesin complexes containing SA2–12xA simply dissociated from chromosomes more slowly than wild-type complexes. To address this possibility, we analyzed cells in which prometaphase was prolonged by treatment with nocodazole. Under these conditions, complexes containing wild-type SA2 dissociated completely from chromosome arms within 3 h but remained at centromeres ( Figure 5 B), confirming earlier observations made for Scc1 [ 15 , 18 ]. In contrast, SA2–12xA could still be detected clearly on chromosome arms after 3 h and even after 10 h of nocodazole treatment ( Figure 5 B), excluding the possibility that the dissociation of cohesin complexes containing this mutant is simply slower than the dissociation of wild-type complexes. Cohesin Complexes Containing Nonphosphorylatable SA2 Are Able to Establish Cohesion It was also possible that the amino acid exchanges in SA2–12xA had rendered the protein nonfunctional, possibly resulting in the formation of complexes that bound chromatin nonspecifically and therefore did not dissociate from chromosomes in mitosis. Thus, we asked whether cohesin complexes containing SA2–12xA are able to establish sister chromatid cohesion. To answer this question, we again treated cells with nocodazole, which normally results in the loss of cohesion between chromosome arms but not at centromeres, resulting in the formation of X-shaped chromosomes with “open arms” that can be seen by chromosome spreading and Giemsa staining [ 18 ]. In this assay, chromosomes from most cells expressing wild-type SA2-myc showed the normal “open arm” phenotype (between 55% and 71%, depending on the cell line analyzed; Figure 6 A). In contrast, only 10% of mitotic cells expressing SA2–12xA contained chromosomes whose arms had lost cohesion during the nocodazole arrest, whereas 87% had maintained cohesion between chromosome arms ( Figure 6 A). This result strongly indicates that the cohesin complexes that contain SA2–12xA and that remain on chromosome arms in prometaphase (see Figure 5 B) are able to establish and maintain cohesion between sister chromatids. Figure 6 The Presence of SA2–12xA on Chromosome Arms Correlates with Cohesion between Sister Chromatid Arms (A) Cells were cultured in the absence or presence of different amounts of doxycycline as indicated. After arrest in nocodazole for 3 h, cells were fixed, spread on glass slides, and stained with Giemsa (photomicrographs, above). Single chromosomes (indicated by a box) are shown at higher magnification in the lower right corners. The number of cells with chromosome arms that had opened (arms open) or that were connected (arms closed) was scored as indicated (bar graphs, below). Scale bar 10 μm. (B) Whole-cell extracts were prepared from HeLa cells expressing SA2–12xA-myc after treatment with increasing amounts of doxycycline (0, 0.2, and 2.0 μg/ml). The ratio of exogenous SA2–12xA-myc to endogenous SA2 was visualized by immunoblotting with antibodies to SA2. The position of molecular weight markers is indicated on the right. Finally, to further confirm this hypothesis, we asked whether the ability of SA2–12xA-expressing cells to maintain arm cohesion during a nocodazole treatment depends on the amount of ectopic protein that is expressed. We therefore treated cells containing SA2–12xA transgenes with different doses of doxycycline and compared the levels of exogenous SA2 with the arm cohesion phenotype. The levels of mutant SA2-myc were well controlled by the amount of doxycycline used ( Figure 6 B), and there was a clear correlation between the amount of SA2–12xA and the number of cells whose chromosomes maintained arm cohesion during treatment with nocodazole, whereas expression of wild-type SA2-myc had no significant influence on arm cohesion ( Figure 6 A). Cells of the SA2–12xA cell line maintained arm cohesion slightly more frequently than control cells even if SA2–12xA expression was not induced, but it is possible that small amounts of the ectopic protein are also synthesized in the absence of doxycycline. These observations rule out the possibility that the different frequencies with which arm cohesion is observed after nocodazole treatment in SA2–12xA and control cells are due to other differences in the cell lines than expression of the different transgenes. We therefore conclude that cohesin complexes containing SA2–12xA are able to maintain and establish cohesion, but that these complexes are unable to dissociate from chromosomes during prophase and prometaphase. Phosphorylation of SA2 at its C terminus therefore appears to be essential for the unloading of cohesin from chromosome arms during early mitosis. Phosphorylation of SA2 Is Required for Efficient Resolution of Sister Chromatids Although the major amount of human cohesin dissociates from chromosomes during early mitosis, the physiological importance of this process is still unknown. One reasonable assumption is that cohesin dissociation might be required for binding of condensin complexes and for condensation of chromatin. Like cohesin, condensin complexes contain subunits that are members of the Smc and kleisin protein families [ 28 , 29 , 30 ], and cohesin dissociation and condensin binding normally coincide in cells [ 3 , 13 ]. However, inhibition of cohesin dissociation by interfering with Plk1/Plx1 or Aurora-B function in human cells or Xenopus egg extracts did not affect the binding of condensin and the overall condensation of chromosomes [ 16 , 17 , 18 , 31 ]. In line with these results, we found that condensin binding, compaction, and shortening of chromosomes in a prolonged mitotic arrest was not detectably influenced by expression of SA2–12xA ( Figure S3 ), further strengthening the notion that condensin binding does not require the phosphorylation-dependent dissociation of cohesin complexes from chromosome arms. Experiments in Xenopus egg extracts showed that in the absence of Plx1 and Aurora-B, chromosome arms on mitotic chromosomes remained very close to each other, i.e., the resolution of sister chromatid arms was impaired [ 17 ]. In these experiments, it was difficult to distinguish whether Plx1 and Aurora-B promoted sister chromatid resolution by inhibiting the dissociation of cohesin or by another, independent pathway. We therefore examined how chromosome structure is influenced by expression of nonphosphorylatable SA2. We noticed that sister chromatid arms often stayed in closer proximity in SA2–12xA-expressing cells than in controls ( Figure 7 A). When we measured the interchromatid distance during prometaphase and metaphase, we found a variation within the same cell line from around 0.4 to 1.1 μm between individual cells ( Figure 7 B). This variability is presumably caused by different periods of time that cells have spent in mitosis; i.e., cells in which the interchromatid distance was small might have spent shorter time in mitosis than those in which sister chromatids were more resolved. We also found that there was some variability between different cell lines (unpublished data). However, cells expressing wild-type SA2 did not show any prominent difference in the average interchromatid distance in the absence or presence of exogenous SA2, whereas the interchromatid distance was progressively shortened when SA2–12xA-myc was expressed in increasing amounts ( Figure 7 B). Figure 7 Phosphorylation of SA2 Is Required for Efficient Resolution of Sister Chromatid Arms during Prometaphase and Metaphase (A) HeLa cells expressing SA2-WT-myc or SA2–12xA-myc were spread on glass slides and chromosomes were stained with Giemsa. Representative cells from SA2 WT-myc or SA2–12xA-myc cell lines after induction with 2 μg/ml doxycycline are shown. Scale bar 10 μm. (B) HeLa cells were induced to express SA2-WT-myc or SA2–12xA-myc by different amounts of doxycycline as indicated, and processed as in (A). More than 50 cells in prometaphase or metaphase were selected randomly from each sample. The distance between sister chromatids was determined for five chromosomes in each cell and averaged. Light gray bars indicate average values that have been measured in one or two cells, and darker gray bars indicate average values that have been measured in three or more cells. Diamonds indicate the average distance for all cells in a given sample. (C) Representative immunofluorescence image of normal anaphase in a cell expressing SA2–12xA-myc. The cell was not extracted prior to fixation, so the soluble pool of SA2–12xA-myc is revealed by myc-staining. Since resolution of sister chromatid arms happens progressively during prophase and prometaphase, a reduction in the average distance between sister chromatids might also be caused by shortening the time up to metaphase (therefore leaving cells less time to resolve sister chromatids). However, when we compared the percentage of different mitotic stages in SA2–12xA- versus wild-type SA2-expressing cells, we found no indication of a shortening of prometaphase (unpublished data). Phosphorylation of SA2 and dissociation of cohesin from chromosome arms therefore appear to be required for the efficient resolution of sister chromatid arms. SA2 Phosphorylation Is Not Essential for Anaphase The prolonged persistence of cohesin on chromosome arms and the closer proximity of sister chromatid arms might cause defects in anaphase, for example because larger amounts of cohesin cannot be cleaved efficiently enough, or because perturbance of chromosome structure might interfere with the separation of sister chromatids. However, we did not observe obvious anaphase defects in cells expressing nonphosphorylatable SA2. Furthermore, in all anaphases that we observed ( n > 50), we found that SA2–12xA had completely disappeared from separating chromatids ( Figures 7 C and S3B ). Scc1 cleavage products could be detected in cells expressing SA2–12xA (unpublished data), consistent with the interpretation that the loss of SA2–12xA-containing cohesin complexes from chromatids in anaphase is mediated by separase. These observations indicate that SA2 phosphorylation is required for the dissociation of cohesin from chromosome arms during prophase and prometaphase, but that this dissociation process is not absolutely essential for the initiation of anaphase. These data also further support the notion that separase is able not only to cleave cohesin at centromeres but on chromosome arms as well [ 18 ]. Discussion It has long been known that cohesin's Scc1 and Scc3/SA1/SA2 subunits are specifically phosphorylated in mitosis [ 2 , 5 , 14 , 16 , 19 , 32 , 33 ]. In budding yeast it has been shown that phosphorylation of Scc1 by Cdc5 enhances the cleavability of cohesin by separase [ 19 , 20 ]. In vertebrates, however, the functional significance of these modifications has remained unclear. The circumstantial evidence in vertebrate systems that has existed so far points to a role of cohesin phosphorylation in controlling the ability of cohesin to bind chromatin [ 2 , 16 ], not in modulating Scc1 cleavage by separase. It was also unclear whether Plk1, a kinase that is essential for cohesin dissociation from chromosome arms during prophase and prometaphase [ 16 , 17 , 18 ], regulates cohesin in early mitosis by directly phosphorylating Scc1 or SA2, or by modifying other proteins that might be required for cohesin unloading. Our analysis of mitotic cohesin regulation in human cells revealed distinct roles for the phosphorylation of Scc1 and SA2. Phosphorylation of human Scc1 enhances the cleavability of this protein by separase, at least at the second of Scc1's two cleavage sites (see Figure 2 ), and thereby shows that this mode of cohesin regulation is conserved from yeast to humans. Furthermore, our data imply that Scc1 phosphorylation is not required for the dissociation of cohesin from chromosome arms (see Figure S2 ), again consistent with the situation in yeast where Scc1 is phosphorylated in mitosis [ 19 ], yet cohesin complexes do not dissociate from chromosomes until separase is activated [ 34 , 35 ]. In contrast to the data for Scc1, our results show that SA2 phosphorylation is essential for the dissociation of at least some cohesin complexes from chromosome arms (see Figure 5 ), but this modification does not seem to be necessary for the cleavage of cohesin complexes by separase ( Figure 7 C and unpublished data). Cells that express nonphosphorylatable versions of SA2 are unable to remove all cohesin complexes from their chromosome arms during prometaphase (see Figure 5 A), even if mitosis is prolonged for many hours by treatment with spindle poisons (see Figure 5 B); presumably as a consequence, cohesion between sister chromatid arms is not lost in these cells during prolonged prometaphase arrest (see Figure 6 ). These phenotypes are virtually identical to the cohesin and cohesion phenotypes of cells in which Plk1 has been depleted [ 18 ]. It has also been observed that Plk1 can phosphorylate Scc1 and SA2 in vitro, can thereby decrease the ability of cohesin to bind chromatin, and is required for the mitosis-specific phosphorylation of Scc1 and SA1/SA2 in Xenopus egg extracts [ 16 ]. Obviously none of these results can exclude the possibility that kinases other than Plk1 contribute to SA2 phosphorylation, nor the possibility that Plk1 may also have to phosphorylate proteins other than SA2 to allow cohesin dissociation, but the simplest interpretation of all the data is that Plk1 is essential for cohesin unloading because it is required for SA2 phosphorylation, which in turn is a prerequisite for cohesin dissociation. It will be interesting to learn whether this type of regulation is restricted to human SA2, or whether it also applies to paralogs and orthologs of SA2. In addition to SA1 and SA2, a meiosis-specific paralog of the Scc3 family exists in mammals, called SA3 (or STAG3) [ 36 ]. Most of the phosphorylation sites that we identified in SA2 are conserved in SA1, whereas SA3 diverges from SA1 and SA2 mostly in its C-terminal sequence (unpublished data). This difference could have important implications for the regulation of meiosis I, where arm cohesion needs to be protected to allow the separation of homologous chromosomes in anaphase I (reviewed in [ 37 ]), and where chromosome arms do not separate even if cells are arrested by treatment with spindle poisons [ 38 ]. It is possible that the replacement of SA1/2 by SA3 renders meiotic cohesin complexes resistant to Plk1-dependent removal from chromosome arms, and thereby allows the maintenance of arm cohesion until separase is activated. Likewise it will be interesting to analyze whether Scc3 is phosphorylated in budding yeast, in which cohesin dissociation from chromosome arms in early mitosis has not been detected. How Important Are Scc1 and SA2 Phosphorylation In Vivo? Somewhat unexpectedly, we found that expression of neither nonphosphorylatable Scc1 nor nonphosphorylatable SA2 blocked progression through mitosis. Although it remains formally possible that our experiments did not identify all mitosis-specific phosphorylation sites and that we therefore did not mutate all critical sites, we consider it more plausible to think that phosphorylation of these proteins is not absolutely essential for progression through mitosis, at least in transformed cultured cells. This notion is supported by the finding that cohesin can be removed from chromosomes (presumably by separase), even in cells in which Plk1 has been depleted and Aurora-B has been inhibited—i.e., under conditions where the early mitotic dissociation of cohesin from chromosome arms is inhibited [ 18 ]. The implication is that the early mitotic dissociation of cohesin from chromosomes is not absolutely essential for mitosis, because separase is able to cleave all cohesin complexes that reside on chromosomes at the metaphase-anaphase transition. In this respect, human cells therefore appear to be more similar to budding yeast than previously suspected, in that HeLa cells can also initiate anaphase without first having to remove cohesin from chromosome arms. Likewise, there are similarities between yeast and HeLa cells in the regulation of Scc1. In both systems, Scc1 phosphorylation enhances its cleavability by separase, but in neither case is this modification essential for viability (this study) [ 19 ]. An interesting hint to the possible function of Scc1 phosphorylation comes from the observation that budding yeast cells lacking the securin Pds1 are viable and are able to undergo anaphase, but this ability is dramatically decreased if phosphorylation sites in Scc1 are mutated [ 19 ]. Since securin not only inhibits separase but is also required for its activation, yeast cells lacking securin may not have enough separase activity to cleave cohesin if Scc1 is not phosphorylated. Phosphorylation of Scc1 might increase its affinity for separase, and this effect may simply enhance the fidelity of anaphase initiation. Securin is also not essential for viability in human cells, but in its absence the specific activity of separase is decreased [ 39 ]. It would be interesting to test whether human cells lacking securin require Scc1 phosphorylation for viability. Similarly, it is possible that SA2 phosphorylation and the resulting dissociation of cohesin from chromosomes in early mitosis, albeit not being essential, increase the fidelity of chromosome segregation. It is also conceivable that removal of cohesin prior to cleavage is not important for mitosis but for the next interphase. Separase-dependent cohesin removal destroys the Scc1 subunit and thereby renders cohesin nonfunctional. In contrast, phosphorylation-dependent dissociation appears to leave cohesin intact and might thereby enable the rapid reloading of cohesin onto chromatin in telophase, i.e., without the necessity for new Scc1 transcription and translation, which is inhibited during mitosis. How Does SA2 Phosphorylation Lead to Cohesin Dissociation? Cohesin is bound to chromatin in an extremely stable manner ([ 8 ]; E.R. and J.M.P, unpublished data), and this may be related to the fact that Smc1, Smc3, and Scc1 form a ring-like complex, at least in budding yeast [ 40 , 41 ]. It has been proposed that this protein ring establishes cohesion by encircling the sister chromatid strands [ 40 ]. In this model, it is easy to imagine how cleavage of Scc1 releases cohesin from chromatin. However, Scc3 binds to Scc1 and is not required for formation of the ring-like complex, and it is therefore not immediately obvious how phosphorylation of SA2 could lead to dissociation of cohesin from chromosomes. One possibility is that SA2 phosphorylation induces a conformational change in cohesin that opens the ring. Bulk phosphorylation of SA2's C terminus, for example, might considerably change its surface charge, thereby affecting interactions between Scc1 and the Smc1/3 subunits. In its simplest form, this model would predict that SA2 phosphorylation is sufficient for opening of the cohesin ring and thus is sufficient for cohesin dissociation. However, in preliminary experiments, we have been unable to observe cohesin dissociation when we added purified active Plk1 to chromatin (I. Sumara and J.M.P, unpublished data), whereas the simultaneous addition of Plk1 and Xenopus egg extracts to chromatin did enable cohesin dissociation [ 16 ]. It is therefore also possible that phosphorylation of SA2 recruits cohesin unloading factors to chromatin (which in the above experiment might have been contributed by the Xenopus extract), which then somehow enable the dissociation of cohesin from chromosomes. In budding yeast and C. elegans, cohesin needs additional factors for its loading onto chromatin [ 42 , 43 ]. Cohesin might similarly need aid for unloading, at least in the absence of Scc1 cleavage. If such additional factors exist and interact with SA2, the cell lines we created might provide a means to isolate the relevant molecules by differential purification of cohesin complexes containing wild-type SA2 and SA2–12xA. If SA2 phosphorylation results in cohesin unloading by somehow enabling the opening of the cohesin ring without its cleavage, it would be conceivable that this reaction is simply the reverse of the loading process, during which the cohesin ring presumably also has to be opened transiently [ 44 ]. However, a prediction of this model would be that SA2 phosphorylation would also be required for the loading of the cohesin complex, whereas we find that complexes containing nonphosphorylatable cohesin can efficiently associate with chromatin and even establish functional cohesion. It is therefore more plausible to hypothesize that SA2 phosphorylation is a modification that is specifically used to remove cohesin from chromosomes in early mitosis by enabling a reaction that is not simply the reverse of the loading reaction. Which Cohesin Complexes Are Regulated by SA2 Phosphorylation? SA2 phosphorylation is required for the dissociation of cohesin from chromosome arms, but it does not seem to affect the behavior of cohesin at centromeres. As a consequence, sister chromatid arms are resolved much farther from each other than centromeres during a normal prometaphase, and they can lose cohesion completely if prometaphase is prolonged, whereas cohesion at centromeres is protected. How is this regulation achieved? Recent work in fission yeast has shown that members of the Mei-S332 family of proteins [ 45 ] are required for the persistence of cohesin at centromeres in meiosis I [ 46 , 47 , 48 , 49 ]. These proteins, called shugoshins or Sgo1, are thought to protect centromeric cohesin in anaphase I from premature cleavage by separase, but they are also found at centromeres in mitotic Drosophila and budding yeast cells [ 49 , 50 ]. In an associated paper by McGuinness et al. [ 51 ], we show that an ortholog of Sgo1 is also required for the persistence of cohesin at centromeres and for the maintenance of sister chromatid cohesion during prometaphase in human cells. Remarkably, Sgo1-depleted cells do not show cohesion defects if a nonphosphorylatable form of SA2 is expressed. This observation implies that cohesin normally persists at centromeres because Sgo1 protects cohesin in this chromosomal domain from phosphorylation. To test this hypothesis it will be important to determine whether SA2 phosphorylation does occur at centromeres. Our identification of in vivo phosphorylation sites on SA2 may be an important prerequisite for achieving this goal, because it should enable the generation of phosphospecific antibodies. Likewise, it will be interesting to learn how Sgo1 prevents or antagonizes SA2 phosphorylation, and whether the same mechanism is able to protect centromeric cohesin from separase in meiosis I. Previous work has highlighted the difference in the regulation of cohesin complexes between chromosome arms and centromeres [ 10 , 18 ], but several observations suggest that there may also be important differences among different populations of cohesin complexes on chromosome arms. When we compared Scc1 cleavage in cells expressing either wild-type or nonphosphorylatable SA2, we noticed that the levels of Scc1 cleavage products in the latter cells were only slightly increased, if at all (unpublished data). If all complexes containing nonphosphorylatable SA2 remained on chromosome arms until prometaphase, we would instead expect to see more Scc1 cleavage in cells containing these complexes than in cells containing wild-type SA2. Furthermore, we noticed that the immunofluorescence intensity of SA2–12xA-myc in interphase cells from which soluble cohesin complexes had been removed by preextraction was clearly higher than the intensity of SA2–12xA-myc on prometaphase and metaphase chromosomes, and in subcellular fractionation and immunoblotting experiments we found that a fraction of SA2–12xA still became soluble in nocodazole-arrested cells (unpublished data). We made similar observations in immunofluorescence experiments in which we analyzed the chromosome association of Scc1-myc in Plk1-depleted cells. Also in these cells, some cohesin still seemed to dissociate from chromosome arms, despite the depletion of Plk1 (T. Hirota and J.M.P, unpublished data). The observation that some cohesin complexes do dissociate from chromosome arms even if Plk1 is depleted or if these complexes contain nonphosphorylatable SA2, whereas others do not, cannot simply be explained by slow dissociation kinetics of cohesin under these conditions, because those complexes that persist on chromosome arms can still be found there after 10 h of prometaphase arrest (see Figure 5 B). We therefore favor the hypothesis that there are, in fact, two distinct populations of cohesin on chromosome arms: one whose dissociation depends on Plk1 activity and SA2 phosphorylation, and one whose dissociation does not. Since the population of cohesin complexes whose dissociation depends on Plk1 and SA2 is able to establish cohesion (see Figure 6 ), it is possible that those complexes that seem to be able to dissociate without Plk1 and SA2 phosphorylation are bound to chromatin in a manner that does not establish cohesion. Such binding modes must exist, because cohesin rebinds to chromatin in telophase [ 3 , 13 ], i.e., long before sister chromatids have been generated by DNA replication. This speculative model makes important predictions, for example that cohesin dissociation from unreplicated DNA should not depend on SA2 phosphorylation; we will attempt to test this prediction in the future. Materials and Methods Cell lines and growth conditions HeLa cells expressing myc-tagged human Scc1 were described previously [ 15 ]. HeLa cells expressing mutated Scc1, wild-type SA2, or mutated SA2 were generated by transfecting HeLa tet-on cells (Clontech, Palo Alto, California, United States) with the pTRE2-hygro vector (Clontech) containing the respective cDNA and a 9xmyc tag fused in frame. Hygromycin-resistant clones were selected by growth in medium containing 400 μg/ml hygromycin, and were tested for expression of the myc-tagged protein by immunoblotting after induction with 2 μg/ml doxycycline for 1–3 d. Untransfected HeLa cells were grown in DMEM supplemented with 10% FCS, 0.2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. For growth of stably transfected HeLa tet-on cells, the medium was supplemented with 200 μg/ml G418 and 100 μg/ml hygromycin. Hydroxyurea and nocodazole were used at concentrations of 2 mM and 330 nM, respectively. cDNA mutagenesis Serine or threonine residues in human Scc1 and SA2 were changed to alanine by mutation of the respective cDNAs using the Quick Change Site-Directed or Multi Site-Directed Mutagenesis kit (Stratagene, La Jolla, California, United States). Where the phosphorylated site could not be assigned to a certain residue within a peptide, we mutated all possible candidate sites to alanine residues. C-terminally truncated versions of SA2 were generated by PCR. For in vitro translation, cDNAs were cloned into pcDNA 3.1 (−)/Myc-His A (Invitrogen) modified to contain a 9xmyc cassette. The expression of cDNAs as 35 S-methionine- and 35 S-cysteine-labeled proteins was performed using a coupled transcription-translation system in rabbit reticulocyte lysate (Promega, Madison, Wisconsin, United States). Antibodies Antibodies specific to phosphorylated threonine were from Cell Signaling Technology (Beverly, Massachusetts, United States; #9381). To detect myc-tagged protein, we used either polyclonal, affinity-purified rabbit anti-myc antibody (Gramsch Laboratories, Schwabhausen, Germany; #CM-100), or monoclonal 9E10 mouse-anti-myc antibody [ 10 ]. Polyclonal antibodies to human Scc1, SA1, SA2, Smc1, Smc3, and Smc2 have been described [ 3 , 10 , 30 ]. Anti-INCENP peptide antibody was raised in rabbit, and the serum was affinity purified. The following peptide was used for immunization: TDQADGPREPPQSARRKRSYC. Human CREST serum was a kind gift of A. Kromminga (Institut für Immunologie, Pathologie und Molekularbiologie, Hamburg, Germany). Protein purification and fractionation Endogenous cohesin was purified from HeLa cell extracts by immunoprecipitation using purified SA2 peptide antibodies (antibody 447) crosslinked to Affi-Prep protein A beads (Bio-Rad, Hercules, California, United States) as described [ 3 , 16 ]. For mass spectrometric analysis, cohesin precipitates were first washed with IP buffer (20 mM Tris-HCl [pH 7.5], 100 mM NaCl, 0.2% Nonidet P-40, 20 mM β-glycerophosphate, 10% glycerol, 1 mM NaF, and 0.5 mM DTT) [ 3 ] supplemented with 400 mM NaCl, followed by a wash step with IP buffer not containing detergent. Cohesin was eluted from the antibody beads by addition of 1.5 bead volumes of 100 mM glycine (pH 2.2). The samples were titrated to pH 8.5 by addition of NH 4 HCO 3 to a final concentration of 200 mM. Cohesin complexes containing myc-tagged subunits were immunopurified with rabbit anti-myc antibodies (Gramsch Laboratories, #CM-100) using similar conditions. Sucrose density gradient centrifugation was performed as described [ 15 ]. Mass spectrometry A volume of 100 μl of immunopurified cohesin (isolated from about 10 mg of total HeLa protein) was reduced with 1 μg of DTT for 1 h at 37 °C and alkylated with 5 μg of iodoacetamide for 30 min at room temperature in the dark. The proteins were digested in solution with one of the following proteases or combination of proteases: 200 ng of trypsin for 4 h at 37 °C, followed by addition of another 200 ng of trypsin and further incubation for 4 h at 37 °C; 400 ng of Glu-C for 8 h at 25 °C; 400 ng of trypsin overnight at 37 °C followed by 400 ng of Glu-C for 8 h at 25 °C; 400 ng of chymotrypsin or elastase for 4 h at 25 °C (all proteases were sequencing grade; Roche, Basel, Switzerland); or 300 ng of subtilisin (Fluka, from Sigma-Aldrich, St. Louis, Missouri, United States) for 30 min at 37 °C. Generated peptides (100-μl sample) were separated by reversed phase nano-HPLC (LC Packings, Sunnyvale, California, United States) and analyzed using an ion trap mass spectrometer (ThermoFinnigan, from Thermo Electric, Waltham, Massachusetts, United States) as described by Mitulovic et al. [ 52 ]. All tandem mass spectra were searched against the human nonredundant protein database by using the SEQUEST program (ThermoFinnigan). Any phosphopeptide matched by computer searching algorithms was verified manually. In vitro Scc1 cleavage assays In vitro cleavage assays were performed as described [ 10 ] with the exception that for the assays in Figure 2 A, in vitro-translated human wild-type or mutant Scc1-myc was used as substrate. To isolate human separase, purified polyclonal antibodies generated against recombinant human separase were used (kindly provided by I. Waizenegger). In some reactions human GST-Plk1 (16) was added in a concentration of approximately 50 ng/μl. SA2 electrophoretic mobility shift assays Xenopus interphase egg extracts were supplemented with 1/20 volume of in vitro-translated 35 S-labeled SA2, which had C-terminal deletions or mutations of phosphorylation sites. All SA2 constructs used in this assay lacked the 69 N-terminal amino acids, because the start codon was initially misassigned. Extracts were induced to enter mitosis by addition of cyclin B Δ90 and 1 μM okadaic acid as described in Sumara et al. [ 3 ]. Samples were collected at the indicated time points and analyzed by SDS-PAGE followed by Phosphorimager analysis (Storm, Amersham Biosciences, Little Chalfont, United Kingdom). In vitro phosphorylation assay SA2-myc or SA2–12xA-myc containing cohesin complexes were purified by immunoprecipitation with affinity-purified rabbit anti-myc antibody (Gramsch Laboratories, # CM-100). Conditions of the in vitro phosphorylation assay have been described [ 26 ]. Microscopy For immunofluorescence microscopy, cells were either grown on coverslips or spun onto glass slides using a Cytospin centrifuge (Shandon brand, available from Thermo Electric). Cells were extracted with 0.1% Triton X-100 prior to fixation to remove the soluble pool of cohesin. Fixation, immunostaining, and image acquisition were performed as described [ 10 ]. Chromosome spreads followed by Giemsa staining were performed as described [ 31 ]. Quantification of interchromatid distance On pictures of chromosome spreads, a line scan was performed across chromosome arms orthogonal to the long axis of the chromosome using MetaMorph software (Universal Imaging, Downingtown, Pennsylvania, United States). On the line scan, the distance from peak to peak was measured. Five chromosomes were thus analyzed per cell, and the resulting distances were averaged. More than 50 cells were randomly picked per cell line, and thus analyzed. Supporting Information Figure S1 Phosphorylation Sites in Human Scc1 and SA2 Sites for Scc1 (A) and SA2 (B) are shown. The sites marked in red and blue were found to be phosphorylated in mitosis by mass spectrometry. Residues marked in red were unambiguously identified, whereas in the regions marked in blue the phosphorylated residue could not be assigned with certainty. The separase recognition sites on human Scc1 (A) are indicated in green. Peptides identified by mass spectrometry are highlighted in yellow. The residues at which SA2 was truncated for the assay in Figure 4 C are also indicated (B). (1.2 MB EPS). Click here for additional data file. Figure S2 Phosphorylation of Scc1 Is Not Required for Dissociation of Cohesin from Chromosome Arms during Prometaphase and Metaphase (A) Extracts were prepared from HeLa cells expressing Scc1-S 454 A-myc or Scc1–9xA-myc and fractionated by sucrose density gradient centrifugation (5%–30% sucrose), followed by immunoblotting with antibodies recognizing the proteins indicated on the right (inp. = input/unfractionated sample of the extract). (B) HeLa cells expressing Scc1 WT-myc or Scc1–9xA-myc were either grown logarithmically (0 h Noc) or arrested in prometaphase for 5 h by nocodazole (5 h Noc). Cells were extracted prior to fixation, and stained with myc-antibodies. Kinetochores were labeled with human CREST (calcinosis, Raynaud phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias) serum, and DNA was counterstained with DAPI. Scale bar, 10 μm. (870 KB JPG). Click here for additional data file. Figure S3 Condensation or Condensin Binding Is Not Impaired in SA2–12xA-Expressing Cells (A) Untransfected HeLa tet-on cells and HeLa cells expressing SA2-WT-myc, or SA2–12xA-myc were arrested with nocodazole for 10 h. Cells were fixed, spread on glass slides, and stained with Giemsa. For each sample, one representative cell is shown. The small bars next to one of the chromosomes in all panels have the same length. (B) HeLa cells expressing SA2-WT-myc or SA2–12xA-myc were spread on glass slides, extracted prior to fixation, and immunostained as indicated, using an antibody against human Smc2 to reveal condensin. Scale bars in (A) and (B), 10 μm. (721 KB JPG). Click here for additional data file. Accession Numbers GenBank accession numbers for proteins discussed in this paper are human Scc1 (NP_006256) and human SA2 (NP_006594).

D:\keerthana\PMC001xxxxxx\PMC1054881.xml

1054882

Shugoshin Prevents Dissociation of Cohesin from Centromeres During Mitosis in Vertebrate Cells

Cohesion between sister chromatids is essential for their bi-orientation on mitotic spindles. It is mediated by a multisubunit complex called cohesin. In yeast, proteolytic cleavage of cohesin's α kleisin subunit at the onset of anaphase removes cohesin from both centromeres and chromosome arms and thus triggers sister chromatid separation. In animal cells, most cohesin is removed from chromosome arms during prophase via a separase-independent pathway involving phosphorylation of its Scc3-SA1/2 subunits. Cohesin at centromeres is refractory to this process and persists until metaphase, whereupon its α kleisin subunit is cleaved by separase, which is thought to trigger anaphase. What protects centromeric cohesin from the prophase pathway? Potential candidates are proteins, known as shugoshins, that are homologous to Drosophila MEI-S332 and yeast Sgo1 proteins, which prevent removal of meiotic cohesin complexes from centromeres at the first meiotic division. A vertebrate shugoshin-like protein associates with centromeres during prophase and disappears at the onset of anaphase. Its depletion by RNA interference causes HeLa cells to arrest in mitosis. Most chromosomes bi-orient on a metaphase plate, but precocious loss of centromeric cohesin from chromosomes is accompanied by loss of all sister chromatid cohesion, the departure of individual chromatids from the metaphase plate, and a permanent cell cycle arrest, presumably due to activation of the spindle checkpoint. Remarkably, expression of a version of Scc3-SA2 whose mitotic phosphorylation sites have been mutated to alanine alleviates the precocious loss of sister chromatid cohesion and the mitotic arrest of cells lacking shugoshin. These data suggest that shugoshin prevents phosphorylation of cohesin's Scc3-SA2 subunit at centromeres during mitosis. This ensures that cohesin persists at centromeres until activation of separase causes cleavage of its α kleisin subunit. Centromeric cohesion is one of the hallmarks of mitotic chromosomes. Our results imply that it is not an intrinsically stable property, because it can easily be destroyed by mitotic kinases, which are kept in check by shugoshin.

Introduction Cohesion between sister chromatids ensures that traction of sister chromatids towards opposite poles (known as bi-orientation) generates a tug-of-war between microtubules attempting to pull sisters apart and cohesion between them resisting this. The resulting tension is thought to stabilize the attachment of kinetochores to microtubules [ 1 ]. Only when all chromosomes have come under tension and have congressed to the metaphase plate do cells destroy the connection holding sister chromatids together, which triggers the simultaneous disjunction of all sister chromatid pairs and traction of sister chromatids to opposite poles of the cell, known as anaphase. Sister chromatid cohesion depends on a multisubunit complex called cohesin, which is composed of a heterodimer of Smc1 and Smc3 proteins associated with an α kleisin protein called Scc1 [ 2 , 3 ], that is in turn associated with either SA1 or SA2 variants of the Scc3 protein. These proteins create a gigantic ring structure within which DNA molecules might be entrapped [ 4 , 5 ]. Sister chromatid cohesion is destroyed at the metaphase-to-anaphase transition because of cleavage of cohesin's α kleisin (Scc1/Rad21) subunit by a protease called separase [ 6 ], whose activity causes cohesin to dissociate from chromatin. For most of the cell cycle, separase activity is inhibited by its phosphorylation by the Cdk1 kinase [ 7 ] and by the binding of an inhibitory chaperone called securin [ 8 , 9 , 10 ]. Separase is eventually activated by proteolysis of the Cdk1 cyclin B subunit and securin, both mediated by a ubiquitin protein ligase called the anaphase-promoting complex or cyclosome (APC/C) [ 11 , 12 ]. Because of the production by unattached kinetochores of an inhibitory form of the Mad2 protein, which binds an essential APC/C cofactor called Cdc20, the APC/C destroys securin and cyclin B only when all chromosomes have successfully bi-oriented [ 13 ]. This surveillance mechanism is known as the mitotic spindle checkpoint. Cells lacking any one of cohesin's four subunits fail to bi-orient chromosomes properly and arrest at least transiently in a mitotic state because of inhibition of APC/C-Cdc20 by Mad2 [ 14 , 15 ]. In yeast, cohesin persists at centromeres and along chromosome arms until the onset of anaphase, whereupon it is removed by the APC/C-separase pathway [ 8 ]. In animal cells, however, the bulk of cohesin associated with chromatin during G2 dissociates from chromosome arms but not from centromeres during prophase and prometaphase [ 16 ]. This so-called “prophase pathway” is thought to be driven not by cleavage of cohesin's α kleisin subunit but instead by hyperphosphorylation of Scc3-SA subunits [ 17 ] mediated (directly or indirectly) by the Aurora B [ 18 ] and Plk1 mitotic kinases [ 19 ]. Crucially, expression of an Scc3-SA2 subunit whose serine and threonine residues phosphorylated during mitosis have been mutated to alanine reduces the dissociation of cohesin from chromosome arms, which, as a consequence, remain more tightly associated [ 17 ]. Surprisingly, this mutation does not dramatically interfere with mitosis, presumably because the extra cohesin associated with metaphase chromosomes is efficiently removed by separase upon activation of the APC/C. The function of the prophase pathway is therefore unclear. Centromeric cohesin is not just more slowly removed by the prophase pathway because cohesin remains at centromeres during the prolonged mitotic arrest that is caused by activation of the spindle checkpoint. Under these circumstances, cohesion between chromosome arms is lost entirely, while that between sister centromeres persists [ 20 ]. Centromere-specific factors presumably protect cohesin from the prophase pathway. An analogous process occurs during meiosis, during which cohesin along chromosome arms must be treated differently from cohesin at centromeres. Because of recombination between maternal and paternal chromatids, cohesin along chromosome arms holds homologous centromeres together and enables their bi-orientation. Cleavage of cohesin's α kleisin subunit by separase along chromosome arms destroys this connection and thereby triggers the first meiotic division [ 21 , 22 ]. However, cohesin at centromeres is refractory to separase during meiosis I and therefore persists until metaphase II, and these cohesin complexes are crucial for the bi-orientation of sister chromatids during the second meiotic division. Recent work has shown that a class of proteins associated with meiotic centromeres, called MEI-S332 in Drosophila melanogaster [ 23 ] but now known as shugoshins, are essential for cohesin's ability to persist at centromeres after anaphase I. Shugoshin protects cohesin from separase during meiosis I in yeast [ 24 , 25 , 26 ]. Surprisingly, shugoshins are also found at centromeres during mitosis in yeast [ 24 , 27 ] and D. melanogaster [ 28 ]. This observation raises the possibility that shugoshins might have an important, albeit different, function during mitosis. We show here that a human shugoshin that is possibly orthologous to fly MEI-S332 and yeast Sgo1 proteins [ 24 , 25 , 26 ] associates with centromeres during prophase and disappears at the onset of anaphase in mitotic tissue culture cells. HeLa cells whose shugoshin has been depleted by RNA interference (RNAi) fail to retain cohesin at centromeres during mitosis, and, as a consequence, their sister chromatids separate asynchronously before anaphase can be initiated, which triggers a prolonged mitotic arrest. Remarkably, expression of nonphosphorylatable Scc3-SA2 alleviates both the precocious loss of sister chromatid cohesion and the mitotic arrest. This suggests that centromeric shugoshin prevents phosphorylation of cohesin's Scc3-SA2 subunit and thereby protects the cohesion between sister centromeres that is essential for mitosis. Results The Sgo1/MEI-S332 Protein Accumulates at Centromeres as Cells Enter Mitosis but Disappears during Anaphase The human and mouse genomes each contain a single gene ( hsSgo1 and mmSgo1, respectively) that encodes proteins with a similar structure and weak homology to fungal Sgo1 proteins and MEI-S332 from D. melanogaster . Analysis of cDNAs from various databases as well as RT-PCR products from mRNA obtained from HeLa cells and mouse testis suggest that variable splicing may give rise to several types of protein ( Figure S1 ). To detect these proteins, we raised two antibodies against peptides from the N and C termini that are common to all human splice variants. To evaluate the specificity of affinity-purified antibodies, we tested whether immunofluorescence signals or bands on Western blots were abolished when Sgo1 mRNA was depleted by transfecting cells with small interfering RNAs (siRNAs) ( Figure 1 ). Because most subsequent experiments addressed Sgo1's role during mitosis, siRNA transfection in HeLa cells was combined with a double thymidine block and release to synchronise cells ( Figure 1 A). Cells were transfected with siRNAs 8 h after release from the first thymidine block, and 4 h later a second block was imposed by repeat addition of thymidine. After a further 12 h, cells were released from the second block and samples taken at specific times thereafter. Under these circumstances, most cells complete S phase within 7 h of the second release and soon thereafter enter mitosis. Both affinity-purified antibodies detected a 72-kDa protein from a chromatin fraction that was depleted by treatment with an Sgo1-specific siRNA but not by a mock treatment ( Figure 1 B). This size is consistent with an approximately 60-kDa protein predicted from the longest of the cDNAs analysed. Figure 1 Centromeric Enrichment of Sgo1 in Mitosis (A) Schematic overview of cell synchronisation and transfection procedure. (B) Characterization of Sgo1 antibodies. Affinity-purified Sgo1 antibodies raised against two different regions of Sgo1 recognize the same 72-kDa protein by Western blotting. Synchronised HeLa cells were transfected with water (Mock) or Sgo1 siRNA. Cells were harvested 7 and 11 h after release from thymidine block, and chromatin fractions were resolved by SDS-PAGE and probed with antibodies 94 and 95. A major band at 72 kDa was detected by both antibodies, which disappeared or was greatly reduced upon Sgo1 knockdown. The blot was reprobed with HP-1β antibody as a loading control. (C) Localisation of Sgo1. Cells were costained for Sgo1 with antibody 94 (shown in green), and with CREST antiserum (shown in red). DNA was counterstained with DAPI. Five different stages of mitosis are shown: (a) prophase, (b) prometaphase, (c) metaphase, (d) anaphase, and (e) telophase. Bar = 10 μm. (f) A single pair of CREST labelled kinetochores are enlarged, showing two adjacent Sgo1 foci as commonly observed for late stage metaphases. Bar = 1 μm. (D) In situ immunofluorescence of (a) mock-treated or (b) Sgo1 siRNA-transfected HeLa cells, demonstrating the disappearance of the Sgo1 signal following siRNA treatment. The latter shows the typical in situ appearance of an Sgo1-depletion arrested cell (see Figures 3 and 6 ). Sgo1 was stained with antibody 94 (shown in green), and CREST is shown in red. DNA was counterstained with DAPI. In situ immunofluorescence of cycling HeLa cells suggested that Sgo1 accumulates at centromeres during mitosis (see Figure 1 C). Very similar staining patterns were observed with antibodies against both peptides ( Figure S2 ). The centromeric staining was eliminated in most cells by siRNA depletion (see Figure 1 D). Only 16% of all mitotic cells remained positive for Sgo1 staining following siRNA treatment, compared to 99% in mock-treated cells. Sgo1 started to form foci on chromosomes as cells entered prophase (see Figure 1 C, part a). During prometaphase, Sgo1 was concentrated, usually as a single focus, on the inner side of the twin (sister kinetochore) structures stained by a CREST (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias) antiserum (see Figures 1 C, part b, and 2 A) thought to detect the kinetochore proteins CENP-A/B or C, and Sgo1 overlapped with and covered a slightly larger area than Aurora B during prometaphase ( Figure 2 B). During metaphase, two Sgo1 foci overlapping with or adjacent to the two CREST signals were observed at many kinetochores ( Figure 1 C, part f). Centromere-associated Sgo1 was much less abundant during anaphase, but it could nevertheless still be faintly detected adjacent to kinetochores at the leading edge of chromatids (see Figure 1 C, part d). No Sgo1 foci could be detected in telophase cells (see Figure 1 C, part e). A similar pattern was observed after transfection with a gene expressing a Sgo1-GFP fusion protein (E. Watrin and J.-M. Peters, personal communication). The distribution of Sgo1 during mitosis in HeLa cells resembles that of MEI-S332, its presumptive orthologue in D. melanogaster [ 28 ]. Figure 2 Immunofluorescent Staining of Chromosome Spreads from HeLa Cells Chromosome spreads were stained for Sgo1 with antibody 94 (shown in green), and counterstained with (A) CREST antiserum (bar = 10 μm), and (B) antibody to Aurora B, (both shown in red). DNA was stained with DAPI (shown in blue). Enlarged views show a single Sgo1 focus located between two sister CREST dots (A), and overlapping with Aurora B focus (B). Bar = 1 μm [applies to images b–d in (A) and (B)]. Sgo1 Is Required to Maintain Sister Chromatid Cohesion in Mitotic Cells To address the function of Sgo1 in HeLa cells, we harvested cells at 2-h intervals after release from a double thymidine block and compared cells whose Sgo1 protein had been depleted by siRNA to mock-treated cells (see Figures 1 A, 1 B, and 3 ). Sgo1 depletion had little or no effect on the kinetics of S phase completion or on entry into mitosis (see Figure 3 A and 3 C). At each time point, chromosomes were spread on glass slides after methanol-acetic acid fixation, and stained with Giemsa. We first measured the fraction of chromosome spreads in a mitotic state, then measured the fraction of seven different categories of mitotic chromosomes amongst mitotic cells (see Figure 3 A and 3 B). In mock-treated cells, the fraction of mitotic spreads peaked 9 h after release and subsequently declined. The most frequent category at early time points was prophase (see Figure 3 B, part a), and the most frequent category at 9 h was metaphase (see Figure 3 B, part b). Early-anaphase cells (see Figure 3 B, part c) were very rare, whereas telophase cells (see Figure 3 B, part d) accumulated later than metaphase cells. Sister chromatids were either tightly associated as during metaphase or fully separated into two equal-sized clusters, as during anaphase or telophase. Figure 3 Sgo1 Depletion Causes Precocious Sister Separation and Mitotic Arrest Synchronised HeLa cells were transfected with Sgo1 siRNA or dH 2 O, harvested at 2-h intervals following release and examined by chromosome spreading and Giemsa staining. 100 cells were scored for mitotic index, and 100 mitotic cells were classified into seven categories based on chromosome configuration as exemplified in (B) for each time point. (A) The frequency of each category of mitotic cell is given as a percentage of total cell numbers, such that the sum of each column represents the mitotic index. (B) Representative pictures of seven categories of mitotic cells. Chromosome spreads: (a) prophase; (b) metaphase/metaphase-like; (c) anaphase; and (d) telophase. (e) Early phase of precocious sister disjunction. At this stage, sisters are beginning to separate and some or all presumptive sister pairs are still discernible. Arrowheads indicate chromosomes whose centromeric cohesion seems to be lost. (f) Later phases of sister chromatid separation. Sister pairs at this stage are no longer discernible, remnants of the metaphase plate are still visible, and sisters have not yet hypercondensed. (g) Scattered single chromatids. Sisters are completely separated and distributed randomly in relation to one another, individual chromatids are hypercondensed, giving a “curly” appearance. Note that chromatid separation in normal anaphases (c) are different from precocious sister chromatid separation (e–g), in that disjoined and paired chromatids coexist in the same cell. (C) A portion of the cells harvested for the analysis in Figure 3 A and 3 B were ethanol-fixed and their DNA content was analysed by flow cytometry. Sgo1-depleted cells accumulated in mitosis with similar kinetics, but many failed subsequently to exit from a mitotic state, with the result that nearly 50% of the cells had accumulated in a mitotic state 15 h after release (see Figure 3 A). FACS analysis showed that around half of the cells failed to undergo cytokinesis (see Figure 3 C). Many Sgo1-depleted cells accumulated at least transiently with chromosomes in a metaphase-like state, but at all time points a large number accumulated with fully condensed or even hypercondensed chromatids that had separated from their sisters (see Figure 3 B, part g). At 9 h, when the frequency of metaphase cells peaked in mock-treated cultures, we observed not only a high fraction of metaphase cells but an equally high fraction of spreads in which most sister chromatid pairs had disjoined but chromatids had not yet hypercondensed (see Figure 3 B, part f). Although many sisters had separated in these cells, we could also detect remnants of a metaphase plate. We also detected a small fraction of spreads in which this abnormal separation of sister chromatids had only commenced—namely, spreads in which sister chromatids were still aligned but much farther apart than during a normal metaphase (see Figure 3 B, part e). These data suggest that Sgo1-depleted cells enter mitosis and align most chromosomes on a metaphase plate, but subsequently fail to undergo anaphase. They then disjoin their sister chromatids asynchronously and in a manner that is not accompanied by directed movement to opposite poles, and, finally, arrest for a prolonged period with hypercondensed, fully separated chromatids. Because the transfected siRNAs could, in principle, interfere with the RNAi machinery that might be necessary for centromeric sister chromatid cohesion [ 29 ], we tested whether other siRNAs produce a similar phenotype. Not one of 12 different siRNAs caused the rapid mitotic arrest with separated sister chromatids that is characteristic of cells treated with Sgo1 siRNAs ( Figure S3 ). The Kinetics of Sister Chromatid Separation in Sgo1-Depleted Cells To analyse the consequences of Sgo1 depletion with greater temporal resolution, we filmed mock-treated and Sgo1-depleted HeLa cells that stably express histone H2B tagged with enhanced green-fluorescent protein (EGFP), which enabled us to observe the movement of individual chromosomes and chromatids ( Figure 4 ). After mock treatment, all cells that entered mitosis subsequently underwent anaphase. The time from nuclear envelope breakdown (NEBD) to anaphase onset was 33 ± 9 min (average ± standard deviation) ( n = 30). In Sgo1-depleted cells, 73% of cells that entered mitosis separated sister chromatids without undergoing anaphase. Although most chromosomes initially congressed to a metaphase plate, many were slow to do so and some never congressed at all ( Figure 4 A). Although 27% of mitotic cells underwent anaphase, we noticed most of these did so with lagging chromatids (unpublished data). The time from NEBD to the point at which the first sister chromatid was pulled out from the metaphase plate in Sgo1-depleted cells was 39 ± 12 min ( n = 60), which is only slightly longer than the time taken by mock-treated cells to commence anaphase or the time taken by the minority of Sgo1-depleted cells that underwent anaphase (34 ± 7 min; n = 23). Cells that disjoined sister chromatids precociously remained in a mitotic state with condensed but fully separated chromatids for a prolonged period (6 h 20 min ± 2 h 3 min) before chromatids decondensed or cells underwent apoptosis. Kinetochore movements in mock-treated and Sgo1-depleted cells were also compared by filming cells expressing a CENP-A protein fused to EGFP ( Figure 4 C). The results confirmed that most sister kinetochore pairs congressed to a metaphase plate before splitting precociously, which led to collapse of the metaphase plate. It also confirmed the congression defect of some sister kinetochore pairs ( Figure 4 C, arrows). Our data imply that the highly abnormal separation of sister chromatids seen in Sgo1-depleted cells occurs at around the same time as cells would normally undergo anaphase and not as soon as cells enter mitosis. This suggests that Sgo1 may be required not to build sister chromatid cohesion during S phase, but rather to maintain sister chromatid cohesion during mitosis, a period during which most cohesin is removed from chromosome arms. Our data also show that the abnormal sister chromatid separation of Sgo1-depleted cells is not merely a response to an extended mitotic arrest. Figure 4 Live Cell Analysis of Sgo1-Depleted Cells (A) Synchronised HeLa cells expressing EGFP-tagged histone H2B were transfected with Sgo1 siRNA or dH 2 O as in Figure 1 A, and analysed with time-lapse confocal microscopy. Stacks of 12 different z-plane images were obtained every 5 min and projected images for several time points are shown. Note that several chromosomes failed to congress to the metaphase plate, which was followed by progressive sister chromatid separation over time. Typically, a pair of disjoining sister chromatids dissociated simultaneously from the metaphase plate towards opposite poles. (B) Summary of live cell analysis. Cells prepared as in (A) were examined by time-lapse immunofluorescence microscopy. Mitotic cells were aligned on the time axis according to NEBD as determined from the loss of a defined nuclear boundary. Time from NEBD to anaphase onset/sister chromatid separation (white portion of bars) and from this point to chromosome decondensation or apoptosis (grey portion of bars) was measured. (C) To follow centromere dynamics, cells expressing EGFP-CENP-A were either mock-treated or Sgo1 siRNA transfected. Images were obtained by time-lapse confocal microscopy. Note that in cells depleted of Sgo1, dots of paired GFP-CENP-A can be found in earlier unaligned chromosomes (arrows). Single dots progressively fell apart in later phases, resulting in collapse of metaphase plate (images in lower row). Sgo1 Prevents Precocious Dissociation of Cohesin from Centromeres During Mitosis To test if the abnormal separation of sister chromatids in Sgo1-depleted cells is accompanied by loss of cohesin from chromosomes, we depleted Sgo1 from synchronised HeLa cells that inducibly express a myc-tagged version of the Scc1 α kleisin cohesin subunit ( Figure 5 ) [ 16 ]. Cells were harvested 9.5 h after release from the second thymidine block when many had already entered mitosis. After mock treatment, Scc1-myc staining along chromosomes was detectable in two-thirds of mitotic cells, which were identified by being positive for histone H3 phosphorylation ( Figure 5 A and 5 B). Scc1-myc was typically more abundant at centromeres in such cells. This centromere enrichment was more clearly detectable ( Figure 5 A) if cells had been incubated in the presence of the spindle poison nocodazole for 4 h prior to harvesting, a treatment that induces a prometaphase arrest during which cohesin is completely removed from chromosome arms. Sgo1 depletion greatly reduced the fraction of mitotic cells with chromosomal Scc1-myc staining, in both the presence and the absence of nocodazole ( Figure 5 A and 5 B). Interestingly, we noticed that Scc1-myc was rarely if ever enriched at centromeres in the few mitotic Sgo1-depleted cells whose chromosomes were still associated with Scc1-myc ( Figure 5 C and 5 D). Importantly, Sgo1 depletion did not reduce the amount of Scc1-myc associated with prophase chromosomes ( Figure S4 ). These results are consistent with the notion that the loss of mitotic sister chromatid cohesion caused by Sgo1 depletion is due to dissociation of cohesin from centromeres before cells initiate anaphase. Figure 5 Sgo1 Is Required for Stable Association of Cohesin at Centromere (A) Synchronised HeLa cells that inducibly express Scc1-myc were transfected with control or Sgo1 siRNA and processed for immunofluorescence microscopy with or without 4-h treatment with nocodazole. Mitotic cells were spun down on glass slides and analysed for cohesin with antibodies to myc epitope (right photomicrographs). Cells were costained with P-H3 (left photomicrographs, red) to identify cells from prophase to metaphase. DNA was counterstained with DAPI (left photomicrographs, blue). Note that with Sgo1 RNAi, Scc1-myc was undetectable in the majority of P-H3 positive cells (arrows). Cells with myc staining are indicated as staining controls (asterisks). Bar = 10 μm. (B) Quantification of Scc1-myc staining. Approximately 200 cells with positive staining for P-H3 were assessed for Scc1-myc staining, which was classified as positive, reduced, or negative, as indicated. (C) Lack of centromere enrichment of cohesin in Sgo1-depleted cells. Mitotic cells were prepared as in (A), and the centromeric enrichment of Scc1-myc was analysed by immunofluorescence microscopy. Merged pictures of Scc1-myc (green) and CREST antigen (red) are shown. Approximately 200 cells were scored for each experiment. Bar = 5 μm. (D) Centromeric cohesin is not maintained in Sgo1-depleted cells. Mitotic cells were prepared and processed for immunofluorescence as in (C). Representative cells with various levels of Scc1-myc are shown. Note that centromeric staining of Scc1-myc emerges as the bulk of cohesin dissociates from chromosomes in controls (upper panels). However, in Sgo1 RNAi cells, centromeric enrichment is hardly seen at any stage of arm cohesin dissociation (lower photomicrographs). Bar = 10 μm. Sgo1-Depleted Cells Arrest in a Prometaphase-Like State To investigate whether the loss of cohesin from centromeres is due to precocious activation of separase, we used in situ immunofluorescence to assess cyclin B1 levels and localization of Mad2 in mock-treated and Sgo1-depleted cells 11 h after their release from the second thymidine block. In mock-treated cells, cyclin B1 was concentrated within the nuclei of prophase cells and on the mitotic spindles of all prometaphase and most metaphase cells, although also throughout their cytoplasm, and was absent from anaphase or telophase cells ( Figure 6 A). To quantitate this result, we classified mitotic cells into five categories based on DAPI (4′,6′-diamidino-2-phenylindole) staining and scored the fraction of each category that was positive for cyclin B1 ( Figure 6 B). In mock-treated cells, cyclin B1 was positive in all prometaphase cells, in about 70% of metaphase cells, and in no anaphase cells. This pattern reflects destruction of cyclin B1 shortly before the onset of anaphase. In Sgo1-depleted cells, cyclin B1 was positive in all prometaphase and most metaphase cells. It was also positive in the vast majority of cells that contained metaphase plates with some dispersed chromatids and in cells whose chromatids had largely dispersed from the metaphase plate ( Figure 6 B). This implies that cyclin B1 destruction never occurs in mitotic Sgo1-depleted cells. This is presumably due to continued activation of the mitotic spindle checkpoint because the chromosomes of mitotic cells lacking Sgo1 always contained foci of Mad2 associated with their centromeres, which is normally seen only in prometaphase in mock-treated cells ( Figure 6 C and unpublished data). These data suggest that separase is not activated in Sgo1-depleted cells. Their loss of centromeric cohesin is therefore unlikely to be due to Scc1 cleavage. Figure 6 Sgo1-Depleted Cells Arrest in a Prometaphase-like State (A) Sgo1 depletion prevents destruction of cyclin B1. Synchronised HeLa cells mock-treated or transfected with Sgo1 siRNA were processed for immunofluorescence microscopy using cyclin B1 antibodies (green) and CREST sera (red) 11 h after release from the second thymidine block. DNA was visualized by DAPI staining (blue). In mock-treated cells, cyclin B1 staining in prophase (a) and in prometaphase (b) disappears as cells segregate their chromosomes upon anaphase entry (c). Sgo1-depleted cells which segregate chromosomes asynchronously and arrest in a mitotic state retain preanaphase levels of cyclin B1 (d–e). Bar = 10 μm. (B) Mitotic cells shown in (A) were classified into five categories based on DAPI-labelled chromosome configuration, and scored for presence or absence of cyclin B1 staining. Black bars represent the frequency of each category, and red bars represent the percentage of cells that fall into each category and are positive for cyclin B1 staining. (C) Recruitment of Mad2 in cells depleted of Sgo1. Cells prepared as in (A) were stained with cyclin B1 antibody, and counterstained with Mad2 antibody (red) and DAPI to visualize DNA (blue). (a) Prometaphase mock-transfected cells show Mad2 staining at kinetochores, which is lost as the metaphase plate assembles and disappears before cells enter anaphase (unpublished data). (b) Sgo1-transfected cells which have prematurely segregated sisters remain positive for Mad2 kinetochore staining. Bar = 10 μm. (D) Aurora B remains at centromeres in Sgo1-depleted cells. Cells prepared as in (A) were stained with Aurora B antibody (green) and counterstained with CREST antiserum (red), and DAPI to visualize DNA (blue). In prometaphase (a) mock-transfected cells, Aurora B is found at centromeres before relocalizing to the central spindle as cells enter anaphase (b). In Sgo1 transfected cells that show precocious sister separation, Aurora B remains localized at centromeres (c). Another event that normally occurs at the onset of anaphase is the disappearance of Aurora B from inner centromeres and its accumulation at the spindle midzone [ 30 ]. Aurora B does not dissociate from centromeres at any stage of the abnormal mitoses of Sgo1-depleted cells, as Aurora B was found adjacent to CREST staining not only in metaphase cells (unpublished data) but also in cells whose chromatids have dispersed from the metaphase plate ( Figure 6 D). A corollary of this finding is that cohesin is not required to maintain Aurora B at centromeres, contrary to a previous suggestion [ 15 ]. Neither Plk1 Depletion nor Aurora B Inhibition Suppresses the Precocious Sister Separation in Sgo1-Depleted Cells Removal of cohesin from chromosome arms during prophase and prometaphase depends on the activity of mitotic kinases, including Plk1. Might this Plk1-dependent process in normal cells be prevented from attacking centromeric cohesin by Sgo1? To test this idea, we analysed whether synchronised cells depleted for both Sgo1 and Plk1 also lose sister chromatid cohesion when they enter and arrest in mitosis. Mock-treated, Plk1-depleted, Sgo1-depleted, and Sgo1- and Plk1-depleted cells were harvested at 7, 9, and 11 h after release from the thymidine block, and the state of spread chromosomes was analysed by Giemsa staining (as described in Figure 3 ). The mitotic index of mock-treated cells peaked at 9 h and declined by 11 h, but the mitotic indices of singly or doubly depleted cells continued to rise after 9 h, reaching around 50% by 11 h ( Figure 7 A). Plk1-depleted cells arrested with hypercondensed, rod-shaped chromosomes whose arms were more tightly associated than those of prometaphase or metaphase mock-treated cells ( Figure 7 B, part d), which is consistent with previous findings [ 20 ]. Sgo1 singly depleted cells arrested with separated chromatids, progressing via the same set of stages described in Figure 3 B. Importantly, cells depleted for both Sgo1 and Plk1 also largely arrested with separated chromatids ( Figure 7 A). The morphology of these chromatids was, however, very different from those seen in Sgo1 singly depleted cells. Of those cells whose sister chromatids had been separated (i.e., those represented by the red bars in Figure 7 A), 100% possessed short, curly (presumably coiled) separated chromatids by 11 h in a culture depleted of Sgo1 alone, as shown in Figure 7 B, part b. In contrast, 80% of such cells that had been depleted for both Sgo1 and Plk1 possessed separated chromatids that were both longer and straighter, as shown in Figure 7 B, part c. Only 12% possessed short, curly chromatids, while 8% possessed chromatids whose arms were still loosely associated with their sisters, as shown in Figure 7 B, part a. Figure 7 Neither Plk Depletion nor Aurora B Inhibition Suppresses the Precocious Sister Separation Seen in Sgo1-Depleted Cells (A and B) Synchronised HeLa cells were mock-treated or transfected with the indicated combination of Sgo1 and Plk1 siRNA and harvested at the time shown following release from thymidine block. Chromosomes were then spread on glass slides and examined by Giemsa staining. As in Figure 3 , the percentage of mitotic cells were calculated out of 200 cells, and mitotic chromosome spreads were then further classified into one of 10 categories ( n = 200 mitotic spreads for each time point). (A) The frequency of each category of mitotic cell (see examples in Figures 3 B and 7 B) is given as a percentage of total cell numbers, such that the sum of each column represents the mitotic index. (B) In addition to the seven categories in Figure 3 A and 3 B, the following categories (illustrated in Figure 7 B) were scored: (a) sister centromeres separated, arms still cohesed; (b) scattered chromatids, same as Figure 3 B, part g; (c) sisters are separated and randomly distributed in relation to one another, but are not hypercondensed; and (d) sister chromatids are cohesed along their length, and rod-shaped chromosomes are hypercondensed, as is characteristically seen with Plk1 knockdown [ 20 ]. For simplicity, the categories illustrated in Figure 3 B, part e and f, and Figure 7 B, part a–c, were combined into a single category representing premature sister separation (shown in red). In smaller pictures, bar = 10 μm; in enlarged portions, bar = 2 μm. (C and D) Synchronised HeLa cells were mock-treated or transfected as indicated. At 6 h after release from the second thymidine block, cells were treated with nocodazole with or without Aurora B inhibitor Hesperadin as indicated, and harvested at hourly intervals up to 3 h thereafter. Mitotic index and chromosome spreads were assessed as in (A). (C) The frequency of each category of mitotic cell (illustrated in Figures 3 B and 7 D) is given as a percentage of total cell numbers, such that the sum of each column represents the mitotic index. (D) The seven categories scored for Figure 3 A and 3 B were also scored for this experiment. (a) Scattered chromatids as in Figure 3 B, part g; (b) early phase of precocious sister disjunction as in Figure 3 B, part e. In addition the following categories (illustrated in Figure 7 D) were also scored: (c) Chromosomes have begun to decondense prior to separation, and sister resolution is defective, characteristic of Hesperadin treatment. (d) Centromeres cohesed, and arms opened and hypercondensed, characteristic of nocodazole treatment. The fact that treatment with Plk1 siRNA dramatically changed the morphology of separated chromatids in Sgo1-depleted cells confirms that both proteins had in fact been effectively depleted. Our data therefore imply (somewhat surprisingly) that Plk1 is not necessary for the precocious separation of sister chromatids induced by Sgo1 depletion. Interestingly, Sgo1 depletion permitted not only sister centromere separation but also that along chromosome arms in cells supposedly lacking Plk1. This raises the possibility that the tight cohesion between sister chromatid arms that persists in Plk1-depleted cells depends on Sgo1. Aurora B kinase activity has also been implicated in the removal of cohesin from chromosome arms. To address whether this kinase is responsible for the precocious loss of sister centromere cohesion in Sgo1-depleted cells, we examined whether inhibition of Aurora B by the small molecule inhibitor Hesperadin can suppress their sister chromatid separation. Because Aurora B and its yeast equivalent Ipl1 are also required to prevent cell cycle arrest of cells with defective sister chromatid cohesion [ 31 ] or cells whose microtubules cannot generate centromeric tension [ 32 , 33 ], it was necessary to maintain Mad2 inhibition of APC/C activity by addition of nocodazole at the same time that Hesperadin was added, 6 h after they had been released from the thymidine block. The former delays by at least 3 h the exit from mitosis of Hesperadin-treated cells [ 32 ]. Chromosomes were spread and examined by Giemsa staining after harvesting mock-treated and Sgo1-depleted cells at 6, 7, 8, and 9 h after release ( Figure 7 C and 7 D). Addition of nocodazole caused both mock-treated and Sgo1-depleted cells to accumulate in mitosis between 6 and 9 h after release, which was largely unaffected by Hesperadin addition. Importantly, Hesperadin had only a modest, if any, effect on the precocious separation of sister chromatids in Sgo1-depleted cells ( Figure 7 C and 7 D). Interestingly, treatment with nocodazole had a clear effect on the arrangement of the separated sisters in Sgo1-depleted cells. By 9 h, up to 73% of mitotically arrested cells resembled the image shown in Figure 7 D, part b, where single chromatids lie in the neighbourhood of their presumptive sisters. In the absence of nocodazole, less than 5% of cells fell into this category ( Figure 3 B, part e). This implies that when sister cohesion is prematurely lost as a result of Sgo1 depletion, spindle pulling forces are not required to separate sisters but they are required to produce the scattered chromatid effect seen in the previous experiments. We conclude that Hesperadin cannot prevent the precocious separation of sister chromatids induced by Sgo1 depletion, at least when such cells are prevented from exiting mitosis by addition of nocodazole. In a very similar experiment in which nocodazole was omitted, Hesperadin did indeed reduce the number of mitotic cells whose sisters had separated precociously (unpublished data). We believe that this effect is probably a statistical artefact caused by the failure of Sgo1-depleted cells treated with Hesperadin to arrest in mitosis. The Precocious Sister Separation and Mitotic Arrest of Sgo1-Depleted Cells Is Suppressed by a Nonphosphorylatable Scc3-SA2 In a related paper, Hauf et al. [ 17 ] describe the mitosis-specific phosphorylation of 12 serine or threonine residues clustered within the C-terminal domain of cohesin's Scc3-SA2 subunit. Remarkably, a large fraction of Scc3-SA2 protein in which all 12 residues have been mutated to alanine (Scc3-SA2 12xA), which is no longer phosphorylated during mitosis, persists on chromosome arms throughout mitosis and even does so when cells are arrested for prolonged periods in a prometaphase-like state due to nocodazole treatment. The persistence of Scc3-SA2 12xA on chromosomes under these circumstances prevents loss of cohesion between sister chromatid arms in nocodazole-arrested cells, but it does not obviously interfere with mitosis in cycling cells. This implies that phosphorylation of Scc3-SA2 may be largely, if not solely, responsible for the removal of cohesin from chromosome arms during prophase and prometaphase. We therefore analysed the effects of Sgo1 depletion in cell lines in which either wild-type SA2 or SA2 12xA protein (both tagged with nine myc epitopes) is expressed from a doxycycline-inducible promoter at levels (when induced) that are comparable to endogenous SA2 protein (see Figures 8 A and S5 ). We conducted the experiments on cycling cells that had been treated with (or without) doxycycline for 72 h prior to transfection. Cells were then transfected either with water (mock) or with the Sgo1 siRNA and harvested 18 or 24 h later, and chromosome spreads were examined after Giemsa staining. In mock-transfected cultures, the frequency of mitotic cells remained constant (at around 4–5%) at 18 and 24 h, regardless of the presence or absence of doxycycline or whether wild-type or mutant protein was induced ( Figure 8 B). In Sgo1-depleted cells, the mitotic indices of cells expressing wild-type Scc3-SA2 myc increased to between 16% and 18% by 24 h; the mitotic index also increased to 14% in cells that expressed only very low levels of Scc3-SA2 12xA-myc in the absence of doxycycline. As expected, most sister chromatids had separated in both sets of these Sgo1-depleted cells. The low frequency of telophase cells suggests that these cells largely failed to undergo anaphase. Remarkably, induction of Scc3-SA2 12xA-myc with doxycycline suppressed the accumulation of mitotic cells caused by Sgo1 depletion. Moreover, only a few of the mitotic cells contained precociously separated sister chromatids, and significant numbers had clearly undergone anaphase and produced telophase cells ( Figure 8 B). Most metaphase cells possessed chromosomes whose sisters were cohesed at centromeres as well as along their arms. Thus, expression of nonphosphorylatable Scc3-SA2 at physiological levels suppresses both the mitotic arrest and the precocious sister chromatid separation caused by Sgo1 depletion. This indicates that loss of cohesin from centromeres in Sgo1-depleted cells may be due to (hyper-) phosphorylation of Scc3-SA2 at centromeres. Figure 8 Expression of Nonphosphorylatable Scc3-SA2 Suppresses Sgo1-Depletion Phenotype (A) Total cell extracts prepared from HeLa cells that, upon doxycycline treatment, inducibly express either a wild-type (SA2-wt) or nonphosphorylatable (SA2–12xA) myc-tagged version of Scc3-SA2 were resolved by SDS-PAGE and probed with anti-SA2 antibody. The lower band represents endogenous Scc3-SA2, the upper band, myc-tagged Scc3-SA2. The first lane contains total cell extract prepared from untagged HeLa cells. (B) Expression of nonphosphorylatable Scc3-SA2 suppresses sister separation caused by Sgo1 depletion. Cycling HeLa cells carrying inducible myc-tagged versions of Scc3-SA2 were treated with (or without) doxycycline at 2 μg/ml for 72 h prior to and during transfection to induce expression, as indicated. Cells were harvested 18 h and 24 h post transfection, chromosomes were spread on glass slides and Giemsa-stained. As in Figure 7 , the percentage of mitotic cells were calculated out of 200 cells, and mitotic chromosome spreads were then further classified into one of five categories ( n = 200 mitotic spreads for each time point). The frequency of each category of mitotic cell (see examples in Figures 3 B, 7 B, and 7 D) is given as a percentage of total cell numbers, such that the sum of each column represents the mitotic index. (C) Nonphosphorylatable Scc3-SA2-mediated suppression is still observed when combined with nocodazole arrest. The indicated treatments were repeated as outlined in (B) in the presence of nocodazole which was added to cultures 4 h posttransfection, i.e., 14 and 20 h prior to harvesting of cells. Samples were processed as in (B), with the addition of a sixth mitotic category. (D) Live cell analysis of Sgo1 depletion of nonphosphorylatable Scc3-SA2 (SA2–12xA) expressing cells. To follow chromosome behaviour, cells stably expressing EGFP-H2B and inducibly expressing SA2–12xA were used. Expression of SA2–12xA was induced 72 h prior to transfection as in (B). Cells were transfected with Sgo1 siRNA as in Figure 1 A and examined by time-lapse fluorescence microscopy. Significant number of cells exit mitosis by expressing nonphosphorylatable Scc3-SA2. (E) Nonphosphorylatable SA2 (SA2–12xA) is found at centromeres even in the absence of Sgo1. HeLa cells containing either the myc-tagged wild-type (a) or SA2–12xA (b) inducible transgene were either uninduced or induced as in (B) 72 h before transfection. Transfection of Sgo1 siRNA was performed prior to the second thymidine block as in Figure 1 A. At 8.5 h after the release from early S phase, mitotic cells were spun down on glass slides and analysed for cohesin localisation by immunofluorescence microscopy using antibodies to the myc epitope (shown in green). Cells were costained with CREST antiserum to label kinetochores (shown in red) DNA was counterstained with DAPI (shown in blue). (F) Quantification of SA2-myc staining. Samples similar to those described in (E) were stained with myc and P-H3 antibodies (the latter to identify cells from prophase to metaphase). Approximately 200 P-H3-positive cells were assessed for SA2-myc staining, and the percentage of cells that were both P-H3- and SA2-myc-positive was plotted. We believe that the apparent drop in the number of SA2–12xA-myc positive cells observed with depletion of Sgo1 relative to mock transfection is a statistical artefact caused by the mitotic arrest and accumulation of those cells that did not express SA2–12xA-myc (∼30%) but that were depleted of Sgo1. It is nevertheless possible that nonphosphorylatable Scc3-SA2 suppresses a mitotic defect of Sgo1-depleted cells responsible for their cell cycle arrest, and as a consequence these cells might not have long enough to lose sister chromatid cohesion. To address this possibility, nocodazole was added to doxycycline-induced cells 4 h after transfection (i.e., 14 or 20 h prior to harvesting), which caused both mock-treated and Sgo1-depleted cells to accumulate in mitosis ( Figure 8 C). In the case of the culture in which Sgo1 had been depleted and wild-type SA2 myc protein induced, a large fraction of mitotic cells (31%) had separated sister chromatids without undergoing anaphase. This fraction was much lower (9%) when SA2 12xA-myc protein had been induced. Most Sgo1-depleted mitotic cells expressing SA2 12xA-myc contained intact arm, as well as centromere, sister chromatid cohesion ( Figure 8 C). We conclude that SA2 12xA-myc expression suppresses loss of sister chromatid cohesion in Sgo1-depleted cells even when cells have been arrested in mitosis for many hours. To examine more carefully whether expression of nonphosphorylatable Scc3-SA2 permits cells lacking Sgo1 to undergo anaphase, we created a new cell line that expressed histone H2B tagged with EGFP as well as doxycycline-inducible Scc3-SA2 12xA-myc. Sgo1 was depleted in induced and uninduced cells from the same cell line and the behaviour of chromosomes followed by time-lapse video microscopy ( Figure 8 D). Three types of mitoses were observed: those in which sister chromatid pairs first congressed to a metaphase plate but then lost cohesion before undergoing anaphase (precocious separation), those in which some chromosomes were slow to congress and failed to undergo anaphase but did not display mass sister separation, and those in which chromosomes congressed to a metaphase plate and then underwent what appeared to be a normal anaphase with unaltered kinetics. Only 23% of uninduced cells depleted for Sgo1 underwent anaphase, and 63% separated their chromatids precociously, while the rest had congression defects. Remarkably, induction of Scc3-SA2 12xA-myc increased the fraction of cells that underwent anaphase from 23% to 60% and reduced the fraction of cells that separated sisters precociously from 63% to 19%. These data suggest that expression of Scc3-SA2 12xA-myc enables cells to undergo anaphase in the absence of Sgo1. Finally, we addressed whether the loss of mitotic phosphorylation sites on Scc3-SA2 enables cohesin to persist at centromeres as well as along chromosome arms in mitotic Sgo1-depleted cells. To do this, we analysed mitotic chromosomes from Sgo1-depleted cells expressing either Scc3-SA2-myc or Scc3-SA2 12xA-myc. Wild-type SA2 myc-tagged protein was associated with chromosomes in very few cells, while nonphosphorylatable SA2 myc tagged protein was found along the axes of all chromosomes in nearly 40% of cells ( Figure 8 E and 8 F). Crucially, Scc3-SA2 12xA-myc was associated with centromeres as well as chromosome arms. Preventing Scc3-SA2 phosphorylation therefore enables cohesin to persist at centromeres in mitotic cells lacking Sgo1 as well as on chromosome arms in otherwise wild-type mitotic cells. This suggests that the loss of cohesin from centromeres in Sgo1-depleted cells is due to mitosis-specific phosphorylation of SA2. Discussion It has long been recognized that cohesion between sister chromatids in the vicinity of centromeres has an especially important role during both meiosis and mitosis. Centromeric cohesion is special during meiosis because it completely resists destruction at the first meiotic division when dissolution of cohesion along chromosome arms triggers the resolution of chiasmata. It is special during mitosis because, unlike cohesion along chromosome arms, centromeric cohesion persists even when cells are arrested in mitosis for prolonged periods by spindle poisons. The work described in this paper suggests, somewhat surprisingly, that both of these special attributes of centromeric cohesion might be conferred by the same protein, namely Sgo1. Recent work suggests that, in both budding and fission yeast, Sgo1 protects centromeric cohesion at meiosis I by preventing cleavage of cohesin's α kleisin subunit by separase. Surprisingly, both Sgo1 in S. cerevisiae and its probable orthologue in D. melanogaster MEI-S332 are associated with centromeres during mitosis as well as meiosis. We show here that a related protein in human cells concentrates at centromeres during mitosis and disappears from chromosomes during anaphase. Remarkably, depletion of human Sgo1 by RNAi prevents HeLa cells from completing mitosis. Though most chromosomes congress to a metaphase plate in Sgo1-depleted cells, cells never undergo anaphase and instead arrest for a prolonged period in a prometaphase-like state with high cyclin B1 levels. Our data suggest that this is caused by a catastrophic separation of sister chromatids at around the time cells should have normally undergone anaphase, and that this event occurs in the absence of APC/C activation or dissociation of Aurora B from centromeres. This precocious separation is accompanied by loss of cohesin from centromeres as well as chromosome arms and activation of the mitotic spindle checkpoint, as documented by persistent recruitment of Mad2 to kinetochores. Remarkably, expression of a version of cohesin's Scc3-SA2 subunit whose C-terminal domain can no longer be phosphorylated as cells enter mitosis (Scc3-SA2 12xA-myc) not only alleviates the precocious loss of sister chromatid cohesion of Sgo1-depleted cells but also permits a large fraction of them to complete what appears, at least superficially, to be a normal mitosis. Our findings together with those described by Hauf et al. [ 17 ] suggest that phosphorylation of Scc3-SA2 normally triggers dissociation of most cohesin from chromosome arms during prophase and prometaphase, but is prevented from doing so at centromeres by Sgo1, which is more abundant at this location than it is along chromosome arms ( Figure 9 A). We suggest that in the absence of Sgo1, Scc3-SA2 is removed by the prophase pathway from centromeres as well as from chromosome arms and that this leads to the complete disjunction of sister chromatids before cells can initiate anaphase, which is meanwhile delayed by the mitotic checkpoint ( Figure 9 A). Figure 9 Model for Sgo1 Function during Mitosis (A) During an unperturbed mitosis (Wild Type), arm cohesin (red circles) is removed in a kinase-dependent manner during prophase/prometaphase. Sgo1 protects centromeric cohesin (brown circles) until the metaphase-anaphase transition. Once all chromosomes have successfully bi-oriented on the metaphase plate, Mad2 inhibition of the APC/C is relieved, allowing separase activation. Separase in turn removes cohesin remaining at centromeres through cleavage of the α kleisin Scc1 subunit, allowing the cell to enter anaphase. In the absence of Sgo1 (Sgo1 Depletion), cohesin is removed from the chromosome arms and at the centromere during prophase/prometaphase before chromosomes have properly bi-oriented and been attached to their full complement of spindles. Thus, Mad2 activity continues to maintain the spindle checkpoint, causing cells to arrest for a prolonged period in a prometaphase-like state with separated sister chromatids. Expression of nonphosphorylatable Scc3-SA2 (Sgo1 Depletion + Cohesin containing SA2–12xA) prevents the prophase removal of cohesin both from arms and centromeres, thus allowing cells to proceed through an apparently normal anaphase even in the absence of Sgo1 activity. (B) The phosphorylation state of cohesin may be the result of a dynamic balance in which Sgo1 somehow functions to antagonize the activity of mitotic kinases, including Plk1 and Aurora B and, potentially, other as-yet unidentified kinases (Kinase X, Y). In this model, when the direction of the reaction is artificially sent towards the hyperphosphorylated state (following Sgo1 knockdown), cohesin efficiently dissociates from chromatin; when the opposite state is favoured (e.g., as a result of Plk1 depletion) cohesin remains tightly associated with chromatin. The suppression of the mitotic arrest induced by Sgo1 depletion by Scc3-SA2 12xA-myc implies that it is the precocious loss of sister chromatid cohesion and not the lack of some other function of Sgo1 that is largely responsible for the pathological mitotic arrest of Sgo1-depleted cells. Our findings demonstrate for the first time that Sgo1-like proteins promote sister chromatid cohesion by regulating the activity of cohesin. This is inconsistent with the suggestion that Sgo1/MEI-S332 proteins confer a form of sister chromatid cohesion during mitosis that is distinct from that conferred by cohesin [ 34 , 35 ]. Salic et al. [ 35 ] have also shown very recently that Sgo1 is required to prevent precocious sister kinetochore splitting. The authors suggested that Sgo1 has an additional role in stabilizing microtubules attached to kinetochores. Their proposal, that a defect in this putative function is the primary cause of the mitotic arrest due to Sgo1 depletion, is inconsistent with our observations. In our view, the simplest, although by no means only, explanation for the mitotic arrest of Sgo1-depleted cells is that loss of sister chromatid cohesion eliminates the tension needed to stabilize the connection between microtubules and kinetochores, which in turn leads to unoccupied kinetochores that recruit Mad2 and generate a form of Mad2 that is capable of inhibiting the APC/C's destruction of securin and cyclin B. Nevertheless, our finding that 14% of Sgo1-depleted cells expressing SA2 12xA-myc had congression defects is consistent with the notion that Sgo1 might also function to regulate kinetochore-microtubule interactions. How might Sgo1 regulate cohesin? One possibility is that Sgo1 enables cohesin to be refractory to the effects of Scc3-SA2 phosphorylation. If cohesin embraces chromatid fibres within its ring structure, then phosphorylation of Scc3-SA2's C-terminal tail might induce the cohesin ring to open, and this process might be blocked by Sgo1. Alternatively, Sgo1 might prevent cohesin's dissociation by blocking Scc3-SA2's phosphorylation in the first place or by recruiting a phosphatase that reverses its phosphorylation ( Figure 9 B). Antibodies capable of specifically detecting phosphorylated Scc3-SA2 would make it possible to distinguish the first from the second and third scenarios. Several lines of evidence suggest that the process against which Sgo1 protects centromeric cohesin does not involve separase-mediated Scc1 cleavage. First, Scc3-SA2 12xA-myc's ability to suppress cohesin's dissociation both from arms in wild-type cells and from centromeres in Sgo1-depleted cells suggests that these two phenomena have a common cause. If the former is separase-independent, then so, presumably, is the latter. Second, the persistence of cyclin B1 throughout the period during which centromeric cohesion is lost in Sgo1-depleted cells implies that the APC/C is not activated and separase must therefore remain associated with its inhibitor chaperone securin. Third, cells expressing Scc3-SA2 12xA-myc proliferate fairly normally, and their timely destruction of sister chromatid cohesion at the onset of anaphase presumably involves the activation of separase by the APC/C. If separase deregulation were responsible for cohesin's precocious disappearance from centromeres in Sgo1-depleted cells, then Scc3-SA2 12xA-myc would have to suppress this defect by delaying Scc1 cleavage, which is difficult to reconcile with the timely onset of anaphase in cells expressing Scc3-SA2 12xA-myc. It is therefore likely that Sgo1 protects centromeric cohesin from different processes during mammalian mitosis and yeast meiosis, from chromosomal dissociation induced by Scc3-SA2 phosphorylation during mitosis, and from Rec8 or Scc1 cleavage during meiosis. Although different, both processes may have a key property in common, namely regulation by Plk1-mediated phosphorylation. It is noteworthy that many shugoshins, including human Sgo1, disappear from centromeres during anaphase. This event might be triggered by their destruction at the hands of the APC/C along with securin and cyclin B [ 35 ]. This event might nevertheless assist in the dissolution of cohesion between sister centromeres at the onset of anaphase by exposing cohesin to the effects of its phosphorylation. Hyperphosphorylation of Scc3-SA2 might promote dissociation of cohesin complexes that had somehow evaded cleavage by separase, while hyperphosphorylation of Scc1 (which could also be triggered by Sgo1's departure) might actually facilitate Scc1's cleavage as occurs in both budding yeast [ 36 ] and human cells. The identity of the mitotic kinases responsible for phosphorylation of Scc3-SA2 remains unclear. Because Plk1 is required for dissociation of cohesin from chromosome arms in HeLa cells containing Sgo1 [ 20 ], it was surprising that Plk1's depletion did not suppress the precocious loss of cohesion caused by Sgo1 depletion. This finding indicates that protein kinases other than Plk1 can phosphorylate the C terminus of Scc3-SA2. Because Plk1 is necessary for dissociation of cohesin from chromosome arms in the presence of Sgo1 [ 20 ] but not in its absence ( Figure 7 A), Sgo1 would appear to have a role in inhibiting dissociation of cohesin from chromosome arms as well as from centromeres. Though we could not detect Sgo1 associated with chromosome arms, modest amounts residing at this location might (at least partly) inhibit the prophase pathway. We suggest that Sgo1 antagonizes phosphorylation of Scc3-SA2 by several mitotic protein kinases, including Plk1 ( Figure 9 ). Our study of human Sgo1 has hitherto been confined to HeLa cells. It will be important in the future to analyse its function in other cell types, especially in untransformed cells, and ultimately in real tissues from mice whose Sgo1 gene could be deleted by homologous recombination. It is curious, for example, that Sgo1's likely orthologue in D. melanogaster, MEI-S332, is not apparently essential for somatic cell divisions, which contrasts with Sgo1's essential function in HeLa cells. It is conceivable that small changes in the kinetics with which cohesin is dissociated from centromeres in Sgo1/MEI-S332-depleted cells is key to whether a cell is able to undergo anaphase before precocious loss of sister chromatid cohesion triggers the mitotic spindle checkpoint. One of the most surprising implications of our work is that centromeric sister chromatid cohesion, which has long appeared to be a very stable state and to be the cornerstone (not to mention the most recognisable hallmark) of mitotic chromosomes, only exists because shugoshins protect centromeric cohesin from mitotic protein kinases that maraud mitotic chromosomes and threaten to destroy their integrity. This may have important implications for meiotic cells where protection of cohesin by Sgo1-like proteins (in this case, presumably from attack by separase) is essential for the second meiotic division. Chromosome mis-segregation during meiosis is responsible for trisomies such as Down's syndrome and possibly for age-related infertility and a large fraction of spontaneous abortions [ 37 ]. Our work raises the possibility that this could be caused by modest misregulation of protein kinases or phosphatases that affect the phosphorylation of centromeric cohesin. Whether Sgo1 protects centromeric cohesin from separase during meiosis I also by preventing cohesin's phosphorylation is an important question for the future. Materials and Methods Antibodies To generate rabbit anti-Sgo1 antibodies, we synthesized two peptides: one (for which the antibody was termed antibody 94) with the sequence YKEPTLASKLRRGDPFTDL (aa 474–492) from the C terminus of a mouse cDNA (identical to corresponding peptide sequence in human protein) and a second (antibody 95) with the sequence QKRSFQDTLEDOKNRMKEKRNKN (aa 7–29) from the N terminus of a mouse cDNA (the corresponding sequence in human protein contains just four substitutions). Each peptide was conjugated with KLH and injected into rabbits. Sera were taken by a standard scheme, and specific antibodies were obtained by affinity purification using antigenic peptide-conjugated columns. For Western blotting, blots were incubated with the primary antibodies overnight at 4 °C at a 1:500 dilution. For immunofluorescence, cells were incubated with the primary antibody overnight at 4 °C at a dilution of 1:100. Other antibodies used in this study were: mouse anti-Aurora B antibody (anti-AIM-1; BD Transduction Laboratories, San Diego, California, United States); CREST serum (gift from A. Kromminga, Hamburg, Germany); mouse anti-cyclin B1 antibody (Santa Cruz Biotechnology, Santa Cruz, California, United States; #GNS1); rabbit anti-Mad2 serum [ 38 ]; mouse anti-myc antibody (clone 4A6, Upstate Cell Signaling Solutions, Charlottesville, Virginia, United States); antibody to phosphohistone H3 (Ser 10 ) [P-H3, a mouse monoclonal antibody that detects histone H3 when phosphorylated at serine 10 (Cell Signaling Technology, Beverly, Massachusetts, United States; #6G3)]; rat anti-HP-1β (Serotec Laboratories, Oxford, United Kingdom; #MCA1946); and rabbit anti-SA2 (antibody 466; see [ 39 ]). Cell culture, synchronisation, and transfection HeLa cells were cultured in DMEM supplemented with 10% FCS, 0.2 mM L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin. For synchronisation, HeLa cells were grown in the presence of 1 mM thymidine (Sigma-Aldrich, St. Louis, Missouri, United States) for 12 h, washed with PBS, and grown in fresh medium for 8 h. Cells were then transfected (see below) in serum-free medium (Opti-MEM, Gibco, San Diego, California, United States) supplemented with 0.3% FCS for 4 h, followed by addition of 1 mM thymidine and FCS to a final concentration of 20%. After a further 12 h, cells were again washed in PBS and transferred to fresh medium. Samples were harvested at various time points up to 15 h after the second release, as described below and in the figure legends. Nocodazole and Hesperadin were each used, as indicated, at a final concentration of 100 ng/ml and 100 nM, respectively. Doxycycline-inducible nonphosphorylatable SA2 HeLa cells were established as described in the accompanying paper [ 17 ]. For generation of EGFP-H2B stably expressing line, cells were transfected with plasmid encoding EGFP-H2B [ 40 ] using FuGene6 reagent (Roche, Basel, Switzerland), and stable expressants were selected in a complete medium containing 2.0 μg/ml blasticidin-S, and were screened by fluorescence microscopy for expression of EGFP-H2B. HeLa cells stably expressing EGFP-CENP-A were established and characterized as described [ 41 ]. For flow cytometric analysis, cells fixed in 70% ethanol were washed with PBS and subsequently stained in PI buffer [10 μg/ml propidium iodide, 10 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , and 200 μg/ml RNase A] for 20 min at 37 °C. RNAi The oligonucleotide sequence used to target human Sgo1 (oligo Sgo1.1A) was 5′-CAGU AGAACCUGCU CAGAA-3′. The oligonucleotide sequence used to target human Plk1 cDNA was 5′- CGAGCUGCUUAAU GACGAG-3′ [ 20 ]. Synthetic sense and antisense oligonucleotides were purchased from VBC Genomics (Vienna, Austria). For mock transfections, dH 2 O was used instead of siRNA. Transfection of siRNA duplexes was performed at a final concentration of 150 nM. Experiments performed with a second duplex targeted to a different region of human Sgo1 cDNA (oligo Sgo1.2B) yielded similar results. The oligonucleotide sequence of Sgo1.2B is 5′-GGAUAU CACCAAUGUCUCC-3′. Preparation of HeLa cell extracts To prepare fractionated HeLa cell extract, cells were harvested by washing in ice-cold PBS and scraping from the plate. Cells were pelleted at 1,200 g for 3 min at 4 °C, and washed three times with ice-cold PBS. The cell pellet was resuspended in 200 μl of extraction buffer, which contained 150 mM NaCl, 10 mM NaF, 40 mM β-glycerophosphate, 20 mM HEPES (pH 7.5), 2 mM MgCl 2 , 10 mM EDTA, 0.5% Triton X-100; 10 ml of this buffer was supplemented with one tablet of Complete Protease Inhibitor (Roche) and 1 mM PMSF. Cells were homogenized with ten strokes in a Dounce homogenizer and incubated on ice for 10 min, followed by another ten strokes. Nuclei and insoluble material were pelleted by centrifugation at 13,000 g for 10 min at 4 °C. The supernatant was completely removed and denatured by addition of an appropriate volume of 4× Laemmli buffer followed by boiling for 10 min at 95 °C. The pellet was resuspended in 4 volumes of 1× Laemmli buffer, boiled at 95 °C for 10 min, and passed two times through a 27-gauge needle to shear DNA. Immunofluorescence microscopy For immunostaining, cells were either cultured (and transfected where applicable) directly on glass coverslips, spun onto glass slides using a Cytospin centrifuge (Shandon brand, available from Thermo Electric, Waltham, Massachusetts, United States) for 5 min at 1,500 rpm, or spread on glass slides as described below. Cells were washed with PBS, (in the case of Scc1-myc and SA2-myc staining only, cells were preextracted with 0.1% Triton X-100 in PBS for 2 min followed by PBS wash for 2 min), and fixed in 4% paraformaldehyde in PBS for 15 min. Subsequently, cells were permeabilized with 0.2% Triton X-100 in PBS for 5 min. Cells were blocked in 3% BSA in PBS-T (0.01% Triton X-100) for 1 h prior to incubation with primary antibodies in the same blocking solution. Secondary antibodies were incubated for 1 h at room temperature. The following secondary antibodies conjugated to Alexa Fluor 488 and 568 (Molecular Probes, Eugene, Oregon, United States) were used: goat anti-rabbit, goat anti-mouse, and goat anti-human. DNA was stained by incubating cells in DAPI (1 μg/ml in PBS) for 10 min. Cells were mounted in Vectashield Mounting Medium (Vector Laboratories, Burlingame, California, United States). Images were captured using MetaMorph software (Universal Imaging, Downingtown, Pennsylvania, United States). Chromosome spreads To prepare chromosome spreads for Giemsa staining, cells were harvested by trypsinization and pretreated with hypotonic buffer containing 40% medium and 60% tap water for 5 min at room temperature. Cells were fixed with freshly made Carnoy's solution (75% methanol and 25% acetic acid), the fixative was changed three times, and the cells were then stored overnight at −20 °C. For spreading, cells in Carnoy's solution were dropped onto glass slides and dried at room temperature. Slides were stained with 5% Giemsa (Merck, Darmstadt, Germany) at pH 6.8 for 10 min, washed briefly in tap water, air-dried, and mounted with Entellan (Merck). Chromosome spreads for immunostaining were prepared using the method previously described for spermatocyte spreads [ 42 ], as adapted for HeLa cells. Briefly, cells were treated with 330 nM nocodazole for 1 h, and mitotic cells were collected by shaking-off. Cells were incubated in a hypotonic buffer [30 mM Tris (pH 8.2), 50 mM sucrose, and 17 mM sodium citrate] for 7 min and then suspended in 100 mM sucrose (pH 8.2). A small volume of the cell suspension was put on a slide glass that had been dipped in fixative [1% paraformaldehyde, 5 mM sodium borate (pH 9.2), and 0.15% Triton X-100] and dispersed on the slide by continuous tilting. After extensive washing with PBS, lysed nuclei were dried and processed for immunofluorescence microscopy. Live cell imaging Cells were grown and synchronised in Lab-Tek chambered cover glasses (Nunc, Roskilde, Denmark) and transfected with siRNA as described in Figure 1 . At 6 h after the release from the second thymidine block, medium was changed to CO 2 -independent medium without phenol red, and the chambers were sealed with silicone grease. Three-dimensional image stacks of live cells were captured over time using either Zeiss 510 confocal microscope ( Figure 4 A and 4 C; Zeiss, Oberkochen, Germany) or Olympus BX51 fluorescence microscope ( Figures 4 B and 8 D; Olympus, Tokyo, Japan). Supporting Information Figure S1 Cloning and Expression Analysis of hsSgo1 and mmSgo1 (A) The expression pattern and primary structures of human and mouse Sgo1 (hsSgo1 and mmSgo1, respectively) transcription products were investigated. Total RNA from HeLa cells, mouse testis and mouse thymus were prepared using TRIzol reagent (Invitrogen, Carlsbad, California, United States) according to the manufacturer's instructions. First-strand cDNA was synthesized by SuperScript II reverse transcriptase (Life Technologies, Carlsbad, California, United States) primed by an oligo dT primer, and PCR was performed using the Expand High Fidelity PCR system (Roche). For mmSgo1 cDNA amplification, we designed primers 5′- CTGAGTGGCCGAGATGAATTTCAC-3′ and 5′- TGTCCAAGAGACCCTCCTGATCAG-3′ according to GenBank cDNA NM_028232. We obtained two major bands at 0.9 and 1.8 kb [lane 1 (testis) and lane 2 (thymus)] and cloned them into the pGEM-T vector (Promega, Madison, Wisconsin, United States). Multiple clones were sequenced and PCR error-free sequences mmSgo1A and mmSgo1B were obtained. Note that when the secondary PCRs were done using the PCR reaction above as templates and primers 5′- AATCTATGACCCACCTGCCTTAGC-3′ and 5′- GGTTCTGCCCATTGTGCACTGTCT-3′, we saw three minor bands between 0.9 and 1.5 kb (unpublished data), suggesting multiple species of splicing variant. For hsSgo1 cDNA cloning, we had found two potential splicing variants in the ENSEMBL database (BC001339 and BC017867) that have different 3′ sequences. When primers 5′- CTGGAGAGCTTCGAAGAGCCTTGA-3′ and 5′- CCTCTCCTGAAGCAACAGAAAGAG-3′ (designed to amplify products from BC001339) were used, 0.9- and 1.0-kb bands were detected (lane 3), and hsSgo1A–hsSgo1D cDNAs were obtained. When primers 5′- CTGGAGAGCTTCGAAGAGCCTTGA-3′ and 5′- CTGAGTGAAACAGACTGTCAACAC-3′ (designed to amplify products from BC017867) were used, six bands between 0.9 and 2.0 kb were detected (lane 4), and cDNA clones obtained from those bands were named hsSgo1E–hsSgo1H and hsSgo1J–hsSgo1L . We could not obtain a clone that encodes the identical open reading frame encoded by BC001339. (B) To examine the expression of Sgo1 in HeLa cells and various other human and mouse tissues, we performed Northern analysis. Total HeLa RNA was prepared as described in (A), and 20 μg was run on a formaldehyde agarose gel and RNA blots were prepared. A blot was probed with a [ 32 P]-labelled probe (probe H1) generated against the common N terminus region of hsSgo1 [nucleotides (nt) 1–647 of hsSgo1E). Many bands, although faint and ambiguous, were seen (lane 1). We could not assign each signal to a specific cloned cDNA variant; however, these data support the notion that many species of splicing variant are expressed in HeLa cells. When the blot was probed by another probe (probe H2) that is designed to hybridize to the central region specific for hsSgo1E and hsSgo1F cDNAs (nt 665–1454 of hsSgo1E), a signal at ∼2.2 kb was seen (lane 2). Using two probes of hsSgo1, we could confirm that the two relatively long transcripts, hsSgo1E and hsSgo1F, that possess the central region, and the other shorter transcripts that lack the central region, are expressed in HeLa cells. (C) The tissue distribution of hsSgo1 mRNA was studied using a commercially available Multi Tissue Northern blot (BD Clontech, Palo Alto, California, United States). Multiple, relatively strong, signals were detected in testis, among other organs, on both blots probed by H1 and H2 (upper left blot and lower left blot, respectively). This higher expression of Sgo1 in testis might reflect the fact that testis contains a higher fraction of proliferating cells. Expression analysis of mouse Sgo1 was also performed in parallel. Both a probe against the common N-terminal region of mmSgo1 (probe M1, corresponding to nt 1–610 of mmSgo1A ) and an mmSgo1A-specific probe (probe M2, corresponding to nt 704–1339 of mmSgo1A ) detected potential mmSgo1 messages in spleen and testis (upper right blot by probe M1, and lower right blot by probe M2, respectively). Mouse Sgo1 messages in spleen were more abundant than in testis, but human Sgo1 message was not detected in human spleen. The biological significance of the different levels of Sgo1 expression observed between human and mouse spleen is unclear. The tested polyA + RNAs are from spleen (SP), thymus (TH), prostate (PR), testis (TE), ovary (OV), small intestine (SI), colon (CO), peripheral blood leukocyte (PL), heart (HE), brain (BR), lung (LU), liver (LI), skeletal muscle (SM) and kidney (KI). In conclusion, we cloned multiple splicing variants of hsSgo1s and mmSgo1s from HeLa cells and mouse organs. The presence of a variety of messages was also confirmed. However, the functional difference between and biological meaning of such variation remains unknown and requires further study. (D) Structures of cDNAs for hsSgo1A–hsSgo1H and hsSgo1J–hsSgo1L . (E) Predicted structures of proteins depicted as cDNAs in (D). (F) Structures of cDNAs for mmSgo1A and mmSgo1B . (G) Predicted structures of proteins depicted as cDNAs in (F). (2.7 MB EPS). Click here for additional data file. Figure S2 Sgo1 Antibody Characterisation Identical Sgo1 localisation patterns were obtained by staining cells with two peptide antibodies raised against different regions of the protein (see Materials and Methods for details). (A) HeLa cells grown on glass coverslips were paraformaldehyde-fixed and costained with Sgo1 antibody 94 (directed against C-terminal peptide, shown in green) and CREST antiserum (shown in red) to label kinetochores. Five different stages of mitosis are illustrated as indicated. (B) As (A), except cells were costained with Aurora B antibody (shown in red) to label inner centromere. (C) As (B), except Sgo1 was stained with antibody 95 (directed against N-terminal peptide, shown in green). (10 MB EPS). Click here for additional data file. Figure S3 Precocious Sister Separation Is Specific to Sgo1 siRNA Treatment Synchronised HeLa cells were transfected as described in Figures 1 and 3 with siRNA oligonucleotides directed against the following proteins: Aurora B, condensin subunits Cap-D2, Cap-D3, Cap-G, Cap-G2 and Smc2, Cyk4, Ect2, Lamin A, myosin heavy chain Type II (MHC), MKLP1, RhoA, and Sgo1 (two oligonucleotide targets; see Materials and Methods ). Cells were harvested 11 h after the second thymidine release and analysed as described for Figure 3 A and 3 B. The histogram indicates the frequency of each observed category as a percentage of total cells, such that the sum of each column represents the mitotic index. The separated sisters category (represented by the red bars) combines the three categories illustrated in Figure 3 B, part e–g. (1.3 MB EPS). Click here for additional data file. Figure S4 No Reduction of Cohesin on Prophase Chromosomes in Sgo1-Depleted Cells HeLa cells expressing Scc1-myc were mock-treated or Sgo1 siRNA-transfected and harvested as described for Figure 5 , cells spun down on glass slides, preextracted prior to fixation, and analysed by immunofluorescence microscopy using myc antibody. Cells were costained with P-H3. Prophase cells, identified by positive P-H3 staining with intact nuclear envelope, were measured for their fluorescence intensities of Scc1-myc and P-H3 using Image-J software. The ratio of fluorescence intensities of myc signals to P-H3 signals was calculated. (583 KB EPS). Click here for additional data file. Figure S5 Induction of SA2-myc Expression upon Doxycycline Treatment A sample of the cells from the experiment described in Figure 8 B were spun onto glass slides and processed for immunofluorescence, staining with myc antibody as described (but cells were not preextracted with 0.1% Triton X-100 prior to fixation). In uninduced cells, a moderate level of leaky expression of the tagged protein is observed in a small number of cells (photomicrographs a, b, e, and f). In induced cells, a high level of the tagged protein is observed in the vast majority of cells (photomicrographs c, d, g, and h). Anti-myc staining of HeLa cells which do not contain any myc-tagged gene expression cassette (photomicrograph i). (4.7 MB TIF). Click here for additional data file. Accession Numbers The NCBI GeneID ( http://www.ncbi.nlm.nih.gov/ ) and ENSEMBL ( http://www.ensembl.org/ ) accession numbers of the genes discussed in this paper are hsSgo1 (GeneID 151648; ENSEMBL gene ENSG00000129810) and mmSgo1 (Gene ID 72415; ENSEMBL gene ENSMUSG00000023940). The pending GenBank/EMBL/DDBJ accession numbers of the cDNA sequences discussed in Figure S1 are hsSgo1A (AB193056), hsSgo1B (AB193057), hsSgo1C (AB193058), hsSgo1D (AB193059), hsSgo1E (AB193060), hsSgo1F (AB193061), hsSgo1G (AB193062), hsSgo1H (AB193063), hsSgo1J (AB193064), hsSgo1K (AB193065), hsSgo1L (AB193066), mmSgo1A (AB193067), and mmSgo1B (AB193068).

D:\keerthana\PMC001xxxxxx\PMC1054882.xml

1054883

Predictive Spatial Dynamics and Strategic Planning for Raccoon Rabies Emergence in Ohio

Rabies is an important public health concern in North America because of recent epidemics of a rabies virus variant associated with raccoons. The costs associated with surveillance, diagnostic testing, and post-exposure treatment of humans exposed to rabies have fostered coordinated efforts to control rabies spread by distributing an oral rabies vaccine to wild raccoons. Authorities have tried to contain westward expansion of the epidemic front of raccoon-associated rabies via a vaccine corridor established in counties of eastern Ohio, western Pennsylvania, and West Virginia. Although sporadic cases of rabies have been identified in Ohio since oral rabies vaccine distribution in 1998, the first evidence of a significant breach in this vaccine corridor was not detected until 2004 in Lake County, Ohio. Herein, we forecast the spatial spread of rabies in Ohio from this breach using a stochastic spatial model that was first developed for exploratory data analysis in Connecticut and next used to successfully hind-cast wave-front dynamics of rabies spread across New York. The projections, based on expansion from the Lake County breach, are strongly affected by the spread of rabies by rare, but unpredictable long-distance translocation of rabid raccoons; rabies may traverse central Ohio at a rate 2.5-fold greater than previously analyzed wildlife epidemics. Using prior estimates of the impact of local heterogeneities on wave-front propagation and of the time lag between surveillance-based detection of an initial rabies case to full-blown epidemic, specific regions within the state are identified for vaccine delivery and expanded surveillance effort.

Introduction Major recommendations from several Centers for Disease Control and Prevention (CDC) and Institute of Medicine reports on emerging diseases underscore the importance of improving and developing new surveillance strategies to better inform interventions limiting the impact of novel and reemerging disease [ 1 , 2 , 3 , 4 ]. By using predictive models of spread we can position ourselves to target surveillance activities and prepare emergency response interventions tailored to the particular feature of an outbreak. Rabies emergence in the eastern United States and Canada provides an excellent example of how predictive models can help guide surveillance and intervention strategies. Though long endemic in the southeastern U.S., raccoon rabies expanded rapidly along the eastern seaboard during the 1980s and 1990s from an initial focus along the West Virginia–Virginia border; the initial focus was linked to the long-distance translocation (LDT) of rabid animals from Florida [ 5 ]. The particular variant of rabies virus associated with raccoons [ 6 , 7 , 8 ] has spread as a heterogeneous wave away from its original site of introduction and now extends as far north as Ontario [ 9 ] and as far west as eastern Ohio, Tennessee, and Alabama [ 10 ]. Understanding the context in which models of rabies spread were first developed requires a brief background explanation of the history of epidemic raccoon rabies and the methods used to control spread. Since late 1997–early 1998, the westward expansion of rabies in raccoons has been curtailed through a massive program of oral rabies vaccine (ORV) delivery focused in eastern Ohio, and later expanded into adjacent border counties in Pennsylvania and West Virginia. Organized by the Ohio Departments of Health, Natural Resources, and Agriculture, in collaboration with the U.S. Department of Agriculture, the CDC, and Canadian agencies, a vaccine cordon sanitaire was constructed to keep raccoon rabies out of Ohio [ 11 , 12 ]; it also served the greater purpose of potentially preventing the raccoon variant's spread throughout the geographic range of raccoons, which, with minor exceptions in the southwest, includes the entire continental U.S. south of Alaska [ 13 ]. The campaign was successful, and through 2003, raccoon rabies in Ohio was limited to a few sporadic cases within the vaccine zone [ 10 ]. In 2004, this comforting streak of success came to an abrupt end. On 21 July 2004, approximately 11 km beyond the western extent of the vaccine corridor, a rabid raccoon was detected in Leroy Township, Lake County, Ohio [ 14 ] ( Figure 1 ). As of 12 August, at least ten additional rabid raccoons had been detected in Lake County and surrounding counties ( Figure 1 ); given the delay of 3–4 mo between first detection of rabies and development of a full-blown epidemic [ 15 , 16 , 17 ], and the 40-km linear extent of this outbreak, we assume this is a new rabies focus with the potential to engender continued spread of raccoon rabies into Ohio and areas to the west. Figure 1 Spatial Location of All Positive Raccoon Rabies Cases in Eastern Ohio as of 11 August 2004 Each sample is identified by number, date of collection, and the laboratory responsible for the positive identification (ODH indicates Ohio Department of Health). To date, more than 300 raccoons have been submitted to the CDC for testing. Ci, Cincinnati; Cl, Cleveland; Co, Columbus; T, Toledo; Y, Youngstown. Various state agencies have already initiated a remedial wildlife vaccination program to limit further expansion of this epidemic. However, for strategic intervention purposes, determining the expected trajectory and velocity of rabies spread and estimating an effective zone for remedial vaccination would be invaluable [ 18 ]. In this paper, we use previously verified mathematical models for predicting rabies spatial dynamics to provide guidance on the likely trajectory and velocity of rabies spread from this emerging focus of infection and to identify specific areas for vaccine delivery and active surveillance. Results/Discussion The Raccoon Rabies Epidemic When the expanding wave front of raccoon rabies reached the borders of Connecticut (CT) and New York (NY), county-level counts of animal rabies data were already being reported monthly to the CDC, as required for this nationally notifiable disease. Previously we analyzed time series data from the county-level data to make predictions about the temporal structure of recurring epidemics and estimated the lag time (approximately 3–4 mo) from detection of raccoon rabies to epizootic development [ 15 , 16 ]. In separate analyses, we arrived at similar estimates for surveillance delays [ 17 ]. The county-level data were suitable for assessing a general rate of front movement across the northeastern U.S. but not sufficiently resolved so as to provide information about local environmental heterogeneities that may alter rates of spread. Concurrent with the collection of county-level data, animal rabies cases occurring within individual townships were also collected by NY and CT state health departments, increasing the spatial resolution of disease reporting data 20-fold above that available from county-level reports. This increase in spatial resolution allowed for the construction of a detailed spatial model of rabies spread. The Model The data from NY and CT have been used to parameterize a stochastic spatial model for rabies spread among townships ( Figure 2 ). The model was used for an exploratory data analysis to quantify the spatial dynamics across CT, and we found a 7-fold reduction in local transmission when geographic regions (i.e., townships) were separated by major rivers. In addition, the LDT of rabid animals was incorporated into the model to accommodate these rare, but significant events, often capable of engendering epizootics well in advance of the wave front when larger spatial domains were examined. The spatial model parameterized for CT was used to hind-cast the time to first appearance of raccoon rabies across NY townships using only knowledge of where the disease was introduced into the state and local heterogeneities that may influence the rates of spread [ 19 ]. The a priori CT model was able to predict the pattern of spread across NY, as well as provide an estimate of the possible effect of ORV intervention. Having demonstrated the predictive power of the spatial stochastic simulator for CT and NY, we can now use the model and our knowledge of the current outbreak to strategically forecast the spatial spread of rabies in Ohio. Figure 2 A Stochastic Model Was Used to Simulate the Heterogeneous Spread of Raccoon Rabies on an Irregular Network, Illustrated Here on a Simple Array An infected township i infects its adjacent neighbor j at rate λ ij . In addition, township j may become infected because of translocation of rabid raccoons at rate μ j . Heterogeneity was incorporated by allowing the local rate of infection from neighboring townships, and the rate of translocation, μ j , to be different in different models. Each algorithm for associating a set of rates with rivers defines a stochastic candidate model. The simulation algorithm involved six steps. (A) For each township, add the rates of infection from all possible routes of infection. (B) Add the townships rates to compute a total rate. (C) Draw a random number to determine the elapsed time. (D) Check to see if any forced townships have become infected in the elapsed interval. (E) If no edges were forced, select a random township to infect. (F) Infect the forced township, update the local rates, and repeat until each township becomes infected (after [ 17 ]). The model transforms Ohio townships into a network; local spread of raccoon rabies among adjacent townships in the Ohio model was predicted where two townships shared at least one common point along their borders. An infected township i was assumed to infect its adjacent neighbor j at rate λ ij . In addition, long-distance dispersal of rabies was incorporated by assuming a low and constant rate of global infection μ j for all uninfected townships regardless of spatial proximity to infected neighbors ( Figure 2 ). Incorporating environmental heterogeneities into local rates of spread is crucial to predicting the local wave-front movement of terrestrial rabies. Rabies transmission was 7-fold lower when townships were separated by a river in CT, while in NY, the Adirondack mountain range was, and remains, an impenetrable barrier to the incursion of raccoon rabies. Ohio lacks the rough mountainous terrain of NY, but it has several major rivers that could influence westward expansion of raccoon rabies from a focus near the Pennsylvania or West Virginia border; five major Ohio rivers, the Miami, Muskingum, Scioto, Maumee, and Cuyahoga Rivers, were incorporated into our simulations ( Figure 3 ). Figure 3 Satellite Image of Ohio Topography Illustrating Major River Systems and Location of Vaccine Corridor Extent and shape of the vaccine corridor is approximated. (Map reproduced with permission from Ohio Department of Natural Resources, Division of Geological Survey.) The Ohio Forecast Transposing the model for use in Ohio merits a discussion of the geographic differences between CT, NY, and Ohio. In CT, the Connecticut River bisects the state, running north to south. The river's effect on the epidemic was maximized because the direction of epidemic spread ran orthogonal to the river, i.e., from west to east. Moreover, CT is much smaller than NY or Ohio. When the model was transposed into NY from CT, the townships in the Adirondack Mountains were excluded a priori because none of them had ever reported a case of raccoon rabies. The general triangular shape of NY and the exclusion of the Adirondack townships from the simulation, in conjunction with there being three initial foci within NY, left few townships far from any single initial focus. Thus, the successful hind-cast was based mostly on the relatively predictable local dynamics. In contrast, in Ohio we know of only one focus, in the northeastern quarter of the state, leaving many townships hundreds of kilometers from the epidemic focus. Given the greater distance between the epidemic focus and far reaching townships, LDT of infected individuals in Ohio could have a much more profound impact on the wave-front dynamics than was possible in NY or CT. Given the unpredictable nature of LDT events, we present two projections. In the first projection, we ignored LDT and focused on local rabies wave-front movement: λ ij = α = 0.66 for townships not separated by a river; λ ij = β = 0.12 for townships separated by a river; and μ = 0. In the second projection, we used the exact parameters fitted to the CT epidemic: λ ij = α = 0.66 for townships not separated by a river; λ ij = β = 0.12 for townships separated by a river; and μ = 0.0002. The local-spread-only projection for Ohio assumed the rabies wave-front origin was from the townships nearest the center of the Lake County outbreak (case 3 in Figure 1 ), and the epidemic was simulated without LDT. As the local dynamics of rabies are fairly predictable, this could be considered the best-case scenario. Retaining the parameters for local spread from CT and disallowing any LDT, the rabies epidemic required 70 mo to reach the western corners of the state ( Figure 4 A). In the absence of LDT, the predicted rate of front propagation is faster than previously reported for other states because of the influence of the permissive zone of central Ohio, in which few environmental impediments exist. Figure 4 Predicted Trajectory of Rabies in Ohio Based on the Current Outbreak (A) Predicted time to first appearance (months) of raccoon rabies spreading from Chardon Township and surrounding townships given no vaccine intervention and with LDT excluded from model predictions. The time course of spread is revealed through the contour plot, where each color band indicates a given time interval to arrival at a township. The width of the bands corresponds to velocity of spread, with wider bands associated with more rapid spread. Major cities are Cleveland (Cl), Youngstown (Y), Toledo (T), Columbus (Co), and Cincinnati (Ci). The two black lines labeled A and B correspond to the area where cases have been detected (A) and the expected position of the wave front given a long-tailed distribution of incubation periods. (B) Predicted time to first appearance (months) of raccoon rabies spreading from Chardon Township and surrounding townships given no vaccine intervention and including estimates of long-distance dispersal modeled in CT and NY . The projection with LDT was approximately one-third faster than the local-only spread, even with the very low levels of long-distance dispersal modeled. In these simulations rabies spread across central Ohio within 33 mo and covered the state by month 41 ( Figure 4 B). Passage across the midsection of Ohio was particularly fast; the estimate of 100 km/y far exceeds previous estimates for the rate of spread of raccoon rabies, which typically ranges between 30 and 60 km/y [ 20 , 21 , 22 ]. By our estimates, if unchecked, rabies will likely spread across Ohio in the same amount of time that it took rabies to transverse CT, even though Ohio is 2.1 times wider than CT. The potential for such rapid spread is quite alarming. Adding the possibility of global LDT to the model, even at the low rate assumed in the original model for CT, served as a massive promoter of rapid spread. This can be envisaged as the worst-case scenario, with LDT greatly increasing the rate of wave-front propagation across the state. Because of the largely unpredictable nature of LDT, it must be assumed that reality is likely to be somewhere in between the two modeled scenarios, but exactly where is difficult to discern. The picture that emerges from both scenarios suggests that rabies will spread rapidly through the middle of Ohio, where there are few environmental impediments to disease front expansion. Rivers and mountains in Ohio do not constitute a major barrier to rabies spread. Similarly, rabies will not encounter any impediment like the Adirondack Mountains in NY. With no effective physical barrier across the middle of Ohio, rabies could move more rapidly through this zone then in any previously recorded epizootic. The design of vaccine barriers to control the Lake County epidemic must allow for the distance the wave front will advance prior to detection of the first laboratory-confirmed case of raccoon variant rabies within a county. This parameter is unknown, although estimates of the delay from the first detected case of raccoon variant rabies to the start of the first epidemic exist [ 15 , 16 , 17 , 18 ]: the delay is approximately 3–4 mo. Thus, the detection of the first case of raccoon rabies in a township through passive surveillance indicates roughly where the front was 3–4 mo before. As no other data exist to help assess the bias in lag between the arrival date of raccoon rabies and the date of detection recorded in the national surveillance database, we set the interval from arrival to detection at 3–4 mo. This parameter defines where to place the inner perimeter line A ( Figure 4 A). Few estimates of incubation time from natural field situations exist; however, Tinline et al. [ 18 ] provided some estimates on the distribution of incubation times for raccoons infected with their homologous variant. We used the 75th percentile in incubation time from Tinline et al. [ 18 ] to derive a confidence line for the expected wave-front advance (line B in Figure 4 A), adding the incubation time to the calculations for line A; thus the area between perimeter lines A and B defines the vaccination corridor required to target susceptible animals ( Figure 4 A). Strategic Planning and Alternative Outbreak Scenarios Any intervention decisions or strategic plans based on model projections require that models deliver a robust outcome. We explored two additional scenarios for raccoon rabies emergence and spread through Ohio based on potential breach points in the vaccine corridor and consider the utility of the model for selecting sights for enhanced surveillance. Alternative scenario 1 A breech in the cordon sanitaire could occur within a similar time frame as a LDT, and we considered the effects of independent and concomitant breeching and LDT events in a series of alternative scenarios. Adding such events to the spatial simulations significantly decreases the times to first appearance of raccoon rabies compared to that predicted solely on local spread. Modeling various LDT scenarios is warranted because LDTs occur frequently [ 23 ] and have demonstrably played a central role in the epizootiology of raccoon rabies, and are regarded herein and elsewhere as a critical element in our efforts to understand and model the epizootic dynamics of raccoon rabies. We simulated the pattern of rabies spread associated with a breach in the vaccine corridor near Youngstown (near the current outbreak location) coupled with a forced LDT event into the Athens-Hocking landfill. The Athens-Hocking landfill is the landfill closest to the current front of raccoon rabies outside of Ohio that receives interstate refuse. The LDT event was incorporated into the computer simulation as an initial condition that occurred at month 12 during the simulated spread, in addition to base levels of LDT derived from CT (μ = 0.0002). Spread from the landfill in conjunction with the advancing front originating near Youngstown generated two advancing fronts that converged in the middle of the state ( Figure 5 A). Figure 5 Combined Effects of LDT and a Break in the Vaccine Barrier on Pattern of Rabies Spread across Ohio (A) Predicted time to first appearance (months) of raccoon rabies given a breach of the cordon sanitaire near Youngstown and the LDT of an infected individual into the Athens-Hocking landfill at month 12. (B) Residual differences between simulations of the Youngstown breach with and without the forced LDT event at the Athens-Hocking landfill. The larger the red squares, the larger the residual difference between the simulations with and without the forced LDT. Ci, Cincinnati; Cl, Cleveland; Co, Columbus; T, Toledo; Y, Youngstown. To explore the overall impact of multiple introductions on the time to first appearance we compared simulations with and without the translocation event into the Athens-Hocking landfill. The forced LDT event caused rabies to reach the southern portion of Ohio far earlier than predicted in the absence of the forced LDT. However, the rivers seem to limit the overall additive effect on time of first appearance in western Ohio by acting as natural barriers to spread. There appears to be no additive effect on the rate of disease movement across the middle of the state associated with the union of the advancing wave fronts, as revealed in a plot of the residual differences between simulations ( Figure 5 B). Alternative scenario 2 Rabies has a variety of alternative paths of entry into Ohio from infected regions across the cordon sanitaire . In this scenario, we consider the spatial trajectories of rabies spread from a variety of single points of entry and compare these alternative trajectories to determine the most common areas of overlap and, hence, where rabies is most likely to emerge. For example, we consider the trajectory of spread given an introduction around Youngstown and compare this with an introduction around Bridgeport, Ohio (near Wheeling, West Virginia), that would be near the southern end of the vaccine corridor in Ohio. It is reasonable to assume that the Ohio River serves as a barrier to raccoon movement, but there is the potential for rabid individuals to cross bridges into Ohio. Given an introduction at Bridgeport, Ohio ( Figure 6 ), the model predicts that the initial westward expansion would be halted by the Muskingum River, while northward expansion would be rapid. Following a crossing of the Muskingum, the model predicts the same sort of rapid westward expansion seen in the first scenario. Figure 6 Combination of Predicted Time to First Appearance Projections from Youngstown Breach and Ohio River Breach Scenarios Thick black line indicates the minimal distance between the two epidemic foci orthogonal to the temporal contours. Given the trajectories, the path corresponds to the most likely location where the waves will first collide. Scenario comparison By jointly considering the predictions from these introduction scenarios, we demonstrate the utility of our model in establishing sentinel points for surveillance. Figure 6 shows the model predictions for spread associated with a river crossing near the southern terminus of the cordon sanitaire overlaid with our projections for spread away from the Youngstown area introduction. From the combination of these two scenarios we see that the first point at which the two epidemic fronts are predicted to meet is not along the Ohio–West Virginia border but further west towards the interior of the state. We can use this juxtaposition of the two trajectories to establish a minimum time path for the epidemic movement ( Figure 6 ) and can suggest sentinel points for surveillance along this line in addition to the surveillance strategies already in place. The three simulation sets—the current outbreak (with LDT) and alternative scenarios 1 and 2—can be compared to assess the sensitivity of rate estimation on at least these initial conditions. A plot of the maximum residual differences in time to first appearance between all three simulation sets suggests that the predicted velocity and overall trajectory of rabies spread are robust to changes in outbreak origin ( Figure 7 ). Though large residuals exist between the three simulations around the introduction sites, a striking common feature of all simulation outcomes is the rapid spread of raccoon rabies across central Ohio given an average effect of LDT. Maximum differences in the estimated time to raccoon rabies arrival between simulations in central and western Ohio varied by less than 1 mo, irrespective of breach location in the ORV barrier. Figure 7 Residual Differences between Time to First Appearance from Simulated Epidemics Initiated by Breaches in Different Geographic Locations The maximum difference between the predicted times to first appearance is indicated in each township by the size and color of the corresponding diamond. (A) represents an epidemic simulated from current outbreak data shown in Figure 1 . (B) represents an epidemic simulated from the movement of an infected animal across the cordon sanitaire near Youngstown. (C) represents the introduction of an infected animal from West Virginia at a bridge across the Ohio River. Conclusions Experience with parenteral rabies vaccination programs and ORV-based attempts to control wildlife rabies in the U.S. and Europe has demonstrated that all vaccine barriers are, to some extent, permeable and vulnerable to breaches [ 24 , 25 , 26 ]. Ohio has proven to be no exception. Considering the vulnerabilities of any vaccine corridor, and the difficulties inherent in anticipating break sites in a vaccine barrier and identifying sites of LDT, the unlimited ability to explore scenarios by mathematical simulation and to compute and compare different design strategies for ORV interventions provide a valuable exercise to plan for real epidemics. From the simulations shown here, a robust pattern of trajectory, an extraordinarily high rate of spread of raccoon rabies through central Ohio, the significant potential impact of LDT, and lack of environmental impediments suggest that a strategy combining early detection and rapid intervention are requisite if control is to succeed. A robust intervention plan is invaluable, even if the exact site or sites to which it is applied prove elusive . The approach discussed here reveals some of the many gaps in our knowledge of wildlife rabies and in the data available to inform initial conditions for simulations. Without a second measure of the arrival–detection delay, obtained from an independent and extremely sensitive surveillance system (e.g., coupling active sampling of road-killed raccoons, purposeful hunter collections, and kill-trap data), we have no tools to calibrate detection times obtained through the passive national surveillance system. In western Europe, reinforcement of natural barriers, such as rivers, lakes, and mountains, have been a staple in the successful campaign to control red fox rabies by vaccination [ 24 ]. In Massachusetts and NY, pre-existing natural (i.e., the Adirondacks) and artificial (i.e., the Cape Cod canal [ 27 ]) impediments to the free movement of raccoons, and, hence, raccoon rabies, were reinforced by vaccine distributions. Unfortunately for the remedial efforts to contain the Lake County focus in Ohio, the benefits derived from enhancement of natural barriers are not an option. With the exception of river systems in southeastern Ohio, such as the Muskingum (for topographic details see Figure 3 ), there are few natural barriers to augment ORV distribution in the Lake County region or in central Ohio, if required. As reinforced natural barriers are not an option for rabies control in much of Ohio, the need for rapid remedial intervention by ORV and intensified, active surveillance to estimate the actual epidemic boundaries is immediate. Enhanced surveillance, perhaps by collection and rabies testing of road-killed raccoons, and stockpiling of ORV sufficient to establish a new ORV barrier in central Ohio, are prudent courses of action. If the Lake County epidemic escapes beyond remedial intervention efforts, the fallback position demands rapid construction of a second cordon sanitaire spanning central Ohio, at a location dictated by the then current position of the epidemic. If the disease is not confined within Ohio, the limits to raccoon variant rabies spread are defined only by the geographic distribution of its host. Nothing short of these activities will contain a rapidly expanding and fast-moving epidemic of raccoon rabies, should ring vaccination and other regional efforts fail to eliminate the current focus around Lake County. The interventions described here would appear as inexpensive alternatives to the uncontrolled westward spread of rabies and the loss of the millions of dollars invested in what would become the vaccine equivalent of the Maginot Line.

D:\keerthana\PMC001xxxxxx\PMC1054883.xml

1054884

Recombination Every Day: Abundant Recombination in a Virus during a Single Multi-Cellular Host Infection

Viral recombination can dramatically impact evolution and epidemiology. In viruses, the recombination rate depends on the frequency of genetic exchange between different viral genomes within an infected host cell and on the frequency at which such co-infections occur. While the recombination rate has been recently evaluated in experimentally co-infected cell cultures for several viruses, direct quantification at the most biologically significant level, that of a host infection, is still lacking. This study fills this gap using the cauliflower mosaic virus as a model. We distributed four neutral markers along the viral genome, and co-inoculated host plants with marker-containing and wild-type viruses. The frequency of recombinant genomes was evaluated 21 d post-inoculation. On average, over 50% of viral genomes recovered after a single host infection were recombinants, clearly indicating that recombination is very frequent in this virus. Estimates of the recombination rate show that all regions of the genome are equally affected by this process. Assuming that ten viral replication cycles occurred during our experiment—based on data on the timing of coat protein detection—the per base and replication cycle recombination rate was on the order of 2 × 10 −5 to 4 × 10 −5 . This first determination of a virus recombination rate during a single multi-cellular host infection indicates that recombination is very frequent in the everyday life of this virus.

Introduction As increasing numbers of full-length viral sequences become available, recombinant or mosaic viruses are being recognized more frequently [ 1 , 2 , 3 ]. Recombination events have been demonstrated to be associated with viruses expanding their host range [ 4 , 5 , 6 , 7 ] or increasing their virulence [ 8 , 9 ], thus accompanying, or perhaps even being at the origin of, major changes during virus adaptation. It remains unclear, however, whether recombination events represent a highly frequent and significant phenomenon in the everyday life of these viruses. Viruses can exchange genetic material when at least two different viral genomes co-infect the same host cell. Progeny can then become hybrid through different mechanisms, such as reassortment of segments when the parental genomes are fragmented [ 10 ], intra-molecular recombination when polymerases switch templates (in RNA viruses) [ 11 ], or homologous or non-homologous recombination (in both RNA and DNA viruses). Quantification of viral recombination in multi-cellular organisms has been attempted under two distinct experimental approaches: in vitro (in cell cultures) [ 12 , 13 , 14 , 15 ], and in vivo (in live hosts) [ 16 , 17 , 18 ]. The in vitro approach, which has so far been applied only to animal viruses, allows the establishment of the “intrinsic” recombination rate in experimentally co-infected cells in cell cultures [ 14 , 15 , 19 ]. However, it does not necessarily reflect the situation in entire, living hosts, where the frequency of co-infected cells is poorly known and depends on many factors such as the size of the pathogen population, the relative frequency and distribution of the different variants, and host defense mechanisms preventing secondary infection of cells. The in vivo experimental approach is closer to biological conditions and may thus be more informative of what actually happens in “the real world.” However, as discussed below, numerous experimental constraints have so far precluded an actual quantification of the baseline rate of recombination. First, many experimental designs have used extreme positive selection, where only recombinant genomes were viable (e.g., [ 13 , 20 , 21 ]). Other studies did not use complementation techniques but detected recombinants by PCR within infected hosts or tissues [ 18 , 22 , 23 , 24 , 25 ], which provides information on their presence but not on their frequency in the viral population. So far, no quantitative PCR or other quantitative method has been applied to evaluate the number of recombinants appearing in an experimentally infected live host. Finally, recent methods based on sequence analysis inferred the population recombination rate, rather than the individual recombination rate [ 1 , 26 , 27 ]. While results from these methods certainly take in vivo recombination into account, there are other caveats: isolates have often been collected in different hosts—sometimes in different geographical regions—and sometimes the selective neutrality of sequence variation on which these estimates are based is not clearly established. Estimates from such studies by essence address the estimation of the recombination rate at a different evolutionary scale. Taken together, the currently available information indicates that no viral recombination rate has ever been estimated directly at time and space scales corresponding to a single multi-cellular host infection, although this level is most significant for the biology and evolution of viruses. This study intends to fill this gap by evaluating the recombination frequency of the cauliflower mosaic virus (CaMV) during a single passage in one of its host plants (the turnip Brassica rapa ). CaMV is a pararetrovirus, which is a major grouping containing hepadnaviruses (e.g., hepatitis B virus), badnaviruses (e.g., banana streak virus), and caulimoviruses (e.g., CaMV). Pararetroviruses are characterized by a non-segmented double-stranded DNA genome. After entering the host cell nucleus, the viral DNA accumulates as a minichromosome [ 28 ] whose transcription is ensured by the host RNA polymerase II [ 29 ]. The CaMV genome consists in approximately 8,000 bp and encodes six viral gene products that have been detected in planta ( Figure 1 ) [ 30 ]. Viral proteins P1 to P6 are expressed from two major transcripts, namely a 19S RNA, encoding P6, and a 35S RNA corresponding to the entire genome and serving as mRNA for proteins P1–P5 [ 31 ]. Using the pre-genomic 35S RNA as a matrix, the protein P5 (product of gene V) reverse-transcribes the genome into genomic DNA that is concomitantly encapsidated [ 30 ]. Figure 1 Genetic Map of CaMV The CaMV genome is a circular double-stranded DNA of 8,024 bp, represented in the figure by a double line. The thick arrows with different textures represent the organization of open reading frames I to VI, encoding proteins detected in planta . Markers a, b, c, and d were engineered at the positions indicated (see Materials and Methods for precise positions). The inner black arrows represent monocistronic 19S RNA and polycistronic 35S RNA produced by the cellular machinery. The nucleotide position 0 (numbering according to [ 44 ]) indicates the origin of replication via reverse transcription, which occurs in the direction indicated by the dotted outermost circle-like arrow. Reverse transcription is accomplished by the viral reverse transcriptase, using the 35S RNA as template [ 49 ]. The detection of CaMV recombinants in turnip hosts has been reported numerous times. Some studies have demonstrated the appearance of infectious recombinant viral genomes after inoculation (i) of a host plant with two infectious or non-infectious parental clones [ 21 , 32 , 33 , 34 , 35 ] or (ii) of a transgenic plant containing one CaMV transgene with a CaMV genome missing the corresponding genomic region [ 36 ]. While the former revealed inter-genomic viral recombination, the latter demonstrated that CaMV can also recombine with transgenes within the host's genome. Another study based on phylogenetic analyses of various CaMV strains has clearly suggested different origins for different genomic regions and, hence, multiple recombination events during the evolution of this virus [ 37 ]. Indirect experimental evidence has indicated that, in some cases, CaMV recombination could occur within the host nucleus, between different viral minichromosomes, presumably through the action of the DNA repair cellular machinery [ 21 , 35 ]. Nevertheless, the mechanism of “template switching” during reverse transcription, predominant in all retroviruses, most certainly also applies to pararetroviruses. For this reason, and on the basis of numerous experimental data, CaMV is generally believed to recombine mostly in the cytoplasm of the host cell, by “legal” template switching between two pre-genomic RNA molecules [ 21 , 35 , 36 , 38 , 39 ], or “illegal” template switching between the 19S and the 35S RNA [ 36 , 40 ]. Under this hypothesis, recombination in CaMV could therefore be considered as operating on a linear template during reverse transcription, with the 5′ and 3′ extremities later ligated to circularize the genomic DNA (position 0 in Figure 1 ). The above cited studies clearly demonstrate that CaMV is able to recombine. However, since these studies are based on complementation techniques, non-quantitative detection, or phylogenetically based inferences of recombination, they do not inform us on whether recombination is an exceptional event or an “everyday” process shaping the genetic composition of CaMV populations. In the present work, we aimed at answering this question. To this end, we have constructed a CaMV genome with four genetic markers, demonstrated to be neutral in competition experiments. By co-inoculating host plants with equal amounts of wild-type and marker-containing CaMV particles, we have generated mixed populations in which impressive proportions of recombinants—distributed in several different classes corresponding to exchange of different genomic regions—have been detected and quantified. Altogether, the recombinant genomes averaged over 50% of the population. Further analysis of these data, assuming a number of viral replications during the infection period ranging from five to 20, indicates that the per nucleotide per replication cycle recombination rate of CaMV is of the same order of magnitude, i.e., on the order of a few 10 −5 , across the entire genome. We thereby provide the first quantification, to our knowledge, of the recombination rate in a virus population during a single passage in a single host. Results Recombinant Frequency in CaMV Populations from Co-Infected Plants From Figure 1 , and supposing that all marker-containing genomic regions can recombine, we could predict the detection and quantification of seven classes of recombinant genotypes: +bcd/a+++, a+cd/+b++, ab+d/++c+, abc+/+++d, ++cd/ab++, a++d/+bc+, and a+c+/+b+d. Indeed, all classes were detected, and their frequencies in the ten CaMV populations analyzed are summarized in Table 1 . Table 1 Quantification of Recombinant Genomes in CaMV Populations a Viral genomes were cloned and analyzed from ten of 24 co-infected plants b Seven possible classes of recombinants were predicted, their respective frequency in the population is expressed in percentage c The proportions of recombinants from the seven classes were added to estimate the total percentage of recombinant genomes within each tested plant Altogether, the proportion of recombinant genomes found in the mixed viral populations was astonishingly high and very similar in the ten co-infected plants analyzed ( Table 1 , last column), ranging between 44% (plant 5) to 60% (plants 7, 12, and 20), with a mean frequency (± standard error) of 53.8% ± 2.0%. This result indicates that recombination events are very frequent during the invasion of the host plant by CaMV and represents, to our knowledge, the first direct quantification of viral recombination during the infection of a live multi-cellular host. Probability of Recombination between Various Pairs of Markers The inferred per generation recombination and interference rates, assuming that CaMV undergoes ten replication cycles during the 21 d between infection and sampling, are given for each of the ten plants in Table 2 . Recombination rates between adjacent markers are large, on the order of 0.05 to 0.1. Taking the distance in nucleotides between markers into account yields an average recombination rate per nucleotide and generation on the order of 4 × 10 −5 . Interestingly, this recombination rate does not vary throughout the genome (Kruskal–Wallis test, p = 0.16). Table 2 Recombination Parameters for the Viral Populations Sampled from Ten Infected Plants The various parameters are as follows: r 1 , recombination rate between markers a and b; r 2 , recombination rate between markers b and c; r 3 , recombination rate between markers c and d; i 12 , interference between crossovers in segments a–b and b–c; i 23 , interference between crossovers in segments b–c and c–d; i 13 , interference between crossovers in segments a–b and c–d; i 123 , second-order interference accounting for residual interference. The recombination rates are the maximum likelihood estimates (± 95% confidence intervals). The interference parameters were obtained numerically as explained in the Materials and Methods To relax the assumption of the number of replications during the 21 d, we calculated the recombination parameters assuming five or 20 generations. The effect of the number of generations on the estimates is linear: doubling the number of generations results in a halving of the recombination rate (detailed results not shown). For example, the average recombination rates r 1 , r 2 , and r 3 assuming 20 generations were equal to 0.05, 0.04, and 0.025, respectively (compare with values in Table 2 ), yielding per nucleotide per generation recombination rates of 1.9 × 10 −5 , 2.2 × 10 −5 and 1.6 × 10 −5 . Inspection of Table 2 also shows that first-order interference coefficients were in general negative, indicating that a crossing over in one genomic segment increases the probability that a crossing over will occur in another genomic segment, while the second-order coefficient parameter had an average value close to zero with a large variance. The mechanism leading to these results will be discussed in the following section. Discussion One major breakthrough in the work presented here lies in the space and time scales at which the experiments were performed. Indeed, the processes occurring within the course of a single infection of one multi-cellular host are of obvious biological relevance for any disease. Previous studies on viral recombination suffered from major drawbacks in this respect, basing their conclusions on experiments relying on complementation among non-infectious viruses or between viruses with undetermined relative fitness, on phylogenetically based analyses, or on experiments in cell cultures. For reasons detailed in the Introduction, the first two methods either do not provide information on the frequency of recombination, but only its occurrence, or address the question at a different temporal, and often spatial, scale. Results from cell cultures, on the other hand, impose cell co-infection by different viral variants, potentially overestimating the frequency of recombination events. Our study circumvents these limitations by analyzing viral genotypes sampled from infected plants after the course of a single infection, and therefore the invasion and co-infection of cells in various organs and tissues is very close to natural. More than half of the genomes (53.8% ± 2.0%; see Table 1 ) present in a CaMV population after a single passage in its host plant were identified as recombinants, and these data allowed us to infer a per nucleotide per generation recombination rate on the order of 2 × 10 −5 to 4 × 10 −5 . The time length of one generation, i.e., the time required for a given genome to go from one replication to the next, is totally unknown in plant viruses. The only experimental data available on CaMV are based on the kinetics of gene expression in infected protoplasts, where the capsid protein is produced between 48 and 72 h [ 40 ]. The reverse transcription and the encapsidation of genomic DNA being two coupled phenomena [ 30 ], we judged it reasonable to assume a generation time of 2 d and, thus, an average of ten generations during our experiments. In case this estimate is mistaken, we have verified a linear relationship between r and the number of generations, thereby allowing an immediate adjustment of r if the CaMV generation time is more precisely established. At this point, we must consider that all cloned genomes may not have been through all the successive replication events potentially allowed by the timing of our experiments. It was previously shown that about 95% of CaMV mature virus particles accumulate in compact inclusion bodies [ 41 ], where they may be sequestered for a long time, as such inclusions are very frequent in all infected cells, including those in leaves that have been invaded by the virus population for several weeks. The viral population may thus present an age structure that could bias the estimation of the recombination rate. In order to minimize this bias, the clones we analyzed were collected in one young newly formed leaf, where the chances of finding genomes from “unsequestered lines” were assumed to be higher. In any case, our data analysis is conservative, since this age structure can only lead to an underestimation of the recombination rate. Our results show that interferences between pairs of loci are negative: a recombination event between two loci apparently increases the probability of recombination between another pair of loci. We believe that the most parsimonious explanation of these negative interferences is based on the way the infection builds up within plant hosts. Indeed, one can divide infected host cells into those infected by a single virus genotype and those infected by more than one viral genotype. In the former, analogous to clonal propagation, recombination is undetectable. In the latter, recombination is not only detectable but, as our results indicate, very frequent. Samples consisting of viruses resulting from a mixture of these two types of host cell infections will thus contain viruses with no recombination and viruses with several recombination events, thus yielding an impression of negative interference. These conceptual arguments are supported by mathematical models. It is indeed easy to show (detailed results not shown) that if a proportion F of the population reproduces clonally, analogous to single infections, while the remaining reproduces panmictically, negative interferences could be inferred even if they do not exist. For example, assuming a three-locus model with real recombination rates r 1 and r 2 and interference i 12 , the “apparent” recombination and interference parameters, would be r 1 = (1 − F )r 1 , r 2 = (1 − F ) r 2 , and i 12 = −( F − i 12 )/(1 − F ). Interestingly, this example also shows that our estimates of the recombination rate are conservative: that a fraction F of host cells are singly infected while others are multiply infected leads to an underestimation of the recombination rate. As judged by r 1 , r 2 , and r 3 , calculated between markers a–b, b–c, and c–d, respectively, we found evidence for recombination through the entire CaMV genome. The values for r 1, r 2 , and r 3 are remarkably similar, hence the recombination sites seem to be evenly distributed along the genome. We considered the template-switching model as the major way recombinants are created in CaMV. As already mentioned in the Introduction, hot spots of template switching have been predicted at the position of the 5′ extremities of the 35S and 19S RNAs [ 21 , 36 , 42 ]. If other recombination mechanisms, such as that associated with second-strand DNA synthesis or with the host cell DNA repair machinery, act significantly, hot spots would be expected at the positions of the sequence interruption Δ1, Δ2, and Δ3 [ 43 ]. Due to the design of our experiment and the position of the four markers, we have no information on putative hot spots at positions corresponding to the 5′ end of the 35S RNA and to Δ1 (at nucleotide position 0). Nevertheless, the putative hot spots at the 5′ end of the 19S RNA and at Δ2 and Δ3 (nucleotide positions 4,220 and 1,635, respectively) fall between marker pairs c–d, b–c, and a–b, respectively. Our results indicate that either these hot spots are quantitatively equivalent—though predicted by different recombination mechanisms—or, more likely, that they simply do not exist. Whatever the explanation, what we observe is that the CaMV can exchange any portion of its genome, and thus any gene thereof, with an astonishingly high frequency during the course of a single host infection. To our knowledge, the viral recombination rate has never previously been quantified experimentally for a plant virus [ 3 ]. In contrast, retroviruses and particularly HIV-1 have been extensively investigated in that sense. As we have already discussed for these latter cases, the quantification of the intrinsic recombination rate was carried out in artificially co-infected cell cultures. The estimated intrinsic per nucleotide per generation recombination rate in HIV-1 is on the order of 10 −4 [ 14 , 15 , 19 ], less than one order of magnitude higher than our estimation for CaMV. Because for various reasons detailed above we probably underestimate the within-host CaMV recombination rate, we believe that the intrinsic recombination rate in CaMV is higher and perhaps on the order of that of HIV. Other pararetroviruses such as plant badnaviruses or vertebrate hepadnaviruses have a similar cycle within their host cells, including steps of nuclear minichromosome, genomic size RNA synthesis, and reverse transcription and encapsidation. Nevertheless, vertebrate hepadnaviruses (e.g., hepatitis B virus) infect hosts that are very different from plants in their biology and physiology, and this could lead to a totally different frequency of cell co-infection during the development of the virus populations. Thus, even though our results can be informative for other pararetroviruses because of the viruses' shared biological characteristics, they should not be extrapolated to vertebrate pararetroviruses without caution. Materials and Methods Viral isolates We used the plasmid pCa37, which is the complete genome of the CaMV isolate Cabb-S, cloned into the pBR322 plasmid at the unique SalI restriction site [ 44 ]. To analyze recombination in different regions of the genome, we introduced four genetic markers: a, b, c, and d, at the positions 881, 3,539, 5,365, and 6,943, respectively, thus approximately at four cardinal points of the CaMV circular double-stranded DNA of 8,024 bp ( Figure 1 ). All markers, each corresponding to a single nucleotide change, were introduced by PCR-directed mutagenesis in pCa37, and resulted in the duplication of previously unique restriction sites BsiWI, PstI, MluI, and SacI in a plasmid designated pMark-S. Because, in this study, we targeted the possible exchange of genes between viral genomes, all markers a, b, c, and d were introduced within coding regions corresponding to open reading frames I, IV, V, and VI, respectively. Another important concern was to quantify recombination in the absence of selection, i.e., to create neutral markers. Consequently all markers consist of synonymous mutations (see below). Production of viral particles and co-inoculation To generate the parental virus particles, plasmids pCa37 and pMark-S were mechanically inoculated into individual plants as previously described [ 33 ]. All plants were turnips ( B. rapa cv, “Just Right”) grown under glasshouse conditions at 23 °C with a 16/8 (light/dark) photoperiod. Thirty days post-inoculation, all symptomatic leaves were harvested and viral particles were purified as described earlier [ 45 ]. The resulting preparations of parental viruses, designated Cabb-S and Mark-S, were quantified by spectrometry using the formula described by Hull et al. [ 46 ]. We fixed the initial frequency of markers to a value of 0.5, and a solution containing 0.1 mg/ml of virus particles of both Cabb-S and Mark-S at a 1:1 ratio was prepared. Plantlets were co-infected by mechanical inoculation of two to three leaves with 20 μl of this virus solution, using abrasive Celite AFA (Fluka, Ronkonkoma, New York, United States). The mixed CaMV population was allowed to grow during 21 d of systemic infection. Estimation of marker frequency within mixed virus populations We designed an experimental protocol for quantifying marker frequency within a mixed Cabb-S/Mark-S virus population after a single passage in a host plant. Twenty-four individual plants, inoculated as above with equal amounts of Cabb-S and Mark-S, were harvested 21 d post-inoculation, when symptoms were fully developed. The viral DNA was purified from 200 mg of young newly formed infected leaves according to the protocol described previously [ 47 ]. After the precipitation step of this protocol, the viral DNA was resuspended and further purified with the Wizard DNA clean-up kit (Promega, Fitchburg, Wisconsin, United States) in TE 1X (100 mM Tris-HCl and 10 mM EDTA [pH 8]). Aliquots of viral DNA preparations were digested by restriction enzymes corresponding either to marker a, b, c, or d and submitted to a 1% agarose gel electrophoresis, colored by ethydium bromide and exposed to UV. Each individual restriction enzyme cut once in Cabb-S DNA and twice in Mark-S, thus generating DNA fragments of different sizes attributable to one or the other in the mixed population of CaMV genomes. After scanning the agarose gels, we estimated the relative frequency of the two genotypes in each viral DNA preparation and at each marker position, by densitometry using the NIH 1.62 Image program. The statistical analyses of the frequency of the four markers are described below. Isolation of individual CaMV genomes and identification of recombinants To identify and quantify the recombinants within the CaMV mixed populations, aliquots from ten of the 24 viral DNA preparations described above were digested by the restriction enzyme SalI, and directly cloned into pUC19 at the corresponding site. In each of the ten viral populations analyzed, 50 full-genome-length clones were digested separately by BsiWI, PstI, MluI, and SacI, to test for the presence of marker a, b, c, and d, respectively. In this experiment, with the marker representing an additional restriction site, we could easily distinguish between the Cabb-S and the Mark-S genotype at all four marker positions, upon agarose gel (1%) electrophoresis of the digested clones. Clones with none or all four markers were parental genotypes, whereas clones harboring 1, 2, or 3 markers were clearly recombinants. Due to the very high number of recombinants detected, markers eventually appearing or disappearing due to spontaneous mutations were neglected. Statistical analysis Here we present the different methods we used to quantify recombination in the CaMV genome. Because all these methods assume that the different markers are neutral, we first discuss assumption. We used two datasets to test the neutrality of markers, both resulting from plants co-infected with a 1:1 ratio of Mark-S and Cabb-S. The first consisted of viral DNA densitometry data derived from 24 plants (described above), where for each plant we have an estimate of the frequency of each marker in the genome population. The second consisted of the restriction of 50 individual full-genome-length viral clones obtained from one co-infected plant (described above), yielding an estimate of the frequency of each marker, and this was repeated on ten different plants. The frequencies of the different markers were 0.508, 0.501, 0.516, and 0.507 for markers a, b, c, and d in the first dataset and 0.521, 0.518, 0.514, and 0.524 in the second dataset. We tested whether these frequencies were significantly different from the expected value under neutrality, 0.5, using either t -tests, for datasets where normality could not be rejected (seven out of eight cases), or Wilcoxon signed-rank non-parametric tests otherwise (marker c in the first dataset). In all cases p -values were larger than 0.05. There are several cautionary remarks regarding these analyses. First, in all cases we found an excess of markers. Unfortunately, the two datasets cannot be regarded as independent because, even though the methods through which the frequency estimates were obtained were different, the plants used in the second dataset were a subset of the plants of the first. We thus have only four independent estimates in each case, and there is minimal power to detect significant deviations from neutrality with such a small sample size. It should be noted at this stage that deviations from the expected value could also be caused either by slight deviations from the 1:1 ratio in the infecting mixed solution, or by deviations from that ratio in the frequency of the viral particles that actually get into the plants. Second, because of the relatively small sample sizes and low statistical power, the tests presented above could have detected only large deviations. The results clearly show, however, precisely that the markers do not have large effects, if any, and that therefore recombination estimates would be affected only very slightly by any hypothetical selective effects of the markers. Because of this, along with the fact that the introduced markers provoke silent substitutions in the CaMV genome, we assumed that markers were effectively neutral in the rest of the analysis. The dataset used to estimate the recombination frequency consisted of the 500 full-genome-length viral clones (50 from each of ten co-infected plants) individually genotyped for each of the four markers. As discussed in detail in the Results, recombination was very frequent and concerned all four markers. Indeed, approximately half of the genotyped clones exhibited a recombinant genotype. It was therefore meaningful to try to obtain quantitative estimates of recombination from our data. Our aim was to analyze viral recombination in a live host. Consequently, we had to deal with the fact that more than one viral replication cycle occurred during the 21 d that infection lasted in our experiment (we had to wait that long for the disease to develop and to be able to recover sufficient amounts of viral DNA from each infection). Based on the kinetics of gene expression [ 40 ], we postulate that each replication cycle lasts between 2 and 3 d, and that therefore seven to ten cycles occurred between infection and the sampling time. In case this assumption is incorrect, we did calculations assuming five, seven, ten, or 20 replication cycles during these 21 d. As shown, the results were not affected qualitatively, and only slightly quantitatively. It is important to note that we assumed that recombination occurred through a template-switching mechanism, and that therefore, from a recombination point of view, the CaMV genome is linear. The reverse transcription starts and finishes at the position 0 in Figure 1 , which is the point of circularization of the DNA genome. This implies that changes between contiguous markers a–b, b–c, and c–d can be considered as true recombination whereas those between a and d cannot, as they may simply stem from circularization of DNA, during the synthesis of which the polymerase has switched template once anywhere between a–b, b–c, or c–d. To estimate the recombination rate between markers, we wrote recurrence equations describing the change in frequency of each genotype over one generation, assuming random mating and no selection (i.e., the standard Wright–Fisher population genetics model). We then expressed the frequency of all possible genotypes n generations later as a function of their initial frequency and of the recombination parameters. Subsequently we calculated the maximum likelihood estimates of the recombination parameters and their asymptotic variances given initial frequencies (we assumed that the two “parental” genotypes, Mark-S and Cabb-S, had equal initial frequencies of 0.5 and that all other genotypes had initial frequencies of zero) and frequencies after n generations (the observed frequencies; as stated above we used different values of n ). All algebraic and numerical calculations were carried out with the software Mathematica. The recombination parameters are the recombination rates between two adjacent loci, e.g., r 1 for the recombination rate between markers a and b, and the interference coefficients, e.g., i 12 for interference between recombination events in the segments between markers a and b and b and c. To define these parameters we followed Christiansen [ 48 ], and in particular the recombination distributions for two, three, and four loci (respectively, Tables 2 .7, 2.8, and 2.9 of [ 48 ]). It is important to realize that given the definitions of these parameters, the estimator of the recombination rate between two loci is not affected by the number of loci considered. In other words, we obtain the same estimation of the recombination rate between markers a and b whether we consider genotypic frequencies at just these two loci, or the frequencies at these two loci plus a third locus, or the complete information to which we have access, the four-marker genotypes. Information on additional loci only affects the estimates of the interference coefficients. It proved impossible to carry out the calculations for four loci algebraically. Instead, we used a computer program to calculate the expected genotypic frequencies at all four loci after n generations, given the above stated initial frequencies and specified recombination parameters. For each combination of recombination parameters we calculated a Euclidean distance between the vector of the expected genotypic frequencies and the observed genotypic frequencies, and considered that the estimated recombination parameters were those yielding the minimal Euclidean distance. In all cases, the estimated recombination rates between pairs of loci were equal to the second decimal to those estimated algebraically from data for three or two loci.

D:\keerthana\PMC001xxxxxx\PMC1054884.xml

1054885

Separating Sisters: Shugoshin Protects SA2 at Centromeres but Not at Chromosome Arms

DNA replication leaves the cell with two identical copies of each chromosome. To ensure their proper segregation during the anaphase stage of mitosis, the members of each pair, called sister chromatids, are held together by a protein complex, aptly named cohesin, that links the two not only at the centromere, but also along the chromatid arms. Anaphase is triggered when cohesin is cleaved, by the equally well-named separase. But cleavage is not the only way to remove cohesin from the chromosome; indeed, in humans and other higher eukaryotes, mitotic kinases such as Plk1 remove the majority of cohesin from chromosome arms—but not from the centromere—during prophase and prometaphase. These facts raise two questions: what is the precise target of Plk1, and what protects centromeric cohesin from removal by the same pathway? Both questions are addressed in a new article in PLoS Biology . Jan-Michael Peters and colleagues show that phosphorylation of the cohesin subunit SA2, presumably by Plk1, is required for cohesin removal from chromosome arms in early mitosis, while data from Kim Nasmyth and colleagues suggest that a protein called shugoshin protects centromeric SA2 from such phosphorylation. Premature loss of sister chromatid cohesion Cohesin is composed of multiple subunits, each of which can be phosphorylated at multiple threonine or serine amino acid residues. These subunits include Scc1 (the target of separase), Smc1, and Smc3, plus Scc3 in yeast, and SA1 or SA2 in humans and other higher eukaryotes. By isolating and analyzing cohesin subunits from cells undergoing mitosis, Peters and colleagues deduced that Scc1, SA1, and SA2 are phosphorylated only during mitosis, suggesting that phosphorylation of one or more of them triggers the breakup of cohesin. Further analysis by mass spectrometry allowed them to identify the exact amino acids that bore the phosphates on each subunit. In Scc1, these were clustered around the known sites of separase cleavage. The researchers showed that phosphorylation at these sites is required for efficient cleavage by the enzyme during anaphase, but is not required to dislodge cohesin specifically from the chromosome arms, as this proceeded essentially normally even after these sites were mutated to prevent their phosphorylation. The same mutation strategy applied to SA2, on the other hand, revealed that phosphorylation of this subunit is essential for dissociating cohesin from the chromosome arms during prometaphase. Interestingly, the mutations did not prevent the ultimate separation of the chromatids at anaphase. This suggests that separase, once it is activated, can cleave cohesin on the arms as well as at the centromere. Cohesin at the centromere is removed later in mitosis than cohesin bound to chromatid arms, namely, at the metaphase-to-anaphase transition, suggesting centromeric cohesin is protected by a centromere-specific molecule. Possible candidates would be members of the shugoshin family, which are known to prevent unloading of centromeric cohesin during the first division of meiosis, thus keeping chromatids together as homologous chromosomes are separated. To investigate human shugoshin's mitotic role, Nasmyth and colleagues depleted shugoshin by RNAi. The result was loss of cohesin not only from the arms but also from the centromere, early separation of chromatids, and failure of anaphase, suggesting that shugoshin protects centromeric cohesin. But how? To find out, the authors examined the effect of shugoshin depletion in cells whose SA2 had been mutated to prevent phosphorylation. Strikingly, these cells underwent mitosis successfully. Together, these results suggest that shugoshin's normal mitotic role is to protect centromeric SA2 from phosphorylation, delaying chromatid separation until the moment when the chromosomes are ready to separate, at which time cohesin is cleaved by separase. The picture that emerges from these two studies is that sister chromatid cohesion is safeguarded throughout early mitosis by shugoshin, which protects centromeric cohesin from the threat of protein kinases that, in the authors' vivid language, “maraud mitotic chromosomes and threaten to destroy their integrity.” This delicate balance of power between kinases and shugoshin means that any upset in the balance may prevent a cell from dividing properly, which often means not dividing at all. (Also see the Primer “Chromosome Cohesion: A Cycle of Holding Together and Falling Apart” [DOI: 10.1371/journal.pbio.0030094 ].)

D:\keerthana\PMC001xxxxxx\PMC1054885.xml

1054886

Location, Location, Location: How Position Affects Gene Expression in the Nucleus

Control of gene expression plays a role in determining cell fate, differentiation, and the maintenance of specific cell lineages. In the absence of regulation, aberrant gene expression can lead to developmental defects and disease. As a result, gene expression is highly regulated and that regulation takes many forms. Control mechanisms may be specific to one gene or operate on a gross chromosomal level, ultimately ensuring that genes are expressed at the right time, in the right place. It is only in the run-up to and during cell division that chromosomes take on the condensed form that enables them to be recognized as discreet structures. During the rest of the cell cycle, interphase chromosomes exist in a relaxed state that at first glance looks like an unraveled ball of wool floating randomly about the nucleus. But a closer look reveals that they are in fact non-randomly organized and compartmentalized, and these groupings have functional ramifications for how genes are expressed or silenced (repressed). Assessing gene expression and gene location in single cells What is actually visible is chromatin, a combination of naked DNA and proteins that associate with it. It can exist in two forms: euchromatin and heterochromatin. Actively expressed chromosomal regions (loci) are predominantly located within euchromatin, while loci within heterochromatic regions are silenced. Genetic and cytological evidence indicates that interaction between euchromatic genes and heterochromatin can cause gene silencing. Getting a gene into position for such an interaction may be achieved in two ways. The first is by changing the gene's position on the chromosome to bring it very close to expanses of centromeric heterochromatin, thereby increasing the likelihood for interaction. The second is by changing the position of a section of heterochromatin to place it close to a euchromatic gene. The small regions of heterochromatin involved in this second process seem sufficient to mediate long-range interactions between the affected gene and the larger heterochromatic regions near the centromere, but not so large or powerful as to mediate silencing by themselves. In this issue, Brian Harmon and John Sedat study the functional consequences of long-range chromosomal interactions—consequences that have been inferred in several different organisms but until now have not been analyzed on a cell-by-cell basis or directly verified. Several Drosophila fruitfly mutants have been identified that exhibit cells in the same organ with varied phenotypes (appearance), though their genotypes (DNA instructions) are the same. This occurs through a phenomenon known as position-effect variegation, in which the expression of variegating genes is determined by their position on the chromosome relative to regions of heterochromatin. Working with fruitflies, the authors labeled three variegating genes and areas of heterochromatin with fluorescent probes and visualized expression of the affected genes in tissues where they are normally expressed. Silenced genes, they discovered, are far closer to heterochromatin than expressed genes, indicating that silenced genes interact with heterochromatin while expressed genes do not. This study of interactions between a gene and heterochromatin in single cells illustrates unequivocally a direct association between long-range chromosomal interactions and gene silencing. The novel cell-by-cell analysis paves the way for further analysis of this phenomenon and will lead to a greater insight into the understanding and functional significance of nuclear architecture.

D:\keerthana\PMC001xxxxxx\PMC1054886.xml

1054887

Lineage-Specific Expansions of Retroviral Insertions within the Genomes of African Great Apes but Not Humans and Orangutans

Retroviral infections of the germline have the potential to episodically alter gene function and genome structure during the course of evolution. Horizontal transmissions between species have been proposed, but little evidence exists for such events in the human/great ape lineage of evolution. Based on analysis of finished BAC chimpanzee genome sequence, we characterize a retroviral element ( Pan troglodytes endogenous retrovirus 1 [PTERV1]) that has become integrated in the germline of African great ape and Old World monkey species but is absent from humans and Asian ape genomes. We unambiguously map 287 retroviral integration sites and determine that approximately 95.8% of the insertions occur at non-orthologous regions between closely related species. Phylogenetic analysis of the endogenous retrovirus reveals that the gorilla and chimpanzee elements share a monophyletic origin with a subset of the Old World monkey retroviral elements, but that the average sequence divergence exceeds neutral expectation for a strictly nuclear inherited DNA molecule. Within the chimpanzee, there is a significant integration bias against genes, with only 14 of these insertions mapping within intronic regions. Six out of ten of these genes, for which there are expression data, show significant differences in transcript expression between human and chimpanzee. Our data are consistent with a retroviral infection that bombarded the genomes of chimpanzees and gorillas independently and concurrently, 3–4 million years ago. We speculate on the potential impact of such recent events on the evolution of humans and great apes.

Introduction Mammalian genomic sequence is littered with various classes of endogenous retroviruses that have populated genomes during the course of evolution [ 1 , 2 ]. In the case of humans, approximately 8.3% of the genome sequence consists of long terminal repeat (LTR) and endogenous retrovirus elements classified into more than 100 separate repeat families and subfamilies [ 3 , 4 ]. The bulk of human endogenous retrovirus elements are thought to have originated as a result of exogenous retrovirus integration events that occurred early during primate evolution. Based on comparative analyses of orthologous genomic sequence and sequence divergence of flanking LTR elements, the last major genomic infection of the human lineage is estimated to have occurred before the divergence of the Old World and New World monkey lineages (25–35 million years ago) [ 5 , 6 , 7 , 8 ]. Since the divergence of chimpanzee and human (5–7 million years ago), only one major family of human endogenous retroviruses (HERVK10) has remained active, and it has generated only three full-length copies with the open reading frame still intact [ 3 ]. While new insertions of endogenous retroviral sequences have been described [ 8 , 9 ], most of these are thought to have originated from other previously integrated retroelements [ 10 ] or longstanding associations with rare source virus [ 11 ]. This apparent wane in activity has led to the view that LTR retroposons have had a history of declining activity in the human lineage and are “teetering on the brink of extinction” [ 3 ]. Endogenous retroviruses may arise within genomes by at least two different mechanisms: retrotransposition from a pre-existing endogenous retrovirus (intraspecific transmission) or infection and integration via an exogenous source virus (horizontal transmission). Many cross-species transmissions have been documented and frequently manifest themselves as inconsistencies in the presumed phylogeny of closely related species. During the 1970s and 1980s, Benveniste and colleagues identified, by DNA hybridization and immunological cross-reactivity, several retroviral elements that could be found among more diverse primate/mammalian species but not necessarily among more closely related sister taxa [ 12 , 13 , 14 ]. Lieber and colleagues, for example, reported the isolation of a particular class of type C retroviruses from a woolly monkey (SSV-SSAV) and gibbon ape (GALV) but not the African great apes [ 13 ]. These viruses shared antigenic properties with previously described type C activated endogenous retroviruses of the Asian feral mouse Mus caroli. Cross-species infection from murines to primates was proposed as the likely origin of the retrovirus. A related endogenous retrovirus was subsequently identified in the koala, suggesting a zoonotic transmission from placentals to mammals [ 15 ]. Evidence of horizontal transmission for other families of retrovirus has been reported among classes of species as distantly related as avians and mammals [ 15 ]. Comparative analyses of closely related genomes have suggested that retroviral cross-species transmissions and genome integrations are a common occurrence during the recent evolutionary history of several species. Murine genomes, in particular, have been bombarded with relatively recent retroviral integrations [ 16 ]. In contrast to humans, there is ample evidence that exogenous retrovirus continues to bombard and fix within the genomes of Old World monkey species. Cross-species transmissions and genome integration of retroviruses as recent as 500,000 years ago have been reported between various simian species [ 17 , 18 ]. Differences in the distribution of endogenous retroviruses have even been noted between feral and domesticated mammalian species. The genomes of domestic cats, for example, harbor specific families of endogenous feline leukemia viruses that are not found in the genomes of wild cats [ 19 ]. Similarly, the PERV-C (porcine endogenous retrovirus type C) is restricted to domesticated pigs and has not been identified in the genomes of the wild boar from which domestication is thought to have occurred approximately 5,000 years ago [ 20 ]. From a functional perspective, the integration of retroviral sequence may have considerable impact. Endogenous retroviruses harbor cryptic mRNA splice sites, polyadenylation signals, and promoter and enhancer sequences. As such, their integration into the genome may significantly alter the expression patterns of nearby genes. Moreover, integrated retroviruses are often preferential sites of methylation and may promote rearrangement of DNA by way of non-allelic homologous recombination between elements. Consequently, these elements have been recognized as potent mutagens [ 2 , 21 ] that may significantly alter the phenotype [ 22 , 23 ]. The mechanism by which such elements originate and differentially spread among closely related species is, therefore, fundamental to our understanding of evolution. Results Distribution of Pan troglodytes Endogenous Retrovirus 1 among Primates During a comparison of human and chimpanzee BAC sequence, we identified several members of a full-length endogenous retrovirus family that were present in chimpanzee but absent in corresponding human genome sequence ( Figure 1 ). Analysis of five full-length insertion sequences revealed that the endogenous retroviral elements (termed Pan troglodytes endogenous retrovirus 1 [PTERV1]) ranged in size from 5 to 8.8 kb in length ( Materials and Methods ). Translation of the sequence showed strong protein similarity to gammaretroviruses (53%–69%), in particular, murine leukemia virus, feline leukemia virus, porcine endogenous retrovirus type C, and baboon (Papio cynocephalus) endogenous retrovirus ( Figure 1 ). Large deletions (1–2 kb) of the reverse transcriptase in some copies as well as the presence of multiple stop codons in all examined full-length copies indicate that this particular family of endogenous retrovirus is not replication competent. Figure 1 Identification and Sequence Analysis of PTERV1 (A) A graphical alignment of chimpanzee genomic sequence (AC097267) and an orthologous segment from human Chromosome 16 (Build 34) depicting an example of a PTERV1 (approximately 10 kb) insertion. Aligned sequences are shown in blue (miropeats) [ 47 ]. (B) The typical retroviral structure of the insert ( gag, pol, env, and LTR) is compared to a baboon (Papio cynocephalus) endogenous retrovirus (PcEV). Regions of nucleotide homology are designated by black blocks and inter-sequence connecting lines. The location of probes (see Table S1 ) used in genomic library hybridizations, Southern blot analyses, and neighbor-joining tree analyses are shown (red). We designed two probes to the gag and env portions of PTERV1 ( Figure 1 ; Table S1 ) and assessed the distribution of PTERV1 among apes and Old World monkeys by Southern analysis. More than 100 copies of the endogenous retrovirus were detected in each African ape and Old World monkey species ( Figures 2 and Figure S1 ). Comparison between DNA digested with methylation-sensitive and -resistant restriction enzymes indicated that most copies were extensively methylated in these species ( Figure S2 ). In contrast, analysis of multiple Asian apes (siamang, gibbon, and orangutan) and a panel of human DNAs showed no hybridization signal. These findings were consistent with early DNA hybrid melting experiments [ 12 ] and DNA hybrid electron microscopic studies [ 14 ] that indicated that DNA from the African great apes harbored sequences homologous to both colobus monkey and baboon exogenous retroviruses while the genomes of man and Asian apes did not. These data were sometimes used as supporting evidence for an Asian origin of modern humans [ 12 ]. Figure 2 Southern Hybridization of PTERV1 among Primates Species represented include human (HSA), common chimpanzee (PTR), bonobo (PPA), gorilla (GGO), orangutan (PPY), siamang (HSY), white-handed gibbon (HLA), Abyssinian black-and-white colobus monkey (CGU), olive baboon (PHA), rhesus macaque (MMU), and Japanese macaque (MFU). Below each panel, a restriction map (chimpanzee sequence AC097267) is presented in relation to the hybridization probe: PstI (closed circles), PvuII (open circles), and HpaII/MspI (triangles) (see Figures S1 and S2 for additional details). (A) The absence of PTERV1 among Asian apes and humans is shown in contrast to a generally accepted catarrhine species phylogeny. Primate DNAs have been digested with PstI restriction enzyme, Southern-transferred to nylon membrane, and hybridized with PTERV1 gag probe number 1. (B) Multiple African great ape species are compared for both the gag probe number 1 and env probe number 3 ( Figure 1 ). Proximity of probe number 1 to the VNTR, which is variable in length between copies (400 bp to 10 kb), reveals hundreds of insertion sites. (C) Multiple individuals from different subspecies of the olive baboon are compared for both gag probe number 1 and env probe number 3. The pattern of Southern hybridization shows limited intra-specific variation, indicative of either polymorphism in restriction enzyme sites or copy number variation. In order to further resolve the evolutionary relationship of these endogenous retroviruses, we compared the sites of retroviral integration in the genomes of chimpanzee, gorilla, macaque, and baboon. To this end, we screened large-insert genomic (BAC) libraries for each species using multiple probes from the PTERV1 reference sequence ( Table S1 ). These data allowed us to estimate copy number in each species and to distinguish clones harboring full-length retroviral inserts versus solo LTR elements ( Table 1 ). In addition, we used BAC end sequences from nonhuman primate clones harboring full-length retroviruses to map their locations back to the human genome. We then compared the locations ( Figure 3 ; Table 2 ) between species to determine whether the sites were non-orthologous. Based on an analysis of 1,467 large-insert clones, we mapped 299 retroviral insertion sites among the four species ( Figure 3 ; Table S2 ). A total of 275 of the insertion sites mapped unambiguously to non-orthologous locations ( Table 2 ), indicating that the vast majority of elements were lineage-specific (i.e., they emerged after the divergence of gorilla/chimpanzee and macaque/baboon from their common ancestor). Figure 3 PTERV1 Insertion Sites Large-insert genomic clones that contained full-length endogenous retrovirus were identified by hybridization from four species: chimpanzee (PTR), gorilla (GGO), baboon (PAN), and rhesus macaque (MMU). End sequencing of large-insert clones ( n = 1,467) and alignment against the human genome reference sequence identified 287 insertion sites (see Table S2 ). A total of 95.8% of these sites were non-orthologous when compared between species. chr, chromosome. Table 1 PTERV1 BAC Library Hybridizations: Number of BACs (Estimated Copy Number) a Large-insert BAC libraries were hybridized with radioactive probes corresponding to the gag (probe number 1), env (probe number 3), and LTR (probe number 2) portions of PTERV1. The number of strongly hybridizing positive BACs was used to estimate the copy number based on the depth of coverage of each library. BACs that hybridized with both the gag and env portions were considered to harbor full-length copies of the retrovirus, while BACs that hybridized with the LTR probe but not with gag + env portions were considered to represent solo LTR copies of PTERV1 b The number of hybridization-positive BACs for each probe is shown. The estimated copy number of PTERV1 in different species is indicated in parentheses Table 2 Cross-Species Retroviral Insertion Mapping A total of 1,467 BACs that hybridized with PTERV1 gag and env probes were end-sequenced and mapped against the human genome assembly by quality sequence alignment (see Table S2). Insertion sites were refined based on the placement of two or more BACs from a species to the same location. The number of mapped loci and the number of non-overlapping locations with respect to the other species are indicated. A total of 275/287 (275 + 12) = 95.8% of the locations were non-orthologous. Twenty-four locations were ambiguous within the limits of resolution of this study and could, in theory, correspond to 12 orthologous sites (see Table S3) Within the limits of this BAC-based end-sequencing mapping approach, 24 sites mapped to similar regions of the human reference genome (approximately 160 kb) and could not be definitively resolved as orthologous or non-orthologous ( Table S3 ). We classified these as “ambiguous” overlap loci ( Figure 3 ). If all 24 locations corresponded to insertions that were orthologous for each pair, this would correspond to a maximum of 12 orthologous loci. The number of non-orthologous loci was calculated as 275/287 (275 + 12) or 95.8%. This is almost certainly a lower-bound estimate owing to the limitation of our BAC-based mapping approach to refine the precise locations of the insertions. We performed two analyses to determine whether these 12 shared map intervals might indeed be orthologous. First, we examined the distribution of shared sites between species ( Table S3 ). We found that the distribution is inconsistent with the generally accepted phylogeny of catarrhine primates [ 5 ]. This is particularly relevant for the human/great ape lineage. For example, only one interval is shared by gorilla and chimpanzee; however, two intervals are shared by gorilla and baboon; while three intervals are apparently shared by macaque and chimpanzee. Our Southern analysis shows that human and orangutan completely lack PTERV1 sequence (see Figure 2 A). If these sites were truly orthologous and, thus, ancestral in the human/ape ancestor, it would require that at least six of these sites were deleted in the human lineage. Moreover, the same exact six sites would also have had to have been deleted in the orangutan lineage if the generally accepted phylogeny is correct. Such a series of independent deletion events at the same precise locations in the genome is unlikely ( Figure S3 ). For the three intervals putatively shared between macaque and chimpanzee, we attempted to refine the precise position of the insertions by taking advantage of the available whole-genome shotgun sequences for these two genomes. For each of the three loci, we mapped the precise insertion site in the chimpanzee and then examined the corresponding site in macaque ( http://www.ncbi.nlm.nih.gov ). In one case, we were unable to refine the map interval owing to the presence of repetitive rich sequences within the interval. In two cases, we were able to refine the map location to single basepair resolution (Figures S4 and S5 ). Based on this analysis, we determined that the sites were not orthologous between chimpanzee and macaque. It is interesting to note that this level of refined mapping in chimpanzee revealed 4- to 5-bp AT-rich target site duplications in both cases. These findings are consistent with an exogenous retrovirus source since proviral integrations typically target AT-rich DNA ranging from 4 to 6 bp in length [ 24 ]. Although the status of the remaining overlapping sites is unknown, these data resolve four additional sites as independent insertion events and suggest that the remainder may similarly be non-orthologous. This apparent independent clustering of retroviral insertions at similar locations may be a consequence of preferential integration bias or the effect of selection pressure against gene regions, limiting the number of effective sites that are tolerated for fixation. Phylogenetic Analysis of PTERV1 We next examined the phylogenetic relationship of the retroviral elements by comparing portions of the gag and env regions. We chose a total of 103 BAC clones representing distinct loci from the four species and PCR-amplified and sequenced noncontiguous gag and env portions (823 bp) from each clone. Based on sequence analysis, each of the 101 BAC clones contained a single copy of the gag and env portions as determined by analysis of the sequence. These were deemed to be linked to the same endogenous retrovirus insert. Two clones showed “heterozygous” sequence signatures consistent with two or more copies clustered within the BAC clone and were excluded from further analysis. We constructed a phylogenetic tree based on the multiple sequence alignment ( Figure 4 ) of the concatenated gag and env regions. (A similar tree topology with less bootstrap support is obtained if env and gag segments are considered separately [ Figure S6 A and S6 B].) While it is clear that this particular class of endogenous retroviruses shares a common origin, the retroviral phylogeny is inconsistent with the generally accepted primate species tree based on molecular data [ 5 ]. The chimpanzee and gorilla PTERV1 elements are most closely related (3%–4% divergence) and belong to a single phylogenetic clade ( Table S4 ). In contrast, macaque and baboon inserts show considerably greater sequence divergence (9%–11%) and a much more stratified phylogeny with little discrimination based on species. This tree topology suggests a polyphyletic origin with at least three groups of Old World virus being distinguished ( Figure 4 ). Interestingly, one of these Old World groups (group 1) shows a possible monophyletic origin with respect to chimpanzee and gorilla. Figure 4 PTERV1 Phylogenetic Tree Portions of the gag and env genes (about 823 bp) were resequenced from 101 PTERV1 elements from common chimpanzee ( n = 42), gorilla ( n = 25), rhesus macaque ( n = 14), and olive baboon ( n = 20). A neighbor-joining phylogenetic tree shows a monophyletic origin for the gorilla and chimpanzee endogenous retroviruses but a polyphyletic origin among the Old World monkey species. Bootstrap support ( n = 10,000 replicates) for individual branches are underlined. Although the retroviral insertions have occurred after speciation, retroviral sequences show greater divergence than expected for a non-coding nuclear DNA element (see Table S4 ). Table S8 provides a clone key for number designation. Phylogenetic trees showing the gag, env, and LTR segments separately are presented in Figure S6 . Sequences 11 and 30 (red) are mapped to one of the 12 ambiguous overlapping loci described in the text (see Table S3 ). They do not cluster in this phylogenetic tree, which indicates that they are unlikely to be true orthologs. Since gag and env regions of viruses may experience vastly different selection pressures, we repeated the analysis by examining 295 bp of LTR sequence from a subset of 55 loci ( Table S4 ). Phylogenetic analysis of the LTR segment between species revealed a virtually identical tree topology ( Figure S6 C) to that of the concatenated gag and env segments. Macaque and baboon interproviral divergence (14%–24%), once again, was significantly greater than that for chimpanzee and gorilla (3%–4%). It should be noted that this degree of divergence is 2-fold greater than the average divergence for a neutral site between gorilla and chimpanzee (approximately 1.6%) [ 25 ] and approximately 10-fold greater than estimated neutral divergence rates between macaque and baboon (1.5%, E. E. E., unpublished data). Identical copies of LTR sequences are created as part of the retrovirus life cycle [ 2 , 8 ]. Consequently, divergence of flanking LTR elements has been used extensively as a metric to estimate the evolutionary age of the source infection. We compared intraproviral LTR sequence divergence from 101 loci in the chimpanzee, gorilla, baboon, and macaque genomes ( Figure 5 ). We designed oligonucleotides within conserved portions of the LTR sequence alignment, PCR-amplified the LTR flanks, sequenced both products from each clone, and compared them for mismatches. Gorilla and chimpanzee LTR elements showed a median divergence of 0.98% and 1.15%, respectively. Using neutral estimates of primate LTR divergence [ 8 ], we estimate that a contemporaneous infection occurred in these ancestral gorilla and chimpanzee lineages 3–4 million years ago (see Materials and Methods ). LTR divergence among baboon and macaque was significantly less (0.051% and 0.058%, respectively; p < 0.007, one-tailed t test), corresponding to a much more recent origin (approximately 1.5 million years ago). These observations may be reconciled with differences in the phylogenetic tree topology if the Old World monkeys were infected by several diverged viruses while gorilla and chimpanzee were infected by a single closely related exogenous source. In this scenario, most of the genetic differences observed among macaque and baboon endogenous retroviruses are not nuclear in origin but arose as part of normal variation between viruses (see Materials and Methods ). Figure 5 LTR Variation A total of 101 loci that contained full-length PTERV1 elements were examined for the number of mismatches between left and right LTR flanks (295 bp). Different distributions were obtained for Old World monkeys (baboon, mean = 1.6 ± 1.4; macaque, mean = 1.6 ± 1.5) and great ape species (chimpanzee, mean = 3.4 ± 2.2; gorilla, mean = 2.9 ± 2.3). We examined the distribution pattern of intraproviral LTR divergence to determine whether the observed pattern was consistent with a single or multiple infections. We modeled the probability of a mutation occurring within a LTR sequence using the Poisson distribution ( Table S5 ). For each of the four species, the distribution of LTR mismatches did not differ significantly from normal statistical fluctuations (Poisson distribution, p > 0.1). While these results are consistent with a single burst of retroviral insertions within each lineage, we cannot exclude the possibility of multiple integrations over a shorter span of 1–2 million years generating a virtually identical pattern. Indeed, based on the presence and absence of hybridizing bands among individuals from the same species (see Figure 2 B and 2 C), we estimate that 5%–10% of the sites are polymorphic within the baboon and chimpanzee populations. This may be the result of deletion or more recent reinfections of the germline. PTERV1 Integration Bias and Gene Expression Integration of retroviral sequence into genomes has long been recognized as a potent mutagen due to the fact that such proviruses frequently alter transcription, disrupt splicing, or become targets of hypermethylation. In this study, for example, the majority of full-length retroviral elements were heavily methylated (see Figure S2 ), suggesting that most copies had been transcriptionally silenced in transformed and peripheral blood lymphocytes. Studies of proviral integrations followed by inbreeding suggest that 5%–10% of novel insertions result in phenotypic change and/or alter gene expression [ 22 , 23 ]. We examined the distribution of full-length retroviruses within the chimpanzee genome and identified 107 sites of integration (Tables S6 and S7 ). We defined an integration site as genic if it mapped between the transcription start and stop site of a known annotated human gene. Only 13% (14/107) of retroviral integrations mapped within the introns of genes. Another 75 of the 107 insertion sites were located more than 50 kb upstream or downstream from a transcription start site. This is a significant departure ( p < 0.001) from a random model of intron integration (29.2%) and shows a distinctly opposite trend to patterns of somatic retroviral insertion, for which gene-rich regions of the genome are strongly favored (34%–60% of murine leukemia virus and HIV infections) [ 26 , 27 ] ( Figure S7 ). We propose that this 3- to 6-fold bias against gene-rich regions is the direct result of strong purifying selection pressure on the ancestral chimpanzee population. In order to determine whether the small subset of genes ( n = 14) that had been targets of insertions also showed significant changes in expression, we examined gene expression data for humans and chimpanzees in five tissues. This dataset was controlled for DNA sequence single-basepair differences between human and chimpanzee by excluding all probes that did not match perfectly between the two species (P. K and I. Hellman, unpublished data; [ 28 ]). Six of the ten genes detected in this dataset showed significant differences in expression levels between human and chimpanzee tissues ( Table 3 ). In five of the six cases, the genes showed reduced levels of gene expression in the chimpanzee when compared to human. A simulation study based on the total number of differentially expressed (approximately 30%) genes revealed that this enrichment is weakly significant ( p = 0.0489). The results suggest that retroviral insertion may have had an influence on expression difference between humans and chimps, but because of the small sample size (ten genes), it should be cautioned that the results are far from definitive. Additional studies, including RT-PCR and Northern analysis, with carefully controlled probes and matched tissue sources will be required to fully address this issue. Table 3 Retroviral Insertions That Map within Genes in the Chimpanzee Genome Assembly a Insertions were mapped based on the chimpanzee genome assembly (CGSC) and by paired-end sequence analysis (see Materials and Methods ) b Expression differences were based on an Affymetrix (U133A) microarray analysis of five tissues (brain, heart, kidney, liver, and testis) from five chimpanzee and six humans (P. K. and S. P., unpublished data). Tissues with a significant difference in expression are indicated as follows: −1 denotes significant upregulation in chimpanzee, while +1 represents significant upregulation in human. Genes with equivalent or those that were not detected (ND) on the microarray are indicated. Six out of ten genes showed significant differences in gene expression when compared to a control set that showed an approximately 30% expression difference (2,459/7,947 genes) between human and chimpanzee Discussion Most human endogenous retroviruses are thought to have emerged as a result of ancient infections more than 25 million years ago [ 7 , 8 ], followed by subsequent retrotransposition events. Several lines of evidence indicate that chimpanzee and gorilla PTERV1 copies arose from an exogenous source. First, there is virtually no overlap (less than 4%) between the location of insertions among chimpanzee, gorilla, macaque, and baboon, making it unlikely that endogenous copies existed in a common ancestor and then became subsequently deleted in the human lineage and orangutan lineage. Second, the PTERV1 phylogenetic tree is inconsistent with the generally accepted species tree for primates, suggesting a horizontal transmission as opposed to a vertical transmission from a common ape ancestor. An alternative explanation may be that the primate phylogeny is grossly incorrect, as has been proposed by a minority of anthropologists [ 29 ]. This seems unlikely in light of the extensive molecular evolutionary data that have been collected over the last few years [ 5 , 25 ] that clearly place orangutan as the outgroup species to the human–chimpanzee–gorilla clade and Old World monkeys as an outgroup to the human/ape lineage. Third, the single nucleotide substitution rate for the viruses is significantly greater than what would be expected for neutral nuclear DNA. The extent of chimpanzee and gorilla substitution, for example, has been estimated at approximately 0.016 ± 0.008 substitutions per site [ 25 ], with approximately half of this variation occurring in each lineage. The endogenous retrovirus sequence that we examined showed significantly greater divergence (0.038 ± 0.003) (62 pairwise comparisons). A similar excess of divergence was observed if only intraspecific retroviral divergence was compared (0.028 ± 0.003 and 0.041 ± 0.003 for chimpanzee and gorilla, respectively) (see Table S4 ). Such an acceleration of neutral substitution would easily be explained if it were composed of a viral and nuclear component (see Materials and Methods ). Fourth, if we partition synonymous and non-synonymous substitution sites (see Materials and Methods ), we observe a deficiency of amino acid replacement sites ( K a / K s = 0.63). We observed a similar result ( K a / K s = 0.44) for one of the ambiguous overlap loci shared between gorilla and chimpanzee (see Table S3 ). This significant departure from neutrality would be an expected residuum if a portion of PTERV1 sequence variation accrued while being propagated as an infectious virus [ 11 ]. If it were solely derived from an ancestral endogenous element, a neutral pattern, as opposed to a relaxed pattern of purifying selection, would be expected. Finally, in the few examples where the insertion sites have been mapped precisely, both the length and the composition of the target site duplications are characteristic of the patterns of retroviral integrations [ 24 ]. While these multiple lines of indirect data indicate that PTERV1 likely emerged from an exogenous source, its source reservoir, if it still exists, is unknown. PTERV1 does not share high sequence identity to any known retrovirus. Translation of the protein-encoding portions shows sequence similarity (approximately 50%) to feline leukemia viruses, murine leukemia viruses, and the baboon endogenous retrovirus. Such sequences are known to transfer frequently between the soma of species and occasionally enter the germline [ 17 , 18 , 19 , 20 ]. It is interesting that one of the three main branches of Old World monkey PTERV1 may actually share a monophyletic origin with the gorilla and chimpanzee elements. One possible scenario may be that this retrovirus was introduced into the great ape lineages by horizontal transmission, perhaps from contact with an ancient Old World monkey species. Our data support a model where ancestral chimpanzee and gorilla species were infected independently and contemporaneously by an exogenous source of gammaretrovirus 3–4 million years ago. While similar infections with a related retrovirus appear commonplace among the Old World monkeys, contemporary human and orangutan populations show no molecular vestiges of this infection (see Figure 2 ). The molecular basis for this historical difference is unclear. While geographic isolation of the African and Asian ape lineages during the Miocene [ 30 , 31 ] might account for part of this difference, the ancestral habitat of early hominids is generally thought to have overlapped, in part, with the African apes [ 32 , 33 ]. Furthermore, both Asian (macaque) and African (baboon) Old World monkeys show evidence of PTERV1 proviral integrations less than 2 million years ago, indicating that the exogenous source virus is either endemic to both continents or that ancestral populations frequented both continents. Several speculative scenarios may be envisioned to explain the absence of retrovirus in both the orangutan and human lineages. It is possible that the African apes evolved a susceptibility, or humans and Asian apes developed resistance to infection, although in either scenario convergent evolution would have had to have occurred with respect to the viral infections. Studies of the retroviral infection of the Lake Casitas mouse population reveal that such susceptibility/resistance genes may emerge very quickly among closely related strains of mice [ 34 ]. Another scenario may be that the lineage that ultimately gave rise to humans did not occupy the same habitat as the ancestral chimpanzee and gorilla lineages. An excursion by early hominids to Eurasia during the time that PTERV1 infected African great apes and then a return to Africa would explain this phylogenetic inconsistency. It is also possible that this effect may have been created by dramatic differences in ancestral population structure. If, for example, the ancestral populations of humans and orangutans were substantially larger than those of the African great apes, the fixation of new insertions (1/2 N ) would occur much more rapidly within small inbred populations even if similar infection rates existed. A similar model has recently been proposed, albeit in the opposite direction, to explain an increase of “apparent” Alu Ya5 and Yb8 retroposition activity in the human lineage but not in chimpanzees and gorillas [ 35 ]. In this regard, it is interesting that documented differences in the patterns of endogenous retrovirus between domesticated and feral species have been attributed to inbreeding [ 19 , 20 ]. There is, however, no evidence to date that the ancestral populations of chimpanzees were smaller than that of humans. Recent studies suggest that ancestral chimpanzee populations, in fact, may have been two to four times larger [ 36 , 37 ] than the effective human population size (greater than 10,000). A dramatic population crash in ancestral gorilla and chimpanzee populations would be required to explain the effect we have observed. Further population genetic studies of contemporary great apes or paleoanthropological work may help to eliminate these and other possible scenarios. Finally, it is not unreasonable to assume that these ancient infections reduced effective population size if fitness of ancestral populations were compromised by the infection. Recently, such an ancient retroviral infection was predicted to have occurred in chimpanzee based on a completely separate line of reasoning. De Groot and colleagues reported a dramatic reduction of genetic variability of intronic sequence from the major histocompatibility complex (MHC) I human leukocyte antigen loci (A, B, and C) among chimpanzee when compared to human populations [ 38 ]. This is a notable exception to other studies that demonstrate 2- to 4-fold greater diversity in chimpanzee populations than in humans [ 39 ]. Based on the evolutionary age of some of the new lineages and comparisons between chimpanzee and bonobo, de Groot and colleagues estimated the loss of MHC I diversity to have occurred before subspeciation, 2–3 million years ago. Due to the central role that pathogens play in eliciting immune response against viral infections and the fact that high-frequency MHC I haplotypes target conserved epitopes of the HIV-1 virus, the authors speculated that the unusual pattern of MHC I diversity might be the result of a pandemic retroviral infection that positively selected a small number of lineages to be swept through the population. Our findings are consistent with the timing of this loss of diversity and may represent the genomic vestiges of this retroviral pandemic. Due to its integration into the germline, this retroviral infection, thus, may have had a double impact. At a genetic level, at least 5% of the retroviral insertions would have resulted in lethality when homozygosed [ 22 ]. The 3-fold integration bias against gene insertions may represent a strong signature of this selection. This distribution is in sharp contrast to patterns of somatic retroviral infection [ 26 , 27 ] as well as recent class II human endogenous retroviral elements that map near or within genes [ 9 ]. In a background of reduced survival and lowered fecundity, genetic bottlenecks may have been frequent occurrences among ancestral chimpanzee and gorilla populations after speciation [ 33 ]. During this time of retroviral crisis, small subsets of retrovirus-induced mutations may have been fixed at an increased frequency. The mutation and fixation of multiple weakly deleterious mutations could, in theory, promote further saltatory and irrevocable changes in phenotypic traits among these progenitor populations. Such episodic mutational events may have simultaneously propelled species differentiation and cemented reproductive barriers between humans and the African great apes. In such a scenario greater sequence divergence over these regions might be expected because of a lack of introgression upon secondary contact among incipient species. It will be interesting to compare patterns of divergence for these sites with those for other genomic regions in humans and African great apes when genome sequence of sufficient quality becomes available. Materials and Methods Chimpanzee genome analysis A computational pipeline was established to identify insertions and deletions between contiguous BAC chimpanzee genome sequence and the human genome assembly (EEE, unpublished data). Using alignment visualization software (PARASIGHT), we detected an 8.45-kb segment present in chimpanzee (AC097267 from 91,434 to 99,886) but absent in human. The sequence was not classified as a repeat (RepeatMasker, version 4.0) but six-frame translation and protein BLAST sequence similarity searches showed significant homology to retroviral sequences including gag, pol, and env proteins (see Figure 1 ). This family of repeat elements, PTERV1, corresponds to the largest of three retroviral repeat families that were discovered in chimpanzee genome sequence but not in humans (Chimpanzee Genome Sequencing Consortium, unpublished data). Sequence analysis of the chimpanzee genome (November 2003) identified 107 putative full-length copies and 90 solo LTR elements (see Table S7 ). Since most full-length copies were incompletely assembled, map positions were confirmed by paired end-sequence analysis (see below) in addition to BLAST sequence similarity searches. Genomic hybridization Primate DNA was restriction-enzyme-digested, transferred to nylon membrane, and hybridized as described previously [ 40 ]. Species included human (Homo sapiens), common chimpanzee (Pan troglodytes), bonobo (Pan paniscus), gorilla (Gorilla gorilla), orangutan (Pongo pygmaeus), siamang (Hylobates syndactylus), white-handed gibbon (Hylobates lar), Abyssinian black-and-white colobus monkey (Colobus guereza), olive baboon (Papio anubis), rhesus macaque (Macaca mulatta), and Japanese macaque (Macaca fuscata) (see Figure 2 ). In addition, we analyzed multiple individuals (three or four) from each ape lineage and a diversity panel of 16 humans and nine baboons. PCR-amplified products (see Table S1 for PCR oligonucleotide sequence and conditions) corresponding to the gag, env, and LTR portions of PTERV1 were used as radioactive probes as described [ 41 ]. Large-insert genomic BAC libraries from chimpanzee (RPCI-43), gorilla (CHORI-255), the olive baboon (RPCI-41), and the rhesus macaque (CHORI-250) were also hybridized. A total of 2,706 BAC clones were obtained that hybridized with PTERV1 gag , env, and LTR probes, while 6,032 clones hybridized solely with the LTR probe (see Table 1 ). The former were classified as full-length insertions of PTERV1, while the latter was classified as solo LTR elements. Sequencing. A total of 1,467 BAC clones containing full-length insertions were end-sequenced and are publicly available from GenBank. In addition, a total of 101 genomic clones (42 chimp, 25 gorilla, 20 baboon, and 14 macaque) corresponding to distinct insertion loci were comparatively sequenced from gag and env portions of the endogenous retrovirus (approximately 823 bp each). Individual clones were also subjected to LTR sequencing, and sequencing variants were analyzed using PolyPhred software [ 42 ]. All PCR products (forward and reverse reactions) were directly sequenced, using a modified dye terminator sequencing protocol [ 41 ]. Fluorescent traces were analyzed using an Applied Biosystems PRISM 3100 DNA Sequencing System (Perkin-Elmer Applied Biosystems, Norwalk, Connecticut, United States), and the quality of the sequence data was assessed with Phred/Phrap/Consed software [ 43 , 44 ]. Expression analysis Gene expression differences between human and chimpanzee were assessed as previously described [ 28 ] with the following exceptions: five tissues (heart, brain, liver, testis, and kidney) were compared among five chimpanzee and six human samples using Affymetrix (Santa Clara, California, United States) HG U133plus2 arrays. Only those oligonucleotide probe sets that showed a perfect match between human and chimpanzee genomic sequence DNA were considered in this analysis (17,617/54,377). Differentially expressed genes were defined as those that met the following criteria: (i) the gene had to be expressed in all individuals from at least one species (detection p- value of less than 0.065); (ii) the gene had to show a change in expression in the same direction (change p -value [two-tailed] of less than 0.5 or greater than 0.5) in all 30 pairwise comparisons; (iii) different probe sets from the same gene did not conflict. The full dataset that we analyzed consisted of a total of 17,617 probes corresponding to 7,947 genes that showed significant levels of expression in both chimpanzee and human. About 70% of the genes (5,488) showed equivalent levels of expression between both species, while approximately 30% (2,459) showed differential patterns of expression. We identified 14 genes in which retroviral insertions had occurred within the chimpanzee lineage. Ten of these were found in the full expression dataset: four genes showed equivalent expression levels, while six showed differential patterns (largely decreases) in expression. To determine whether this 2-fold increase might be significant, we performed a simulation study. We randomly sampled ten genes from the full dataset (7,947) and calculated the number of genes for which significant differences were observed for each replicate. We repeated this 10,000 times and determined that 9,511 of the replicates showed less than six differentially expressed genes, 374 showed precisely six differentially expressed genes, and 115 showed more than six. Based on this analysis of the control dataset, we determined that enrichment was weakly significant ( p = 0.0489). Evolutionary analyses End sequences generated from BAC clone inserts (T7 and SP6) were used to position retroviral insertions with respect to the human genome reference (Build 34). We considered only optimal placements after sequence quality rescoring (Phred Q > 25) and masking of common repeats (at least 50 bp of unmasked sequence was required to place an end sequence). We required two independent BAC clones per locus, which allowed mapping intervals to be refined and eliminated potential false positives. A total of 275/287 locations mapped to independent locations. Twelve locations could not be distinguished based on the resolving power of this mapping approach (100–150 kb). Estimates of genetic distance (pairwise deletion) were calculated using Kimura's two-parameter model (where s / v was approximately 2.0) [ 45 ]. The elevated substitution rate for PTERV1 is consistent with the total number of substitutions ( K total ) being partitioned between a viral component ( K v ) and a nuclear component ( K n ). For protein-encoding portions of the retrovirus, the average number of synonymous ( K s ) and nonsynonymous ( K a ) substitutions per site was estimated using the modified Nei–Gojobori method [ 46 ]. We calculated evolutionary times of retroviral insertion based on the divergence of LTR flanks (0.13% divergence per million years and r = K /2 T ) [ 8 ]. Phylogenetic trees of multiple aligned sequences (ClustalW) were generated using neighbor-joining distance estimates (MEGA2). Only bootstrap values greater than 80% are indicated in the tree topology (see Figure 4 ). We modeled the genic distribution of retroviral insertions by random simulation of 107 full-length map positions within the human genome reference sequence. An insertion was classified as intronic if the insertion site mapped within the transcription start and transcription end of a Refseq gene annotation (920 Mb). By this measure, 29.7% (with gaps) and 32.1% (without gaps) of the genome is transcribed. Not once in 10,000 replicates was the measured gene distribution (14/107) observed. Similarly, we estimated the probability that six of the ten genes that showed significant expression differences between human and chimpanzee would have occurred by random chance. Supporting Information Figure S1 Southern Analysis of PTERV 1 among Primates A panel of African great ape DNAs is compared for both the gag probe number 1 by PstI and PvuII restriction enzyme digest. Proximity of probe number 1 to the VNTR sequence and its variable copy number allows the detection of hundreds of insertion sites, showing limited intraspecific variation. Species represented include human (HSA), common chimpanzee (PTR), bonobo (PPA), gorilla (GGO), and orangutan (PPY). Below each panel, a restriction map (chimpanzee reference sequence AC097267) is presented in relation to the hybridization probe: PstI (closed circles), PvuII (open circles), and HpaII/MspI (triangles). (475 KB PDF). Click here for additional data file. Figure S2 Methylation Status of PTERV1 Sequence Methylation status of the retroviral insertions is compared between peripheral blood DNA and transformed lymphoblast cell lines for baboon (PHA), chimpanzee (PTR), and gorilla (GGO). HpaII does not digest methylated restriction enzyme sites, while MspI is insensitive. The differential patterns suggest most PTERV1 insertions are methylated. (317 KB PDF). Click here for additional data file. Figure S3 Ambiguous Overlap Loci and Primate Phylogeny The Distribution of Ambiguous Overlap Loci for Human and Great Apes Is Shown (Arcs) with Respect to a Generally Accepted Phylogeny of Catarrhine Primates If these sites are orthologous, then at least six deletion events would have had to occur independently in orangutan and human at precisely the same positions in both genomes. (10 KB PDF). Click here for additional data file. Figure S4 Resolution of Ambiguous Overlap Locus (RP43–114h16) (A) Multiple alignment of sequence flanking chimp insertion RP43–114h16 against chimpanzee whole-genome shotgun sequence reads. Precise site of retrovirus integration is indicated by the red arrow, and a 5-bp target site duplication is indicated in the blue box. Note that no chimpanzee sequence reads are contiguous across the region because of the 6-kb retrovirus insertion that extends from the insertion site (data not shown). (B) Sequence similarity search of sequence flanking chimp insertion against macaque whole-genome shotgun sequence reads. The analysis shows that the chimpanzee insertion site is not shared with macaque. In contrast, human and macaque sequence are contiguous across this site. A macaque retroviral insertion maps to a different location within this 160-kb interval of the genome. This overlap locus is, therefore, resolved as non-orthologous. (33 KB PDF). Click here for additional data file. Figure S5 Resolution of Ambiguous Overlap Locus (RP43–151j13) (A) Multiple alignment of sequence flanking chimp retroviral insertion RP43–151j13 against chimpanzee whole-genome shotgun sequence reads. Precise site of retrovirus integration is indicated by the red arrow, and a 4-bp target site duplication is indicated by the blue box. Note that no chimpanzee sequence reads are contiguous across the region because of the 6-kb retrovirus insertion that extends from the insertion site (data not shown). (B) Sequence similarity search of same sequence flanking chimp insertion against macaque whole-genome shotgun sequence reads. The analysis shows that the chimpanzee insertion site is not shared with macaque. Human and macaque sequences are contiguous across the chimpanzee insertion site. A macaque retroviral insertion maps to a different location in this 160-kb region of the genome. This overlap locus is, therefore, resolved as non-orthologous. (34 KB PDF). Click here for additional data file. Figure S6 PTERV1 Phylogenetic Trees Neighbor-joining phylogenetic trees were constructed for (A) env, (B) gag, and (C) LTR portions of the retrovirus independently, as described in Figure 4 . Bootstrap support ( n = 10,000 replicates) for individual branches is indicated in italics and is substantially lower owing to the limited number of basepairs compared within each tree (especially in Figure S6 B). The general topology of each of the three trees is comparable and shows gorilla and chimpanzee retroviral insertions as a single monophyletic clade. (47 KB PDF). Click here for additional data file. Figure S7 A Random Simulation of 107 Retroviral Insertions within the Human Genome Genic space was determined as the distance encompassed from transcription start to transcription end of Refseq gene annotations (920 Mb). By this measure, 29.7% (with gaps) and 32.1% (without gaps) of the genome is transcribed. Not once in 10,000 replicates was the measured gene distribution (14/107) observed. (13 KB PDF). Click here for additional data file. Table S1 PCR Primers and Probes (11 KB PDF). Click here for additional data file. Table S2 Mapping of Full-Length Retroviral Insertion Sites on Human Genome (Build 34) (760 KB PDF). Click here for additional data file. Table S3 Retroviral Map Intervals That Potentially Overlap between Species (50 KB PDF). Click here for additional data file. Table S4 Inter- and Intra-Specific Genetic Distance Estimation among PTERV1 Elements (A) Genetic distance for gag–env portion. (B) Genetic distance for LTR portion. (13 KB PDF). Click here for additional data file. Table S5 Distribution of LTR Mismatches for Each Species (Observed) Compared with a Poisson Distribution (Expected) (24 KB PDF). Click here for additional data file. Table S6 Mapping of Solo LTR and Full-Length PTERV1 Elements on Human Genome (Build 34) (66 KB PDF). Click here for additional data file. Table S7 PTERV1 Distribution in Chimpanzee Genome (9 KB PDF). Click here for additional data file. Table S8 The Keys (Numbers) and Their Corresponding Clones for Figure 4 (24 KB PDF). Click here for additional data file. Accession Numbers The sequence data described in this paper have been submitted to GenBank ( http://www.ncbi.nlm.nih.gov/Genbank/ ) under accession numbers AY758600–AY760022, AY760471–AY760631, AY760098–AY760470, and AY760640–AY761013. The GenBank accession numbers for the sequences discussed in this paper are baboon endogenous retrovirus (AF142988), feline leukemia virus (AAC31801), murine leukemia virus (AAA79427), porcine endogenous retrovirus type C (CAC39617), and chimpanzee BAC basepair positions 91,434–99,886 (AC097267).

D:\keerthana\PMC001xxxxxx\PMC1054887.xml

1054888

The Few, the Strong: Rat Cortex Features Small Numbers of Powerful Connections

How is the brain wired up? Each neuron may connect with hundreds or even thousands of others, and the human brain has a hundred billion neurons. Determining the connection diagram for a whole brain is a truly daunting prospect, and currently well beyond reach. But one way into this thicket is to look for patterns in a small region. In this issue, Dmitri Chklovskii and colleagues show that in the rat visual cortex, some kinds of connection patterns are much more common and much stronger than chance would predict. To determine the pattern of connections, the researchers placed electrodes into randomly chosen quartets of neurons near each other. They stimulated each in turn, and determined which members responded, and how strongly. Sampling over 800 such quartets, they found 931 actual connections out of a possible 8,050, for an average rate of connectivity of 11.6%. From the group of connected neurons, they then asked about reciprocal connections: what was the likelihood that, if A stimulated B, B stimulated A as well? They found that bidirectionally connected cells were four times as common as expected by chance, a pattern previously observed in other regions of cortex. They asked the same question for groups of three cells, for which there are 16 possible connection patterns. Two patterns stood out as especially significant: (1) A and B talk back and forth with each other, and both listen to C; and (2) A, B, and C all talk with one another. For four cells, although the numbers were too small for statistical analysis, a common over-represented class was chain connections, a kind of a path connecting all four cells that can be drawn without lifting the pencil from the page. Recording multiple neurons simultaneously Because the strength with which one neuron stimulates another can be just as important to network function as whether a connection exists at all, the authors examined connection strength as well. They found that connection strengths are distributed broadly, with some connections ten times stronger than the average connection and the strongest 17% of connections contributing half of total synaptic strength. They found that, on average, connections that were part of bidirectional pairs were about 50% stronger than unidirectional ones, and because of this, despite being fewer in number, they disproportionately contributed to the total amount of excitation in the neural network. A similar pattern was found for neuronal triplets—the most highly connected groups of neurons had the strongest connections among them. Taken together, these results show that neural networks, at least in this portion of the rat brain, are characterized by a vocal minority of unexpectedly strong and reliable connections amidst a large number of weak ones, which suggests the strong ones may play more central roles in local computation or communication. This stands in strong contrast to the usual starting assumption of neural modelers, that connectivity is random. The exact pattern of connectivity seen here for excitatory neurons in one cortical layer (layer 5) may not be universal, and indeed, different patterns have been described in the cerebellum. Nonetheless, the essential feature seen here—“a skeleton of stronger connections in a sea of weaker ones,” as the authors put it—may be an important and common functional feature of brain wiring.

D:\keerthana\PMC001xxxxxx\PMC1054888.xml

1054889

Forecasting the Path of a Raccoon Rabies Epidemic

Rabies recently hit the national headlines when a Wisconsin teenager survived after showing full-blown symptoms. Even more remarkable, the girl—who was bitten by a bat—recovered after receiving a novel therapy, since doctors felt her case was too advanced for the standard rabies inoculations to work. Rabies is nearly always fatal if not treated immediately, and continues to pose a serious public health threat. Though most rabies fatalities in the United States stem from bat bites, far more people are treated for raccoon rabies. A new strain of raccoon rabies started spreading throughout the eastern United States in the mid-1970s, after raccoons caught in Florida were released along the West Virginia–Virginia border to replenish hunting stocks. Some of the imports carried a rabies variant that caused an outbreak in local populations and has been steadily expanding ever since. In 1990, raccoons topped the list of most often reported rabid mammal. Predicting the spread of raccoon rabies across Ohio Controlling this re-emerging public health threat depends on predicting the spatial dynamics of the disease—where new outbreaks might occur and how the virus might spread. Toward this end, Leslie Real and colleagues work on probabilistic simulation models that calculate the effects of various factors, such as local transmission rates between townships, ecological barriers to transmission, and long-distance “translocation” rates between townships. (The deliberately released Florida raccoons were one such translocation, but raccoons have also been known to hitch rides on garbage trucks.) As reported elsewhere, these models previously accurately reflected rabies spread in both Connecticut and New York. In a new study reported in PLoS Biology , Real and colleagues apply their model to the likely spread of rabies in Ohio—a potential gateway for spread throughout the Midwest—and find that raccoon rabies could spread throughout the state in just three years. One strategy for limiting rabies spread is to establish vaccine corridors by distributing vaccine baits—vaccine doses hidden in fishmeal—to wild raccoons. This cordon sanitaire strategy limited rabies in Ohio to sporadic cases from 1997 until 2004, when a rabid animal was detected—11 kilometers beyond the buffer zone—in northeastern Ohio. The authors had previously shown that local transmission was significantly reduced when townships were separated by geographical barriers—the Connecticut River in Connecticut and the Adirondack Mountains in New York. In modeling the likely transmission path in Ohio, the authors incorporated the likely effect of Ohio's five major rivers on transmission from local points along the Pennsylvania or West Virginia border. Given Ohio's topography (three of its rivers run along the southern and eastern border) and a single point of emergence in the northeast, the authors adjusted their simulations to estimate the potential impact of translocations. Even without the occasional garbage truck ride, because of the lack of ecological barriers in central Ohio, the simulations predict that rabies will spread far faster in Ohio than in New York and Connecticut. Factoring in those garbage truck rides, the scenario is considerably bleaker: rabies would take just 33 months to spread across central Ohio—compared to 48 months to cross the much smaller state of Connecticut—and cover the state in 41 months. This transmission rate—100 kilometers/year—significantly surpasses previous estimates, which range from 30 to 60 kilometers/year. The potential for such rapid spread, if unchecked, “is quite alarming,” the authors warn. But they also point out that the path of a real epidemic would likely fall somewhere between these two scenarios, given the unpredictable nature of translocations. The authors also simulated potential breech points in the vaccine corridor and found that the Ohio and Muskingum rivers halted viral advance initially. But a raccoon can certainly cross a bridge when the opportunity arises, so any delays would likely be temporary. Given the unpredictable nature of rabies transmission—challenging efforts to identify potential leaks in vaccine corridors and sites of dispersal—the authors' simulations provide a valuable resource for anticipating alternate outbreak scenarios and preparing multiple game plans to prevent or contain them. They also indicate the best sites for establishing a new vaccine barrier. And given how fast raccoon rabies could spread, Real and colleagues make a strong case that halting its western march depends on a strategy based on early detection and high-powered intervention programs—a sensible approach for any infectious disease.

D:\keerthana\PMC001xxxxxx\PMC1054889.xml

1054890

Recombination as a Way of Life: Viruses Do It Every Day

In theory, a cell's nuclear membrane guards its contents by barring access to potential foes. In reality, pathogens employ a diverse bag of tricks to circumvent this barrier. The murine leukemia virus (a retrovirus), for example, waits until the nuclear membrane degrades during cell division. Other retroviruses, like HIV and so-called pararetroviruses, enlist protein escorts that help them slip through undetected. Pararetroviruses include both animal viruses, such as hepatitis B, and plant viruses, such as the cauliflower mosaic virus (CaMV). Once inside the nucleus, the double-stranded DNA genome of the CaMV is transcribed into an RNA transcript (called 35S RNA), thanks to the activity of the 35S promoter. (This CaMV promoter is widely used to drive transgenic expression in plants.) Replication proceeds through reverse transcription as a viral enzyme reverse transcribes the 35S RNA into genomic DNA that is then packaged into viral particles. Turnip infected by cauliflower mosaic virus During replication, genetic material can pass between different viral genomes when two viral particles infect the same host cell. These exchanges can create novel viruses, much like mutations in bacteria can produce new bacterial strains that show resistance to host defenses and antibiotics. But with little data on viral recombination rates in multicellular organisms, it's unclear how these recombinant viral genomes are influencing host infection. In a new study, Yannis Michalakis and colleagues follow the course of the cauliflower mosaic viral infection in one of its natural hosts, the turnip plant ( Brassica rapa ), to measure the frequency of viral recombination. Recombination was evident in over half of the recovered viral genomes, suggesting that recombination is routine for this plant virus. It's thought that CaMV recombination occurs mostly outside the nucleus, in the host's cytoplasm, during reverse transcription. To quantify the frequency of such events, Michalakis and colleagues generated a CaMV genome with four genetic markers and then infected 24 turnip plants with equal amounts of marked and unaltered viruses. Recombination between the two “parent” genomes would produce viral populations with genetic material from both parents. The plants were harvested when full-blown symptoms developed, 21 days after inoculation, and viral DNA was extracted from their leaves to evaluate the occurrence and frequency of recombination. Assuming that all marker-containing genomes could recombine, the authors predicted that the viruses should produce seven classes of recombinant genotypes, which is what they found. These recombinant genotypes showed up in over 50% of the viral populations—which the authors call an “astonishingly high” proportion. Though little information exists on the length of viral replication cycles in plants, the authors assumed a generation time of two days, which would amount to ten replication cycles over the 21-day experimental period. From this assumption, the authors calculated the recombination rate on the order of 4 × 10 −5 per nucleotide base per replication cycle—hardly a rare occurrence. Certain CaMV genomic regions have been predicted as recombination hot spots, but the authors found that the virus “can exchange any portion of its genome… with an astonishingly high frequency during the course of a single host infection.” By evaluating the recombination behavior of a virus in a living multicellular organism, Michalakis and colleagues created a realistic approximation of recombination events during infection in the field. And since recombination events are linked to both expanded viral infection and increased virulence, understanding the rate of recombination could help shed light on mechanisms underlying the evolution and pathology of a virus—insight that could prove critical for developing methods to inhibit or contain an infection.

D:\keerthana\PMC001xxxxxx\PMC1054890.xml

1054891

The Chimp Genome Reveals a Retroviral Invasion in Primate Evolution

It's been known for a long time that only 2%–3% of human DNA codes for proteins. Much of the rest of our genomes—often referred to as junk DNA—consists of retroelements: genomic elements that are transcribed into RNA, reverse-transcribed into DNA, and then reinserted into a new spot in the genome. Human endogenous retroviruses make up one class of these retroelements. Retroviruses can insinuate themselves into the host's DNA in either soma (nonreproductive cells) or the germline (sperm or egg). If the virus invades a nonreproductive cell, infection may spread, but viral DNA will die with the host. A retrovirus is called endogenous when it invades the germline and gets passed on to offspring. Because endogenous retroviruses can alter gene function and genome structure, they can influence the evolution of their host species. Over 8% of our genome is made of these infectious remnants—infections that scientists believe occurred before Old World and New World monkeys diverged (25–35 million years ago). In a new study, Evan Eichler and colleagues scanned finished chimpanzee genome sequence for endogenous retroviral elements, and found one (called PTERV1) that does not occur in humans. Searching the genomes of a subset of apes and monkeys revealed that the retrovirus had integrated into the germline of African great apes and Old World monkeys—but did not infect humans and Asian apes (orangutan, siamang, and gibbon). This undermines the notion that an ancient infection invaded an ancestral primate lineage, since great apes (including humans) share a common ancestor with Old World monkeys. Eichler and colleagues found over 100 copies of PTERV1 in each African ape (chimp and gorilla) and Old World monkey (baboon and macaque) species. The authors compared the sites of viral integration in each of these primates and found that few if any of these insertion sites were shared among the primates. It appears therefore that the sequences have not been conserved from a common ancestor, but are specific to each lineage. PTERV1 contains three structural genes— gag , pol , and env —and regulatory sequences called long terminal repeats (LTRs). To further explore the evolutionary history of the retroviral elements, the authors compared the sequences of gag and pol , as well as the LTR sequences, for each infected primate species. The sequence history, they discovered, did not comport with the established evolutionary history of the primates themselves. Divergence between macaque and baboon was significantly greater than between gorilla and chimp—even though slightly more evolutionary time separates gorilla and chimp than macaque and baboon. When a retrovirus reproduces, identical copies of LTR sequences are created on either side of the retroviral element; the divergence of LTR sequences within a species can be used to estimate the age of an initial infection. Eichler and colleagues estimate that gorillas and chimps were infected about 3–4 million years ago, and baboon and macaque about 1.5 million years ago. The disconnect between the evolutionary history of the retrovirus and the primates, the authors conclude, could be explained if the Old World monkeys were infected by “several diverged viruses” while gorilla and chimpanzee were infected by a single, though unknown, source. Phylogenetic tree of retroviral insertions in primates As for how this retroviral infection bypassed orangutans and humans, the authors offer a number of possible scenarios but dismiss geographic isolation: even though Asian and African apes were mostly isolated during the Miocene era (spanning 24 to 5 million years ago), humans and African apes did overlap. It could be that African apes evolved a susceptibility to infection, for example, or that humans and Asian apes evolved resistance. A better understanding of the evolutionary history and population genetics of great apes will help identify the most likely scenarios. And knowing how these retroviral elements infiltrated some apes while sparing others could provide valuable insights into the process of evolution itself.

D:\keerthana\PMC001xxxxxx\PMC1054891.xml

1059451

Basal Immunoglobulin Signaling Actively Maintains Developmental Stage in Immature B Cells

In developing B lymphocytes, a successful V(D)J heavy chain (HC) immunoglobulin (Ig) rearrangement establishes HC allelic exclusion and signals pro-B cells to advance in development to the pre-B stage. A subsequent functional light chain (LC) rearrangement then results in the surface expression of IgM at the immature B cell stage. Here we show that interruption of basal IgM signaling in immature B cells, either by the inducible deletion of surface Ig via Cre-mediated excision or by incubating cells with the tyrosine kinase inhibitor herbimycin A or the phosphatidylinositol 3-kinase inhibitor wortmannin, led to a striking “back-differentiation” of cells to an earlier stage in B cell development, characterized by the expression of pro-B cell genes. Cells undergoing this reversal in development also showed evidence of new LC gene rearrangements, suggesting an important role for basal Ig signaling in the maintenance of LC allelic exclusion. These studies identify a previously unappreciated level of plasticity in the B cell developmental program, and have important implications for our understanding of central tolerance mechanisms.

Introduction B lymphocytes follow a highly ordered program of development in the bone marrow (BM), beginning with the commitment of lymphoid progenitors to the B lineage and the somatic recombination of heavy chain (HC) immunoglobulin (Ig) alleles [ 1 ]. Following an initial diversity (D H ) to joining (J H ) gene segment rearrangement, usually on both alleles, pro-B cells then rearrange one of many upstream variable (V H ) region segments to the D-J H segment, creating the V(D)J joint. These rearrangements require the action of the lymphoid-specific recombination activating genes Rag1 and Rag2, together with a number of ubiquitously expressed DNA repair proteins [ 2 ]. Cells with a productive protein-encoding HC rearrangement express HC together with invariant surrogate Ig light chains VpreB and lambda 5 (λ5), and then undergo clonal expansion before efficient initiation of rearrangements at light chain (LC) loci (kappa, κ, or lambda, λ) [ 3 ]. A productive LC rearrangement results in the cell surface expression of IgM, which defines the immature B cell stage (IgM + IgD − ). Due to the stochastic nature of V(D)J recombination, B cells express an extremely diverse Ig receptor repertoire (more than 10 9 specificities). To reduce the potential for autoimmune antibody responses, cells bearing strongly self-reactive Ig receptors are tolerized, either by clonal deletion, functional inactivation through the induction of anergy, or by receptor editing where new LC rearrangements revise the antigen (Ag) specificity of the receptor [ 4 , 5 ]. The maintenance of tolerance also requires that individual B cells express a single Ig HC and LC, since cells bearing multiple receptors could have significant autoimmune potential. In addition, cells bearing receptors in which the two antibody binding sites are not identical would have a reduced ability to bind certain antigens, which could, in turn, compromise downstream antibody effector functions such as complement activation [ 6 ]. The process by which cells express a single receptor is called allelic exclusion [ 3 ], with a functional Ig rearrangement likely providing a “stop” signal that blocks further rearrangements. In general, the mechanisms that initiate and maintain allelic exclusion are not well understood. HC allelic exclusion requires the expression of a functional membrane-bound HC protein, since mice lacking the Cμ transmembrane domain show a complete block in B cell development at the pro-B stage, and B cells fail to establish HC allelic exclusion [ 7 ]. HC allelic exclusion also requires the Ig receptor-associated signaling proteins Igα and Igβ [ 8 , 9 , 10 , 11 ]. Less is known about the signaling requirements for LC allelic exclusion, where the situation is complex due to the presence of two κ and two λ alleles and the potential for multiple rearrangements at each locus. LC receptor editing occurs in immature B cells with self-reactive Ig receptors, and continues until a suitable receptor is formed, whereupon further rearrangements are suppressed. Recent studies indicate that receptor editing at LC loci is a common theme in normal B cell development, occurring in approximately 20% or more of B cells during their maturation [ 12 ]. Despite the importance of receptor editing in shaping the B cell immune repertoire, our understanding of the mechanisms that drive editing are rudimentary. It is clear that Rag proteins can be re-induced in immature B cells following B cell receptor (BCR) crosslinking by self-antigen, and that this can lead to new rearrangements at LC loci [ 13 , 14 ]. The prevailing view is that positive signaling through crosslinked BCRs drives the editing response. However, in experiments investigating LC receptor editing responses to soluble self-antigen, we found an inverse relationship between levels of surface IgM and LC editing (i.e., low levels of IgM associated with high levels of editing) [ 15 ]. These data were consistent with the hypothesis that tonic signals provided by surface BCRs are important for suppressing editing responses in immature B cells. The experiments here were designed to examine this possibility, and the results suggest that basal signaling from the BCR is critical for immature B cells to suppress Rag expression and, surprisingly, to maintain developmental stage. Results Model for Inducible Deletion of Surface IgM in Immature B Cells For the initial series of experiments, we used transgenic mice specifically generated to examine the consequences of inducible deletion of the BCR on B cell function [ 16 ]. These animals carry a “floxed” B1-8 HC knock-in allele (B1-8f, flanked by LoxP recombination sites), a 3-83κ LC knock-in allele (3-83κ) [ 17 ], and an interferon (IFN)-inducible Cre recombinase transgene (Mx-Cre) [ 18 ]. In previous experiments, in vivo treatment of mice carrying these transgenes with type I IFN led to the induction of Cre, followed by the efficient deletion of the B1-8f HC and loss of BCR surface expression. These studies demonstrated that the survival of mature B cells was critically dependent on basal signaling by the BCR [ 16 ]. Bone marrow from B1-8f/3-83κ/Mx-Cre and control B1-8f/3-83κ mice was incubated in vitro with interleukin-7 (IL-7), a potent growth factor for early B cell progenitors, for 5 d. At the end of the culture period, flow cytometry revealed an expanded population of IgM + IgD − immature B cells that expressed the B1-8 IgM a allotype and the 3-83κ LC [recognized by the S27 antibody ( Figure 1 A)]. All of the IgM a allotype-positive cells also stained with an anti-B1-8 idiotype antibody (data not shown). Additional cell surface profiling of these cells is presented in box 4, A small population of B220 + cells was negative for the BCR in both cultures. Figure 1 Inducible Cre-Mediated Deletion of the B1-8f HC Allele Leads to Loss of Surface Ig from Immature B Cells (A) Flow cytometry showing surface phenotype of B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre BM after 5-d IL-7 culture. (B) Flow cytometric analysis of B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre BM culture cells incubated with IFNαβ (1,000 or 5,000 units/ml), in the absence of IL-7, for 1, 2, or 3 d. The cell populations shown are gated on lymphocytes by forward and side scatter, and then for B220. At the end of the 5-d IL-7 culture, more than 90% of the cells were viable and B220 + . The numbers shown indicate the percentage of gated cells. Cells were then washed and incubated with either 1,000 or 5,000 units/ml of IFN αβ for 1, 2, and 3 d, in the absence of IL-7. There was efficient IFN-dependent deletion of surface IgM in B1-8f/3-83κ/Mx-Cre B cells that was apparent by day 1 and was approximately 90% complete at day 3 ( Figure 1 B). Deletion was more efficient with 5,000 units/ml of IFNαβ than with 1,000 units/ml. Control experiments demonstrated that the cells lacking IgM receptor derived from the IgM hi population following Cre-mediated deletion of HC (described in detail below). Microarrays to Evaluate Gene Expression in Cells Undergoing Inducible Deletion of IgM IFN-treated B1-8f/3-83κ IgM hi (Ctrl-M hi ), B1-8f/3-83κ/Mx-Cre IgM hi (Cre-M hi ), and B1-8f/3-83κ/Mx-Cre IgM-deleted (Cre-M lo ) cell populations were sorted from 2-d IFN cultures ( Figure 2 A), total RNA was isolated, and biotinylated cRNA probes were generated (see Materials and Methods ). Probes were hybridized with Affymetrix U74A mouse GeneChip microarrays, which represent approximately 11,000 transcripts, and expression levels were quantified using Affymetrix MicroArray Suite 5.0 software. The data from each array were scaled to correct for minor differences in overall array hybridization intensity, and analyzed to identify genes differentially expressed between Ctrl-M hi and Cre-M lo cells. Figure 2 Microarray Gene Expression Analysis Demonstrates Co-Clustering of Cre-Deleted IgM lo Cells with IgM − Cell Populations (A) FACS sorting strategy for Ctrl-M hi , Cre-M hi , and Cre-M lo immature B cells incubated with IFNαβ (1,000 units/ml) for 2 d. The numbers shown indicate the percentage of gated cells. (B) Affymetrix microarrays were used to identify genes differentially expressed between IFN-treated Ctrl-M hi and Cre-M lo cells. Analysis identified 327 transcripts that met the following criteria: 2-fold or greater change in mean expression level, a more than 200-unit difference in mean expression values, and Student's t-test p < 0.01. Individual expression values for each gene were divided by the mean of expression levels for three control IgM + cell populations: IFN-treated Ctrl-M hi ; IgM hi cells sorted from 5-d IL-7 HEL-lg BM cultures (HEL-M hi ); and sorted from normal Balb/c BM (FxE). Other populations included Cre-M hi , Hardy Fraction D pre-B cells (FxD) sorted from normal Balb/c BM, and lgM − cells sorted from 5-d IL-7 cultures of control B6 BM (B6-M − ). Data were transformed into log 2 space, and represent fold-differences relative to the IgM + cell populations (see scale bar). Data from 293 transcripts (duplicates and all but four representative Ig HC and LC transcripts removed) were clustered and visualized using CLUSTER and TREEVIEW [ 51 ]. Red represents genes expressed at higher levels, while green represents genes expressed at lower levels, than the mean of IgM + cells. Each column represents an individual sorted cell population. Analysis of the array data confirmed that Ig HC gene expression dropped dramatically in the Cre-M lo cell population ( Table 1 ; two representative probesets are shown), while κ LC gene expression showed a significant approximately 2-fold up-regulation in Cre-M lo cells. Strikingly, Cre-M lo cells had elevated levels of mRNA for Rag1 and Rag2, the DNA repair enzyme Ku70, the surrogate light chains VpreB and λ5, as well as the terminal deoxynucleotidyl transferase (TdT), an enzyme thought to play a particularly important role in the generation of junctional diversity of B cell HC alleles. Table 1 Representative Gene Expression in Cells Undergoing Cre-Mediated Deletion of IgM a Each value represents the mean expression level from four independent experiments b Value of 20 is assigned to transcripts with no detectable expression by Affymetrix algorithms Surface BCR Expression Is Required to Maintain B Cell Developmental Stage Further examination of the microarray data revealed a large number of genes that were differentially expressed between Ctrl-M hi and Cre-M lo cells. A total of 327 transcripts met the following criteria: a 2-fold or greater difference in expression levels, expression value difference more than 200 units, and T-test p < 0.01. In order to better understand the significance of the gene expression changes observed between Ctrl-M hi and Cre-M lo cells, we prepared cRNA probes and performed microarray analyses on several additional sorted, control cell populations: (a) IgM − pro-B cells from 5-d IL-7 cultures of BM from nontransgenic C57Bl/6J mice (B6-M − ); (b) IgM hi cells from 5-d IL-7 BM cultures from hen egg lysozyme (HEL) Ig transgenic mice [ 19 ] (HEL-M hi ); (c) Hardy fraction (Fx) [ 20 ] “D” pre-B cells (B220 + CD43 − IgM − ) from normal Balb/c BM (FxD); and (d) Fx “E” immature B cells (B220 + IgM + IgD − ), also from normal Balb/c BM (FxE). To visualize the microarray data, expression values for each transcript were first converted to fold-difference ratios by dividing each value by the mean expression level for the IgM hi cell populations (Ctrl-M hi , HEL-M hi , and FxE). These ratios were then converted into log 2 space, and the data were analyzed by unsupervised hierarchical clustering. Visualization of the data showed that approximately 60% of the genes were overexpressed (red in Figure 2 B) in Cre-M lo relative to Ctrl-M hi or Cre-M hi cells, while approximately 40% were down-regulated (green in Figure 2 B). Strikingly, the Cre-M lo cells clustered with normal pre-B cells (FxD) and with surface IgM-negative cells (B6-M − ), rather than with the IgM + cell populations. This suggested the possibility that Cre-M lo cells had differentiated back to an earlier stage in B cell development as a consequence of losing IgM receptor expression. Of the genes that were differentially expressed between Ctrl-M hi and Cre-M lo , many are developmentally regulated during normal B cell differentiation ( Figure 3 and Table 1 ). In addition to the gene expression changes noted above, Cre-M lo cells showed up-regulated mRNA levels for CD43 and the IL-7 receptor (IL-7R), genes selectively expressed in pro-B cells. Cre-M lo cells also showed elevated levels of mRNA for the cell surface proteins insulin-like growth factor II receptor, CD98, Notch1, transforming growth factor β receptor, and the thrombin receptor; the intracellular signaling molecules serine-threonine kinase 3, protein phosphatase 2Cβ, Son of Sevenless 2, and mitogen-activated protein kinase; and transcription factors such as B lymphocyte-induced maturation protein (BLIMP), c-jun, sex determining region Y-box 4, lymphoid enhancer factor 1, myb, and elk4. Many of the genes expressed at high levels in Cre-M lo cells were also expressed at elevated levels in FxD and B6-M neg cell populations, compared with the lower expression levels found in the various IgM + populations ( Figure 3 ). Figure 3 Genes Differentially Expressed between Ctrl-M hi and Cre-M lo Cell Populations Shown are representative genes that were generally either (A) up-regulated or (B) down-regulated in Cre-M lo cells compared with Ctrl-M hi cells. See Figure 2 legend for details. A similar situation held for genes that were down-regulated in Cre-M lo cells compared to Ctrl-M hi . Transcripts for a large number of surface markers that characterize mature B cells (CD20, CD22, CD23, CD82, CD83, integrin β7, paired-Ig-like receptor-A3 and -B, Mac-2, CXCR5, Fcγ receptor IIB, and ICAM-1/CD54) were expressed at significantly lower levels in Cre-M lo than in Ctrl-M hi or Cre-M hi cells. Other genes down-regulated in Cre-M lo cells included those encoding class II major histocompatibility complex (MHC) molecules, intracellular signaling molecules (e.g., Vav, Src homology 2 phosphatase-1 (SHP-1), the inositol 1,4,5-trisphosphate receptor, phosphatidylinositol phosphate 5-kinase, MAP4K, mitogen- and stress-activated protein kinase 2, hemopoietic cell kinase, phosphodiesterase 7A, and protein kinase Cγ), and adaptor proteins (e.g., SH3-domain binding protein-2 and -5, and LNK). Finally, the mRNAs for several interesting nuclear proteins were also down-regulated in Cre-M lo cells, including the transcription factors Oct-2, class II transactivator (CIITA), and early growth response 1, as well as the cell cycle regulators cyclin D2 and cyclin-dependent kinase inhibitor 1a. Nearly all of the down-regulated genes in Cre-M lo cells were also expressed at low levels in B6-M − pro-B cells and FxD cells. From these data, we conclude that the B cell developmental program has shifted into reverse in immature B cells that have undergone inducible loss of surface IgM expression. Cre-M lo Cells Derive from Cre-M hi Cells It was important to rule out the possibility that the back-differentiation observed in Cre-M lo cells might be an artifact of selective expansion and/or survival of IgM − cells that were present at the initiation of the IFN cultures. Cell counts throughout the culture period revealed that the overall number of viable cells was not significantly different between control and BCR-deleting populations ( Figure 4 A). This was confirmed by similar annexin V and 7-amino-actinomycin D (7-AAD) flow cytometric staining profiles (box 4 (nd data not shown). CFSE [5(6)-carboxyfluorescein diacetate succinimidyl ester] labeling indicated that cells cultured with IFNαβ were not proliferating at significant levels ( Figure 4 B). We also isolated highly purified populations of IgM + immature B cells prior to incubation with IFN, and found that these cells similarly underwent the back-differentiation response ( Figure S1 . Thus, these experiments ruled out a selective expansion of pre-existing IgM − cells in the IFN-treated B1-8f/3-83κ/Mx-Cre cultures. Figure 4 Gene Expression Phenotype of Cre-M hi Cell Populations Is Intermediate between Ctrl-M hi and Cre-M lo Populations (A) Viable cell counts were performed in control B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre cell populations at the initiation of culture with medium alone (nil) or with IFNαβ (IFN) 1,000 units/ml, and daily for 3 d. Data are presented as percent of the day 0 cell numbers, and represent 4–7 experiments for each population. Standard errors were less than 10%, and are not shown. (B) BM cells were cultured for 5 d in IL-7, and were then labeled with CFSE and incubated with 1,000 units/ml IFNαβ or, as a proliferation control, 16 ng/ml IL-7 for an additional 3 d. Flow cytometric analysis of CFSE dye dilution of B220 + cells indicated no significant proliferation of IFN-treated cell populations. Numbers represent percent of gated cells. (C) A PCR-based assay was used to quantitate the extent of B1-8f deletion in Ctrl-M hi , Cre-M hi , and Cre-M lo populations. Ctrl-M hi cells contained 100% intact B1-8f alleles, Cre-M lo cells were 100% deleted, and Cre-M hi cells had 34 ± 16% average deletion of the B1-8f allele. Expression values for the Ctrl-M hi transcripts (significantly different between Ctrl-M hi and Cre-M lo ) were normalized to 1, and relative expression levels of transcripts up- ( n = 184) and down-regulated ( n = 143) in the three populations were calculated. In both groups of genes, Cre-M hi cells showed intermediate levels of gene expression between Ctrl-M hi and Cre-M lo , indicating that Cre-M lo cells originated from the Cre-M hi population. Additional strong evidence that Cre-M lo cells were derived from Cre-M hi cells was the observation that Cre-M hi cells showed an “intermediate” gene expression phenotype between Ctrl-M hi and Cre-M lo cell populations. As shown in Table 1 , the level of transcript expression for most of the genes in Cre-M hi cells was intermediate between Ctrl-M hi and Cre-M lo cell populations. Using a PCR-based assay (see Materials and Methods ), we determined that about a third of the cells in the Cre-M hi population (mean 34 ± 16%, range 14–48%) had undergone deletion of the floxed B1-8f allele ( Figure 4 C). When we compared the 327 transcripts that best discriminated Cre-M lo from Ctrl-M hi cells (see Figure 2 B), the “intermediate” gene expression phenotype of Cre-M hi cells was generally observed across the entire spectrum of genes ( Figure 4 C). The difference in gene expression between Cre-M hi cells and the other populations was highly statistically significant ( Figure 4 C). Reversal in Differentiation Is Confirmed by Analysis of Cell Surface Proteins The changes in gene expression in Cre-M lo cells were also reflected at the level of protein expression ( Figure 5 ). After 3–4 d of culture, flow cytometry demonstrated that Cre-M lo cells up-regulated surface expression of the activation antigens CD69 and CD86, as well as the IL-7R. Proteins down-regulated to varying degrees included B220, integrin β7, CD22, CD21/35, CD24, CD54, PirA/B, and Class II IA/IE. A few of the genes that showed differences at the mRNA level did not show measurable differences in protein expression (e.g., CD23). In nearly every case, the up- or down-regulation of surface protein expression mirrored the changes in mRNA expression. Importantly, the cells within each culture showed a similar cell surface phenotype, confirming that the back-differentiation was occurring across the entire population and not in just a minor subset. Additional control experiments supported the conclusion that back-differentiation of these cells was occurring in response to Cre-mediated loss of BCR expression, and not simply to elevated Cre levels (Figure S2 and S5 ). Figure 5 Protein Expression Confirms Reversal of Development in Immature B Cells Losing Surface IgM B1-8f/3-83κ control and B1-8f/3-83κ/Mx-Cre cell populations were harvested at the end of a 3- or 4-d culture with IFNαβ (3,000 units/ml), stained with mAbs for cell surface proteins, and analyzed by flow cytometry. Shown are the expression levels of B220-gated Ctrl-M hi (thick line) and Cre-M lo (thin line) cells at the end of the culture period. Essentially identical results were observed when Cre-M hi and Cre-M lo cells were compared (data not shown). Herbimycin A and Wortmannin also Induce Back-Differentiation of Immature B Cells To begin to explore the mechanisms by which basal IgM signaling maintains developmental stage, we first examined the influence of blocking tyrosine kinase signaling pathways downstream of the BCR with the inhibitor herbimycin A. For these experiments we used BM from mice carrying both a conventional Ig transgene (HEL-Ig) [ 19 ], and a bacterial artificial chromosome transgene containing green fluorescent protein (GFP) under the regulatory control of the Rag2 promoter/enhancer (Rag2-GFP) [ 14 , 21 ]. Previous studies have shown that the expression of GFP from this transgene reflects endogenous Rag2 protein expression in lymphocytes, but with a longer half-life [ 21 ]. Bone marrow from HEL-Ig/Rag2-GFP double transgenic (Tg) mice was cultured in IL-7 for 5 d, which generated a population of cells highly enriched for Rag2-GFP − , IgM + IgD − immature B cells ( Figure 6 A). Cells were then washed and incubated with the tyrosine kinase inhibitor herbimycin A or medium alone for 8 or 24 h. Herbimycin A induced strong Rag2-GFP expression, with about half of the cells GFP + at 24 h ( Figure 6 B). The observed GFP response was not due to nonspecific toxicity of herbimycin A treatment, since no GFP expression was observed in the dead or dying annexin-V + or 7-AAD + cells in the cultures (data not shown). Incubation of cells with other tyrosine kinase inhibitors (e.g., genistein and piceatannol) also resulted in significant induction of Rag2-GFP (L. E. Tze et al., unpublished data). BCR signaling and B cell development are highly dependent on phosphatidylinositol 3-kinase (PI3K) activity at the plasma membrane [ 22 ]. Incubation of HEL-Ig/Rag2-GFP immature B cells with the PI3K inhibitor wortmannin (30 μM) led to a strong induction of Rag expression at 24 h ( Figure 6 C). Thus, in a single cell assay for induction of Rag protein, interruption of proximal BCR signaling pathways with either herbimycin A or a PI3K inhibitor led to a derepression of Rag2 expression. Figure 6 Immature B Cells Undergo a Reversal in Development after Blockade of Basal Ig Signaling with Herbimycin A or Wortmannin (A) Flow cytometric profile of HEL-Ig/Rag2-GFP BM at the end of a 5-d IL-7 culture. Cells are electronically gated by forward and side light scatter, and represent all lymphoid cells in the culture. The majority of cells were Rag-GFP − IgM + IgD − immature B cells. (B) After a 5-d IL-7 culture, HEL-Ig/Rag2-GFP cells were washed and incubated with medium alone (containing DMSO carrier) or with 300 ng/ml herbimycin A for 8 or 24 h. Cells were then analyzed by flow cytometry, with cells gated based on size and annexin-V exclusion. Numbers represent percent of gated cells. (C) A representative experiment of Rag2-GFP expression levels in HEL-Ig/Rag2-GFP immature B cells incubated for 24 h with medium alone (black), herbimycin A (300 ng/ml) (green), or wortmannin (30 mM) (pink). (D and E) HEL-Ig/Rag2-GFP BM was cultured in IL-7 for 5 d, and cells were then incubated with herbimycin A (400 ng/ml) for 1 d. IgM + GFP + [GFPpos (D)] cells were sorted from herbimycin A treated populations, and IgM + GFP − [GFPneg (D)] cells were sorted from control cultures. In parallel, immature B cells were sorted from the BM of bcl-2 transgenic mice, and cultured with medium alone for 24 h (FxE Ctrl 1 and -2) or 48 h (FxE Ctrl 3 and 4), or with herbimycin A for 24 h (FxE HA-1, -2, -3, and -4) or 48 h (FxE HA-5 and -6). Total RNA was isolated from the cell populations and biotinylated cRNA probes were generated and hybridized to Affymetrix chips (E). All other cell populations were as described in Figure 2 . The array data were clustered using the 101 transcripts from the Ctrl-M hi /Cre-M lo gene list that were also differentially expressed between normal FxD (pre-B) and FxE (immature B) cells (using the following criteria: 2-fold or greater change in mean expression level, a greater than 200-unit difference in mean expression values, and Student's t-test p < 0.05). HC and LC transcripts were not used for this clustering analysis. The herbimycin A-treated immature B cells, both cultured and from bcl-2 Tg BM, cluster with more primitive cell populations. Data represent log 2 fold-differences, with individual expression values for each gene divided by the mean expression value of the Ctrl-M hi , HEL-M hi , FxE, GFP − , and FxE Ctrl populations. Representative genes are annotated. We then tested whether immature B cells treated with herbimycin A showed similar evidence for global back-differentiation as observed in Cre-M lo cells. HEL-Ig/Rag2-GFP BM was grown in IL-7 for 5 d, and cells were washed and incubated either in medium alone or with herbimycin A for 24 h. Total RNA was extracted from sorted control Rag2-GFP − (GFPneg in Figure 6 D) and herbimycin-treated Rag2-GFP + (GFPpos in Figure 6 D) cells and examined by microarrays. As shown in Figure 6 D and 6 E, the gene expression patterns of Rag2-GFP + cells coclustered with those of Cre-M lo , FxD, and B6-M − cells, consistent with a reversal in development of immature B cells after blockade of signaling pathways with herbimycin A. Next, we determined whether normal polyclonal immature B cells sorted from BM were capable of moving backward in development in response to loss of basal Ig signaling. This was important because it was possible that the ability of the cultured cells to back-differentiate might, in part, be a consequence of the accelerated development experienced by B cells carrying a pre-rearranged Tg Ig receptor [ 23 ]. Because of the increased sensitivity of immature B cells to apoptosis, we sorted immature B cells (B220 + IgM + IgD − ) to high purity (over 95%) from the BM of mice carrying a B cell-restricted bcl-2 transgene [ 24 ]. These immature B cells were then cultured for 1 or 2 d either in medium alone (FxE Ctrl) or with herbimycin A (FxE HA), then total RNA was harvested for microarrays. Strikingly, polyclonal immature B cells treated with herbimycin A showed a reversal in development similar to that observed for the in vitro cultured Ig Tg B cells ( Figure 6 D and 6 E), and clustered with cell populations that lacked an Ig receptor (B6-M − , FxD, Cre-M lo ). The polyclonal immature B cells did not appear to back-differentiate to quite the same degree as the cells carrying a prerearranged BCR. For instance, Rag and Ku70 genes were turned on in the polyclonal FxE cells in response to herbimycin A, similar to Cre-M lo cells, while TdT, VpreB, and λ5 were not ( Figure 6 ). We conclude that normal immature B cells undergo a reversal in development when treated with herbimycin A. New LC Rearrangements in Cells Undergoing Back-Differentiation Finally, we wished to explore the possibility that cells losing basal Ig signaling and showing subsequent induction of Rag might also lose LC allelic exclusion. GFP + cells were sorted from 24-h herbimycin A cultures of HEL-Ig/Rag2-GFP BM, and GFP − cells were sorted from parallel control cultures. Genomic DNA was isolated and assayed for new endogenous Ig LC rearrangements using quantitative PCR. We observed strong induction of new DNA rearrangements at both κ and λ LC loci following incubation with herbimycin A ( Figure 7 A). In addition, double-stranded DNA signal end breaks, which are intermediates of active Rag-mediated recombination, were strongly induced in GFP + herbimycin-treated cells ( Figure 7 B). Thus, the interruption of basal tyrosine kinase signaling pathways in immature B cells leads to the induction of the recombination machinery (e.g., Rag1 and Rag2) and new endogenous LC rearrangements. Figure 7 Immature B Cells That Lose Basal Signaling Show Induction of LC Rearrangements (A) PCR analysis of endogenous Ig light chain rearrangements (V-Jλ3, V-Jλ1, RS, and V-Jκ1) in genomic DNA of FACS-sorted HEL-Ig/Rag2-GFP BM cells incubated with medium alone or with 300 ng/ml herbimycin A for 24 h. IgM a+ GFP + cells were sorted from herbimycin A-treated cultures, and IgM a+ GFP − were sorted from control cultures. Data are from three independent experiments. CD14 is a loading control. −, negative control (C57Bl/6J tail DNA); +, positive control (C57Bl/6J spleen DNA). (B) Genomic DNA from the same cell populations described in (A) was subjected to ligation-mediated PCR to detect double-strand signal end DNA breaks at Jκ1. Controls in right three lanes of blot: H 2 O, control; −, negative control (S17 stroma); +, positive control (C57Bl/6 BM). (C) Genomic DNA was extracted from B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre immature B cells 3 d following incubation with IFNαβ. Quantitative PCR analysis was used to determine the fold-induction of LC rearrangements in B1-8f/3-83κ/Mx-Cre immature B cells treated with IFN compared to medium alone. Data represent the mean ± standard deviation of two (V-Jλ1), three (V-Jλ3), or four experiments (V-Jκ1, RS). Similar rearrangement assays were performed in cells undergoing Cre-mediated deletion of the B1-8f HC. We noted some “leakiness” of basal LC allelic exclusion in that culture system, with low levels of ongoing LC rearrangements even in the absence of deletion of the floxed Ig receptor (unpublished data). Nevertheless, cells losing BCR surface expression following Cre-mediated excision of the B1-8 HC showed a significant induction of LC rearrangements over background after 3 d of culture, ranging from 2- to 5-fold ( Figure 7 C). These data suggest that basal signaling through the Ig receptor in immature B cells is important for suppressing Rag gene expression and preventing further rearrangements at LC loci. Discussion We report here the surprising result that inducible deletion of the BCR from immature B cells using Cre–lox-mediated excision leads to a global movement of cells to an earlier stage in B cell development. Similar findings were observed in immature B cells incubated with the tyrosine kinase inhibitor herbimycin A or the phosphatidylinositol 3-kinase inhibitor wortmannin. Our interpretation of these data is that immature B cells actively maintain their stage in development by constitutive basal Ig signaling through protein tyrosine kinases (PTKs). Current models for signal transduction in lymphocytes envision a dynamic equilibrium between membrane-proximal protein tyrosine phosphatases (PTPs) and protein tyrosine kinases (PTKs) that serves to maintain cells in a resting state [ 25 ]. The BCR-associated molecules Igα and Igβ are phosphorylated on immunoreceptor tyrosine-based activation motifs (ITAMs) by the Src family PTK Lyn [ 26 , 27 ]. Phosphorylated ITAMs, in turn, lead to the recruitment and activation of the PTK Syk, which activates the Tec family PTK Btk. These kinases are counterbalanced by the activity of PTPs such as SHP-1 [ 28 ]. In the experiments presented here, incubation of immature B cells with the kinase inhibitors herbimycin A or wortmannin presumably shifts the equilibrium to one dominated by PTPs, resulting in an interruption of basal tyrosine kinase activity. Loss of tonic BCR signaling in immature B cells, either through deletion of the BCR or incubation with kinase inhibitors, led to rapid changes in gene expression, with the overall picture suggesting a reversal in differentiation to an earlier stage in B cell development. It is intriguing that many transcription factors known to be important in controlling B cell development were modulated by the interruption of basal signaling. For example, the proto-oncogene myb, known to positively transactivate the Rag2 promoter [ 29 ], was up-regulated in Cre-M lo cells, as was BLIMP, a transcriptional repressor that shuts off many B cell genes during terminal differentiation to plasma cells [ 30 ]. Oct-2, CIITA, and Egr-1, all known to transactivate B lineage genes, were strongly down-regulated in immature B cells following BCR deletion or treatment with herbimycin. Further dissection of these genetic pathways should provide new insights into the regulatory circuits that serve to actively maintain developmental stage in immature B cells. A number of control experiments were performed to rule out the possibility that the back-differentiation observed could have been artifactual due either to a selective expansion of IgM − cells present at the initiation of the cultures, or due to nonspecific effects of high-level Cre expression. First, in the Cre deletion experiments, cell counts, annexin V staining, and CFSE labeling demonstrated no evidence for a selective expansion or death of any subpopulation. Second, the Cre-M hi population, in which approximately a third of the floxed alleles had been deleted, showed an intermediate gene expression phenotype between Ctrl-M hi and Cre-M lo , indicating that the changes in gene expression were due to changes within the IgM + population, since these cells were sorted based on being IgM hi . Third, flow cytometry showed that cell surface protein expression was similar across the entire population of cells for most of the markers, and no small subpopulations of cells were observed that could have contributed to the changes in gene expression profiles. Fourth, at the end of the some of the IL-7 cultures, B cells were isolated to high purity (more than 99% of cells B220 + , IgM interm/hi ) before treatment with IFN. After culture with IFN, these cells showed efficient BCR deletion and evidence for back-differentiation (see Figure S3 ). Finally, to rule out the possibility that nonspecific Cre-mediated DNA breaks at cryptic genomic loxP sites could be inducing DNA damage that was, in turn, responsible for the developmental changes observed, we showed that Cre recombinase had no effect on developmental status when expressed at high levels in control B cells lacking a floxed BCR (Figures S4 and S5 ). Taken together, these experiments point to a robust back-differentiation of immature B cells upon interruption of basal Ig signaling in this model system. The Cre-M hi cells still expressed surface IgM, yet showed early evidence for the back differentiation response ( Table 1 ). One possibility to explain this finding is that Cre-M hi cells have lowered levels of the BCR in intracellular compartments (e.g., endoplasmic reticulum), and that some portion of the basal signal derives from intracellular BCR complexes. Alternatively, and perhaps more likely, the basal level of signaling provided by the knock-in BCR in this model may be very close to the threshold required for suppressing Rag and the back-differentiation program, and flow cytometry is insufficiently sensitive to detect the small decreases in BCR surface expression levels that accompany the early loss of the floxed HC allele. In support of this, we note that Rag1 levels are nearly 10-fold higher in sorted Ctrl-M hi cells (mean = 2,330 expression units) by microarray as compared to normal polyclonal immature B cells (FxE; mean = 243), and over 30-fold higher than conventional HEL-Ig Tg immature B cells (mean = 71). Similar differences were noted for Rag2. Suppression of endogenous LC rearrangements was also less efficient in the knock-in model as compared to the HEL system ( Figure 7 ). Thus, these data are consistent with the notion that the B1-8f/3-83κ knock-in receptor provides a relatively weak basal signal, and that cells bearing these receptors are sensitive to very small changes in BCR density. The idea that threshold signaling through the antigen receptor is required to suppress V(D)J recombination is supported by several observations in T and B cells. In cortical thymocytes, the interaction of the T cell receptor with MHC molecules down-regulates the expression of Rag1 and Rag2 and thereby shuts off further α chain rearrangements [ 31 , 32 ]. Interruption of this interaction with antibodies to the MHC resulted in up-regulation of Rag gene expression [ 33 ], indicating that the process is reversible in T cells. In an important series of experiments, Roose and colleagues [ 34 ] have recently shown that Jurkat T cells carrying mutations in proximal signaling molecules, including the adaptor molecules LAT and SLP76, have elevated Rag expression and altered gene expression programs. Using a panel of reconstituted Jurkat mutants and a variety of pharmacologic signaling inhibitors, the authors demonstrated that tonic signals through the ERK and Abl kinase pathways are important for suppressing Rag expression in Jurkat T cells and normal mouse thymocytes. B cells bearing a single copy of the anti-class I 3-83 HC and LC knock-in alleles showed poor LC allelic exclusion in mice lacking the cognate class I self-antigen (approximately 10% of B220 + cells idiotype-positive) [ 35 , 36 ]. Conversely, B cells from mice homozygous for both knock-in receptors (i.e., allelic inclusion) showed no evidence for additional LC rearrangements. One possibility is that the “dose” of receptor in mice carrying a single knock-in HC and LC allele did not provide sufficient basal signaling to suppress the recombination machinery, while two copies was sufficient. A similar Ag-receptor mediated suppression of Ig rearrangements was reported in mature IgD + peripheral B cells treated with LPS and IL-4 [ 37 ]. These data support the hypothesis that developing B cells require threshold signals from Ig receptors expressed on the cell surface to block further Ig gene rearrangements. The exquisite sensitivity of immature B cells to cell surface IgM basal signaling, as demonstrated by the early changes in gene expression in the Cre-M hi population ( Table 1 ), may be important for normal B cell development at several levels. First, this mechanism may serve as an important quality control mechanism. Cells that fail to rearrange a functional LC will continue to undergo rearrangements until an LC is produced that can be efficiently translated and stably expressed with HC at appropriate levels on the cell surface. This mechanism would also test the ability of HC and LC to pair efficiently and signal, ensuring that the B cell, once mature, can be activated in the periphery upon exposure to antigen. Second, this mechanism might be important for the selection of the naïve B cell V H /V L repertoire [ 38 ]. V L genes with strong promoters that drive high-level protein expression in immature B cells might be selectively recruited into the mature pool, while V L genes driven by weaker promoters might fail to provide sufficient basal signals and would be replaced by secondary rearrangements. Similarly, particular individual V H or V L chains, or V H V L pairs, with an enhanced ability to provide basal signaling might be relatively overrepresented in the repertoire. Finally, we believe that these data have important implications for understanding the regulation of receptor editing in immature B cells. It has been suggested that self-Ag induced receptor editing is driven by the crosslinking of surface Ig on immature B cells, which activates Rag proteins and ultimately results in new LC rearrangements [ 5 ]. An alternative hypothesis is that cells triggered to undergo receptor editing may do so primarily in response to the down-regulation of Ig receptors following engagement of Ig with antigen. Our earlier data showing a reciprocal relationship between surface Ig density of immature B cells and receptor editing following Ig crosslinking by soluble antigen [ 15 ], and data demonstrating receptor internalization during receptor editing in immature thymocytes [ 39 ], are consistent with this idea. We envision that some developing B cells at the immature stage will initially express high levels of the BCR, and only later during the immature stage encounter self-reactive antigens. These cells would receive a strong initial signal from self-antigen; however, the primary functional consequence of the self-antigen encounter would be the down-regulation of surface Ig via endocytosis. A sufficient loss of the basal signal would result in the re-initiation of Rag and induction of the back-differentiation program, which may be required for an efficient editing response. In support of this, we now have evidence that a back-differentiation program, similar to that observed in the models characterized by a loss of basal signaling, is initiated when immature B cells are incubated with specific antigen and undergo prolonged down-regulation of surface Ig (unpublished data). Together, these findings suggest that basal signals from the BCR control key developmental decisions at the pre-B to immature B cell transition. Materials and Methods Mice Rag2-GFP transgenic mice [ 21 ] were kindly provided by Dr. M. Nussenzweig (Rockefeller University, New York, United States), and were bred with HEL-Ig transgenic mice [ 19 ] to generate double-Tg animals. The Mx-Cre transgenic [ 18 ], B1-8f [ 16 ] and 3-83κ [ 17 ] mice were intercrossed to generate B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre animals. Each mouse used in these experiments carried a single copy of the knock-in HC and LC Ig alleles. Additonal breedings produced HEL-Ig/Mx-Cre and B1-8f/3-83κ/Rag2-GFP animals. Bcl-2 mice [ 24 ] were obtained from Jackson Laboratories (Bar Harbor, Maine, United States). All mice were maintained in specific pathogen-free conditions, and were generally 4–8 wk of age at the time of the experiments. IL-7 BM culture Single-cell suspensions of BM cells from HEL-Ig/Rag2-GFP, B1-8f/3-83κ, and B1-8f/3-83κ/Mx-Cre mice were prepared as described [ 15 ], and placed into IL-7 BM culture [ 40 , 41 , 42 ]. Cells were cultured at a concentration of 1.5–2 × 10 6 cells/ml in complete medium consisting of 1:1 RPMI 1640:EHAA (Mediatech, Washington, DC, United States, and Biofluids, Rockville, Maryland, United States, respectively), 10% heat-inactivated FBS (Life Technologies, Rockville, Maryland, United States), L-glutamine (BioWhittaker, Walkersville, Maryland, United States), penicillin and streptomycin (Mediatech), in the presence of 16 ng/ml recombinant murine IL-7 (R&D Systems, Minneapolis, Minnesota, United States) for 5–7 d. Cells were then washed and re-cultured in complete medium with either herbimycin A (Life Technologies) or wortmannin (Sigma, St. Louis, Missouri, United States) for 8 or 24 h, or with mouse fibroblast IFNαβ (Sigma) or recombinant IFNβ (Biosource, Camrillo, California, United States) for 1–4 d. For some experiments, IgM hi cells were purified at the end of the IL-7 culture using the MACs bead system (Miltenyi Biotech, Bergisch Gladbach, Germany). Recombinant TAT-Cre was used as previously described [ 43 ]. For CFSE labeling, cells were incubated in 1× HBSS (BioWhittaker) and 3 μM CFSE (Molecular Probes, Eugene, Oregon, United States) for 5 min at 37 °C, washed, and then cultured in the presence or absence of IFNαβ for 1–4 d. Flow cytometry and cell sorting Cells harvested at the end of IL-7, herbimycin A, wortmannin, or IFNαβ cultures were stained in FACS buffer (PBS containing 2.5% FBS and 0.2% sodium azide) with FITC-, PE-, CYC-, APC-, or biotin-conjugated monoclonal antibodies to B220, IgM, IgM a , IgM b , IgD a , CD21/CD35, CD22, CD23, CD24, CD54, CD62 l, CD69, CD86, IL-7Rα, integrin β7, PirA/B, and IA/IE (BD Pharmingen, San Diego, California, United States) or 3-83κ (S27 mAb). Staining with biotinylated antibodies was revealed by SA-PE or SA-APC (BD Pharmingen). In some staining conditions, annexin-V-PE and/or 7-AAD (BD Pharmingen and Calbiochem, San Diego, California, United States) were used to exclude dead cells. For cell sorting, cells treated with IFNαβ or herbimycin A were stained with IgM a -PE and B220-CYC in staining buffer (PBS containing 10% FBS) and sorted by FACSVantage (Becton Dickinson, Mountain View, California, United States). Flow cytometry analyses were performed using CellQuest (Becton Dickinson) and Flowjo (Treestar, San Carlos, California, United States) software. PCR analysis Genomic DNA was extracted from unsorted BM cells treated with IFNαβ for 3 d as described [ 15 , 44 ], and from sorted BM cells treated with herbimycin A for 1 d or with IFNαβ for 2 d using TRIzol (Life Technologies). Genomic DNA from the equivalent of 40,000 cells, or serial dilutions thereof, was then subjected to quantitative PCR amplifications using primers specific for CD14, and degenerate Vκ primers and a primer downstream of Jκ1 as described [ 45 , 46 ], with slight modifications for the V-Jκ1 PCR conditions: 45 s at 94 °C, 1 min at 63 °C, and 1.5 min at 72 °C for 27 or 30 cycles [ 15 ]. RS recombinations were detected with the following primers: forward primers VDEG1, 5′- GCGAAGCTTCCCTGATCGCTTCACAGGCAGTGG-3′; VDEG2, 5′- GCGAAGCTTCCCW GCTCGCTTCAGTGGCAGTGG-3′; and VDEG3, 5′- GCGAAGCTTCCCAKM CAGGTTCAGTGGCAGTGG-3′; and reverse primerRS3′, 5′- CTCAAATCTGAGCTCAACTGC-3′. The following cycling conditions were used: 45 s at 94 °C, 1 min at 64 °C, and 1 min at 72 °C for 27 or 30 cycles [ 23 ]. V-Jλ rearrangements were detected with: forward primer, 5′- AGGCTGTTGTGACTCAGGAATCTGCA-3′, and reverse primer (JN) 5′- ACTTACCTAGGACAGTGA-3′ or (J16) 5′- ACTCACCTAGGACAGT-3′ using the following cycling conditions: 30 s at 94 °C, 1 min at 62 °C, and 1.5 min at 72 °C for 30, 33, or 36 cycles [ 47 , 48 ]. PCR products were resolved on 1% or 1.5% agarose gels and visualized with ethidium bromide staining. Specificity of bands was confirmed by Southern hybridization of nitrocellulose blots with radiolabeled internal oligonucleotides. Double-stranded signal end DNA breaks at Jκ1 were identified as described [ 49 ]. The extent of deletion of the B1-8f allele was quantified using a PCR-based assay to amplify across the 3′ loxP site. TRIzol-extracted genomic DNA (5 ng) from sorted cell populations was amplified with: forward primer, 5′- GAAAGTCCAGGCTGAGCAAAACACCAC-3′; and a reverse primer conjugated with 6-FAM (Integrated DNA Technologies, Coralville, Iowa, United States), 5′- GGAGACCAATAATCAGAGGGAAGAATAATA-3′. The cycling conditions were 15 cycles of 15 s at 95 °C, 30 s at 55 °C, and 1 min at 72 °C, followed by another 12 cycles of 15 s at 89 °C, 30 s at 55 °C, and 1 min at 72 °C. PCR product for the wild-type allele was 295 bp in length, while the product from the B1-8f allele was 378 bp. The extent of deletion was calculated by the loss of the B1-8f allele. These results were confirmed using the same reverse primer with a forward primer upstream of the 5′ loxP site. The assay was validated using Ctrl-M hi cells (100% intact B1-8f allele) and Cre-M lo cells (100% deleted B1-8f allele). PCR products were resolved by electrophoresis on a 3100 Genetic Analyzer with GeneScan-500 ROX size standards (Applied Biosystems, Foster City, California, United States). Quantitation of product intensities was performed using GeneScan Analysis 3.7 (Applied Biosystems). Microarray gene chip analysis Total RNA from sorted BM cells treated with IFNαβ or herbimycin A was extracted using TRIzol (Life Technologies) and further purified using an RNAeasy kit (Qiagen, Valencia, California, United States). Total RNA was then subjected to double-stranded cDNA synthesis following the standard Affymetrix protocol (Expression Analysis Technical Manual P/N 700218 rev. 2, Affymetrix, Santa Clara, California, United States). FxD and FxE samples were collected from independent sorts (each sample approximately 2 × 10 5 cells) and total RNA extracted using TRIzol as described above. Total RNA was then subjected to two rounds of amplification to generate cRNA probes for hybridization, essentially as described [ 50 ]. Biotin-labeled cRNA probes (generated using High Yield RNA Transcript Labeling Kit, Enzo Diagnostics, Farmingdale, New York, United States) were chemically fragmented, hybridized to murine U74A and U74Av2 probe arrays (Affymetrix), and scanned at the University of Minnesota Biomedical Genomics Center facility. The U74Av2 probe arrays were subjected to U74A mask corrections, and all target hybridization intensities were scaled to 1,500 arbitrary units using Microarray Suite 5.0 (Affymetrix). Four independent sorts were performed for each of the following samples: Ctrl-M hi , Cre-M hi , and Cre-M lo ; two or three sorts were performed for all other samples. Data with probe-set detection p -values less than 0.1 were included in the statistical analysis, and individual probe-set measurements with detection p -values greater than 0.1, and/or signal intensity values 20 or less, were set to a value of 20. We performed all our statistical analyses using MS Excel (Microsoft Office X). To identify genes that were differentially expressed between the Ctrl-M hi and Cre-M lo populations, we used the Student's t-test analysis with the assumptions that the sample groups followed a two-tailed distribution and had unequal variance. For visualization of the arrays, each expression value was divided by the mean of the expression values for the relevant IgM + samples. These ratios were transformed into log 2 space, and subjected to centered-average linkage clustering using CLUSTER and visualized by TREEVIEW software [ 51 ]. Supporting Information Figure S1 Cell Surface Profile of B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre B Cells B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre B cells were collected at the end of 5-d IL-7 cultures, stained with labeled mAbs, and analyzed by flow cytometry. Profiles are compared with wild-type B6 mature IgM + B220 hi splenic B cells, and the various compartments of B cells in BM. Second row, IgM hi B220 hi recirculating mature B cells; third row, IgM + B220 interm immature B cells; and fourth row, IgM − B220 lo pro- and pre-B cells). (1.6 MB JPG). Click here for additional data file. Figure S2 Annexin V Staining of B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre B Cells After IFN Culture BM from control B1-8f/3-83κ and B1-8f/3-83κ/Mx-Cre mice was grown in IL-7 for 5 d. Cells were then washed, and treated with either medium alone or IFNαβ 1,000 units/ml for 1, 2, or 3 d. Cells were harvested and stained for B220, IgM a , and annexin V, and analyzed by flow cytometry. Shown are B220-gated cells. Similar results were obtained using 7-AAD in parallel to detect dead cells. Representative of five experiments. (2 MB JPG). Click here for additional data file. Figure S3 Back-Differentiation of Highly Purified IgM hi Cells To further address the issue of possible outgrowth or selective survival of IgM lo cells present at the initiation of the IFN cultures, we isolated purified populations of IgM hi (A) B1-8f/3-83κ and (B) B1-8f/3-83κ/Mx-Cre B cells at the end of the 5-d IL-7 cultures by first staining with predetermined optimal concentrations of biotinylated anti-IgM Abs, then isolating cells with streptavidin MACs beads. This reduced the IgM lo population to approximately 0.6% of the B220 + cells prior to the secondary IFN cultures (compare postcolumn with precolumn flow profiles). Purified cell populations were incubated with either medium alone or 5,000 units of IFN for 2 d on irradiated S17 feeder cells, and then analyzed by flow cytometry. Numbers refer to percent of viable gated lymphocytes. IFN induced efficient Cre-mediated deletion of the BCR, and the IgM lo cells in the cultures showed evidence for a reverse in differentiation as determined by elevated IL-7Rα expression and diminished CD22 expression levels. We noticed higher levels of deletion of the BCR in the medium-alone treated cells (in this experiment, 10.1%) compared with earlier experiments, likely due to crosslinking of IgM during the column purification step, and subsequent low level induction of Cre. There was approximately a 30% overall cell loss during the IFN culture and no cell proliferation (data not shown). We conclude that the IgM lo cells present after the IFN culture began as immature IgM hi cells. Representative of three experiments. (369 KB JPG). Click here for additional data file. Figure S4 TAT-Cre Treated HEL-Ig B Cells Do Not Back-Differentiate To rule out the possibility that the back-differentiation observed was occurring in response to DNA damage due to high level Cre, we used a TAT-Cre fusion protein that was previously shown to induce Cre-mediated deletion of floxed alleles [ 43 ]. (A) In initial experiments, we performed titrations with the fusion protein and found dose-dependent deletion of the BCR at 24 h in B1-8f/3-83κ/Rag2-GFP immature B cells. In cells that lost the BCR, there was a significant induction of Rag2-GFP reporter activity, and up-regulation of IL-7Rα (data not shown). These data demonstrate the potency of the TAT-Cre preparation in this system and provide a single-cell analysis for the induction of Rag2-GFP following BCR deletion. Cell counts were similar in TAT-Cre and mock-treated cultures. (B) The TAT-Cre fusion protein (200 μg/ml) was then introduced into HEL-Ig/Rag2-GFP immature B cells. Intracellular staining with anti-Cre antibodies in permeabilized cells confirmed that the TAT-Cre protein was introduced with good efficiency (note levels at both 2 and 24 h). B1-8f/3-83κ cells were tested in parallel for deletion of IgM (data not shown). Despite high intracellular levels of Cre, HEL-Ig immature B cells showed no evidence for back differentiation as determined by induction of Rag2-GFP, or up-regulation of IL-7R (data not shown). Representative of five independent experiments. (437 KB JPG). Click here for additional data file. Figure S5 HEL-Ig/Mx-Cre Immature B Cells Fail to Back-Differentiate following IFN Treatment HEL-Ig mice were bred with Mx-Cre animals to generate HEL-Ig/Mx-Cre double transgenic mice. IL-7 BM cultures were established from two mice, and after 5 d IgM + B cells were purified, washed, and then cultured for 2 d in 5,000 units/ml IFN. In parallel, a BM culture was established from a B1-8f/3-83κ/Mx-Cre mouse. After 2 d, cells were harvested, RNA was isolated, and cRNA probes were generated and hybridized to Affymetrix chips. Data analysis was performed identically to that described for Figure 2 . (A) CD22 levels were measured by flow cytometry for the various cell populations after IFN treatment. (B) Shown are mean gene expression data for Ctrl-M hi ( n = 4) and Cre-M lo ( n = 4) samples for selected genes (see Figure 2 and Table 1 ). Fold differences were calculated as mean Cre-M lo /mean Ctrl-M hi . Also shown are the average raw data from the IFN-treated HEL-Ig/Mx-Cre samples ( n = 2), and the parallel IFN-treated B1-8f/3-83κ/Mx-Cre sample. Fold differences were calculated as IFN-treated B1-8f/3-83κ/Mx-Cre/HEL-Ig/Mx-Cre. Overall, the gene expression patterns of the HEL-Ig/Mx-Cre samples closely mirrored the Ctrl-M hi samples and showed no evidence for back-differentiation. In the same experiment, analysis of the cell surface profile and gene expression in HEL-Ig/Mx-Cre cells incubated with a PI3K inhibitor showed the expected back-differentiation response (data not shown), indicating that the HEL-Ig/Mx-Cre cells were capable of the response. (C) Correlation coefficients were measured between the overall gene profiles of the two IFN-treated HEL-Ig/Mx-Cre samples compared with Ctrl-M hi and Cre-M lo mean values. Gene profiles of the IFN-treated HEL-Ig/Mx-Cre samples were significantly correlated with the Ctrl-M hi samples ( r = 0.655 and 0.637), and poorly correlated with the Cre-M hi samples ( r = 0.252 and 0.280), in the same range as Ctrl-M hi vs. Cre-M lo ( r = 0.284). (673 KB JPG). Click here for additional data file. Accession Numbers The Locuslink, or GeneID ( www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene , accession numbers of the genes and proteins discussed in this paper are λ5 (16136), Abelson murine leukemia (11350), B220 (19264), BLIMP (12142), Btk (12229), Burkitt lymphoma receptor 1/CXCR5 (12145), CD20 (12482), CD21/35 (12902), CD22 (12483), CD23 (14128), CD24 (12484), CD43 (20737), CD69 (12515) CD82 (12521), CD83 (12522), CD86 (12524), CD98 (20539), CIITA (12265), c-jun (16476), cyclin D2 (12444), cyclin-dependent kinase inhibitor 1a (12575), early growth response 1 (13653), elk4 (13714), Fcγ receptor IIB (14130), hemopoietic cell kinase (15162), ICAM-1/CD54 (15894), Igα (12518) and Igβ (15985), IGF2R (16004), IL-4 (16189), IL-7 (16196), IL-7R (16197), inositol 1,4,5-trisphosphate receptor (16438), integrin β7 (16421), Ku70 (14375), LAT (16797), LNK (16923), lymphoid enhancer factor 1 (16842), Lyn (17096), Mac-2 (16854), mitogen- and stress-activated protein kinase 2 (56613), mitogen-activated protein kinase (29857), MAP4K (26411), myb (17863), Notch1 (18128), Oct-2 (18987), paired-Ig-like receptor-A3 (18726), paired-Ig-like receptor-B (18733), phosphatidylinositol phosphate 5-kinase (18718), phosphodiesterase 7A (18583), PI3K (18708), protein kinase Cγ (18752), protein phosphatase 2Cβ (19043), Rag1 (19373), Rag2 (19374), serine-threonine kinase 3 (56274), sex determining region Y-box 4 (20677), SLP76 (3937), SH3-domain binding protein 2 (24055), SH3-domain binding protein 5 (24056), SHP-1 (15170), Son of Sevenless 2 (20663), Syk (20963), TdT (21673), thrombin receptor (14062), transforming growth factor β receptor (21812), Vav (22324), and VpreB (22362). The microarray data have been deposited at GEO ( http://www.ncbi.nlm.nih.gov/geo/ ; accession numbers are GSE2227.

D:\keerthana\PMC001xxxxxx\PMC1059451.xml

1059452

Two-Way Traffic in B Cell Development: Implications for Immune Tolerance

Development generally proceeds in one direction. Undifferentiated, pluripotent cells, which can become several different cell types, first of all become committed to restricted cell lineages. Then, under the control of developmental signals, committed cells gradually take on specialized characteristics, eventually producing mature, functioning cell types. To date, there has been little evidence to suggest that this process is ever reversed during normal development. Now, however, Timothy Behrens and his colleagues report that the development of B lymphocytes, the antibody-producing cells of the immune system, can be switched into reverse by blocking or removing basal immunoglobulin signaling activity from immature B cells. Their findings have important implications for our understanding of how the immune system is tailored to respond efficiently to foreign antigens while ignoring self antigens and thus avoiding harmful autoimmune reactions. B lymphocyte development, which occurs in the bone marrow, starts with the commitment of lymphoid progenitors to the B lineage and the somatic rearrangement of the heavy chain (HC) immunoglobulin (Ig) alleles. By stitching together diversity (D H ), joining (J H ), and variable (V H ) region DNA segments, many pro-B cells, each with a single but unique HC allele, are produced. Those cells in which the stitched-together HC allele encodes a functional protein undergo clonal expansion and proceed to the pre-B stage, before repeating the whole rearrangement process for the light chain (LC) Ig alleles. A productive LC rearrangement results in surface expression of IgM, which acts as the B cell receptor (BCR) for antigen for the immature B cell. During development, any B cells bearing strongly self-reactive Ig receptors are removed—this process is called tolerization—either by clonal deletion, by functional inactivation, or by receptor editing. In this last process, new LC rearrangements revise the antigen specificity of the receptor. Little is known about the mechanisms driving receptor editing, but these new data from Behrens and colleagues suggest that signals provided by surface BCRs might suppress receptor editing in immature B cells. To test this hypothesis, the researchers used a genetic system to remove the BCR from the cell surface of immature B cells in an inducible manner in vitro, and then compared gene expression patterns in these cells, control immature B cells, and pre-B cells. They discovered that the BCR-deleted cells had a gene expression pattern similar to that of pre-B cells, indicating that the BCR-deleted cells had gone back to an earlier stage of B cell development as a consequence of losing their BCR. The researchers saw a similar effect on B cell differentiation state when they blocked downstream signaling from the BCR by the use of the tyrosine kinase inhibitor herbimycin A or the phosphatidylinositol 3-kinase inhibitor wortmannin. Finally, the researchers showed that cells undergoing “back-differentiation” also restarted LC rearrangement or receptor editing. These data, suggest Behrens and co-workers, indicate that immature B cells actively maintain their developmental state by constitutive basal Ig signaling through protein tyrosine kinases. Their findings, they say, throw new light onto how receptor editing might be regulated in immature B cells in order to ensure that tolerance to self antigens develops. The researchers propose that when immature B cells encounter self antigens, down-regulation of surface Ig (BCR) via endocytosis might induce a back-differentiation program and thus induce an efficient editing response to the self antigen. Further experiments are now needed to show that these processes do indeed happen not just in vitro, as revealed here, but also in vivo.

D:\keerthana\PMC001xxxxxx\PMC1059452.xml

1062889

The Global Threat of Counterfeit Drugs: Why Industry and Governments Must Communicate the Dangers

The production of substandard and fake drugs is a vast and underreported problem, particularly affecting poorer countries. Cockburn and colleagues argue that the pharmaceutical industry and governments must both take action

Introduction The production of substandard and fake drugs is a vast and underreported problem, particularly affecting poorer countries. It is an important cause of unnecessary morbidity, mortality, and loss of public confidence in medicines and health structures. The prevalence of counterfeit drugs appears to be rising (see “The Scale of the Problem”) and has not been opposed by close cooperation between drug companies, governments, or international organizations concerned with trade, health, customs and excise, and counterfeiting. In this article we suggest that many pharmaceutical companies and governments are reluctant to publicize the problem to health staff and the public, apparently motivated by the belief that the publicity will harm the sales of brand-name products in a fiercely competitive business. Publicly, at least, several industry sources say the justification for secrecy is to avoid any alarm that could prevent patients taking their genuine medicines. We argue that this secrecy, and the subsequent lack of public health warnings, is harming patients and that it is also not in the long-term interests of the legitimate pharmaceutical industry. We urge a change to mandatory reporting to governmental authorities, which should also have a legal duty to investigate, issue appropriate public warnings, and share information across borders. This is not a role for the pharmaceutical industry, which has a serious conflict of interest. While some drug companies have given public warnings to protect patients, others have been criticized for withholding information and, in a recent development in the United States, taken to court for failing to act. The industry is addressing the problem. In 2003, US pharmaceutical companies made an agreement with the US Food and Drug Administration (FDA) that they would report suspected counterfeit drugs to the FDA within five days of discovery (see “Companies That Have Warned”), although this remains a voluntary arrangement. In many poorer countries, where the problem is at its worst, there are no similar governmental and industry initiatives. The Scale of the Problem It has been estimated that up to 15% of all sold drugs are fake, and in parts of Africa and Asia this figure exceeds 50% ([ 1 , 2 , 3 , 4 , 5 , 6 , 7 ]; R. Jones, FDA spokesperson, E-mail statement, 18 November 2004). The FDA estimates that fake drugs comprise approximately 10% of the global medicine market (R. Jones, FDA spokesperson, E-mail statement, 18 November 2004). This estimate suggests annual criminal sales in excess of US$35,000,000,000 [ 1 , 2 ]. The number of investigations of possible counterfeit drugs by the FDA has jumped from about five per year in the 1990s to more than 20 per year since 2000 ( Figure 1 ). Figure 1 The Number of Investigations of Possible Counterfeit Drugs by the FDA Has Been Rising (Figure: Margaret Shear, Public Library of Science, adapted from [ 39 ]) Most of the literature on fake drugs derives from local investigative journalism [ 6 , 8 , 9 , 10 , 11 , 12 , 13 , 14 ], with little scientific public health enquiry relative to the enormous scale of this criminal enterprise. The effects on patients of counterfeit medicines are difficult to detect and quantify, and are mostly hidden in public health statistics. The estimate of 192,000 patients killed by fake drugs in China in 2001 gives an indication of the scale of human suffering (see Sidebar). Secrecy and Counterfeit Medicines Most data on the epidemiology of counterfeit drugs are kept secret by the pharmaceutical industry and by governmental agencies. Drug companies employ investigators to track down and facilitate the shutting down of counterfeit industries, but this occurs very much in private. There are no reliable accessible databases whereby health workers or the public can access current details of which products are being faked in a locality. It is obviously correct that information on anti-counterfeiting strategies and the sources of undercover intelligence should not be released, but we believe that the information on what drug is being counterfeited, and where, should be public knowledge [ 1 ]. Government Reluctance Governments are also often reluctant to publicize problems with the quality of the drug supply in their country. This is reflected in the lack of action in much of the world regarding the problem of counterfeits, relative to their large impact on public health. The World Health Organization (WHO) has a reporting system and some of the information is publicly available [ 15 ]. The public information, crucially, does not include the country or region where the fakes were identified. However, the WHO has received no reports of counterfeit drugs from member countries after 2002, and it received only 84 reports between 1999 and 2002 [ 16 , 17 ]. In some countries, government officials have been accused of involvement in the false certification of counterfeit drugs, while in others, governmental agencies have been criticized for suppressing information [ 9 , 18 ]. The WHO in the Western Pacific region, an area severely affected by counterfeit drugs, is planning a rapid alert system for expediting the sharing of warnings and information between governments in the region. Pharmaceutical Industry Survey We wrote to the Pharmaceutical Security Institute (PSI) (see Box 1 ), which collates information on fake drugs obtained by the industry, asking whether they currently forwarded reports of counterfeit drugs to the relevant governments and the WHO. This question was not answered, but the PSI (in a letter dated 29 July 2003) informed us that, “Since its inception, it was recognized that a great deal of this information it [the PSI] contains would remain confidential and would not be disseminated. There is proprietary information that cannot be disclosed, either to peer member companies or to the general audience. Consequently, at this time the dissemination of information…is restricted and limited.” The letter added that the PSI encourages its members to report counterfeiting incidents to the appropriate authorities, and that it fully supports the voluntary reporting to the FDA. We also wrote to 21 major companies, of the more than 70 pharmaceutical companies with offices in the United Kingdom, asking for information on the companies' policies on what action should be taken and who should be told when one of their products was found to be counterfeited. We have received replies from six companies; one (Merck Sharp and Dohme) declined to give any information, while three (GlaxoSmithKline [GSK], Bristol-Myers Squibb, and Novartis) stated that they would inform the local drug regulatory authority if they were notified that one of their products was being counterfeited. Box 1. The Pharmaceutical Security Institute The PSI is a not-for-profit corporation formed by the major drug companies to collate their fake drug information to cooperate in fighting the racket. Based in Vienna, Virginia, United States, the PSI holds the only known comprehensive and updated source of fake drug information. The PSI Web site ( www.psi-inc.org ) states, “On a daily basis, many individuals unknowingly risk death or serious injury to their health by taking counterfeit pharmaceuticals.” But its databank, which health workers see as holding key information to prevent patients from taking life-threatening fakes, is not accessible to the WHO, health authorities, or the public. Such is the secrecy of the PSI's information, that access is restricted even between its member companies, which include the 15 largest drug manufacturers. Figure 2 Genuine and Fake Guilin Pharma Artesunate Blister Pack Holograms Found in Mainland Southeast Asia (A) is the genuine hologram attached to the blister packs of the genuine Guilin Pharma artesunate. The red arrow points to a legend stating “GUILIN PHARMA”, which is visible with the naked eye as a thin strip below the waves, but can only be read with a microscope (letters are about 0.1 mm high). (B) is a fake artesunate blister pack hologram: the upper red ring shows that the hologram has crescents, rather than a continuous blank line, between mountain and waves, and the lower ring shows that there is no “GUILIN PHARMA” legend. (C) is also a fake artesunate blister pack hologram: the red ring shows that the “GUILIN PHARMA” legend is present but the letters are of larger font than those on the genuine hologram and can be read with the naked eye (letters are about 0.3 mm high). A warning sheet giving more details and photographs is available in [ 47 ]. (Photos: Paul Newton, Wellcome Trust SE Asian Tropical Medicine Research Units) Paucity of Warnings about Fake Drugs That many pharmaceutical companies, professional organizations, and governments, both in developed and developing countries are not releasing warnings is manifested by the paucity of warnings relative to the scale of the problem. The industry's history of secrecy over data about fake drugs, and claims of a commercial motivation, go back over 20 years. In 1982, a spokesperson for the Association of the British Pharmaceutical Industry said, “It is difficult to declare a [fake drug] problem without damaging legitimate business” [ 13 ]. This impression of secrecy is supported by historical statements, such as the following: “The Society [Royal Pharmaceutical Society of Great Britain] is not issuing press releases [about counterfeit drugs] because it believes that as much as possible should be done behind the scenes…and that no great publicity should be sought because it could damage public confidence in medicines” [ 19 ]. But the Royal Pharmaceutical Society of Great Britain has recently revised its position. David Pruce, Director of Practice and Quality Improvement for the organization, told us (E-mail letter, 14 February 2005), “If there is a risk that a patient has been dispensed a counterfeit medicine, then it is vital that they are informed. There have been two recent cases in Great Britain where counterfeit medicines appeared in the legitimate pharmacy supply chain. The public announcement of the problem of the counterfeit medicines was therefore entirely proper and necessary.” He added, “It is important that news stories of this type are handled responsibly so that the public's confidence in their medicines is not undermined. This could deter patients from taking genuine medicines.” Recent Examples of Counterfeit Drugs Approximately one-third to one-half of packets of artesunate tablets, the pivotal, life-saving anti-malarial drug, recently bought in Southeast Asia were fakes, containing no active ingredient at all. A nongovernmental organization in a Southeast Asian country bought 100,000 inexpensive “artesunate” tablets only to find that they were counterfeit [ 7 , 39 ]. See Figure 2 for examples of fake artesunate being sold in mainland Southeast Asia. A total of 192,000 Chinese patients are reported to have died in 2001 from fake drugs, and in the same year Chinese authorities “closed 1,300 factories while investigating 480,000 cases of counterfeit drugs worth 57 million USD” [ 12 ]. In 2004, Chinese authorities arrested 22 manufacturers of grossly substandard infant milk powder and closed three factories after the death of over 50 infants [ 40 ]. In North America, counterfeit atorvastatin [ 41 ], erythropoietin [ 41 ], growth hormone [ 33 ], filgrastim [ 33 , 41 ], gemcitabine [ 36 , 37 ], and paclitaxel [ 36 , 37 ] have been reported recently. Nigeria recently threatened to ban the import of all drugs from India, a major supplier, because of the high prevalence of counterfeits amongst the imports [ 42 ]. In Haiti, Nigeria, Bangladesh, India, and Argentina, more than 500 patients, predominantly children, are known to have died from the use of the toxin diethylene glycol in the manufacture of fake paracetamol syrup [ 43 , 44 , 45 ]. During the 1995 meningitis epidemic in Niger, the authorities received a donation of 88,000 Pasteur Merieux and SmithKline Beecham vaccines from neighbouring Nigeria. The drugs were found to be counterfeit, with no traces of active product. Some 60,000 people were inoculated with the fake vaccines [ 24 ]. The recent discovery of counterfeit antiretrovirals (stavudine-lamivudine-nevirapine and lamivudine-zidovudine) in central Africa [ 46 ] raises the prospect of a disastrous setback in the treatment of AIDS in sub-Saharan Africa, unless vigorous action is taken now. This assessment, that the dangers of causing alarm amongst the general public could outweigh the benefits of disclosure, remains widespread in public statements. A spokesperson for the Association of British Pharmaceutical Industries, Marjorie Syddall, wrote (E-mail letter, 20 October 2003), “A company should be completely satisfied that a medicine is counterfeit before informing the authorities, but more importantly still, before it makes this information known to the public—so that no unnecessary alarm is caused.” Commercial Motivation—“Cut-Throat Competition” Chris Jenkins, a founding member of the PSI, now Associate Director of Pinkerton Consulting and Investigations, told us (E-mail statement, 9 December 2004), “It is necessary to keep fake drug information confidential for commercial reasons…to avoid media leaks and to prevent the possibility of rival drug companies taking unfair commercial advantage of a victim company.” He explained, “At the outset, we [the PSI] were against having data online that anyone could interrogate…If a patient came to harm as a result of a counterfeit product, the company's good reputation is in danger of disappearing, together with a loss of confidence in the products… The one thing we were trying very hard to do was to keep it [data] out of the hands of the commercial people in any of the companies…The importance of meeting sales' targets is such that you can even find cut-throat competition between different operating divisions of the same company, let alone between two companies competing in the same market with similar drugs.” The WHO 1999 guidelines for the development of measures to combat counterfeit drugs states that “the reluctance of the pharmaceutical industry, wholesalers and retailers to report drug counterfeiting to the national drug regulatory authorities could impede the national authorities from successfully taking measures against counterfeiting”, and suggests “the compulsory reporting to the relevant authorities of any incidents in which counterfeits are detected or involved” [ 20 ]. A recent review of the law and counterfeit drugs calls for the “eradication of the clandestine status of records and counterfeit drug information” [ 21 ]. At the International Conference of Drug Regulatory Authorities in Madrid in February 2004, it was stated by the WHO that “the drugs industry had a great deal of data but was ‘very reluctant to make them available’” [ 17 ]. Information Strictly Confidential In the US it was reported that it had been “very difficult to obtain citable factual information about the extent of the problem of counterfeit drugs. Drug companies keep the information they have strictly confidential” [ 22 ]. In 1989, the British Department of Health and Glaxo (now a part of GlaxoSmithKline) were criticized for not publicizing information about the discovery in Britain of fake Glaxo Ventolin asthma inhalers. London's The Times obtained the fake Ventolin's licence and batch numbers for a story, prompting the release of the information. Warning letters, drafted by Glaxo and the Department of Health, were sent to all 14,000 pharmacists in Britain five weeks after the fake's discovery [ 8 ]. In 1998, the company Schering do Brasil was accused of keeping secret the discovery of oral contraceptive pills made of wheat flour for 30 days while they carried out their own investigation [ 23 ]. According to the Far Eastern Economic Review , the company was fined US$2.5 million by the Brazilian government [ 6 ]. Schering do Brasil informed us (E-mail letter, 17 February 2005) that “Federal Justice cancelled the fine in 2002 after the company appealed”. In Niger, in 1995, one of the fake meningitis vaccines originating from Nigeria was labelled as made by SmithKline Beecham, but Le Monde reported that the company did not act against the counterfeiters, afraid that it might damage trade [ 24 ]. Fake Paediatric Anti-Malarial Drugs The need to release fake drug information is acute in Africa, where a resurgence of malaria is killing an estimated one million people a year, the vast majority of them children under five [ 25 ]. One example highlights the problems encountered. One of us (K. Agyarko) found counterfeits of the GSK paediatric anti-malarial syrup halofantrine (Halfan) in August 2002 in Ghana. That month he prepared a public health warning. Agyarko and his deputy told the BBC [ 26 ] that he also alerted GSK's Ghana agent, who visited him with staff from GSK's London headquarters and took away samples of the fake Halfan. Agyarko publicly stated (on 23 September 2002, at the First Global Forum on Pharmaceutical Anticounterfeiting in Geneva, Switzerland) [ 26 ] that he was asked by GSK to withhold his public warning because it would “damage” their product. After his meeting with GSK, no warning was issued. In a written statement (E-mail letter, 24 October 2003), GSK denied receiving Agyarko's fake Halfan alert and said the company was “not provided with any samples of fakes by the authorities in Ghana”. After a year of enquiries, resulting in a BBC Radio programme (BBC Radio 4, “File on 4”, 5 October 2004) [ 26 ], GSK reversed its position and said that its local agent had “bumped into” Agyarko and had received his alert and samples of fake Halfan syrup. In a new statement (E-mail letter, 5 October 2004) GSK said: “At no point was any pressure put on the Ghanaian authorities not to issue a public warning on fake Halfan.” GSK's vice president of communications, Louise A. Dunn, told us (E-mail letter, 6 October 2004), “There was some confusion over the interactions with Mr Agyarko. The key point here is that there was no wrong doing…” However, the Ghana incident needs to be viewed in the context of the wider illegal trade in fake Halfan syrup identified in West Africa, and GSK's reluctance to give us details about this trade. We asked GSK whether it had issued any public warnings about fake Halfan syrup, but the question was not answered. The only reference to counterfeit halofantrine syrup that we have been able to find in the public domain was published in a specialist technical journal that described the mass spectroscopy analysis of fake halofantrine syrups by the GSK Medicines Research Centre [ 27 ] and demonstrated that the fake syrups contained two potentially harmful sulphonamide drugs, but no halofantrine. We wrote to GSK (letter, 20 June 2003) asking when and where discoveries of fake Halfan were made, and whom GSK had informed about them. GSK told us only that “counterfeit Halfan is present in Nigeria and Sierra Leone” (letter, 21 July 2003). It gave no details of preparation type or discovery dates. Fake GSK Halfan syrup was discovered in Nigeria in June 2002 by the Nigerian National Agency for Food and Drug Administration and Control. NAFDAC alerted GSK and issued a public health warning in June 2002 in the regular NAFDAC fake drug bulletin [ 28 ], giving the fake Halfan syrup's identifying details. The NAFDAC's Dora Akunyili told BBC Radio (5 October 2004): “It is more dangerous not to alert the public. We will still issue a warning even if we find it in only one shop. If you find any fake drug product in only one shop you can be sure it is in many villages…We don't defend companies. We are defending the people” [ 26 ]. Figure 3 Poster Advertising the Second Global Forum on Pharmaceutical Counterfeiting (Figure: Ian Lancaster, Reconnaissance International) The Pharmaceutical Board of Sierra Leone, which handles fake drug cases, was not informed by GSK of any discoveries of fake GSK Halfan syrup, according to its director Michael J. Lansana (E-mail letter, 21 January 2004), although it did receive a report of counterfeit adult Halfan caplets from GSK. Later, GSK told us (E-mail letter, 6 October 2004) the fake Halfan syrup it had tested was found in Sierra Leone in late 2001, and that it had informed Sierra Leone's Minister of Health and Sanitation of the find. Only a single report of counterfeit halofantrine, which does not specify details of preparation type or location, is given in the WHO Counterfeit Drug Reports for 1999–October 2000 [ 15 ]. Cross-Border Threats and Cooperation The fake Halfan syrup cases highlight the importance of communication and cross-border cooperation, and the need for industry and governments to inform neighbouring countries when a fake is found. The global distribution and the scale of the racket in fake adult Halfan capsules was clear in December 2000, when Belgian customs seized 57,600 packs of fake GSK Halfan capsules (and 4,400 packs of fake GSK Ampiclox [ampicillin] and 11,000 packs of fake GSK Amoxil [amoxicillin]) en route from China to Nigeria. The counterfeiters in China were found to be preparing to export 43 tons of 17 brands of drugs from seven international pharmaceutical companies [ 29 ]. Companies That Have Warned Sometimes pharmaceutical companies have publicized information to alert health workers and patients and governments to the dangers of counterfeited or tampered products. For example, Johnson and Johnson, Serono, Hoechst, Wellcome Foundation (now part of GSK), GSK, and Genentech have publicized information on their drugs that have been counterfeited or tampered with. In 1982, cyanide-laced paracetamol killed seven people in the US. The pharmaceutical company whose product had been tampered with, Johnson and Johnson, issued alerts and cooperated with the investigation, and although the financial cost to the company was large, its long-term reputation was probably enhanced. Other companies, at least initially, did not take advantage of the disaster for their own financial gain [ 30 ]. In 2002, Johnson and Johnson issued 200,000 letters to health-care professionals in the US warning them of fake Procrit (erythropoetin) within one week of being notified of a severe counterfeit problem [ 31 ]. In 1982, Hoechst voluntarily took out magazine adverts in Lebanon to warn pharmacists and customers of a fake of its drug Daonil (glibenclamide) for the treatment of diabetes mellitus [ 13 ]. In 2001, Serono was told by the FDA to issue a public warning to hospitals, clinics, and patients in seven US states after the discovery of a counterfeit of its drug Serostim, a human growth hormone used in the treatment of AIDS and other conditions [ 32 ]. In 1984, in Thailand, the Wellcome Foundation (now part of GSK) publicized the discovery of fakes of its antibiotic Septrin (co-trimoxazole) that lacked any active ingredients, and the company's efforts to stop its production. Wellcome also had reports that the fakes were being imported into the UK, which it made public along with the warning that it sent to the British Embassy in Bangkok [ 14 ]. In 2001, GSK made public the discovery of fakes of its AIDS treatment Combivir (zidovudine + lamivudine) [ 32 ], and Genentech publicized information on fakes of Neupogen (filgrastim) [ 33 ]. The Pharmaceutical Research and Manufacturers of America announced in April 2003 that, from 1 May 2003, its 60 members would voluntarily report to the FDA “within five working days of determining that there is a reasonable basis to believe their product has been counterfeited” [ 34 ]. This is an important local development but it should be mandated by law and become a global standard. Indeed, we have not found one country where drug companies have a legal duty to report discoveries of counterfeits of their products to public health or trade authorities. The Sharing of Information on Counterfeit Medicines We suggest that the pharmaceutical industry, which is such a benefit to our health, is harming both patients and itself by not vigorously warning the public of fake products when they arise. Apart from the moral imperative, there is the prospect of growing legal pressure on drug companies to take responsibility for fakes of their products. In Britain, there are proposals to introduce a charge of “corporate killing” for companies who have contributed to the deaths of customers [ 35 ] that could also apply to drug companies if they do not take reasonable steps to warn the public of a fake product. Drug Companies Sued in the US Already, the US has seen the first court case brought against two drug companies for allegedly failing to act to protect customers over a fake drug discovery. In 2002, a Kansas City pharmacist was jailed for diluting the anticancer drugs Gemzar (gemcitabine) and Taxol (paclitaxel). The victims and dead patients' families sued the drug companies, Eli Lilly and Myers Squibb, for not taking steps to stop him. The companies argued that they had no duty to protect the plaintiffs from the pharmacist's criminal acts, but a newspaper reported that Eli Lilly and Myers Squibb settled out of court, apparently for US$72 million, avoiding a legal precedent that would hold drug companies liable for not disseminating such information [ 36 , 37 ]. Chris Jenkins suggests that the PSI could face a legal challenge to open its fake drug databases (E-mail, 9 December 2004): “Only the PSI had an overview of the known racket…In theory, every fake drug case reported by the companies should be on there.” He is concerned that private investigators could be liable for fake drug data they obtain for client companies. Governments Must Enforce a Legal Responsibility We believe that the industry, along with pharmacists, health workers, and governments, needs to extend the “behind the scenes” fight against fakes to a public collaborative approach with a legal responsibility to report suspected counterfeits to drug regulatory authorities, in a similar way to the reporting of “notifiable” infectious diseases. The drug regulatory authorities, accountable to the consumers of drugs, should have a statutory duty to investigate and disseminate the information, with the interests of patients as the prime concern. Drug regulatory authorities in economically poor countries will need additional financial support. We recognize that false information could seriously damage a company and that information should be verified and used prudently. We also recognize that careful public information measures will be needed to prevent patients from stopping the use of genuine products, but suggest that this is possible as pharmaceutical companies can, and have, alerted the public in collaboration with government agencies (see above). However, the decision to warn the public should not be made by the pharmaceutical industry alone, which has a serious conflict of interest. We believe that the long-term interests of both the industry and patients are best served by more openness and social responsibility to public health. Company staff and shareholders should not be in a position to adjudicate conflicts between commercial gain and public health—such adjudication should be in the hands of government departments accountable to the public. Aviation Industry Model The UK Civil Aviation Authority provides a model: suspected unapproved aircraft parts must, by law, be reported to it [ 38 ]. When a report of a counterfeit drug is confirmed, the drug regulatory authorities should be responsible for assessing the public health importance of the information and deciding when and how to alert the country's police, trade, customs authorities, and public, and also the drug regulatory authorities of other countries that may be affected, with the assistance of Interpol as required. If a drug regulatory authority is confident, for example, that the fake drug has been intercepted before it has reached the pharmacies, a public alert may not be necessary. The “confusion” reported in the GSK Halfan syrup case also illustrates the great importance for both companies and government departments to keep a secure paper trail of information so that it is clear what has happened and when. The pharmaceutical company is also a victim of the counterfeiter and should be supported by governmental authorities if it reports promptly. Individuals who report information on counterfeit drugs should remain anonymous and be protected from the criminal counterfeiting underworld, which may exact retribution. International agreements between companies to avoid taking advantage of competitors' misfortunes, when precipitated by rumors or confirmed reports of fake drugs, may facilitate enhanced cooperation within the pharmaceutical industry. International Convention against Counterfeit Drugs The Madrid meeting in 2004 considered a proposed international framework convention on counterfeit drugs, presented by the WHO, to promote international cooperation and the exchange of information [ 17 ]. If enacted this could be a very important contribution to improving drug quality. The effective control of the global epidemic of counterfeit and substandard drugs will not be easy, and will need a multifaceted approach: the provision of effective, available, and inexpensive drugs; the enforcement of drug regulation; more openness by governments as to the scale of the problem; more effective police action against the counterfeiters and those who may be corrupt allies within government and industry; enhanced cooperation between the industry, police, customs, and drug regulators; and enhanced education of patients, drug sellers, and health workers [ 4 , 5 , 20 ]. We urge the industry and governments to act, through the sharing of crucial public health information, to facilitate the protection of patients and improve the quality of an apparently deteriorating drug supply. Counterfeit Drug Conference in Paris On 15–17 March 2005, the Second Global Forum on Pharmaceutical Anticounterfeiting will convene in Paris, where representatives of the major pharmaceutical companies, governments, medical and scientific professionals, law enforcement agencies, nongovermental organizations, and private investigators will meet to discuss the growing problem that threatens patients and the pharmaceutical industry ( Figure 3 ). A collection of counterfeit pharmaceutical drugs seized by the NAFDAC in Nigeria (Photograph: NAFDAC/International Chamber of Commerce Counterfeiting Intelligence Bureau)

D:\keerthana\PMC001xxxxxx\PMC1062889.xml

1064074

The Na+–H+ exchanger-1 induces cytoskeletal changes involving reciprocal RhoA and Rac1 signaling, resulting in motility and invasion in MDA-MB-435 cells

Introduction An increasing body of evidence shows that the tumour microenvironment is essential in driving neoplastic progression. The low serum component of this microenvironment stimulates motility/invasion in human breast cancer cells via activation of the Na + –H + exchanger (NHE) isoform 1, but the signal transduction systems that underlie this process are still poorly understood. We undertook the present study to elucidate the role and pattern of regulation by the Rho GTPases of this serum deprivation-dependent activation of both NHE1 and subsequent invasive characteristics, such as pseudopodia and invadiopodia protrusion, directed cell motility and penetration of normal tissues. Methods The present study was performed in a well characterized human mammary epithelial cell line representing late stage metastatic progression, MDA-MB-435. The activity of RhoA and Rac1 was modified using their dominant negative and constitutively active mutants and the activity of NHE1, cell motility/invasion, F-actin content and cell shape were measured. Results We show for the first time that serum deprivation induces NHE1-dependent morphological and cytoskeletal changes in metastatic cells via a reciprocal interaction of RhoA and Rac1, resulting in increased chemotaxis and invasion. Deprivation changed cell shape by reducing the amount of F-actin and inducing the formation of leading edge pseudopodia. Serum deprivation inhibited RhoA activity and stimulated Rac1 activity. Rac1 and RhoA were antagonistic regulators of both basal and stimulated tumour cell NHE1 activity. The regulation of NHE1 activity by RhoA and Rac1 in both conditions was mediated by an alteration in intracellular proton affinity of the exchanger. Interestingly, the role of each of these G-proteins was reversed during serum deprivation; basal NHE1 activity was regulated positively by RhoA and negatively by Rac1, whereas RhoA negatively and Rac1 positively directed the stimulation of NHE1 during serum deprivation. Importantly, the same pattern of RhoA and Rac1 regulation found for NHE1 activity was observed in both basal and serum deprivation dependent increases in motility, invasion and actin cytoskeletal organization. Conclusion Our findings suggest that the reported antagonistic roles of RhoA and Rac1 in cell motility/invasion and cytoskeletal organization may be due, in part, to their concerted action on NHE1 activity as a convergence point.

Introduction Tumour invasion and metastasis associated with neoplastic progression are the major causes of cancer deaths. The invasive process occurs through a complex series of interactions with the host tissue, resulting in infiltration and penetration of normal tissue by cancer cells [ 1 ]. Recent advances have highlighted the importance of the acid component of the tumour microenvironment in driving invasive capacity and subsequent malignant progression [ 2 - 4 ]. The activity of the Na + –H + exchanger (NHE) isoform 1 is known to play a role in acidifying the tumour microenvironment [ 5 ], and the nutrient-deprived conditions that are common to the tumour microenvironment activate tumour cell NHE1, which in turn stimulates increased motility and invasive capability [ 6 ]. This role played by Na + –H + exchanger activity in driving tumour cell motility has been corroborated in transformed renal cells [ 7 , 8 ] and in ascites hepatoma cells [ 9 ], emphasizing the importance of understanding the mechanisms that control Na + –H + exchanger activity in tumour cells. The Rho family of small GTPases has been shown to play a major role in regulating the rearrangement of the actin cytoskeleton in response to cell stimulation [ 10 ] and to be involved in the regulation of a variety of other cellular processes, such as organization of the microfilamental network, cell–cell contact, motility and apoptosis [ 11 ]. Early studies in fibroblasts suggested that Cdc42, Rac1 and RhoA are organized in hierarchical cascades in which activated Cdc42 activates Rac1, in turn activating RhoA [ 10 ]. It is now known that these G-proteins can be organized in many other ways. For example, Rac1 or Cdc42 have been shown to oppose or modify RhoA action in the regulation of a number of processes, including actin cytoskeleton remodelling [ 12 ], axon guidance [ 13 ], dendrite branching [ 14 ] and cell migration [ 15 - 17 ]. Thus, it is clear that the interactions and associations of the various members of the Rho family are cell/tissue-specific or function-specific, or both. Despite the evidence that Rho GTPases are overexpressed in tumours [ 18 ] and play an important role in mitogenesis, proliferation and invasiveness [ 19 ], our understanding of how they interact among themselves in tumour cells to regulate these processes is still incomplete and represents an important issue in oncology [ 20 ]. Molecular control of actin organization is probably at the core of cell motility, and it is of importance that we gain an understanding of the mechanisms that underlie the regulation of actin dynamics so that we may appreciate how motility is regulated. Recent studies conducted in smooth muscle cells have led to a model (for review see [ 21 , 22 ]) that associates RhoA with contraction via activation of Rho kinase (ROCK) and ROCK-dependent phosphorylation of myosin phosphatase, thereby inactivating it and resulting in an increase in the phosphorylation state of myosin light chain and enhancement of myosin binding to actin filaments. Rac1 via activation of the p21-activated kinase (PAK) antagonizes this process by blocking the activity of myosin light chain kinase. As recently discussed [ 23 ], this mechanism pertains to smooth muscle cell contraction and cannot fully explain the effects of Rho family proteins on the actin cytoskeleton in other cell types, such as epithelial cells. Two other mechanisms have been described in which Rac1 downregulates RhoA activity via a redox-dependent mechanism [ 17 ] and by stimulating RhoA degradation via Smurf1 [ 24 ]. Recently, however, a further possible cross-regulatory mechanism has emerged. Pioneering studies in fibroblasts have shown that NHE1, a regulator of intracellular pH (pHi), can play a direct role in controlling actin dynamics and subsequent motility through a protein–protein interaction with the cytoskeletal adaptor protein ezrin, and that, in those cells, RhoA-dependent modulation of cytoskeletal dynamics and motility occurred via direct regulation of NHE1 activity [ 25 , 26 ]. There is increasing evidence that activity of the NHE1 is essential for motility in various cell types [ 6 - 9 , 26 ]. Accordingly, the questions are whether RhoA and Rac1 reciprocally regulate motility in tumour cells of epithelial origin, and if so then do they act via a coordinated regulation of NHE1 activity? We previously reported that serum deprivation, a common component of the tumour microenvironmental, stimulates NHE1 in human epithelial breast cancer cells and drives increased cellular motility and invasive ability via the activated NHE1 [ 6 ]. In light of this essential role played by NHE1 in regulating motility in these cells, the present study was undertaken to characterize the role and pattern of regulation by Rho GTPases of this serum deprivation-dependent activation of both NHE1 and subsequent basic invasive characteristics, such as pseudopodia and invadiopodia protrusion, directed cell motility and penetration of normal tissues [ 27 , 28 ]. The study was conducted in a well characterized human mammary epithelial cell line that represents a late phase in metastatic progression, namely MDA-MB-435 [ 6 ]. We observed that serum deprivation inhibits RhoA activity and stimulates Rac1 activity and, using dominant negative and constitutively active mutants, that Rac1 and RhoA are antagonistic regulators of tumour cell NHE1 activity. As was observed in the regulation of this phenomenon by phosphoinositide-3 kinase (PI3K) [ 6 ], a reversal of RhoA and Rac1 regulatory action on NHE1 activity found in serum replete conditions was found during serum deprivation-dependent upregulation of NHE1. Although serum deprivation reversed the regulatory actions of Rac1 and RhoA on NHE1 activity, the basic pattern of antagonism of action between these two G-proteins was maintained. Furthermore, the same pattern of RhoA and Rac1 regulation found for NHE1 activity was observed in both basal and serum deprivation-dependent increases in cell invasion and motility. Together with the recent reports demonstrating the direct role of NHE1 in controlling actin dynamics in fibroblasts [ 25 , 26 ], our findings suggest that the reported antagonistic roles of RhoA and Rac1 in cell motility and cytoskeletal organization may also be due, in part, to their concerted action on NHE1 activity as a convergence point. Methods Cells and construction of expression vectors containing RhoA mutants MDA-MB-435 cells were cultured as previously described [ 6 ]. Minus serum growth medium was complete Dulbeccos modified Eagle's medium (DMEM; GibcoBRL, Milano, Italy) without serum, in which the osmolarity, if necessary, had been adjusted to be identical to the plus serum medium by the addition of the necessary amount of mannitol. To deprive cells of serum, monolayers that had grown to confluency in complete DMEM with 10% foetal calf serum were washed two times in the minus serum DMEM and then the same volume of minus serum medium was added and the monolayers replaced in the incubator. After the indicated times the experiments were conducted. pCEFL plasmids containing the dominant negative N19RhoA, N17Rac1 and N17Cdc42 mutants were kindly provided by Dr PP Di Fiore (European Institute of Oncology, Milan, Italy) and pEXV plasmids containing the constitutively active V14RhoA and V12Rac1 mutants by Dr MH Symons (Picower Institute for Molecular Research, Manhasset, NY, USA). These cDNAs were subcloned into the pBabe puro expression vector containing a haemagglutinin (HA) tag. Ten micrograms of plasmid cDNA or empty vector was incubated with 100 μl LipoTaxi reagent (Stratagene, La Jolla, CA, USA) in 1 ml of simple DMEM growth medium for 30 min at room temperature. Serum was added to 3% and 200 μl of this mixture was pipetted onto confluent monolayers on glass coverslips and placed in an incubator at 5% CO 2 and 37°C for 6 hours. This mixture was then replaced with fresh complete medium (10% serum) for 24 hours in normal growth conditions. Cells were then treated and NHE1 activity, motility, or invasion were measured. The level of transfection was determined using Western blot of whole cell extracts using anti-HA antibody (Santa Cruz, Santa Cruz, CA, USA). The percentage of transfection with different concentrations of cDNA (0, 2.5, 5 and 10 μg plasmid cDNA) was evaluated by immunofluorescence microscopy of cell monolayers on coverslips using the same anti-HA antibody. As shown in Fig. 1 the transfection efficiency for both the dominant negative (dn) N17Rac1 cDNA and N19RhoA cDNA was commensurate with cDNA concentration and reached approximately 90% when cells were transfected with 10 μg plasmid cDNA. The transfection efficiencies observed with the other cDNAs utilized in the study were very similar to those for N19RhoA and N17Rac1 (data not shown). Intracellular pH and H + efflux rate determinations Intracellular, cytoplasmic pH (pHi) was measured spectrofluorimetrically at 37°C with the fluorescent pH sensitive probe 2',7'-bis(carboxyethyl)-5, 6-carboxy-fluorescein (BCECF), and trapped intracellularly in cell monolayers grown on glass coverslips as previously described [ 6 ]. NHE1 activity was measured by monitoring pHi recovery after an intracellular acid load produced with the NH 4 Cl prepulse technique. The initial rate of Na + -dependent alkalinization was determined by linear regression analysis of the first 15 points taken at 4 s intervals after the readdition of sodium. The use of CO 2 /HCO 3 free solutions minimizes the likelihood that Na + -dependent HCO 3 transport was responsible for the observed changes in pHi. The pHi dependence of intracellular buffer capacity (βi; i.e. the buffering power of all non-HCO 3 , non-CO 2 buffers) was computed using the NH 4 pulse method, and the actual activity of the exchanger in terms of proton flux rate (mmol/l H + /min) was determined by multiplying the rates of pHi change by the cells' intrinsic buffering capacity (βi) at the pHi at which the measurement was taken. Preliminary experiments demonstrated that 2 μmol/l of the specific NHE1 inhibitor 5-(N, N-dimethyl)-amiloride (DMA; Sigma, Milano, Italy) almost completely blocked NHE1 activity, whereas transfection with empty pBabe vector had no effect on NHE1 activity (data not shown). Motility and invasion A quantitative measure of the degree of in vitro motility of the cells was obtained in BW25 Boyden Chambers (Neuro Probe Inc., Cabin John, MD, USA) using 8 μm polycarbonate membranes (Poretics, Livermore, CA, USA) coated with 5 μg collagen I as previously described [ 6 ]. After 4 hours of incubation, cells were fixed and stained (DiffQuick; Baxter, Oakland, CA, USA), those cells that had not transversed the filter were removed, and randomly chosen fields were photographed at 100× magnification using a Nikon Eclipse E800 microscope equipped with an MRC-1024 imaging system (Bio-Rad Laboratories, Milano, Italy). The number of cells traversing the filter in four random fields for each filter were counted from these images. A quantitative measure of the degree of in vitro invasion was measured as the ability to infiltrate into live, confluent MCF-10A monolayers, essentially as previously described [ 6 , 29 ]. MDA-MB-435 cells were metabolically loaded with 3 H-thymidine for the 24 hours prior to the experiment, cells were trypsinized, and a standard curve of incorporated 3 H-dT/cell measured for each treatment. Tumour cells (80,000) were added in suspension to the complete medium of the confluent MCF-10A monolayers. Culture dishes were returned to the incubator for 8 hours and then unattached cells were removed by vigorous washing and mechanical agitation two times with phosphate-buffered saline (PBS). The number of MDA-MB-435 cells that had invaded the MCF-10A monolayer was calculated by measuring the incorporated 3 H-thymidine for each MCF-10A monolayer in a Packard TopCount NXT ® microplate scintillation counter (Packard Instruments, Inc., Palo Alto, CA, USA), and the number of cells present computed using the standard curve of disintegrations per min/cell for each treatment. Experiments standardizing this assay against the Boyden Chamber matrigel assay demonstrated high correlation between the two techniques. Expression and activity of RhoA and Rac1 RhoA and Rac1 activity were assessed using the RhoA-binding domain of Rhotekin or Rac1-binding domain of PAK-1, respectively, in kits supplied from Upstate Biotechnology (Lake Placid, NY, USA). In brief, 3 × 10 6 cells were plated onto 10 cm cell culture dishes and after 24 hours they were treated as indicated. After the indicated time, cells were extracted using RIPA buffer (50 mmol/l Tris, pH 7.2, 500 mmol/l NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 1% SDS, 10 mmol/l MgCl 2 , 0.5 μg/ml leupeptin, 0.7 μg/ml pepstatin, 4 μg/ml aprotinin, and 2 mmol/l PMSF). After centrifugation at 14,000 g for 3 min, the extracts were incubated for 45 min at 4°C with glutathione beads coupled with glutathione S-transferase (GST)–RBD (Rho-binding domain of Rhotekin) fusion protein or GST-PAK-1 (Upstate Biotechnology), and then washed three times with Tris buffer (pH 7.2), containing 1% Triton X-100, 150 mmol/l NaCl and 10 mmol/l MgCl 2 . The RhoA or Rac1 content in these samples or in 50 μg protein of cell homogenate was determined by immunoblotting samples using rabbit anti-RhoA antibody (Santa Cruz) or anti-Rac1 antibody (Upstate Biotechnology). Fluorescence microscopy For analysis of actin cytoskeleton organization, cells were plated on coverslips until they were approximately 60% confluent, at which time they were transfected with empty vector or dn-RhoA or dn-Rac1 and, after 24 hours of incubation, they were either subjected or not subjected to serum starvation in the presence or absence of 2 μmol/l DMA. After 24 hours, cells were fixed with 4% paraformaldehyde in PBS for 15 min, and after residual formaldehyde had been quenched with 50 mmol/l NH 4 Cl in PBS for 10 min the cells were permeabilized with 0.2% Triton X-100 in PBS for 10 min. Filamentous actin (F-actin) was stained with fluorescein isothiocyanate-conjugated phalloidin (Sigma) in PBS (0.5 U/ml) for 1 hour, whereas HA-positive cells were determined with anti-HA antibody (Santa Cruz) at a 1:100 dilution. Microscopy was performed using an MRC-1024 imaging system (Bio-Rad Laboratories) equipped with a Nikon Eclipse E800 microscope and a Nikon Plan Apo 40-by-1.0 or 60-by-1.4 oil immersion objective lens. Analysis of F-actin content Actin polymerization was detected using fluorescent phalloidin and analyzed by confocal microscopy and fluorescence activated cell sorting analysis. Analytic flow cytometric measurements were performed basically as previously described [ 30 ], using a Hewlett-Packard 9153C flow cytometer with argon laser excitation at 488 nm and detection through a 515–540 nm bandpass filter. Ten thousand cells in each sample were analyzed. For determination of F-actin amount the gate was defined as corresponding to scatter parameters on a dot plot of viable control cells (nondeprived). Phalloidin–fluorescein isothiocyanate fluorescence intensities from the gated population were presented as a histogram and the geometrical mean value was used as a measure of F-actin content. Kinetic and statistical analysis The kinetic coefficients for proton dependency of exchange activity were estimated using nonlinear curve-fitting regression utilizing the iterative Marquardt procedure on the KALIDIOGRAPH ® (Abelbeck Software, Reading, PA, USA) program. The proton data were fitted to the Hill equation – V = (Vmax [H] n )/(Km + [H] n ) – in which V is the experimental transport velocity, Vmax is the calculated maximum transport velocity, [H] is the cytosolic proton concentration, Km is the apparent affinity for protons, and n is the apparent Hill coefficient. Results NHE1 lies upstream of serum deprivation-dependent actin cytoskeletal reorganization We previously showed that serum deprivation selectively stimulates NHE1 activity in breast cancer cells and their subsequent NHE1-dependent motility and invasion [ 6 ]. Recently, mounting evidence has shown that NHE1-dependent regulation of motility in normal cells occurs by direct NHE1-dependent reorganization of the actin cytoskeleton [ 25 , 26 ]. To evaluate whether NHE1 is upstream of actin reorganization during serum deprivation in cancer cells, we followed the serum deprivation-dependent remodelling of the F-actin cytoskeleton (Fig. 2a ) and quantity of F-actin (Fig. 2b,2c ) in the presence or absence of the specific NHE1 inhibitor DMA. As can be seen in Fig. 2a , control cells were primarily fusiform in shape with thin, uniform, parallel actin filaments (stress fibres) departing from single points and extending throughout the length of the cell. Serum deprivation provoked a complex and dramatic reorganization of the actin cytoskeleton, resulting in a reduction in the number of stress fibres with an uneven thickening of the remaining stress fibres. Consistent with the model proposed by the group of Barber [ 25 , 26 ], inhibition of NHE1 during serum deprivation with 2 μmol/l DMA completely abrogated the observed serum deprivation-dependent cytoskeletal reorganization, and depolymerization suggesting that stimulation of NHE1 is indeed up-stream of actin cytoskeletal reorganization. As can be seen in Fig. 2c , treatment with the actin cytoskeleton disruptor cytochalasin B (5 μmol/l for 15 min) reduced the amount of F-actin by approximately 75%. The lack of significant inhibitory effect by cytochalasin B treatment on NHE1 activity (absence versus presence of cytochalsin B: 16.2 ± 2.7 versus 20.6 ± 4.7 mmol/l intracellular H + concentration/min [27% increase], n = 7; P = 0.58) further supports this hypothesis [ 31 ]. Serum deprivation activates Rac1 and inactivates RhoA The usual model for RhoA-dependent alterations in motility is linked to RhoA action upon the actin cytoskeleton via the phosphorylation/inactivation of myosin phosphatase by the RhoA effector p160ROCK [ 21 , 22 ]. However, mounting evidence has suggested the existence of an alterative system in which RhoA/ROCK directly alters NHE1 activity [ 32 ] followed by the NHE1-dependent reorganization of the actin cytoskeleton [ 25 , 26 ]. It was recently reported that the tight junction protein NZO-3 increases kidney cell motility via a decrease in stress fibre number due to an inhibition of RhoA activity [ 33 ], and dihydromotuporamine C decreases cancer cell motility and invasion via an increase in stress fibre number that was due to a stimulation of Rho activity [ 34 ]. As a starting point for our study of the regulatory pattern of RhoA and Rac1 in the serum deprivation regulation of NHE activity and the migration/invasion of MDA-MB-435 cells, we measured the effect of serum deprivation on the activation status of RhoA and Rac1. To assess RhoA and Rac1 activation states, we used the GST fusion proteins of RBD to capture GTP-bound RhoA, and PAK-1 Rac-binding domain to capture GTP-bound Rac-1 from cell extracts. As shown in Fig. 3 , serum deprivation resulted in a significant decrease in the amount of RhoA retained by RBD (panel a) and in a significant increase in Rac1 retained by PAK-1 Rac-binding domain (panel b). In contrast, serum deprivation had no effect on total cellular expression of either RhoA or Rac-1, demonstrating that it is indeed changes in their activity and not expression that drives their regulatory control of NHE activity. A similar pattern of inverse correlation between levels of active RhoA and active Rac1 was previously described in epithelial cells and fibroblasts [ 35 , 36 ]. RhoA and Rac1 reciprocally regulate the activity of NHE1 in basal and serum deprivation stimulated conditions To address the question of whether Rho proteins are involved in the regulation of tumour cell NHE1 activity, we compared the effect of transient transfection of HA-tagged dn (Fig. 4a ) or constitutively active (ca; Fig. 4b ) mutants of Rho family proteins on both the basal activity of the NHE1 and its activation by serum deprivation. The relative level of transfection was determined using Western blot of whole cell extracts using anti-HA antibody (the inserts in the respective bar graphs in Fig. 4 ). The transfection efficiency for the various cDNAs was commensurate with cDNA concentration and reached approximately 90% when the cells were transfected with 10 μg plasmid cDNA (see the Methods section, above). In cells not subjected to serum deprivation, the basal level of NHE1 activity was regulated reciprocally by these mutant constructs; specifically, NHE1 was slightly but significantly stimulated by inactivation of Rac1 induced by transfection with dn-N17Rac1 (stripped bar), and slightly but significantly inhibited by inactivation of RhoA induced by transfection with dn-N19RhoA (hatched bar). In serum deprived conditions NHE1 activity was still regulated reciprocally but the role of these G-proteins was reversed; inactivation of RhoA with N19RhoA potentiated the serum deprivation-dependent stimulation of the NHE1 (hatched bar), whereas inactivation of Rac1 with N17Rac1 blocked this stimulation (stripped bar). Inactivation of Cdc42 with dn-N17Cdc42 inhibited NHE1 activity to a similar degree in both nondeprived and deprived conditions (stippled bars). These data suggest that RhoA and Rac1 play antagonistic roles in the regulation of both basal and serum deprivation-induced NHE1 activity, and that there is a reversal of their regulatory action with serum deprivation. To confirm this hypothesis, cells were transfected with constitutively active mutants of RhoA and Rac1. Activating RhoA with V14RhoA blocked serum deprivation-mediated stimulation of the NHE1 whereas the constitutively active V12Rac1 potentiated NHE1 stimulation by serum deprivation (Fig. 4b ). Treating the cells with either a RhoA-specific pharmacological inhibitor (C3 exotoxin) or activator (CNF-1) produced a pattern identical to that observed with the RhoA mutated constructs (data not shown). NHE1 activity is finely regulated by intracellular proton concentration via an allosteric proton regulatory site, and we previously showed that serum deprivation upregulated tumour cell NHE activity via an increased H + affinity of this site, resulting in an alkaline shift of the pK value for intracellular H + [ 6 ]. To determine whether RhoA and Rac1 modulate the serum deprivation-dependent upregulation of the NHE1 via this same mechanism, we analyzed the effect of the dominant negative mutants of RhoA or Rac1 on the dependence of NHE1 activity on pHi in serum deprived cells, as previously described [ 6 ]. Fig. 5 shows the relationship of NHE1 activity (Δmmol/l intracellular H + concentration/min) to pHi in 24 hr serum complete (open circles) and serum deprived cells minus (closed circles) or plus transfection with dn-N17Rac1 (squares) or dn-N19RhoA (triangles) in a typical experiment. Serum deprivation shifted the pK value for intracellular H + to alkaline values (alkaline shift) and increased the slope of the relationship. The observed alkaline shift in serum deprived conditions in the tumour cells is consistent with an increased capacity for net acid extrusion. In the cells transfected with dn-N17Rac1 (squares) serum deprivation no longer had any significant effect on the alkaline shift produced by serum deprivation, whereas transfection with dn-N19RhoA potentiated this alkaline shift. Kinetic analysis on five independent experiments for each treatment demonstrated that the Vmax and H + affinity increased in MDA-MB-435 cells upon serum deprivation (Table 1 ) without a change in the Hill coefficient (n app remained approximately 2). Also, transfection with dn-N17Rac1 blocked the serum deprivation-dependent kinetic shifts whereas transfection with dn-N19RhoA potentiated them. These data suggest that one of the mechanisms by which RhoA and Rac1 regulate the serum deprivation-dependent increases in NHE1 activity is by modifying the affinity of the internal proton regulatory site of the NHE1 for protons. Effect of dominant negative RhoA and Rac1 on serum deprivation-induced alterations in cell shape and actin organization In order to determine the alterations in F-actin cytoskeletal organization associated with the above changes in cell shape, we next incubated the cells with tetramethylrhodamine 5-isothiocyanate (TRITC)-labelled phalloidin in order to stain F-actin in the different treatments. As can be seen in Fig. 6a , control cells were primarily fusiform in shape with thin, uniform, parallel actin filaments (stress fibres) departing from single points and extending throughout the length of the cell, with the presence of lamellipodia (arrow), but there were few dominant, elongated pseudopodia and few actin containing, fine cell extensions (termed invadopodia [ 37 , 38 ]). Serum deprivation provoked a complex and dramatic reorganization of the actin cytoskeleton that resulted in a reduction in number of stress fibres and lamellipodia, with an uneven thickening of the remaining stress fibres and an increase in both dominant leading-edge pseudopodia (arrowhead) observed at the ends of the cells and in invadopodia (asterisk). Interestingly, transfection with mutant constructs that increase either basal or deprivation-stimulated NHE1 activity and motility (e.g. dn-Rac1, ca-RhoA, dn-RhoA plus deprivation or ca-Rac1 plus deprivation) greatly augmented all of the serum deprivation-induced alterations in cell shape and actin organization. There was an increase in invadopodia in all conditions such as to produce a halo-like effect around the cell. Furthermore, in all conditions that increased NHE1 and motility, except for dn-Rac1, there was increased length and complexity of the pseudopodia that reached its maximum development in the condition of dn-RhoA plus deprivation (see also the larger magnification of the end of an elongated dominant leading-edge pseudopodia; Fig. 6b ). A densiometric analysis (AutoDeblur 9.1; AutoQuant Imaging, Inc., New York, NY, USA; data not shown) confirmed that F-actin pixel density decreased with serum deprivation and in those treatments that increased invasion and motility (dn-Rac1, ca-RhoA, dn-RhoA plus deprivation or ca-Rac1 plus deprivation). RhoA and Rac1 reciprocally regulate directed motility and invasive capacity of MDA-MB-435 cells via the activity of NHE1 We previously demonstrated that NHE1 activity plays a fundamental role in motility and invasion in MDA-MB-435 cells [ 6 ], and so we analyzed the ability of dominant negative and constitutively active mutants of RhoA or Rac1 to modify the serum deprivation-induced motility and invasive ability of MDA-MB-435 cells. As differential apoptosis could affect the outcome of these measurements, the effects of transfection of these mutants of RhoA and Rac1 on apoptosis were analyzed and we observed that neither 1 day of serum deprivation nor transfection of any plasmid had any significant effect on apoptosis in either basal or serum deprived states (data not shown). As shown in Fig. 7 , 24 hours of serum deprivation (D) significantly increased basal, non-serum-deprived (ND) MDA-MB-435 cell motility, measured as their ability to cross a collagen I layer in a Boyden Chamber (Fig. 7a,7b ) and their capacity to invade a confluent MCF-10A cell monolayer (Fig. 7c,7d ). This serum deprivation-dependent stimulation of motility and invasion was strongly inhibited in the cells transfected with either dn-N17Rac1 or ca-V14RhoA and was potentiated in the cells transfected with dn-N19RhoA or ca-V12Rac1. Furthermore, in non-serum-deprived conditions (ND), the basal level of cellular motility was regulated opposite to that observed during deprivation by these mutant constructs. That is, motility was slightly but significantly stimulated by transfection with either dn-N17Rac1 or ca-V14RhoA and slightly but significantly inhibited by transfection with dn-N19RhoA or ca-V12Rac1 (values expressed as cells traversed/filter [mean ± standard error of the mean]: control 2148 ± 80, dn-Rac1 3456 ± 263, ca-RhoA 3055 ± 104, dn-RhoA 1727 ± 94, ca-Rac1 1287 ± 55; n = 5 for each treatment). Altogether, these data strongly suggest that regulation of both motility and invasion closely followed regulation of NHE1 activity. Furthermore, incubation with 2 μmol/l of the specific NHE1 inhibitor DMA almost completely abrogated both basal motility and invasion and their stimulation by serum deprivation, and the potentiation by either dn-N19RhoA or ca-V12Rac1, supporting our previous observation [ 6 ] and the findings of others [ 7 - 9 ] that the NHE1 plays a fundamental role in tumour cell motility and invasion. Serum deprived cells that had traversed the collagen coated filter also exhibited a more motile shape than did the serum replete cells (Fig. 7a ), and this increase was potentiated in the cells transfected with dn-N19RhoA and attenuated in cells transfected with dn-N17Rac1. Discussion Rho family GTPases are greatly overexpressed in breast tumours [ 18 ] and RhoA is necessary for Ras-mediated transformation [ 19 ] and metastatic spread [ 39 ]. This suggests that these G-proteins play a critical role in neoplastic progression, but the mechanistic basis of their action in tumour cells is still not completely clear [ 20 ]. Here we demonstrate that the Rho family of GTPases are involved in the regulation of basal NHE1 activity and especially in the serum deprivation-induced stimulation of its activity. The regulation of both basal NHE1 activity and its upregulation by serum deprivation are linked to a reciprocity of RhoA and Rac1 action (Fig. 4a,4b ). We propose that in these breast tumour cells this reciprocal antagonism between RhoA and Rac1 precisely coordinates the specificity and/or magnitude of the pathophysiological response of NHE1 to serum deprivation, with concomitant increases in motility and invasion and hence malignant progression [ 6 ]. The reciprocal antagonism between the actions of Rac1 and RhoA on breast cancer cell NHE1 activity, motility and invasive capacity is identical to that reported for movement/migration [ 21 , 22 ] and actin cytoskeleton remodelling [ 12 , 40 ] in other cell systems. Recent evidence supports a prominent role for NHE1 in coordinating motility [ 6 - 9 ] by selectively regulating cytoskeletal events such as focal adhesion assembly and turnover, cortical cytoskeleton dynamics and tyrosine phosphorylation of focal adhesion kinase, with consequent impaired recruitment and assembly of integrins, paxillin and vinculin at focal contacts [ 41 ]. Importantly, it was recently demonstrated that NHE1 can regulate these processes through direct interaction with the ERM family cytoskeletal linker protein ezrin [ 25 , 26 ]. In the present study we observed that elevated NHE1 activity is upstream from serum deprivation-dependent actin remodelling (Fig. 2 ) and the consequent increase in motility and invasion (Fig. 6 ); furthermore, we found that RhoA and Rac1 act on these processes through a modification of NHE1 activity. This association of increased motility with a decrease in stress fibre number driven by inhibition of RhoA activity is similar to that recently reported for the tight junction protein NZO-3 in controlling renal cell motility [ 33 ], and inhibition of motility and invasion by dihydromotuporamine C [ 34 ] or by degradation of RhoA by Smurf1 in controlling Mv1Lu cell motility [ 24 ], suggesting a common mechanism. In the latter work RhoA degradation was observed primarily in the cell protrusions, and we are currently determining whether inhibition of RhoA by serum deprivation occurs preferentially in the leading edge pseudopodia of the MDA-MB-435 cells. Altogether these studies strongly suggest that in our cell system the reported antagonistic roles of RhoA and Rac1 in cytoskeletal organization, motility and invasion may also be due to their concerted, convergent action on NHE1 activity. Thus, RhoA and Rac1 integrate cytoskeletal reorganization, which is necessary to direct motility via protein–protein interactions of the NHE1 with other focal adhesion components, and thus they facilitate or inhibit downstream signalling and/or cytoskeletal cascades. What remains unclear is whether the RhoA-dependent and Rac1-dependent regulation actin organization via NHE1 forms an integral part of the myosin light chain kinase/myosin phosphatase cycle [ 21 , 22 ], or whether it is part of an independent, parallel system that may cooperate to produce the characteristic integration of cell movement by RhoA and Rac1 antagonism. As discussed previously [ 23 ], this mechanism concerns smooth muscle cell contraction and cannot fully explain the effects of Rho family proteins on the actin cytoskeleton in other cell types, such as epithelial derived tumour cells. The studies performed to date focused on specific cell types and/or experimental systems, and so it is difficult to determine whether these two mechanisms coexist and to interpret possible interactions between the RhoA/Rac1 modulation of motility via the regulation of NHE1 activity [ 25 , 26 ] or via the myosin phosphatase/myosin light chain kinase cycle [ 21 , 22 ]. Furthermore, it will be necessary to determine whether the redox-dependent downregulation of RhoA by Rac1 with subsequent regulation of cytoskeletion and motility described in HeLa cells [ 17 ] has NHE1 as a downstream target. In this way we can begin to acquire an organized view of how the actin cytoskeleton and associated process of motility are regulated by the Rho family of GTPases, and to gain further insights into the mechanisms that underlie the variability of action of these GTPases on actin structures. Conclusion The present study defines the roles of RhoA and Rac1 in the regulation of Na + –H + exchange, and consequently of motility/invasion in tumour cells and in the stimulation of these processes that is unique to tumour cells when confronted with serum deprivation – a common environmental condition in tumours. The regulation of both basal NHE1 activity and its upregulation by serum deprivation are linked to a reciprocity of the actions of RhoA and Rac1. Interestingly, the role of each of these G-proteins is reversed during serum deprivation; basal NHE1 activity is regulated positively by RhoA and negatively by Rac1, whereas RhoA negatively and Rac1 positively directs the stimulation of NHE1 during serum deprivation (Figs 4 and 5 ). The findings presented here extend those of our earlier studies documenting the effects of serum deprivation and the role of PI3K on pH regulation, in which we postulated that the observed inversion of PI3K regulatory action contributes, in part, to the upregulation of the NHE1 by serum deprivation in tumour cells [ 6 ]. That deprivation-induced switch from positive to negative PI3K regulation may reflect a shift in the coupling of active Rho GTPases to different signalling pathways. We postulate that it is this inversion of RhoA/Rac1 function, together with PI3K regulatory action, that contributes, in part, to the upregulation of NHE1 by serum deprivation in MDA-MB-435 cells, and hence to increased invasion and malignant progression. To date, the mechanism that drives this shift in RhoA/Rac1 function is not clear but growing evidence indicates that a delicate balance between positive and negative actions of a signal cascade can determine the specificity and/or magnitude of a physiological or pathophysiological response. A particularly important area in biology is the study of mechanisms that underlie dynamic organization of signalling networks in response to specific cellular signals. In the currently developing paradigm, it is becoming ever clearer that specificity of response in signal transduction is attained through the compartmentalization of signalling proteins and effectors close to their activators, regulatory elements and targets to ensure tight regulation and specificity of action of signalling cascades [ 39 ]. The assembly of these multimolecular complexes or modules at the cellular site of action is accomplished by scaffolding proteins that have the capacity, through simultaneous interaction with multiple signalling proteins, to integrate diverse signalling pathways [ 40 ]. A possibility is that serum deprivation alters the balance of expression or activity of a scaffolding protein, resulting in the creation of new spatiotemporal combinations and/or altered integration of signal components. Finally, the reversal in the roles of Rac1 and RhoA in the regulation of NHE1 activity and motility observed here might explain some seemingly contradictory reports in which activating RhoA can lead to either increased or decreased motility and invasiveness [ 18 ]. These discrepancies possibly reflect different serum treatments of the tumour cells after RhoA activation and point out the peril of an across-the-board inhibition of RhoA as a possible therapy. Importantly, these results also unify recent advances concerning the mechanisms underlying tumour invasive potential: the RhoA-ROCK activation of NHE1 [ 41 , 42 ], the RhoA-ROCK mediation of invasion [ 43 , 44 ] and role of the NHE1 in driving tumour cell motility [ 6 - 9 ] and invasion [ 6 ]. Further studies are now needed to elucidate the signal transduction components that lie upstream and downstream of RhoA and Rac1 action, and to clarify the interactions between the RhoA-and Rac1-dependent signal transduction pathways that are involved in their reciprocal antagonism. It will be of key interest to investigate the interrelationships between these activities and to define the effectors and downstream mediators that are involved in RhoA-and Rac1-transduced mitogenesis, transformation and invasiveness. Competing Interests None declared. Abbreviations ca = constitutively active; dn = dominant negative; DMA = 5-(N, N-dimethyl)-amiloride; DMEM = Dulbecco's modified Eagle's medium; GST = glutathione S-transferase; HA = haemagglutinin; NHE = Na + /H + exchanger; PAK = p21-activated kinase; PBS = phosphate-buffered saline; pHi = intracellular pH; PI3K = phosphoinositide-3 kinase; RBD = Rho-binding domain of Rhotekin; ROCK = Rho kinase.

D:\keerthana\PMC001xxxxxx\PMC1064074.xml

1064075

The progesterone receptor Val660→Leu polymorphism and breast cancer risk

Background Recent evidence suggests a role for progesterone in breast cancer development and tumorigenesis. Progesterone exerts its effect on target cells by interacting with its receptor; thus, genetic variations, which might cause alterations in the biological function in the progesterone receptor ( PGR ), can potentially contribute to an individual's susceptibility to breast cancer. It has been reported that the PROGINS allele, which is in complete linkage disequilibrium with a missense substitution in exon 4 (G/T, valine→leucine, at codon 660), is associated with a decreased risk for breast cancer. Methods Using a nested case-control study design within the Nurses' Health Study cohort, we genotyped 1252 cases and 1660 matched controls with the use of the Taqman assay. Results We did not observe any association of breast cancer risk with carrying the G/T (Val660→Leu) polymorphism (odds ratio 1.10, 95% confidence interval 0.93–1.30). In addition, we did not observe an interaction between this allele and menopausal status and family history of breast cancer as reported previously. Conclusion Overall, our study does not support an association between the Val660→Leu PROGINS polymorphism and breast cancer risk.

Introduction Until recently, the role of progesterone on mammary gland tumorigenesis was not well understood. Data from epidemiological studies revealed a higher risk for breast cancer in postmenopausal women who used a combination of estrogens and progestins, in comparison with those women who used estrogens alone [ 1 , 2 ]. As demonstrated in the progesterone receptor knockout mouse, the physiological effects of progesterone are completely dependent on the presence of its receptor gene, PGR , which exists as a single-copy gene. The PGR gene uses separate promoters and translational start sites to produce two protein isoforms, hPR-A and hPR-B [ 3 - 5 ], that are identical except for an additional 165 amino acids present only in the amino terminus of hPR-B [ 6 , 7 ]. Although hPR-B shares many important structural domains with hPR-A, the two isoforms are functionally distinct transcription factors [ 8 ] that mediate their own response genes and physiological effects with little overlap [ 9 , 10 ]. The progesterone receptor knockout mouse, in which the functional activity of both hPR-A and hPR-B were simultaneously ablated, revealed that progesterone is required for the formation of ductal and alveolar structures during pregnancy [ 11 , 12 ]. Selective ablation of PR-B in a mouse model, resulting in the exclusive production of PR-A, revealed that PR-B is necessary for breast formation [ 13 ]. Given the evidence described above for the role of progesterone in breast cancer causation, we proposed that variations in the PGR gene might predispose women to breast cancer. Several studies have investigated the Val660 →Leu G/T polymorphism and the PROGINS Alu insertion, which are in complete linkage disequilibrium [ 14 ], in association with breast cancer [ 15 - 17 ]. In this study we focused on the Val660→Leu G/T polymorphism that has been reported to be associated with a decreased risk of breast cancer [ 17 ]. Materials and methods Detailed information about this nested case-control study and exposure data has been reported previously [ 18 ]. The protocol was approved by the Committee on Human Subjects, Brigham and Women's Hospital. Genotyping assays were performed by the 5' nuclease assay (TaqMan ® ) by the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). TaqMan primers, probes, and conditions for genotyping assays are available from the authors on request. Genotyping was performed by laboratory personnel blinded to case-control status, and blinded quality control samples were inserted to validate genotyping procedures. Concordance for the blinded samples was 100%. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated by using conditional and unconditional logistic regression. In addition to the matching variables, we adjusted for breast cancer risk factors: body mass index (BMI) (kg/m 2 ) at age 18 years, weight gain since age 18 years, age at menarche, parity/age at first birth, duration of postmenopausal hormone use, first-degree family history of breast cancer, and history of benign breast disease. We also adjusted for age at menopause in analyses limited to postmenopausal women. Indicator variables for all genotypes were created by using the wild-type genotype as the reference category in the regression models. Because of the low prevalence of homozygous variants, we combined heterozygotes and homozygotes in the logistic regression analysis. Interactions between genotypes and breast cancer risk factors were evaluated by including appropriate interaction terms in unconditional logistic regression models. The nominal likelihood ratio test was used to assess the statistical significance of these interactions. We used SAS version 8.0 (SAS Institute, Cary, NC) for all analyses. We tested Hardy–Weinberg agreement by using a χ 2 test. Results and discussion Our study included a total of 1323 incident breast cancer cases, diagnosed after blood draw to 1 June 2000, and 1854 matched controls. Of these, 1134 cases and 1640 controls were postmenopausal at blood draw, and 112 cases and 121 controls were premenopausal; menopausal status was uncertain in 77 cases and 93 controls. The mean age of cases at blood draw was 57.3 years; for controls it was 57.9 years. Cases and controls had similar mean BMI at blood draw (25.5 versus 25.5 kg/m 2 ) and weight gain since age 18 years (11.6 versus 11.3 kg). In comparison with controls, cases had similar ages at menarche (12.5 versus 12.6 years), first birth (23.0 versus 23.0 years) and age at menopause (48.2 versus 47.9 years). The proportion of women with a first-degree family history of breast cancer was significantly higher among the cases (20.0% versus 15.0%). Cases were also more likely to have a history of benign breast disease (64.0% versus 51.0%) and a longer duration of postmenopausal hormone use (50.3% versus 49.7% current users for five or more years). The study population was predominantly Caucasian (89% of cases, 86% of controls). The prevalence of the variant carriers was similar to that in a previous report for Caucasian women [ 14 ]: 31% for the cases and 29% for the controls. The genotype distribution of the Val660→Leu polymorphism among the cases and controls was in Hardy–Weinberg equilibrium. We did not observe a statistically significant association of breast cancer among carriers of the Val660→Leu G/T polymorphism. Too few homozygote variants were available in which to analyze the heterozygotes and homozygotes separately. Compared with the G/G wild-type genotype, the adjusted OR for women with G/T and T/T was 1.10 (95% CI 0.93–1.30) (Table 1 ). Because the previously reported inverse association was confined to premenopausal women, we stratified by menopausal status and observed no association among premenopausal women (adjusted OR 1.21 [95% CI 0.64–2.28]) although we had a relatively small number of women for this analysis (Table 1 ). The Val660→Leu polymorphism has been suggested to modify the association between family history of breast cancer and breast cancer [ 17 ]. We observed no statistically significant interactions between the Val660→Leu polymorphism and first-degree family history of breast cancer. In addition, we selected BMI, history of benign breast disease, and hormone replacement therapy use among postmenopausal women as potential effect modifiers based on biological plausibility. We observed no significant interactions between the Val660→Leu polymorphism and any of these risk factors. Table 1 Association between the progesterone receptor exon 4 (Val660→Leu) G/T polymorphism and breast cancer risk Genotype a Cases, n (%) Controls, n (%) Crude OR (95% CI) Adjusted OR (95 % CI) G/G (Val/Val) 869 (69) 1186 (71) 1.0 b 1.0 c G/T + T/T (Val/Leu+Leu/Leu) 383 (31) 474 (29) 1.08 (0.92–1.27) 1.10 (0.93–1.30) Premenopausal women G/G (Val/Val) 68 (65) 79 (72) 1.0 d 1.0 e G/T + T/T (Val/Leu+Leu/Leu) 36 (35) 31 (28) 1.35 (0.75–2.41) 1.21 (0.64–2.28) Postmenopausal women G/G (Val/Val) 745 (69) 1044 (71) 1.0 f 1.0 g G/T + T/T (Val/Leu+Leu/Leu) 330 (31) 427 (29) 1.09 (0.92–1.29) 1.08 (0.91–1.28) a Numbers may vary because of missing genotypes. b Conditional logistic regression adjusted for the following matching variables: age, menopausal status, postmenopausal hormone use at blood draw, date at blood draw, time at blood draw, and fasting status. c Conditional logistic regression adjusted for matching variables and age at menarche, age at menopause, age at first birth/parity, Body Mass Index (BMI) at age 18 years, weight gain since age 18 years, benign breast disease, first-degree family history of breast cancer, and duration of postmenopausal hormone use. d Unconditional logistic regression adjusted for matching variables listed in footnote b. e Unconditional logistic regression adjusted for matching variables in footnote b and other covariates listed in footnote c. f Unconditional logistic regression adjusted for the following matching variables: age, menopausal status, postmenopausal use at blood draw, date at blood draw, time at blood draw, and fasting status. g Unconditional logistic regression adjusted for matching variables in footnote f and age at menarche, age at first birth/parity, BMI at age 18 years, weight gain since age 18 years, benign breast disease, first-degree family history of breast cancer, and duration of postmenopausal hormone use. Conclusions Our data do not support an inverse association between the Val660→Leu polymorphism and breast cancer risk as reported previously [ 17 ]. These results are consistent with recent studies of mostly Caucasian women in which no association was observed between this polymorphism and breast cancer risk in either premenopausal or postmenopausal women [ 15 , 16 , 19 , 20 ]. Most notable was the study by Spurdle and colleagues, in which a substantial number of premenopausal cases ( n = 769) were evaluated [ 19 ]. We had limited power to study this association in premenopausal women, but this is the largest study of postmenopausal women reported so far. The large sample size, prospective design and extensive relevant life-style information are among the strengths of this study. In conclusion, our results suggest that there is no association between the Val660→Leu polymorphism and breast cancer risk despite the substantial power of the study (more than 80% power to detect an OR of 0.75 or less for the carrier genotype). Competing interests None declared. Abbreviations BMI = body mass index; CI = confidence interval; LD = linkage disequilibrium; OR = odds ratio PGR = progesterone receptor.

D:\keerthana\PMC001xxxxxx\PMC1064075.xml

1064076

Genetic polymorphism in the manganese superoxide dismutase gene, antioxidant intake, and breast cancer risk: results from the Shanghai Breast Cancer Study

Introduction It has been suggested that oxidative stress and mitochondrial DNA damage play important roles in breast cancer carcinogenesis. Manganese superoxide dismutase (MnSOD) is a major enzyme that is responsible for the detoxification of reactive oxygen species in the mitochondria. A T → C substitution in the MnSOD gene results in a Val → Ala change at the -9 position of the mitochondrial targeting sequence (Val-9Ala), which alters the protein secondary structure and thus affects transport of MnSOD into the mitochondria. Methods We evaluated this genetic polymorphism in association with breast cancer risk using data from the Shanghai Breast Cancer Study, a population-based case–control study conducted in urban Shanghai from 1996 to 1998. The MnSOD Val-9Ala polymorphism was examined in 1125 breast cancer cases and 1197 age-frequency-matched control individual. Results Breast cancer risk was slightly elevated in women with Ala/Ala genotype (odds ratio [OR] 1.3, 95% confidence interval [CI] 0.7–2.3), particularly among premenopausal women (OR 1.8, 95% CI 0.9–3.7), as compared with those with Val/Val genotype. The increased risk with the Ala/Ala genotype was stronger among premenopausal women with a higher body mass index (OR 2.5, 95% CI 0.9–7.0) and more years of menstruation (OR 2.6, 95% CI 0.8–8.0). The risk among premenopausal women was further increased twofold to threefold among those with a low intake of fruits, vegetables, vitamin supplements, selenium, or antioxidant vitamins, including carotenes and vitamins A, C, and E. However, the frequency of the Ala allele was low (14%) in the study population, and most of the ORs provided above were not statistically significant. Conclusion The present study provides some evidence that genetic polymorphism in the MnSOD gene may be associated with increased risk of breast cancer among Chinese women with high levels of oxidative stress or low intake of antioxidants. Studies with a larger sample size are needed to confirm the findings.

Introduction More than 90% of the body's oxygen is consumed by the electron transport chain in mitochondria [ 1 ], and about 1–5% of it is released as superoxide (O 2 •- ) and hydrogen peroxide (H 2 O 2 ) [ 2 ]. Reactive oxygen species (ROS) may also be generated from estrogen metabolism through catechol estrogen redox cycling [ 3 , 4 ]. Because of a high level of internally generated ROS, lack of histone protection, and a low level of DNA repair, mitochondrial DNA is particularly vulnerable to oxidative damage [ 5 ]. It has been suggested that mitochondrial DNA damage may play an important role in breast carcinogenesis [ 5 , 6 ]. Manganese superoxide dismutase (MnSOD) and glutathione peroxidase (GPX) 1 are two major enzymes that are responsible for ROS detoxification in mitochondria [ 7 , 8 ]. The MnSOD gene, which is composed of five exons and four introns, is localized to chromosome 6q25 [ 9 , 10 ]. A T → C substitution, resulting in a Val → Ala change at the -9 position (Val-9Ala), which alters the secondary structure of the protein [ 11 ], has been noted to affect transport of MnSOD into the mitochondria [ 12 , 13 ]. Another polymorphism (Ile58Thr) in exon 3 affects the stability of MnSOD and reduces protein amount and enzyme activity [ 14 , 15 ]. Cells that over-expressed the Ile58 allele had higher MnSOD activity than did cells that over-expressed the Thr58 allele [ 16 ]. It is biologically plausible that the Val-9Ala and Ile58Thr polymorphisms play an important role in ROS detoxification, thus affecting the risk for developing cancer, particularly among individuals with a higher level of oxidative stress or who are deprived of other antioxidative protection, such as through a low level of antioxidant intake. Four studies examined the association of the Val-9Ala polymorphism with breast cancer risk, with mixed results [ 17 - 20 ]. Two hospital-based case–control studies found a moderate elevation in risk among women carrying the Ala/Ala genotype [ 17 , 18 ]. A population-based case–control study found no overall association with this polymorphism [ 19 ]. Recently, a case–control study conducted using data from the Ontario Familial Breast Cancer Registry found no association with this polymorphism [ 20 ]. All four studies were conducted among Caucasian women and had relatively small sample sizes, and most of them did not include a comprehensive assessment of environmental exposures or lifestyle data. Using data from the Shanghai Breast Cancer Study, a large-scale population-based case–control study conducted in urban Shanghai from 1996 to 1998, we evaluated the association of the MnSOD gene polymorphism with breast cancer risk, in conjunction with conditions related to oxidative stress and dietary intake of antioxidants. Methods Study participants The cases and control individuals evaluated in this study were participants of the Shanghai Breast Cancer Study, a population-based case–control study. Detailed study methods were published elsewhere [ 21 ]. Briefly, the study included 1459 women aged between 25 and 64 years, who were diagnosed with breast cancer between August 1996 and March 1998, and 1556 age-frequency-matched control individuals. The study protocol was approved by committees of all relevant institutions for the study of humans in research. All study participants were permanent residents of urban Shanghai who had no prior history of cancer and were alive at the time of interview. Through a rapid case ascertainment system, supplemented by the population-based Shanghai Cancer Registry, a total of 1602 eligible patients with breast cancer were identified during the study period, and interviews were completed in-person by 1459 (91%) of them. The major reasons for nonparticipation were refusal (109 [6.8%]), death before the interview (17 [1.1%]), and inability to locate the person (17 [1.1%]). Cancer diagnoses for all patients were confirmed by two senior study pathologists by review of tumor slides. Control individuals were selected using the Shanghai Resident Registry, a population registry containing demographic information for all residents of urban Shanghai, and were frequency matched for age (5-year intervals) to the expected age distribution of the cases in a 1:1 ratio. The inclusion criteria for control individuals were identical to those for the cases but with the exception of a diagnosis of breast cancer. Of the 1724 eligible women, 1556 (90.3%) completed interviews in-person. The remaining women were not included in the study either because of refusal (166 [9.6%]) or death before the interview (2 [0.1%]). A structured questionnaire was used to elicit detailed information on demographic factors, menstrual and reproductive histories, hormone use, dietary habits, prior disease history, physical activity, tobacco and alcohol use, weight, and family history of cancer. For all participants, current weight, circumference of the waist and hips, and height while sitting and standing were measured. Blood samples were obtained from 1193 (82%) cases and 1310 (84%) control individuals who completed in-person interviews. These samples were processed on the same day, typically within 6 hours of sample collection, and stored at -70°C until bioassays were performed. Usual dietary habits over the past 5 years were assessed by in-person interview, using a validated quantitative food frequency questionnaire [ 22 , 23 ]. The food frequency questionnaire included 76 food items or groups, 30 fresh vegetables and nine fruits, covering over 85% of foods commonly consumed in Shanghai. During the in-person interview, each study participant was first asked how frequently she consumed a specific food or group of foods (daily, weekly, monthly, yearly, or never), followed by a question on how many liangs (= 50 g) of food were eaten per unit time (day, week, month, or year) during the previous 5-year period, ignoring any recent changes in usual dietary intake within the 5-year period. Total dietary intakes of vitamin A (mg), carotene (mg), vitamin C (mg), vitamin E (mg), and selenium (μg) were calculated based on data from the Chinese Food Composition Table [ 24 ]. Genotyping method Genomic DNA was extracted from buffy coat fractions. The MnSOD genotypes were determined with PCR–restriction fragment length polymorphism methods, as reported previously [ 17 , 25 ] but with some modification. Briefly, the primers for the Val-9Ala polymorphism were 5'-ACCAGCAGGCAGCTGGCGCCGG-3' (forward) and 5'-GCGTTGATGTGAGGTTCCAG-3' (reverse). The primers for the Ile58Thr polymorphism were 5'-AGCTGGTCCCATTATCTAATAG-3' (forward) and 5'-TCAGTGCAGGCTGAAGAGAT-3' (reverse). The PCR was performed in a PTC-200 Peltier Thermal Cycler (MJ Research Inc., Waltham, MA, USA). Each 20 μl of PCR mixture contained 5 ng DNA, 1× PCR buffer with 1.5 mmol/l MgCl 2 , 0.16 mmol/l each of dNTP, 0.5 μmol/l of each primer, and 1 unit of HotStarTaq™ DNA polymerase (Qiagen Inc., Valencia, CA, USA). The reaction mixture was initially denatured at 95°C for 15 min, followed by 35 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s. The PCR was completed by a final extension cycle at 72°C for 7 min. The PCR products were digested by the Ngo MIV and Eco RV restriction endonucleases for the Val-9Ala and Ile58Thr polymorphisms, respectively. The DNA fragments were then separated using 3% Nusieve/Agarose gel and visualized by ethidium bromide staining. For the Val-9Ala polymorphism, the PCR product (107 bp) with C allele (Ala) was digested to two fragments (89 bp and 18 bp), whereas the PCR product with T allele (Val) cannot be cut by Ngo MIV. For the Ile58Thr polymorphism the PCR product (139 bp) with T allele (Ile) was digested to two fragments (117 bp and 22 bp), whereas the PCR product with C allele (Thr) cannot be cut by Eco RV. No Ile58Thr polymorphism was found in 400 individuals in our study, and we did not perform genotyping for this polymorphism in all participants. The laboratory staff was blind to the identity of the participants. Quality control (QC) samples were included in the genotyping assays. Each 96-well plate contained one water, two CEPH 1347-02 DNA, two blinded QC DNA, and two unblinded QC DNA samples. The blinded and unblinded QC samples were taken from the second tube of study samples included in the study. The Val-9Ala genotypes determined for the blinded QC samples were in complete agreement with the genotypes determined for the study samples. Genotyping data were obtained from 1125 cases and 1197 control individuals. The major reasons for incomplete genotyping were insufficient DNA and unsuccessful PCR amplification. Statistic analysis To evaluate case–control differences for categorical data, including genotype distribution and continuous variables, χ 2 test and t-test were used, respectively. Conditional logistic regression models adjusted for age were applied to estimate odds ratios (ORs) and 95% confidence intervals (CIs) to measure the strength of the association between the MnSOD gene Val-9Ala polymorphism and breast cancer risk [ 26 , 27 ]. Analyses stratified by menopausal status and age were conducted to check homogeneity of the association. Additional analyses stratified by body mass index (BMI), years of menstruation, and intake of fruits, vegetables, vitamin supplements, selenium, or antioxidant vitamins were conducted to evaluate the potential modifying effects of these variables on the association between the MnSOD genotypes and breast cancer risk. A composite dietary antioxidant index was derived to incorporate information on intake of four antioxidant nutrients (i.e. selenium and vitamins A, C, and E) [ 28 ]. Intake of each antioxidant nutrient was first standardized by subtracting the mean and dividing by the standard deviation, and then the antioxidant index was created by summing the standardized intake of these four antioxidant nutrients with equal weight [ 28 ]. All statistical tests were two-sided. Results The distribution of selected demographic characteristics and major risk factors for breast cancer are shown in Table 1 . Breast cancer cases and controls were similar in age and level of education. Elevated risks were observed for all known major breast cancer risk factors [ 29 ], including a prior history of breast fibroadenoma, physical inactivity, higher waist-to-hip ratio, higher BMI, early onset of menarche, late onset of menopause, and late age at first live birth. No apparent differences were found between individuals with genotyping data and those included in the whole study in any of the major known risk factors, demographic characteristics, and dietary antioxidant intake (data not shown), indicating that the chance of selection bias in this study is likely to be small. The frequencies of the Ala allele were 14.3% and 14.0% in cases and control individuals, respectively. Genotype frequencies were 73.9% (Val/Val), 23.6 (Val/Ala), and 2.5% (Ala/Ala) for cases, and the respective frequencies were 73.9%, 24.2%, and 1.9% for control individuals (Table 2 ). The MnSOD Val-9Ala polymorphism was in Hardy–Weinberg equilibrium for both cases and control individuals. No Ile58Thr polymorphism was found in our study population. Thus, we report here the results for the Val-9Ala polymorphism. Compared with women with the Val/Val genotype, breast cancer risk was slightly but not statistically significantly elevated for women with the Ala/Ala genotype (age-adjusted OR 1.3, 95% CI 0.7–2.3; Table 2 ). The Val/Ala genotype was unrelated to risk and was combined with the Val/Val genotype in some subsequent analyses. Additional adjustment for physical activity, BMI, waist-to-hip ratio, age at menarche, number of pregnancies, age at first birth, and family history of breast cancer had no appreciable effect on age-adjusted ORs, regardless of whether the analyses were done among all participants or stratified by menopausal status or age. Thus, only the age-adjusted ORs are presented. The elevated risk associated with the Ala/Ala genotype was restricted to premenopausal women (OR 1.8, 95% CI 0.9–3.7; for postmenopausal women OR 0.8, 95% CI 0.3–2.0) and women younger than 45 years (OR 1.8, 95% CI 0.7–4.3; for women older than 45 years OR 1.1, 95% CI 0.5–2.2). Further analyses were conducted to evaluate the association of the MnSOD Val-9Ala polymorphism and breast cancer risk by duration (years) of menstruation and BMI, factors that are related to the duration and level of estrogen exposure (Table 3 ). The ORs for the Ala/Ala genotype were higher among premenopausal women with a higher BMI (OR 2.5, 95% CI 0.9–7.0) or a longer duration of menstruation (OR 2.6, 95% CI 0.8–8.0). Again, the ORs were not statistically significant. This increased risk was not observed in postmenopausal women (Table 3 ). We further evaluated the association of the MnSOD Val-9Ala polymorphism with breast cancer risk by dietary antioxidant intake (Table 4 ). Intriguingly, the positive association with the Ala/Ala genotype among premenopausal women was consistently found to be stronger among those who had a low intake of fruits, vegetables, selenium, or antioxidant vitamins, including carotenes and vitamins A, C, and E, than among those who had a higher intake of these dietary factors. This pattern of association suggests a modifying effect, although neither the ORs nor the interaction tests were statistically significant, perhaps as a result of the small numbers of participants in the subgroups. The modifying effect of these antioxidant intakes on the association between the MnSOD Val-9Ala polymorphism and breast cancer risk was less apparent in the analyses among postmenopausal women. However, the postmenopausal case–control numbers were small. To illustrate the joint effects of dietary antioxidant intake, we further evaluated the association of the MnSOD Val-9Ala polymorphism with breast cancer risk by dietary antioxidant index. ORs for the Ala/Ala genotype were 2.4 (95% CI 0.6–9.5) and 2.2 (95% CI 0.5–9.4) among premenopausal and postmenopausal women, respectively, who had a lower dietary antioxidant index. Discussion Ambrosone and colleagues previously reported that Val/Val genotype was significantly associated with an increased risk of breast cancer among premenopausal Caucasian women, particularly those who had a low intake of fruits and vegetables and of dietary ascorbic acid and α-tocopherol [ 17 ]. In this large-scale, population-based, case–control study, we found that breast cancer risk was slightly but not significantly elevated in Chinese women with the Ala/Ala genotype as compared with women with the Val/Val genotype, particularly among premenopausal women who had a low intake of fruits, vegetables, selenium, or antioxidant vitamins. A significant association of this polymorphism with breast cancer risk was not observed in present study, which might partly be due to low Ala allele frequency in Chinese women. We observed in this study that women carrying the Ala/Ala genotype who had a higher BMI or a longer duration of menstruation were at higher risk of breast cancer, particularly among premenopausal women. Some ROS may be generated from estrogen metabolism through catechol redox cycling [ 3 , 4 ]. Mitrunen and coworkers [ 18 ] conducted a case–control study among 483 cases and 482 controls in a Finnish Caucasian population, and reported that the Ala allele was associated with breast cancer risk, with an OR of 1.5 in the Ala/Ala or Val/Ala groups compared with the Val/Val group. Postmenopausal women who had used estrogen replacement therapy and carried either the Ala/Ala or Val/Ala genotype had a 2.5-fold higher risk for breast cancer. Women who had used oral contraceptives and carried the Ala/Ala or Val/Ala genotype had a 3.0-fold higher risk for breast cancer [ 18 ]. More recently, Egan and coworkers [ 19 ] conducted a population-based case–control study among 476 cases and 502 controls in an American population. Overall, relative risks were not significantly elevated in women with one (OR 1.27, 95 CI 0.91–1.77) or two (OR 1.18, 95% CI 0.81–1.73) Ala alleles, as compared with the Val/Val genotype. Risk, however, was increased among premenopausal women carrying the Val/Ala genotype (OR 1.88), but not among women carrying the Ala/Ala genotype (OR 0.94) [ 19 ]. Women carrying the Ala/Ala or Val/Ala genotype who had used oral contraceptives or had higher BMI also had an increased risk for breast cancer [ 19 ]. Because of the low frequency of the Ala/Ala genotype and low percentage of estrogen use among Chinese women, we are unable to perform the analyses stratified by these exogenous estrogen exposures. Several studies have evaluated the association of the MnSOD Val-9Ala polymorphism with other cancers, although the results were inconsistent among cancer sites. Recently, Woodson and coworkers [ 30 ] reported that the Ala/Ala genotype was associated with a 1.7-fold (95% CI 0.96–3.08) increased risk for prostate cancer as compared with the Val/Val genotype. No association of this polymorphism with colorectal adenomas was found in a sigmoidoscopy-based case–control study [ 31 ]. Recently, Lin and coworkers [ 32 ] reported no association of the MnSOD Val-9Ala polymorphism with lung cancer risk in a case–control study conducted in Taiwan. Interestingly, Wang and coworkers [ 33 ] reported that the Val allele was associated with lung cancer risk with ORs of 1.34 (95% CI 1.05–1.70) and 1.67 (95% CI 1.27–2.20) in the Val/Ala and Val/Val group, respectively. These differences in the associations of the MnSOD genotype with breast and lung cancers suggest that the role played by MnSOD in carcinogenesis may vary for different tumors. The mechanism underlying this tumor type difference remains to be investigated. The SODs are the first and most important line of antioxidant enzyme defense against ROS and particularly O 2 •- radical. It was predicted that MnSOD Val-9Ala polymorphism might alter transfer of the MnSOD enzyme into mitochondria [ 11 , 12 ]. Recently, Sutton and coworkers [ 13 ] reported that the Ala-MnSOD precursor generated 30–40% more of the active, matricial, and processed MnSOD homotetramer in mitochondrial matrix than did Val-MnSOD. Some human tumor cells lost MnSOD activity, and this loss has been shown to be responsible for at least part of the malignant phenotype [ 34 , 35 ]. MnSOD knockout mice exhibited increased oxidative DNA damage [ 36 ]. Over-expression of MnSOD in MCF-7 cell suppressed the malignant phenotype, as evidenced by decreased cell proliferation, clonogenic fraction in soft agar culture, and tumor growth in nude mice [ 37 ]. Recently, Soini and coworkers [ 38 ] reported that MnSOD expression is less frequent in the tumor cells of invasive breast carcinomas than in in situ carcinomas or non-neoplastic breast epithelial cells. Three distinct isoforms of SOD have been identified in mammals. These three isoforms are encoded by three distinct genes located on different chromosomes [ 10 ]. All three SOD genes are polymorphic. It would be interesting to analyze jointly all of the three genes to evaluate any gene–gene interaction in relation to breast cancer risk. O 2 •- , if not scavenged by SOD, may react with nitric oxide radical (NO • ) to form the strong oxidant peroxynitrite (ONOO - ) [ 39 ]. MnSOD dismutates O 2 •- to H 2 O 2 , which is further detoxified by GPX1 in mitochondria. If not be quenched, H 2 O 2 will be converted to the more toxic hydroxyl radical ( • OH). Thus, study of the joint effect of the MnSOD , nitric oxide synthase, and GPX1 gene polymorphisms might provide further information on the role played by oxidative stress in cancer risk. Frequencies of the Val and Ala alleles of the MnSOD Val-9Ala polymorphism in our control population were 86.0% and 14.0%, respectively. The minor allele (Ala) frequency (14.0%) is comparable with that in Japanese (14.1%) [ 40 ] and Chinese (11.5%) [ 41 ] populations, but substantially lower than that in Caucasian populations [ 17 - 20 ]. The low frequency of the Ala allele in our study population contributes to a wider range of 95% CIs in OR estimations and limits the statistical power for stratified analyses. Small numbers of individuals, such as in the subgroups stratified by menopausal status in the present study, may result in unstable OR estimates. Studies with a larger sample size are needed to confirm the findings. The current study has many strengths. First, the high participation rate and the population-based study design substantially reduce selection bias. Second, Chinese women living in Shanghai are relatively homogenous in their ethnic background; over 98% are from a single ethnic group (Han Chinese). Therefore, the potential confounding effect of ethnicity is not a major concern in this study. Third, the extensive information collected on lifestyle factors allowed comprehensive evaluation of their interaction with genetic polymorphisms. The risk estimates derived from age-adjusted and multivariable adjusted analyses were similar, indicating that a confounding effect is unlikely to be a concern in this study. Conclusion In this population-based case–control study conducted in Chinese women, we found that the MnSOD Ala/Ala genotype was associated with a slightly but nonsignificantly elevated risk of breast cancer. The positive association was more evident among premenopausal women, particularly among those who consumed a low level of antioxidant vitamins or with high levels of oxidative stress. However, the study is limited by the low frequency of the Ala allele in the Chinese population, and most of the ORs were not statistically significant. Studies with a larger sample size are needed to confirm the findings. Author contributions QC, X-OS, and WZ participated in interpretation of results and writing the manuscript. QC participated in laboratory assays. WW conducted the statistical analyses. J-RC, QD, and Y-TG participated in field operation. Competing interests The authors declare that they have no competing interests. Abbreviations BMI = body mass index; bp = base pairs; CI = confidence interval; GPX = glutathione peroxidase; MnSOD = manganese superoxide dismutase; OR = odds ratio; PCR = polymerase chain reaction; QC = quality control; ROS = reactive oxygen species.

D:\keerthana\PMC001xxxxxx\PMC1064076.xml

1064077

Proteotypic classification of spontaneous and transgenic mammary neoplasms

Introduction Mammary tumors in mice are categorized by using morphologic and architectural criteria. Immunolabeling for terminal differentiation markers was compared among a variety of mouse mammary neoplasms because expression of terminal differentiation markers, and especially of keratins, provides important information on the origin of neoplastic cells and their degree of differentiation. Methods Expression patterns for terminal differentiation markers were used to characterize tumor types and to study tumor progression in transgenic mouse models of mammary neoplasia (mice overexpressing Neu ( Erbb2 ), Hras , Myc , Notch4 , SV40-TAg , Tgfa , and Wnt1 ), in spontaneous mammary carcinomas, and in mammary neoplasms associated with infection by the mouse mammary tumor virus (MMTV). Results On the basis of the expression of terminal differentiation markers, three types of neoplasm were identified: first, simple carcinomas composed exclusively of cells with a luminal phenotype are characteristic of neoplasms arising in mice transgenic for Neu , Hras , Myc , Notch4 , and SV40-TAg ; second, 'complex carcinomas' displaying luminal and myoepithelial differentiation are characteristic of type P tumors arising in mice transgenic for Wnt1 , neoplasms arising in mice infected by the MMTV, and spontaneous adenosquamous carcinomas; and third, 'carcinomas with epithelial to mesenchymal transition (EMT)' are a characteristic feature of tumor progression in Hras- , Myc- , and SV40-TAg- induced mammary neoplasms and PL/J and SJL/J mouse strains, and display de novo expression of myoepithelial and mesenchymal cell markers. In sharp contrast, EMT was not detected in papillary adenocarcinomas arising in BALB/cJ mice, spontaneous adenoacanthomas, neoplasms associated with MMTV-infection, or in neoplasms arising in mice transgenic for Neu and Wnt1 . Conclusions Immunohistochemical profiles of complex neoplasms are consistent with a stem cell origin, whereas simple carcinomas might originate from a cell committed to the luminal lineage. In addition, these results suggest that the initiating oncogenic events determine the morphologic features associated with cancer progression because EMT is observed only in certain types of neoplasm.

Introduction Architectural and cytological patterns have been the basis for mammary tumor categorization for more than a century [ 1 ]. Over the past 20 years, immunohistochemistry has added a molecular dimension to the categorization of mammary neoplasms: altered expression of oncogenes, oncosuppressor genes, hormone receptors, and cytoskeletal proteins have been identified as useful indicators of disease outcome. Gene microarray analysis has confirmed the importance of many of these histological prognosticators [ 2 - 4 ], which – in conjunction with clinical features such as lymph node status, tumor size, and tumor grade [ 5 ] – are undeniably useful. However, these prognosticators do not seem to be useful to identify the pathways that lead to tumorigenesis and cancer progression. Transgenic mouse models of mammary cancer have added considerably to our knowledge of human breast tumorigenesis. More specifically, in addition to dissecting the molecular mechanisms underlying mammary development and the initial steps of mammary neoplastic transformation, mouse models have shown that activation of a particular pathway is reflected in specific histologic patterns [ 6 , 7 ]. In view of the evidence that terminal differentiation markers are important pathway markers [ 7 ] and prognostic predictors [ 4 , 8 - 10 ], we compared the expression of terminal differentiation markers and some morphologic features of spontaneous and transgene-induced mammary tumors in mice. This study corroborates molecular [ 2 , 3 , 11 - 14 ] and immunohistochemical [ 4 , 7 , 15 ] studies in human cancer patients and mouse models of human cancer recommending that, in addition to the broadly accepted and well-documented use of architectural features [ 6 , 7 , 16 - 19 ], diagnostic criteria should also take into account the cell types into which neoplastic cells differentiate. We have identified neoplastic cells with luminal, myoepithelial, and mesenchymal phenotypes. Myoepithelial differentiation is 'constitutive' to specific tumor types, for example adenosquamous carcinomas, adenomyoepitheliomas, and adenocarcinomas associated with the active mouse mammary tumor virus (MMTV) and the wingless-related MMTV integration site 1 ( Wnt1 ) transgene. In contrast, in neoplasms dependent on the myelocytomatosis oncogene ( Myc ), Harvey rat sarcoma viral oncogene 1 ( Hras ), and SV40 T antigen ( SV40-TAg ) transgenes, myoepithelial differentiation is an important event in the specific type of tumor progression characterized by epithelial to mesenchymal transition (EMT). The occurrence in some transgenic models of distinct proteotypic expression patterns, defined as groups of neoplasms expressing a specific set of terminal differentiation markers, suggests an as yet unreported degree of variability in transgenic models, perhaps reflective of the diversity of second-hit mutations in these tumors. Finally, specific types of spontaneous mammary tumors are histologically and immunohistochemically indistinguishable from some of the transgene-induced mammary tumor models. This provides the background information to test the hypothesis that tumor phenotype is a predictor of tumor genotype. Methods Mice The pathology database [ 20 ] of The Jackson Laboratory was searched from 1987 to 2001 for spontaneous mammary tumors in production and research colonies. Slides were reviewed and categorized on the basis of standardized criteria [ 17 - 19 , 21 - 23 ]. In addition, newborns of the following strains were obtained from The Jackson Laboratory Induced Mutant Resource (Bar Harbor, ME): C57BL/6J-Tg(WapTAg)1Knw, FVB/N-Tg(WapNotch4)10Rnc/J, B6SJL-Tg(Wnt1)1Hev/J (hereafter abbreviated B6SJL-Wnt1), FVB/NJ-Tg(Wnt1)1Hev/J (hereafter abbreviated FVB-Wnt1), FVB/N-Tg(MMTVneu)202Mul/J, FVB/N-Tg(WapMyc)212Bri/J, FVB/N-Tg(WapHRAS)69Lln Y SJL /J, and FVB/NJ-Tg(MMTVTGFA)254Rjc/J and B6D2-Tg(MMTVTGFA)254Rjc/J. Mice were weaned at 3 weeks of age and genotyped by polymerase chain reaction. Detailed information on the genotyping protocols and breeding conditions are available online at . These mice were aged until they developed mammary masses, which were fixed by immersion in Fekete's acid–alcohol–formalin. Slides were sectioned at 5–6 μm and stained with hematoxylin and eosin (H&E). Immunohistochemistry Consecutive sections (5–6 μm thick) on Superfrost Plus slides (Fisher Scientific, Fair Lawn, NJ) of paraffin-embedded neoplasms were immunolabeled for α-smooth muscle actin (SMA; Sigma, St Louis, MO), keratin 1 (KRT2-1, hereafter abbreviated K1; BabCo, Richmond, CA), keratin 5 (KRT2-5, hereafter abbreviated K5; BabCo), keratin 6 (KRT2-6, hereafter abbreviated K6; BabCo), keratin 10 (K1-10, hereafter abbreviated K10; Babco), keratin 13 (KRT13, hereafter abbreviated K13; Sigma), keratin 14 (KRT1-14, hereafter abbreviated K14; BabCo), keratin 17 (KRT1-17, hereafter abbreviated K17; PA Coulombe, The Johns Hopkins University, Baltimore, MD) [ 24 ], keratins 8/18 (KRT8 and KRT18, hereafter abbreviated K8/18; Progen, Heidelberg, Germany), vimentin (Biomeda, Foster City, CA), filaggrin (BabCo), involucrin (Babco), loricrin (Babco), and trichohyalin (T-T Sun, The New York University School of Medicine, New York, NY) as described previously [ 25 ]. All these antibodies were mouse-specific except the antibodies against K13, trichohyalin, and vimentin, which were raised against human proteins, the antibody against K8/18, which was raised against bovine keratins, and the antibody against SMA, which was raised against avian proteins. The anti-SMA antibody recognizes aortic α2 smooth muscle actin (ACTA2) and enteric γ2 smooth muscle actin [ 26 ]. In brief, the method was as follows. Deparaffinized slides were gradually hydrated. Endogenous peroxidase activity was quenched by incubation with 3% hydrogen peroxide in methanol for 20 min at room temperature (20–25°C). Heat-mediated antigen retrieval in citrate buffer at pH 6.0 was used for the antibodies targeted at K8/18 and vimentin. Slides were washed and were incubated for 30 min with blocking serum (10% normal fetal calf serum diluted in phosphate-buffered saline). Excess blocking serum was blotted and the slides were incubated overnight at 4°C with primary antibodies diluted in phosphate-buffered saline. Secondary biotinylated anti-mouse or anti-rabbit antibody was applied for 30 min at room temperature, followed by incubation with the avidin-biotin complex (45 min). The reaction was developed with the substrate diaminobenzidine (Sigma) and the slides were counterstained with Mayer's hematoxylin. All spontaneous tumors were labeled for MMTV gp27 Gag, gp36 Env, and gp52 Env [ 27 ]. A mouse was identified as positive for MMTV if immunolabeling for at least one of the markers was detected in the tumor or nearby tissues. Images were captured with a DP70 digital camera (Olympus, Melville, NY) on an Olympus BX41 microscope and color enhanced and balanced for contrast with Photoshop 6.0 (Adobe, San Jose, CA). Additional photomicrographs of H&E-stained and immunolabeled slides are archived in the Mouse Tumor Biology Database, where they can be viewed on-line at [ 28 ]. Slide interpretation and scoring of immunolabeling intensity Cell types were defined on the basis of a previous study that evaluated the expression of the same terminal differentiation markers in the developing mammary gland [ 29 ]. Expression of K5, K14, K17, and/or SMA was associated with a direct contact with a basement membrane and the morphology of a myoepithelial cell defined myoepithelial cell differentiation. Absence of labeling for any of the markers, or labeling for K8/18 along with a polygonal morphology, defined luminal cells. Image analysis (Photoshop 6.0; Adobe, San Jose, CA) was used to evaluate the ratio of spindloid neoplastic cells on H&E-stained sections on all tumors with EMT. EMT was considered significant when more than 1% of the neoplasm was composed of neoplastic cells having a spindloid to fusiform shape and blending into the stroma. These cells had an oval to elongated nucleus with less clumped chromatin than epithelial neoplastic cells. Their cytoplasm was more acidophilic than epithelial neoplastic cells. Squamous differentiation was defined by the presence of one or more of the following features: the formation of a large core of cornified material, the formation of keratin pearls, the cornification of individual cells, or the presence of trichohyalin granules or keratohyalin granules. Squamous differentiation was confirmed by immunohistochemical expression of at least one of the following markers in the areas of squamous differentiation: K1, K6, K10, filaggrin, trichohyalin, involucrin, or loricrin, none of which is normally expressed in cycling mammary glands [ 29 ]. The intensity of immunolabeling was scored on a scale of 0 to 3 (immunolabeling levels: 0 = none; 1 = weak; 2 = moderate; 3 = intense). The grade for each cell type was the product of average labeling intensity and the relative percentage of cells (0%, 0; up to 1%, 1; 2–5%, 2; 6–25%, 3; 26–50%, 4; more than 50%, 5) labeled at the predominant intensity. The outcome for each neoplasm was the sum of the grades for each cell type. If immunolabeling was detected in less than 10 cells in a tumor, the tumor was considered to be negative for this marker. Morphologic criteria All tumors were evaluated (presence/absence) on H&E-stained sections for myoepithelial differentiation, EMT, squamous differentiation, vascular invasion, proteinaceous or lipid droplets in neoplastic cells, and ductal differentiation (defined as tubular structures lined by a basal layer of cells with myoepithelial differentiation and a luminal layer of cuboidal cells). Type P tumors are neoplasms composed of a branching network of blind ducts lined by an epithelium at least two cells thick and terminated by structures resembling the terminal end buds of the pubertal mammary gland [ 7 ]. Macrocysts are large cysts lined by an epithelium one or two cells thick that occasionally forms small glands. Lactation-responsive plaques are discoid masses composed of contoured and anastomosed mammary acini at the periphery and ducts at the center. Results Histologic tumor types (Table 1 ) Tumors were categorized histologically using the recommendations of the Mouse Models of Human Cancers Consortium categorization scheme [ 17 - 19 ] and the specific nomenclature applying to certain models [ 22 ] or tumor types [ 21 , 23 ] when applicable. Spontaneous tumors were papillary adenocarcinomas with papillae lined by a single layer of cells ( n = 6; Fig. 1 ), papillary adenocarcinomas with papillae lined by two layers of cells ( n = 2; Fig. 2 ), glandular adenocarcinomas with EMT ( n = 4; Fig. 3 ), microacinar adenocarcinomas in C3H/HeJ mice ( n = 8; Fig. 4 ), type P tumors in C3H/HeJ mice ( n = 3), adenomyoepitheliomas in BALBc/J mice ( n = 5; Fig. 5 ), and adenosquamous carcinomas ( n = 7; Fig. 6 ). All Hras -induced ( n = 4; Fig. 7 ) and Neu -induced ( n = 10) tumors as well as some Myc -induced ( n = 4; Fig. 8c,8e,8f ) and SV40-TAg-induced tumors (data not shown) had a solid pattern. Some SV40-TAg-induced ( n = 5; Fig. 9a,9b ) and Myc -induced tumors ( n = 9; Fig. 8a,8b,8d ) as well as all Notch4 -induced tumors ( n = 6) had a glandular pattern. Some Myc -induced ( n = 13; Fig. 8 ), SV40-TAg-induced ( n = 3; Fig. 9 ), and Hras -induced ( n = 3) carcinomas displayed areas of EMT. Some SV40-TAg-induced tumors ( n = 4) had a papillary pattern with papillae lined by epithelial cells that piled up in a disorderly fashion (data not shown). Preneoplastic lesions of Tgfa -transgenic mice consisted of macrocysts ( n = 8; Fig. 10a ) and lactation-responsive plaques ( n = 3; Fig. 10b ). All Wnt1 -induced tumors ( n = 13; Fig. 11 ) exhibited a type P tumor pattern. All spontaneous type P tumors ( n = 9), all microacinar carcinomas in C3H/HeJ mice ( n = 8), and one of two papillary carcinomas in one C3H/HeJ mouse showed immunolabeling for MMTV (data not shown). Proteinaceous and/or lipid secretion was identified in a small number of tumors in most transgenic models and spontaneous tumor types (Fig. 2a,2b ), although it was observed most consistently in macrocysts and in adenosquamous carcinomas (Fig. 6a,6c,6d,6e ). Terminal differentiation proteins expression patterns The neoplasms were categorized into three groups: those with a pure luminal phenotype (simple carcinomas), those that constitutively expressed myoepithelial and luminal markers, and those that were characterized by EMT. Hampe and Misdorp [ 30 ] defined the term complex as 'any type of neoplasm or proliferation composed of cells resembling both secretory epithelial and myoepithelial cells'. We therefore used the term 'constitutively complex carcinomas' to indicate the tumors in which luminal and myoepithelial cells were arranged as they would be in a normal mammary gland, suggesting that these two types of cells originated from the same progenitor cell. The tumors that expressed myoepithelial markers in the areas of EMT only are designated 'acquired complex carcinomas' to indicate that myoepithelial differentiation was absent in the areas representing the original phenotype of the neoplasms, before the occurrence of EMT. Tumor categorization was then redesigned to take into account not only the architecture of the neoplastic process [ 17 - 19 ] but also myoepithelial differentiation and EMT, which were not always apparent on H&E-stained sections. Keratins 5, 14, and 17, which are generally considered to be markers of myoepithelial cells [ 4 ], were occasionally identified in suprabasal cells in a variety of neoplasms (Figs 6f,6h,6i , 8c,8e,8f , 11a ). α-Smooth muscle actin, another marker of myoepithelial cells, was expressed by suprabasal cells only in MMTV-associated and Wnt1 -induced type P tumors (Fig. 12c ), a feature previously noted by Li and colleagues [ 13 ]. Neoplasms with a pure luminal phenotype (simple carcinomas) were identified in all transgenic models of mammary carcinogenesis (Fig. 7 , Table 1 ) with the notable exception of Wnt1 -induced carcinomas that had a constitutively complex phenotype. Most (six of nine) glandular carcinomas arising in mice transgenic for Myc showed a few areas of myoepithelial differentiation that were interpreted as early EMT rather than the neoplasms arising from a progenitor cell common to the luminal and myoepithelial phenotypes; hence these neoplasms with minimal myoepithelial differentiation were categorized as 'simple' carcinomas. The only spontaneous neoplasms composed exclusively of cells with a luminal phenotype were four of six papillary carcinomas that, on H&E sections, were characterized by papillae lined by an epithelium one cell thick. Constitutively complex carcinomas consisted of spontaneous papillary carcinomas lined by an epithelium two cells thick (Fig. 2 ), microacinar adenocarcinomas arising in MMTV-infected C3H/HeJ mice (Fig. 4 ), type P tumors of MMTV-infected C3H/HeJ mice (Fig. 12 ) and Wnt1 -transgenic mice (Fig. 11 ), adenomyoepitheliomas (Fig. 5 ), adenosquamous carcinomas (Fig. 6 ), and lactation-responsive plaques and most (six of eight) macrocysts of Tgfa -transgenic mice (Fig. 10 ). Neoplastic cells with a myoepithelial phenotype were abundant and formed an almost continuous layer surrounding cells with a luminal phenotype in microacinar carcinomas (Fig. 4 ) and in spontaneous papillary carcinomas lined by an epithelium two cells thick (Fig. 2a ). In type P tumors, myoepithelial differentiation was most prominent in the areas of ductal metaplasia and was minimal or absent in the frond-like areas of the neoplasms (Fig. 12a ). In adenosquamous carcinomas, myoepithelial differentiation was most prominent in the areas of squamous differentiation (Fig. 6f,6i ), and was often absent in the glandular areas. In adenomyoepitheliomas ( n = 5), large proportions of suprabasal cells were labeled for K5, K6, K8/18, K14, and K17, with K1 being expressed in only two tumors, and with duct formation and cornification being identified in four and three neoplasms, respectively. The morphology of neoplastic cells in adenomyoepitheliomas varied, ranging from fusiform to cuboidal. The proteotypic pattern of these cells is not characteristic of any specific cell type, and these cells were categorized as 'cells with myoepithelial differentiation' when in contact with a basement membrane and as 'luminal cells' when not in contact with a basement membrane. Neoplastic cells in adenomyoepitheliomas were difficult to differentiate from stromal myofibroblasts and, as reported earlier [ 31 ], they were negative for SMA. In lactation-responsive plaques of Tgfa -transgenic mice, myoepithelial differentiation was most prominent in the areas of ductal differentiation located at the center of these lesions (Fig. 10b ). The peripheral portions of lactation-responsive plaques, which morphologically resemble mammary acini, generally lacked myoepithelial differentiation. Macrocysts were predominantly composed of cells with a luminal phenotype (Table 1 ) that occasionally rested directly on the basement membrane. Segmental portions of six of eight macrocysts had myoepithelial cells (Fig. 10a ). Immunohistochemistry indicated that the descriptive term 'papillary carcinoma' encompasses a variety of neoplasms. First, most (four of six) spontaneous papillary carcinomas of BALB/cJ mice (Fig. 1 ) and all (four of four) papillary carcinomas with a pseudostratified epithelium in SV40-TAg-transgenic mice were exclusively composed of cells with a luminal phenotype (simple carcinomas). Cells with myoepithelial differentiation, identified in two of six spontaneous papillary carcinomas of BALB/cJ mice, were scarce and might represent entrapped remnants of the normal mammary gland, because they were located at the periphery of the neoplasm. In spontaneous papillary adenocarcinomas characterized by two layers of cells lining neoplastic papillae, the basal layer was continuous and had a myoepithelial phenotype (Fig. 2a ), whereas cells of the luminal layer expressed K8/18 (Fig. 2b ). Type P tumors, adenomyoepitheliomas, and adenosquamous carcinomas were characterized by the presence of subpopulations of cells with a high mitotic index that did not express terminal differentiation markers, or expressed them only weakly. In type P tumors, these cells were grouped at the ends of the fronds, the morphology and immunohistochemistry of which mimicked terminal end buds of the pubertal mammary gland [ 29 ]. Adenomyoepitheliomas (Fig. 5a,5b ) and adenosquamous carcinomas (Fig. 6f,6g,6h,6i ) were populated by individual cells and clusters of cells with a large open-faced nucleus and a moderate amount of pale cytoplasm with the appearance of ground glass. These cells were haphazardly distributed in the solid areas of adenomyoepitheliomas and in the viable epithelium enclosing cornified debris in adenosquamous carcinomas. Similar populations of neoplastic cells with low expression of terminal differentiation markers were not detected in the other neoplasms. Carcinomas in B6SJL- Wnt1 -and FBV- Wnt1 -transgenic mice were essentially similar and had the classical appearance of a type P tumor. However, they differed slightly with regard to their proteotypic patterns: squamous differentiation and expression of K5 and K8/18 was higher in B6SJL- Wnt1 -than in FBV- Wnt1 -transgenic mice (Table 1 ; Fig. 11a,11b ). Tumor progression EMT, an important morphologic marker of tumor progression [ 32 , 33 ], was detected in 23 tumors. It was most commonly observed in Myc -induced carcinomas although it was also identified in Hras -and in SV40-TAg-induced carcinomas, and in carcinomas spontaneously arising in SJL/J and PL/J mice. Various proportions of neoplastic cells with a mesenchymal phenotype expressed vimentin (Figs 3 , 8b ), K5 (Fig. 8c ), K14 (Fig. 8e ), and K17 (Figs 8f , 9b ). These cells consistently expressed SMA but could not be differentiated from myofibroblasts on the basis of the expression of this antigen alone (Fig. 8a ). EMT was also associated with increased expression of K8/18 (Figs 8d , 9a ). In most cases (20 of 23), a small number of cells with a myoepithelial phenotype were present in the glandular or solid areas in the immediate vicinity of the areas of EMT (Fig. 8c,8f ). Morphologic progression was also observed in microacinar adenocarcinomas associated with MMTV infection in C3H/HeJ mice, where it was characterized by focal to multifocal acquisition of a solid pattern. Acquisition of the solid pattern was associated with a loss of immunolabeling for MMTV proteins (data not shown) by neoplastic cells, but the expression of terminal differentiation markers remained unaltered. Neu -induced carcinomas did not display a morphologic alteration suggestive of tumor progression, although the tumors evaluated included both metastatic and non-metastatic neoplasms (data not shown). The markers tested in this study did not segregate the three cell types classically described in Neu -and Hras -induced tumors [ 7 ]. EMT was not observed in Notch4 -induced carcinomas, although the glandular pattern of these neoplasms is reminiscent of Myc -and SV40-TAg-induced carcinomas. Pathway pathology Three histologic features of mammary tumors have recently been related to the activation of specific pathways: myoepithelial differentiation has been associated with activation of the Wnt pathway [ 7 ], squamous metaplasia has been attributed to β-catenin stabilization [ 34 - 37 ], and alveolar differentiation and milk secretion are dependent on signal transducer and activator of transcription 5a [ 38 ]. As expected, all Wnt1 -induced carcinomas and spontaneous type P tumors displayed myoepithelial differentiation. In addition, myoepithelial differentiation, as assessed by routine H&E histology or immunohistochemistry, was identified in spontaneous papillary carcinomas with a bi-stratified epithelium, MMTV-associated microacinar carcinomas, adenomyoepitheliomas, and adenosquamous carcinomas. Squamous metaplasia associated with expression of the full array of terminal differentiation markers of the suprabasal layers of the epidermis was found only in adenosquamous carcinomas (Fig. 6a,6b,6c,6d,6e,6f,6g,6h,6i ). However, squamous metaplasia was also identified in some MMTV-associated (7 of 9) and Wnt1 -induced (12 of 13) type P tumors, Myc -induced carcinomas (7 of 19), adenomyoepitheliomas (3 of 5), and spontaneous carcinomas exhibiting EMT (1 of 4). In spite of areas histologically consistent with squamous differentiation, MMTV-associated and transgene-induced type P tumors express only a limited set of suprabasal markers of the epidermis (Table 1 ). For example, K1 and K10, two markers of the stratum spinosum, were seldom expressed (4 of 30 and 6 of 30 carcinomas, respectively) in neoplasms other than adenosquamous carcinomas that exhibited cornification, whereas involucrin, a marker of the stratum corneum, was expressed in 19 of 30 such neoplasms. Discussion In the present study, interpretation of immunohistochemistry results determined the following: first, that expression of terminal differentiation markers categorizes spontaneous and transgenic models of mammary neoplasia into 'simple' carcinomas, 'complex' carcinomas, and carcinomas with EMT; second, that most neoplasms are composed of multiple cell populations; and third, that some transgenic and spontaneous neoplasms are histologically and immunohistochemically indistinguishable, and they provide an opportunity to test the hypothesis that tumor phenotype predicts tumor genotype [ 6 , 7 ]. Proteotypic patterns: luminal, luminal/myoepithelial, and luminal with EMT Like human breast tumors [ 14 ], most mouse mammary tumors are exclusively composed of neoplastic cells with a luminal phenotype, suggesting that they developed from a progenitor cell committed to the luminal lineage. Progression to EMT through a myoepithelial stage was observed in Myc -, Hras -, and SV40-TAg-induced carcinomas, a subset of spontaneous mammary carcinomas in PL/J and SJL/J mice, and has also been reported in mice transgenic for Tgfa [ 39 ] and matrix metalloproteinase 3 [ 40 ]. Three major pathways have been implicated in EMT: the pathway controlled by the activation of Hras and Src, the transforming growth factor β pathway, and the Wnt pathway [ 32 ]. Evaluation of the expression of target genes of each of these pathways is needed to determine the molecular mechanisms associated with EMT in mouse models of mammary carcinogenesis. Type P tumors and microacinar carcinomas constitutively express myoepithelial markers, whereas expression of these markers is an indicator of progression in Myc -, Hras -, and SV40-TAg-induced carcinomas, as it is in some types of human breast cancer [ 4 , 15 , 41 ]. As a consequence, misinterpretations are likely to occur if tumors are evaluated for EMT with whole-tumor methodologies such as gene or protein array technologies. The genetic background of transgenic mouse models of mammary tumors might account for differences in the penetrance, latency, and phenotype of mammary proliferating lesions [ 42 - 44 ]. The histological phenotype of tumors arising in Wnt1 -transgenic mice was similar, regardless of the background. However, the background strain influenced the proteotypic pattern of these tumors. This observation emphasizes the importance of the background strain in genetically engineered mice, suggests the presence of tumor modifier genes, and indicates possible similar differences in humans. Cell populations The presence of cells negative or weakly positive for terminal differentiation markers establishes the presence of compartments of less differentiated cells in specific types of mammary carcinomas. This observation supports the hypothesis that undifferentiated cells, so-called 'mammary stem' cells, are the source of mammary tumors [ 45 , 46 ]. The histologic phenotype and the proteotypic pattern of type P tumors constitutes the most striking evidence for this hypothesis: the fronds in these carcinomas caricatured terminal end buds of the pubertal mammary gland, one of the compartments in which mammary stem cells are found [ 47 , 48 ]. Gene array data also support a stem cell origin for neoplasms arising in mice transgenic for Wnt1 [ 13 ]. In adenomyoepitheliomas and adenosquamous carcinomas, the location of cells that did not express terminal differentiation markers is suggestive of a ductal origin for these neoplasms. Interestingly, a second compartment of mammary stem cells is located in ductal suprabasal cells [ 49 ]. In addition to viable cells, a large proportion of neoplasms contained areas of cornification and necrosis. This heterogeneous group of non-viable cells has recently been named 'non-tumorigenic cancer cells' [ 50 ]. These cells need to be differentiated from neoplastic cells with self-renewal potential, including those cells with invasive or metastatic potential. Because the populations of 'non-tumorigenic cancer cells' might account for a significant proportion of some neoplasms, they might interfere with studies aiming at evaluating RNA or protein expression or genetic damage at the whole-tumor level. Pathway pathology Cardiff and colleagues [ 6 ] initially proposed that tumor architecture and terminal differentiation markers can predict tumor genotype. The concept has subsequently been refined [ 7 ] and is supported by molecular data [ 12 , 13 , 51 ]. The present study confirmed that simple carcinomas with a solid pattern are typical of tumors driven by the Neu and Ras pathways, whereas type P tumors are a feature of neoplasms arising in mice transgenic for Wnt1 . That type P tumors are also found in MMTV-infected mice was expected because Wnt1 is activated in 75% of mammary carcinomas arising in C3H/HeJ mice [ 52 ], and this observation supports the validity of the pathway pathology hypothesis. Although promising, the pathology pathway concept has some limitations that are illustrated by our data. For example, Myc -induced carcinomas express large amounts of whey acidic protein and casein β [ 53 ], which should correlate with histologic evidence of proteinaceous secretion. However, only 5 of 20 Myc -induced carcinomas examined in this study had evidence of proteinaceous or lipid secretion. In addition, the pathology pathway concept predicts that Myc -, Hras -, and Neu -induced carcinomas should share similar pathways, distinct from those of Wnt1 -induced carcinomas. However, Myc -and Wnt1 -induced neoplasms are independent of cyclin D1, whereas Hras -and Neu -induced carcinomas are dependent on cyclin D1 [ 12 , 54 ], indicating that tumor phenotype and tumor genotype are not consistently matched. Conclusions The observations that mammary neoplasms of mice transgenic for H-Ras , Myc , Neu , Notch4 and SV40-Tag , and that papillary carcinomas of BALB/cJ mice are exclusively composed of cells with a luminal phenotype, are consistent with the hypothesis that these neoplasms arise from a cell committed to the luminal lineage. The fact that adenomyoepitheliomas, adenoacanthomas, microacinar carcinomas, and type P tumors have a complex phenotype is consistent with a mammary stem cell origin. In addition, it is apparent that cancer progression in mammary carcinomas of mice transgenic for H-Ras , Myc , and SV40-Tag is often associated with EMT. However, EMT does not occur in mammary tumors of mice transgenic for Tgfa , Neu , Notch4 , and Wnt1 , which indicates that the morphologic features associated with cancer progression are determined by the initiating oncogenic events. Competing interests The authors declare that they have no competing interests. Abbreviations B6SJL-Wnt1 = B6SJL-Tg(Wnt1)1Hev/J; EMT = epithelial to mesenchymal transition; FVB-Wnt1 = FVB/NJ-Tg(Wnt1)1Hev/J; H&E = hematoxylin and eosin; Hras = Harvey rat sarcoma viral oncogene 1; K = keratin; MMTV = mouse mammary tumor virus; Myc = myelocytomatosis oncogene; SMA = α-smooth muscle actin; Wnt1 = wingless-related MMTV integration site 1.

D:\keerthana\PMC001xxxxxx\PMC1064077.xml

1064078

Birthweight, parental age, birth order and breast cancer risk in African-American and white women: a population-based case–control study

Introduction Much recent work has focused on hypotheses that very early life exposures influence adult cancer risk. For breast cancer it has been hypothesized that high in utero estrogen exposure may increase risk. Methods We used data from the Carolina Breast Cancer Study, a population-based case–control study of incident breast cancer in North Carolina, to examine associations for three possible surrogates of high prenatal estrogen exposure: weight at birth, maternal age, and birth order. We also examined paternal age. Birthweight analyses were conducted for white and African-American women born in North Carolina on or after 1949 (196 cases, 167 controls). Maternal age was analyzed for US born participants younger than 49 years of age (280 cases, 236 controls). Results There was a weak inverse association between birthweight in the highest tertile and breast cancer overall (odds ratio [OR] 0.7, 95% confidence interval [CI] 0.4–1.2), although associations differed by race (OR 0.5, 95% CI 0.2–1.0, and OR 1.0, 95% CI 0.5–2.1 for African-American and white women, respectively). For maternal age there was an approximately threefold increase in risk in women whose mothers were older than 22 years of age, relative to 19–22 years of age, when the women were born. After adjustment for maternal age, older paternal age increased risk in the oldest and youngest age categories (relative to 23–27 years of age at the woman's birth: OR 1.6, 95% CI 0.8–3.1 for age 15–22 years; OR 1.2, 95% CI 0.7–2.2 for age 28–34 years; and OR 1.5, 95% CI 0.7–3.2 for age 35–56 years). There was no association with older paternal age for white women alone. After adjustment for maternal age (265 cases, 224 controls), a birth order of fifth or higher relative to first had an inverse association with breast cancer for women younger than 49 years old (OR 0.6, 95% CI 0.3–1.3). Conclusion Although the CIs are wide, these results lend support to the possibility that the prenatal period is important for subsequent breast cancer risk, but they do not support the estrogen hypothesis as a unifying theory for the influence of this period.

Introduction Recent epidemiologic studies have investigated the possibility that very early life exposures increase adult cancer risk. Trichopoulos [ 1 ] postulated that a highly estrogenic intrauterine environment would create a 'fertile soil' for carcinogenesis in breast tissue and lead to higher risk for breast cancer later in life. Because retrospective prenatal hormone measurements cannot be obtained for large numbers of people, he and others proposed that birth and maternal characteristics be investigated as surrogates for a highly estrogenic intrauterine environment. These birth characteristics include high birthweight, maternal age 20–24 years at birth, and low birth rank. Much work during the past decade has been done on birthweight in particular. There is an apparent modest positive association between high birthweight and breast cancer that is stronger in younger women, which is consistent with the estrogen hypothesis. Data on other surrogates of intrauterine estrogen levels have been less consistent [ 2 ]. Despite an overall higher incidence of breast cancer in white women, incidence and mortality rates are higher in young African-American women than in young white women [ 3 , 4 ]. This crossover in incidence rates, occurring at about age 40 years for women diagnosed between 1950 and 1969, was documented in the Third National Cancer Survey [ 4 ]; SEER (Surveillance, Epidemiology and End Results) data from 1997 document a shift in the crossover to approximately 45 years of age [ 3 ]. Consequently, it is important to investigate relationships between putative causes of breast cancer and breast cancer incidence in younger African-American women. To our knowledge no studies published to date have specifically addressed the relationships between prenatal or birth characteristics and breast cancer in African-American women. The goal of this study was to characterize the relationships of birthweight, maternal age, paternal age, and birth order with breast cancer in African-American and white women in a population-based study. We analyzed data from a subset of women participating in the Carolina Breast Cancer Study (CBCS) [ 5 ], a population-based case–control study that over-sampled younger women and African-American women. Methods Study design and supplemental data collection The CBCS (phase I) is a population-based, case–control study of incident invasive breast cancer conducted between May 1993 and December 1996 in 24 counties of central and eastern North Carolina [ 5 ]. Participants gave informed consent using forms approved by the Institutional Review Board of the University of North Carolina School of Medicine, which were in compliance with the Helsinki Declaration. Cases ( n = 861) were women aged 18–74 years, who were mentally competent and resident in the study area at the time of selection with a first diagnosis of histologically confirmed primary invasive breast cancer. They were identified in cooperation with the North Carolina Central Cancer Registry [ 6 ] using a randomized recruitment protocol [ 7 ] to over-sample African-American women and women under 50 years of age. Potential controls were identified by North Carolina Division of Motor Vehicles (women aged 20–64 years) and/or Health Care Financing Administration (women aged 65–74 years) lists and had no previous or current history of breast cancer. Controls ( n = 790) were frequency matched by race and 5-year age group to cases. Trained nurse interviewers collected information and obtained height and weight measurements during interviews conducted at the participant's home. To obtain birthweight and parental ages we requested birth records for all study participants born in the USA on or after 1 January 1948. Analytic datasets Maternal age was analyzed in the subset of women for whom birth records with maternal age were available (Table 1 ) and was categorized as 15–18 years, 19–22 years, 23–27 years, or 28–44 years, based on homogeneity of risk apparent in smoothed lowess curves [ 8 ]. Paternal age was available for 92.7% of women with maternal age data, and was categorized as 15–22 years, 23–27 years, 28–34 years, or 35–56 years, by the same method. Maternal and paternal age distributions did not permit identical categorizations. Birthweight analyses were restricted to women born in North Carolina (NC-born; Table 1 ). Of case and control birth records, 96% and 97%, respectively, were located and nearly all (97.0% and 94.9%, respectively) recorded birthweight. A restricted birthweight dataset was constructed that excluded women who had any of the following indicators of a possibly poorly measured birthweight: noninstitutional birth, birth attendant other than a physician, and a birthweight recorded only in pounds. Overall, birthweight was recorded only in pounds more often for African-American than for white women (28% versus 7.4%). African-American women were more likely than white women to have been born at home, were less likely to have been delivered by a physician at home, and were less likely to have had a birthweight recorded in pounds and ounces under any birth circumstances. Hence, a disproportionate number of African-American women were excluded from the restricted dataset. Birthweights were converted from pounds and ounces to grams for analysis. Race-specific tertiles were derived from white or African-American controls. Birth order was analyzed twice, first in the full CBCS dataset and then in the subset of women for whom there was information on maternal age (Table 1 ). Birth order was self-reported and was a categorized as first, second to fourth, and fifth or higher. Statistical analysis Unconditional logistic regression was used for all analyses. Odds ratios (ORs) and 95% confidence intervals (CIs) were the primary measure of association. PROC GENMOD in SAS (SAS Institute Inc., Cary, NC, USA) was used with an offset term to account for the age and race specific sampling probabilities used to identify eligible participants [ 7 ]. All estimates presented are adjusted for, at minimum, age and sampling fractions. Except for birthweight, parental age, and body mass index (BMI), all variables were based on self-report. Age at diagnosis (cases) or selection (controls) was categorized using the same 5-year age groups as the sampling protocol (20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, and 65–69 years). The proportion of non-African-American participants who classified themselves as non-white was under 1%, and so, for the purposes of this study, non-African-Americans were classified as white. Age at menarche (≤12 years or >12 years), age at first full-term pregnancy (nulliparous, <26 years, ≥26 years), lactation (nulliparous, ever breastfed, or never breastfed), parity (none, one, two, or three or more), BMI (≤25 kg/m 2 or >25 kg/m 2 , calculated from nurse interviewers' measurements of height and weight), first degree family history of breast cancer (positive if a mother, father, or full sibling had breast cancer), and menopausal status were considered potential confounders. Women younger than 50 years were considered postmenopausal if they had undergone natural menopause, bilateral oophorectomy, or irradiation to the ovaries. Multivariable logistic models were used to adjust for potential confounders [ 9 ]. A potential confounder was included in the model based on a >15% change in the β coefficient for any level of the birth characteristic relative to the referent in either white or African-American women. Lowess curves were generated using Stata version 7.0 (Stata Corporation, College Station, TX, USA). All other analyses were done using SAS version 8.01. Results Racial distributions for analytic datasets are presented in Table 1 . Overall, the proportion of African-American women was higher for the younger NC-born women (i.e. those eligible for the birthweight analysis) than for the full CBCS or for younger CBCS participants (born on or after 1 January 1948). Those eligible for the maternal and paternal age analyses were under 48 years of age at selection/diagnosis. Consequently, the proportion of postmenopausal women was much lower in this group than in the full dataset (11% versus 55%), as was mean age at menopause (controls 39.2 ± 6.1 years versus 44.4 ± 7.3 years). There were somewhat higher proportions of women with first births at age greater than 26 years, no family history of breast cancer, household income above the study median, higher educational level, and nonrural childhoods. Only minor differences emerged between the women eligible for the parental age analyses and those for whom parental age was obtained. This subgroup of women was slightly more likely to have had rural childhoods and lower education. The birthweight analysis was restricted to younger NC-born women. These women reported, on average, only slightly lower educational level, more rural childhoods, lower household income, lower age at first birth, and higher BMI than did younger women overall. No important differences in breast cancer risk factors emerged between those eligible for the birthweight study and those for whom birthweight was obtained. ORs for age at menarche, age at first pregnancy, and lactation were virtually identical in the full CBCS and all analytic datasets. Differences in ORs between the full CBCS dataset and the birthweight dataset were, as expected, due to age restriction in the latter. In the birthweight dataset, ORs for family history were slightly higher (OR 1.6, 95% CI 1.0–2.7 versus OR 1.4, 95% CI 1.0–1.9), whereas ORs for the following were slightly lower: BMI greater than 25 kg/m 2 (OR 0.6, 95% CI 0.4–0.8 versus OR 0.8, 95% CI 0.6–1.0), postmenopausal status (OR 0.7, 95% CI 0.4–1.5 versus OR 0.9, 95% CI 0.7–1.2), and parity of three or greater (OR 0.5, 95% CI 0.2–1.1 versus OR 0.8, 95% CI 0.6–1.1). Risk factor distributions The distributions of birthweight, parental age, and birth order are presented in Table 2 . Birth records were unavailable from some states, increasing the proportion of NC-born women in the maternal age dataset. The birthweight dataset was restricted to NC-born participants because of unavailability of birthweight on most out-of-state birth records. Ages at diagnosis/selection were similar in analytic datasets and relevant subgroups of CBCS cases and controls. Maternal age, paternal age, and birth order were also similarly distributed in datasets including only younger women. As expected, the frequency of birth orders higher than fourth was lower among the younger women. Birthweight Tables 3 and 4 present ORs and 95% CIs for the association between birthweight categories and breast cancer in the full and restricted datasets, combined and by race, respectively. Overall, there was a weak inverse association between higher birthweight and breast cancer in the full dataset but not in the restricted dataset. Higher birthweight was inversely associated with breast cancer among African-American women in the full and restricted datasets, although CIs were wide. There was no association for higher birthweight in white women for the full dataset and a modest but statistically nonsignificant positive association for the restricted dataset. There was no association between lower birthweight and breast cancer for white or African-American women in the full birthweight dataset. As has historically been the case in North Carolina and elsewhere [ 10 ], mean and median birthweight and lower and upper limits of birthweight distributions among controls were higher for whites than for African-Americans. Neither prenatal characteristics nor adult BMI were strongly correlated with birthweight (data not shown). Maternal and paternal age ORs for maternal and paternal age and breast cancer are presented in Table 5 . After adjustment, older maternal age (>22 years of age) increased ORs approximately threefold, whereas older paternal age (>27 years of age) was more weakly associated with breast cancer. Maternal and paternal ages, as categorized, were moderately correlated (Spearman correlation coefficient 0.73, 95% CI 0.69–0.78). Parental ages were somewhat correlated with birth order (Spearman correlation coefficients 0.47, 95% CI 0.40–0.54 and 0.43, 95% CI 0.35–0.50 for maternal and paternal ages, respectively). After full adjustment, the OR was elevated and of borderline statistical significance for maternal age 15–18 years among African-American women but not among white women (Table 6 ). ORs for maternal age over 22 years were increased twofold to fivefold, with 95% CIs usually excluding the null, for both white and African-American women. The odds of breast cancer for all categories of maternal age were slightly stronger for first-born participants, although this was not statistically significant (data not shown). After adjustment for maternal age, birth order, adult BMI, and household income, there was no association between paternal age and breast cancer among white women. For African-American women, ORs were elevated for both younger (15–22 years of age) and older (35–56 years of age) paternal ages, although CIs were wide. There was no substantial difference in results when the maternal/paternal age datasets were restricted to women born on or after 1 January 1949. Among controls, parental ages were distributed similarly by race, with African-American participants having slightly higher mean maternal and paternal ages than whites. Birth order Results for analyses of birth order are shown in Tables 7 and 8 . In the full CBCS dataset, birth order (categorized as first, second to fourth, or fifth or higher) was not associated with breast cancer, overall or by race. In the full CBCS, mean birth order was higher for African-Americans than for whites; this pattern was stronger in younger CBCS participants. No potential confounders met the criteria for model inclusion. Finer categorization of birth order did not change the results. For younger women, adjustment for maternal age, adult BMI, and household income revealed a weak, statistically nonsignificant, inverse relationship between higher birth order and odds of breast cancer, which did not differ appreciably by race. Discussion We examined the relationships between breast cancer and birthweight, parental age, and birth order among women younger than 49 years of age residing in North Carolina. Overall, there was a weak inverse association with higher birthweight, which was stronger in the full dataset than in the restricted dataset. For white women in the study there was no overall association between birthweight and breast cancer. Among white women born in medical facilities, birthweight in the highest tertile was positively associated with breast cancer, but CIs were wide and included the null. Higher birthweight was inversely associated with breast cancer for African-American women regardless of delivery setting, but again CIs were wide for all estimates. Most previous studies have reported weak to modest positive associations between higher birthweight and breast cancer [ 11 - 22 ], with some showing a positive dose response [ 11 , 16 - 19 ]. The only previous report of an overall weak inverse association between higher birthweight and breast cancer was from an Asian population [ 23 ], although similar results were reported among older, white women [ 16 , 19 ]. Two studies have shown no association [ 24 , 25 ]. With the exception of the Asian case–control study [ 23 ], previous studies of birthweight have been done in exclusively or predominantly white populations: seven large record-based, nested case–control studies in Scandinavian cohorts [ 11 , 12 , 14 , 16 , 17 , 22 , 24 ]; five population-based, case–control studies in the USA [ 13 , 19 , 20 , 25 , 26 ]; one US cohort study [ 18 ]; and one cohort study [ 21 ] and a cross-sectional study [ 15 ] conducted in the UK. All previous studies of birthweight and breast cancer included younger women; 10 presented results for premenopausal or younger women separately [ 13 , 14 , 16 - 21 , 23 , 26 ]. Associations were generally positive in younger women, with ORs ranging from 1.25 (95% CI 1.0–1.6) [ 17 ] to 3.5 (95% CI 1.3–9.4) [ 16 ] for birthweights of 4000 g or greater, as compared with the ORs of 1.4–1.5 for birthweight greater than 3458 g among younger white women with reliable birthweights observed in the present study. No previous studies have estimated ORs for African-American women. In the USA maternal report and self-report have been the most common sources of birthweight information, rather than birth records. Andersson and coworkers [ 27 ] conducted an analysis of agreement between self-reported birthweight and birth records. They found that, despite good overall agreement (Spearman correlation coefficient 0.76), 31% of self-reported birthweights differed from birth record data by 500 g or more, and that this level of misclassification led to both underestimation and overestimation of the magnitude and significance of various effect estimates. Moreover, nonresponse can be sizable (ranging from 12% [ 19 ] to 24% [ 28 ]) and may reflect bias toward healthier [ 29 ], more educated, and/or more communicative mothers. In the USA, birthweight was only routinely recorded on birth records of younger women, the group in which the association between birthweight and breast cancer appears to be strongest [ 2 ]. The two US studies that used birth records (conducted in Hawaii [ 26 ] and New York state [ 13 ]) employed a design similar to that of the present study – a population-based, case–control study using cases born in the state where they were recruited. Both observed minimal, statistically nonsignificant, increased risks for breast cancer among women in the highest tertile of birthweight relative to those in the central tertile. The present study is intermediate in sample size between these two studies, which included 74 and 484 cases, respectively. In addition to using birth records to decrease misclassification, we performed a separate birthweight analysis for the subset of women who were delivered by physicians in hospitals or doctors' clinics and had their birthweight recorded in pounds and ounces. During the 1950s, home birth and delivery by lay midwives was common practice in North Carolina, particularly among African-Americans and in rural areas, and this could have affected data collection [ 10 ]. Additionally, participants delivered in a medical setting comprise a subgroup of women more closely comparable to previous study participants than do women born at home. Results for white women from this subset were in agreement with previous literature, with a small positive association between birthweight and breast cancer, whereas birthweight remained inversely associated for African-American women in the restricted dataset with comparable delivery circumstances. Although analyses in this restricted group potentially reduce birthweight misclassification, results may have limited generalizability to less medically advantaged populations. Although the search strategy employed in this study limited nonlocatable birth records to 3.6% of the study population, and records missing birthweight to 4.0% of locatable records (i.e. data available for 92% and 93% of eligible cases and controls, respectively), the missing records were predominantly those of African-American women who self-reported birth in the more rural counties of North Carolina. The inverse associations between higher birthweight and breast cancer seen in this study could also partly be explained by selection bias in the full CBCS dataset if either the case group under-represented the proportion of high birthweight women in the underlying case population or the control group under-represented the proportion of normal weight births in the underlying population. Because birthweights in North Carolina have been increasing over time, more strongly in whites than in African Americans [ 30 ], younger white women would be expected on average to have the highest birthweights, and this group is slightly over-represented rather than under-represented in the case population. Some under-representation of African-American women in the control population (36.5% response rate for younger African-Americans) could have contributed to an upward bias in the control birthweights. In the context of a relatively disadvantaged population such as this one, a higher birthweight may be a surrogate for a different constellation of prenatal and postnatal influences than in a relatively advantaged population. Rather than viewing birthweight solely as an indicator of a highly estrogenic prenatal environment, or even specific physiologic processes, birthweight can also be viewed more globally as an indicator of the prepregnancy health of the mother [ 31 , 32 ]. In this context, it could be considered predictive of the general overall health of the daughter as well and perhaps of a decreased susceptibility to some etiologic agents. Socioeconomic status, based on study participants' self-reported current household income, was not found to be a confounder in this study, but it is probably a poor surrogate for complex environmental influences such as early diet, physical activity, or childhood residence. If there is either a general or breast cancer specific survival advantage to having a higher birthweight, then one would expect to see an inverse association between birthweight and breast cancer among older women. This was found in one study of birthweight and breast cancer [ 19 ] but not in another [ 18 ], although an apparent protective effect of higher birthweight has been found for other chronic diseases [ 19 , 33 ]. Inasmuch as birthweight is a good surrogate for higher intrauterine estrogen levels, these data do not support the hypothesis that in utero estrogen exposure increases risk for breast cancer. One limitation of our birthweight analysis is that, lacking an accurate measure of gestational age, it is not possible to interpret fully the association between lower birthweight and breast cancer. Although Andersson and coworkers [ 11 ] reported increases in risk associated with birthweight after adjustment for gestational age, particularly after additional adjustment for age at menarche, evidence is inconsistent for gestational age as a strong confounder [ 2 , 14 ]. In the birthweight analyses, power (the probability of correctly rejecting the null hypothesis) to detect an OR of 1.5 at the 95% confidence level was low for whites and African-Americans (0.33 and 0.61, respectively). Similarly, power to detect an OR of 0.5 was low for whites and African-Americans (0.25 and 0.50, respectively). Therefore chance cannot be ruled out as an explanation for the results. Although stratification allowed us to characterize the relationships between these early life factors and breast cancer in African-American women, power to detect an overall effect was decreased. Because this is the only study to date that presents data on African-American women, further research should be undertaken. Older maternal age exhibited a moderate positive association with breast cancer in this study. Study participants whose mothers were aged 23 years or older at the participant's birth had approximately twofold to fourfold higher odds of breast cancer than did women whose mothers were between 19 and 22 years of age. Although African-American women whose mothers were aged under 19 years also had elevated odds of breast cancer, the pattern did not differ appreciably between white and African-American women. Although the magnitude of OR for women whose mothers were aged 23–27 years was greater than in previous studies, the findings were consistent with the majority of previous reports: weak positive associations with older maternal age [ 13 , 24 - 26 , 34 - 42 ], with stronger associations (approximate doubling in the oldest categories) found for younger women [ 13 , 26 , 43 ]. Several studies, however, reported no association with older maternal age [ 12 , 17 , 19 , 43 , 44 ]. Innes and coworkers [ 13 ] reported a similar J-shaped relationship between maternal age and breast cancer for women diagnosed before age 33 years; the lowest risk was for those aged 20–24 years, with an approximate doubling of odds for women whose mothers were older than 35 years old at their birth. Collectively, these data do not support highest risk being associated with maternal age 20–24 years, as predicted by the estrogen theory [ 1 ]. Recent evidence, however, indicates that the association between maternal age and levels of pregnancy estrogens may be weaker than was previously thought [ 45 , 46 ]. Alternatively, older oocytes may have sustained more genetic damage over time and/or DNA repair may be deficient in older mothers [ 35 , 47 ]. The pattern of association between having an older father (older paternal age) and breast cancer was somewhat different for white and African-American women. After adjustment for maternal age and birth order, paternal age was not associated with breast cancer for white women. This is consistent with previous reports for white women showing little to no effect of paternal age [ 34 , 38 , 41 , 42 ]. For African-American women in the present study, positive associations with breast cancer were seen for those with the youngest (age 15–22 years) and oldest (age 35–56 years) fathers at their birth, even after adjustment for maternal age and birth order. This is broadly consistent with the only previous study of paternal age to include African-Americans (10.1 % of participants, 52 matched case–control pairs) [ 13 ]. In that study Innes and coworkers found an elevated risk associated with older paternal age, after adjustment for maternal age and birth order (OR 1.3, 95% CI 0.9–1.7 for age 30–34 years; OR 1.2, 95% CI 0.8–1.7 for age 35–39 years; and OR 1.3, 95% CI 0.8–2.0 for age ≥40 years), and there was some suggestion of effect modification by race. Although speculative, the differing risk patterns for paternal age in white and African-American women may reflect differing exposures for white and African-American men at that time and place, which could have affected mutation rates. Birth records with parental age could only be obtained for 76% of CBCS participants born in or after 1948, and were missing almost exclusively by participants' state of birth. Breast cancer mortality is generally higher in the midwest and northeast regions of the USA than in the southeast [ 48 ], and so if participants with higher birthweights from those areas were systematically excluded then bias toward the null would be expected. However, there was no regional pattern to the missing birth records; therefore, this was unlikely to have introduced substantial bias. Paternal information was collected only when the mother was married. Although the proportion of unmarried parents is small, this could have introduced bias. In women younger than 50 years of age, higher birth order exhibited a weak inverse association with breast cancer only after adjustment for maternal age. No association was seen in the full CBCS, which included women aged up to 74 years, although data were not available to adjust for maternal age. This is consistent with the majority of previous studies, which have shown either a weak inverse association with breast cancer [ 19 , 26 , 36 , 49 ] or no association [ 22 , 24 , 37 ]. A weak positive relationship (OR 1.05, 95% CI 1.01–1.10 per 1 unit increase in birth order) was found by Hemminki and Mutanen [ 50 ]. Few studies were able to adjust for maternal age. Because pregnancy estrogens appear to be highest in first pregnancies and decline in successive pregnancies [ 51 , 52 ], these results lend some support to the theory that prenatal estrogen exposure may influence breast cancer later in life. Birth order was self-reported and may have been misclassified. Although the number of previous maternal pregnancies could be a better measure of prenatal estrogen exposure than live birth order, we could not assess this because of poor data quality on the birth records. Younger African-American women are at higher risk for breast cancer than younger white women [ 3 , 4 ]. Birthweight, patterns of parental age at birth, and birth order continue to vary by race [ 53 ]. Our study has several important strengths. Use of birth records as the source of birthweight information improved accuracy of the exposure measurement, eliminated recall bias caused by self-report, and reduced possible selection bias from maternal report. Similarly, birth records improved data quality for parental age. Using a population-based case–control study made it possible to evaluate a wider range of adult-life risk factors as potential confounders and/or effect modifiers than is generally possible in a registry-based study. Conclusion Taken as a whole, the results for birthweight, parental age, and birth order from the present study do not support the estrogen exposure hypothesis as a unifying theory for prenatal influence on adult breast cancer. This emphasizes the importance of further investigating the influence of prenatal factors on breast cancer risk, particularly in multiple populations. Additional hypotheses must be pursued, including the association between birthweight and other hormonal exposures such as insulin-like growth factor I, and between maternal age and endogenous and exogenous mutagenic exposures. Methodologic difficulties involved in investigating prenatal exposures in nonwhite and/or disadvantaged populations are not trivial; nonetheless, this type of investigation must be done to fully understand life course processes that can culminate in breast cancer among women of any background. Author contributions MEH, BN and RCM participated in the interpretation of results and writing of the manuscript. MEH performed data collection, data entry, and statistical analyses. Competing interests The authors declare that they have no competing interests. Abbreviations BMI = body mass index; CBCS = Carolina Breast Cancer Study; CI = confidence interval; OR = odds ratio.

D:\keerthana\PMC001xxxxxx\PMC1064078.xml

1064080

Frequency of CHEK2 mutations in a population based, case–control study of breast cancer in young women

Introduction The cell-cycle checkpoint kinase (CHEK)2 protein truncating mutation 1100delC has been associated with increased risk for breast or prostate cancer. Multiple studies have found an elevated frequency of the 1100delC variant in specific stratifications of breast cancer patients with a family history of the disease, including BRCA1 / BRCA2 negative families and families with a history of bilateral disease or male breast cancer. However, the 1100delC mutation has only been investigated in a few population-based studies and none from North America. Methods We report here on the frequency of three CHEK2 variants that alter protein function – 1100delC, R145W, and I175T – in 506 cases and 459 controls from a population based, case–control study of breast cancer conducted in young women from western Washington. Results There was a suggestive enrichment in the 1100delC variant in the cases (1.2%) as compared with the controls (0.4%), but this was based on small numbers of carriers and the differences were not statistically significant. The 1100delC variant was more frequent in cases with a first-degree family history of breast cancer (4.3%; P = 0.02) and slightly enriched in cases with a family history of ovarian cancer (4.4%; P = 0.09). Conclusion The CHEK2 variants are rare in the western Washington population and, based on accumulated evidence across studies, are unlikely to be major breast cancer susceptibility genes. Thus, screening for the 1100delC variant may have limited usefulness in breast cancer prevention programs in the USA.

Introduction Cell-cycle checkpoint kinase (CHEK)2 has been shown to play a role in cell cycle regulation, apoptosis, and DNA repair, at least in part through phosphorylation of p53 and BRCA1 in response to DNA damage [ 1 , 2 ]. Several studies have reported associations of germline mutations in CHEK2, especially the 1100delC mutation, with increased susceptibility to breast and prostate cancer [ 3 - 8 ]. Although CHEK2 germline variants other than 1100delC have been associated with prostate cancer risk, these have not yet been shown to be enriched in breast cancer cases [ 3 , 4 , 9 , 10 ]. The association between the CHEK2 1100delC variant and risk for breast cancer was initially reported by the CHEK2 Breast Cancer Consortium [ 5 ]. They found that the frequency of the variant was greater among breast cancer patients with a positive family history of breast cancer who do not carry germline mutations in the BRCA1 or BRCA2 genes, and in families with male breast cancer, as compared with healthy control individuals from the UK, The Netherlands, and North America [ 5 ]. Additionally, they noted that the frequency of the 1100delC variant did not differ significantly between breast cancer patients and matched control individuals from a population-based series of young women from the UK (age < 45 years) and of older women from The Netherlands (age ≥ 55 years) [ 5 ]. However, neither population-based series included frequency data after stratifying for family history characteristics. Several additional studies have addressed the association of the 1100delC variant and breast cancer risk in unique populations. In a Finnish study conducted by Vahteristo and coworkers [ 6 ], the frequency of the 1100delC mutation was observed to be slightly but not significantly higher in an unselected cohort of breast cancer patients than in control individuals (identified from the Finnish Red Cross Blood Service). Significant enrichment of the variant was found among index cases with a first-degree or second-degree relative with breast or ovarian cancer, and in women with bilateral breast cancer as compared with patients with unilateral disease. Finally, analysis of the variant in a set of patients with positive family history who were not BRCA1 or BRCA2 germline mutation carriers demonstrated a significantly elevated frequency of the 1100delC variant as compared with controls. These findings from Vahteristo and coworkers [ 6 ] and recent work from Oldenburg and colleagues [ 7 ] are similar to data from the CHEK2 Breast Cancer Consortium, and suggest a significant role played by the 1100delC variant in breast cancer among women with a positive family history of breast cancer whose disease is not attributable to germline mutations in BRCA1 or BRCA2 . Finally, a study from New York by Offit and coworkers [ 8 ] reported a lower frequency of 1100delC carriers in both breast cancer cases and controls as compared with previous studies that largely included Northern European individuals. The 1100delC mutation was identified in 1.0% of cases, which was not statistically different ( P = 0.10) from that observed among controls (0.3%), who were volunteers from the New York Cancer Project. Compared with the general population frequency in New York, the 1100delC variant appears to be even rarer among breast cancer patients from Spain [ 11 ] and India [ 12 ], where studies to date have reported no individuals with the 1100delC variant. To understand better the association of CHEK2 variants and breast cancer risk in the general population in the USA, we analyzed the frequency of three CHEK2 variants – 1100delC, R145W, and I175T, each of which reportedly alters CHEK2 protein function – in a population based, case–control study of 506 breast cancer cases diagnosed before age 45 years from western Washington state, and a set of 459 frequency matched control individuals. Methods Study population A characterization of the study population has previously been reported and is summarized only briefly [ 13 , 14 ]. Cases were identified through the Cancer Surveillance System of Western Washington, a population-based cancer registry and a participant in the National Cancer Institute's Surveillance, Epidemiology, and End Results Program (SEER). Control individuals were identified through random digit dialing and were frequency matched to the cases on 5-year age group and reference year [ 15 ]. The study identified all incident first primary breast cancer cases diagnosed before age 45 years, from May 1 1990 to December 31 1992, in women of all races and ethnic backgrounds, who were residents of King, Pierce and Snohomish counties at the time of diagnosis. Information on potential risk factors for breast cancer, including family history, was obtained through a structured in-person interview. The reference date for the interview, a date beyond which exposure information was not collected, was the month and year of diagnosis for cases and a randomly assigned date for controls. Interviews were completed for 642 cases (84.0%) and 608 controls (73.8% overall response rate). Blood was collected from 540 interviewed cases and 476 interviewed controls. Tested cases tended to be older than untested cases from the study ( P = 0.001) whereas no such age-related differences were seen in controls. Untested cases were more likely to have advanced stage disease (51.0% of tested and 41.2% of untested cases had local stage disease, 30.2% and 40.4% had regional disease, and 1.4% and 5.9% had distant disease; P = 0.001) and were more likely to be deceased at the last follow up in June 2002 (16.6% of tested and 48.8% of untested cases were deceased; P < 0.001). For 40% of participants, blood collection was not attempted until after the initial interview, probably accounting in part for these differences. We observed no difference in cases or controls between those tested and untested with regard to family history. Molecular methods Batches of DNA for genotyping were constructed to contain both case and control samples, and genotyping personnel were blinded as to the case–control status of samples. Previously described specific primers for CHEK2 exon 10 were used for PCR amplification [ 16 ]: 5'-TTA ATT TAA GCA AAA TTA AAT GTC-3' and 5'-GGC ATG GTG GTG TGC ATC-3'. Genomic DNA (25 ng) was amplified using the AccuPrime TAQ DNA polymerase system (Invitrogen, Carlsbad, CA, USA). Touchdown PCR conditions for the 1100delC amplicon were as follows: denaturation at 94°C for 1 min then 94°C for 30 s, 60°C for 30 s, and 68°C for 30 s for seven cycles with the annealing temperature decreasing by 1°C for each cycle, followed by an additional 28 cycles of 94°C for 30 s, 54°C for 30 s, and 68°C for 30 s. The resulting 556 bp amplicon was analyzed by unidirectional DNA sequencing with the reverse primer. The R145W and I175T variants were sequenced from a 409 bp amplicon generated using the following primers: 5'-TTG CCT TCT TAG GCT ATT TTC C-3' and 5'-AAA GGT TCC ATT GCC ACT GT-3'. As above, 25 ng genomic DNA with AccuPrime TAQ DNA polymerase was amplified by touchdown PCR, in which the starting annealing temperature was 64°C and the final annealing temperature was 58°C. For sequencing, the Applied Biosystems Big Dye Terminator Ready Reaction Mix (Foster City, CA, USA) was used in accordance with the manufacturer's recommended protocol. Genotyping was conducted in 506 cases and 459 controls. Valid results for all participants were obtained for the R145W and I175T variants, whereas results for one case and one control were not obtained for the 1100delC variant. Analysis To assess the relationship between CHEK2 variants and breast cancer risk, logistic regression was used to obtain odds ratios as estimates of the relative risk and 95% confidence intervals [ 17 ]. All analyses were completed using Stata statistical software (StataCorp LP, College Station, TX, USA). Because reference age and year were matching variables for the frequency matching employed in the original study, all risk estimates presented are age (continuous)-and reference year (exact)-adjusted. A subset of the samples analyzed in the study had been screened previously for germline mutations in the breast cancer susceptibility genes BRCA1 and BRCA2 [ 18 , 19 ]. Cases were selected for BRCA1 / BRCA2 screening on the basis of an age of diagnosis under 35 years and/or a first-degree family history of breast cancer ( n = 134). In addition, 235 controls were tested for mutations in the BRCA1 gene, and 37 of these controls were additionally tested for BRCA2 . Overall, 110 cases and 33 controls were available for consideration of CHEK2 variant frequencies in BRCA1 / BRCA2 mutation negative subjects. Results Study population characteristics Tested cases and controls were generally similar with regard to age, menopausal status, and racial distribution (Table 1 ). Approximately 90% of all participants were Caucasian. Cases more often reported a family history of breast cancer than did controls, particularly a first-degree family history, which was reported by 19.0% of cases and 8.1% of controls. Variants and risk for breast cancer Overall, no statistically significant differences were observed in frequency between cases and controls for any of the three variants tested (Table 2 ) and all three variants studied were uncommon. For the 1100delC mutation, a 2.9-fold increased risk was observed among cases compared with controls, because six (1.2%) out of 505 cases and two (0.4%) out of 458 controls carried the variant. However, the confidence interval did not exclude 1 and thus chance cannot be excluded as an explanation (95% confidence interval 0.6–14.6). We examined the frequencies of the CHEK2 variants according to age, race, and family history features of the probands (Table 3 ). All 1100delC deletion carriers were Caucasian. Among the cases, 0.7% (2/280) of those with no family history of breast cancer, none (0/120) with only a second-degree family history, and 4.3% (4/94) of those with a first-degree family history were found to carry the 1100delC variant ( P = 0.02). Among controls, 0.3% (1/294) of those with no family history, 0.9% (1/115) of those with only a second-degree family history, and none (0/36) of the controls with a first-degree family history carried the 1100delC variant. One case, or 2.4% of those with a family history of bilateral breast cancer, and one control, or 3.2% of those with a similar family history, were carriers. Cases with a positive first-degree or second-degree family history of ovarian cancer carried an 1100delC variant more frequently (4.4% [2/45]) than did cases with no such family history (0.9% [4/461]; P = 0.09). No controls (0/27) with such family history were carriers. Furthermore, 9.1% (2/22) of the cases with a positive family history of both breast and ovarian cancer were found to carry the 1100delC variant ( P = 0.07). In the overall dataset, only three cases and two controls reported a family history of male breast cancer, and none carried the 1100delC variant. The R145W variant was rare in this data set, with only one case and no controls carrying the variant (Table 2 ). The carrier case was diagnosed before age 30 years, was Caucasian, and reported one first-degree relative with breast cancer who was diagnosed after age 45 years and no family history of ovarian cancer, bilateral breast cancer, or male breast cancer (data not shown). The I175T change was observed in two (0.4%) of 506 cases and four (0.9%) of 459 controls (odds ratio 0.5, 95% confidence interval 0.1–2.6; Table 2 ). One control carrier was non-Caucasian. One case (0.4%) and three controls (1.0%) with no family history of breast cancer carried the I175T change. The other case carrying this variant was in the group with a first-degree family history of breast cancer (1/94 [1.1%]) and the other control carrying this variant was among controls with a second-degree family history (1/115 [0.9%]). None of the I175T carriers had a family history of ovarian cancer, bilateral breast cancer, or male breast cancer. Non- BRCA1 / BRCA2 carriers and CHEK2 mutations Because some studies suggest that the CHEK2 1100delC variant acts as a breast cancer modifier in non- BRCA1 / BRCA2 families only [ 5 - 7 ], we considered the subset of women known not to be BRCA1 or BRCA2 germline mutation carriers. Within this subset of 110 cases and 33 controls, four (3.6%) cases and no (0%) controls carried the 1100delC variant ( P = 0.27), one case and no controls carried the R145W change, and one case and no controls carried the I175T change. No CHEK2 variants were observed in any known BRCA1 or BRCA2 mutation carrier. Discussion We analyzed three CHEK2 variants that are known to disrupt protein function (1100delC, R145W, and I175T) in a population based, case–control study of breast cancer among young North American women. The 1100delC variant is a protein truncating mutation that abrogates CHEK2 kinase activity [ 20 ]. R145W has been shown to have disrupted kinase activity [ 20 , 21 ] and I175T is deficient in binding and phosphorylation of Cdc25A and in binding to BRCA1 and p53 [ 20 - 22 ]. Although an enrichment in the 1100delC variant and a reduction in I175T carriers in the cases were noted, no statistically significant association between any of the CHEK2 variants and breast cancer risk was observed. The absolute number of participants carrying CHEK2 variants was relatively small, and thus there was limited power to examine frequencies according to family history features. Nonetheless, among cases there was some suggestion that the 1100delC variant may be slightly more frequent in those with a positive first-degree family history of breast cancer ( P = 0.02) and in those with any family history of ovarian cancer ( P = 0.09). However, in agreement with two other breast cancer studies [ 9 , 10 ], we observed no suggestive correlation between the R145W and I175T CHEK2 variants and breast cancer risk. No CHEK2 variants were seen among women found previously to carry a BRCA1 / BRCA2 mutation. Our overall frequency results for 1100delC of 1.2% for cases and 0.4% for controls are similar to the frequencies reported previously from the UK, Philadelphia, and New York [ 5 , 6 , 8 ]. In a population based series of individuals from the UK and The Netherlands, the frequency of the 1100delC variant was higher among cases, but did not differ significantly from a set of matched controls (1.3% and 2.5% for cases and 0.3% and 1.2% for controls, respectively) [ 5 ]. Likewise, the frequency of 1100delC in a Finnish series of breast cancer patients was similar to that reported among control individuals from the Finnish Red Cross Blood Transfusion Service (2.0% and 1.4%, respectively; P = 0.18) [ 6 ]. Finally, in North America the 1100delC variant was identified in 1.6% of index cases from breast cancer families in Philadelphia and in 0.6% of control individuals (from the same neighborhood or spouses marrying into a breast cancer family from the same area) [ 5 ]. In New York examination of the 1100delC variant in 192 women with a family history of breast cancer, 92 women with a personal history of breast cancer, and 16 male breast cancer patients [ 8 ] revealed a mutation frequency of 1.0%, which did not differ significantly from the frequency of 0.3% found in volunteers for the New York Cancer Project ( P = 0.10). Several previously published studies [ 5 - 7 , 23 ] reported an elevated frequency of the CHEK2 1100delC variant in specific stratifications of breast cancer patients. Specifically, individuals with positive family history (especially those who are BRCA1 / BRCA2 mutation negative), patients with bilateral disease, and patients with a family history of male breast cancer had a higher occurrence of 1100delC variants as compared with control individuals. We found no CHEK2 variants in women with a family history of male breast cancer, but there were only five individuals with such a history in our entire sample. This finding is similar to those of other recent studies that did not find an association between 1100delC and risk for male breast cancer [ 24 - 26 ]. Although the frequency of 1100delC carriers was higher in cases (2.4%) and controls (3.2%) with a family history of bilateral disease as compared with cases (0.7%) and controls (0.3%) with no family history and cases (1.8%) and controls (0) with only a family history of unilateral disease, this was based on sparse data and family history of bilaterality contributed no insights beyond family history overall. After stratifying by family history, we did observe an elevated frequency of 1100delC carriers among cases with a first-degree family history (4.4%; P = 0.02). Although our numbers are small, this frequency is similar to the frequencies reported by others. Vahteristo and coworkers [ 6 ] reported that, among 1035 breast cancer patients, 3.1% of those with at least one affected first-degree or second-degree relative were 1100delC carriers. Additionally, in index cases with a family history of breast cancer, Meijers-Heijboer and coworkers [ 5 ] observed that 3.0% (31/1036) were 1100delC carriers. Thus far, the most convincing evidence for an association between the 1100delC variant and breast cancer risk is in families who do not carry BRCA1 / BRCA2 germline mutations [ 5 - 7 ]. However, the 1100delC frequency in BRCA1 / BRCA2 mutation positive families did not differ significantly from the frequency observed among controls [ 5 , 6 ]. In this study we observed that four of 110 cases (3.6%) and none of 33 controls who were known to be BRCA1 or BRCA2 negative carried the CHEK2 1100delC variant. The number of women in the present study with a first-degree family history of breast cancer who tested negative for BRCA1 / BRCA2 mutations (71 cases, 27 controls) does not offer adequate power to detect differences in the frequency of CHEK2 variants within this stratification. The significance of the CHEK2 1100delC mutation in individuals with a family history of ovarian cancer is not as well understood. Vahteristo and coworkers [ 6 ] found no association between the 1100delC variant and ovarian cancer family history among women with familial breast cancer (0/40). However, Meijers-Heijboer and colleagues [ 5 ] reported that 4.0% of index cases or 4.3% of all cases with at least one family member with ovarian cancer carried the 1100delC variant ( P = 0.016). This is compatible with the frequency we observed (2/45 [4.4%]) among breast cancer cases with a family history of ovarian cancer. Although the numbers are small, our data suggests that further investigation into the association between the CHEK2 1100delC mutation and ovarian cancer risk is warranted. The results of our study should be assessed with regard to its limits. Specifically, there are differences between tested and untested women. Tested cases were more likely to be alive, older, and have a less advanced stage of cancer than untested cases. Thus, the generalizability of these results, although from a population-based study, must be viewed within that context. As noted earlier, the literature is diverse in terms of its estimates of CHEK2 mutation frequency. Although the overall sample size of our study was generous (965 women), the frequency of the CHEK2 variants turned out to be quite low. As a result, the study had somewhat reduced power, particularly for assessing mutation frequency according to various family history characteristics. Conclusion The population based, case–control study of young women (age at diagnosis < 45 years) presented here does not identify any of the 1100delC, R145W, and I175T variants as major factors in breast cancer susceptibility in western Washington. After stratification by family history characteristics, an association with first-degree family history of breast cancer and possibly family history of ovarian cancer was observed. However, no particular relationship was found with family history of bilateriality or family history of male breast cancer. These results suggest that incorporation of any CHEK2 variants into a breast cancer screening program among Caucasian women in the US would be premature. Additional studies, particularly of women with a family history of breast cancer who do not carry mutations in the BRCA1 or BRCA2 genes, are warranted. Competing interests None declared. Abbreviations bp = base pairs; CHEK = cell-cycle checkpoint kinase; PCR = polymerase chain reaction.

D:\keerthana\PMC001xxxxxx\PMC1064080.xml

1064081

Ratios of involved nodes in early breast cancer

Introduction The number of lymph nodes found to be involved in an axillary dissection is among the most powerful prognostic factors in breast cancer, but it is confounded by the number of lymph nodes that have been examined. We investigate an idea that has surfaced recently in the literature (since 1999), namely that the proportion of node-positive lymph nodes (or a function thereof) is a much better predictor of survival than the number of excised and node-positive lymph nodes, alone or together. Methods The data were abstracted from 83,686 cases registered in the Surveillance, Epidemiology, and End Results (SEER) program of women diagnosed with nonmetastatic T1–T2 primary breast carcinoma between 1988 and 1997, in whom axillary node dissection was performed. The end-point was death from breast cancer. Cox models based on different expressions of nodal involvement were compared using the Nagelkerke R 2 index (R 2 N ). Ratios were modeled as percentage and as log odds of involved nodes. Log odds were estimated in a way that avoids singularities (zero values) by using the empirical logistic transform. Results In node-negative cases both the number of nodes excised and the log odds were significant, with hazard ratios of 0.991 (95% confidence interval 0.986–0.997) and 1.150 (1.058–1.249), respectively, but without improving R 2 N . In node-positive cases the hazard ratios were 1.003–1.088 for the number of involved nodes, 0.966–1.005 for the number of excised nodes, 1.015–1.017 for the percentage, and 1.344–1.381 for the log odds. R 2 N improved from 0.067 (no nodal covariate) to 0.102 (models based on counts only) and to 0.108 (models based on ratios). Discussion Ratios are simple optimal predictors, in that they provide at least the same prognostic value as the more traditional staging based on counting of involved nodes, without replacing them with a needlessly complicated alternative. They can be viewed as a per patient standardization in which the number of involved nodes is standardized to the number of nodes excised. In an extension to the study, ratios were validated in a comparison with categorized staging measures using blinded data from the San Jose–Monterey cancer registry. A ratio based prognostic index was also derived. It improved the Nottingham Prognostic Index without compromising on simplicity.

Introduction Breast cancer is the most common neoplasm in women. Nodal status as determined by pathologic examination of lymph nodes has repeatedly been shown to be the single most important predictor of survival in breast cancer [ 1 ]. The absolute number of pathologically involved nodes has also been shown to be an important prognostic factor in breast cancer survival [ 2 - 6 ]. The extent of lymph node involvement is incorporated into prognostic indices such as the Nottingham Prognostic Index [ 7 - 9 ] (see Additional files 1 , 2 , 3 , 4 , 5 , 6 ). Old lymph node stage measures categorized cases according to whether they had none, one to three, or four or more involved nodes, and recently according to more detailed subdivisions [ 10 , 11 ] (see Additional files 2 , 3 , 4 , 5 and 7 ). However, several authors have noted the inherent confounding by the number of excised nodes [ 12 , 13 ]. To address the variability of nodal examination, an intuitive approach is to use the proportion or the percentage of involved nodes, as was suggested by Rostgaard and coworkers [ 14 ]. The proportion can immediately be derived from pathology reports that clearly state the total number of lymph nodes examined and the total number of involved nodes [ 1 , 15 ]. The proportion has received increasing attention in the literature, providing a reference base on which its clinical relevance may be discussed [ 13 , 14 , 16 - 27 ]. In this report the modeling utility of the proportion of involved nodes is compared with the absolute numbers of involved nodes and of examined nodes. There is a one-to-one correspondence between proportion and ratio between involved and uninvolved nodes. A previous study hinted at an apparent linear relationship with survival between involved and uninvolved nodes (Fig. 1 ) [ 28 ], and therefore this report also examines the utility of expressing ratios as odds instead of proportions. The absolute numbers considered in the study were the number of nodes examined (excised; nx), the number of involved nodes (np), and the number of uninvolved nodes (nn). Methods The SEER (Surveillance, Epidemiology, and End Results) program of the USA [ 29 ] provides extensive cancer incidence data from 11 population-based registries. The data used in the present study were extracted from nine of those registries: San Francisco-Oakland, Connecticut, Metropolitan Detroit, Hawaii, Iowa, New Mexico, Seattle (Puget Sound), Utah, and Metropolitan Atlanta. Selected patients were women without a previous history of cancer who presented with a noninflammatory invasive breast carcinoma, which was histologically confirmed and diagnosed between 1988 and 1997, with specified tumor size no larger than 50 mm (T1 and T2), strictly confined to breasts without distant metastasis, and in which curative surgery and axillary lymph node dissection were performed with removal of at least one node. Cases with involvement of skin, hypodermis or pectoral muscles, or with deep fixation were excluded. Patients who had undergone subcutaneous mastectomy, radical mastectomy, or preoperative or intraoperative radiotherapy were excluded. Data on systemic treatment were not available and therefore could not be taken into account. Certain records were rejected because of data quality concerns: uncertain sequence of treatment, nonhospital based data records, month of diagnosis unknown, or race unknown. Examination of outliers (scarce and extreme values) resulted in further exclusion of cases with more than 50 nodes examined, '0 months' of follow up, and age at diagnosis under 25 years or older than 95 years. The follow-up cut-off date was 31 December 1999. The survival end event was defined as death from breast cancer. The proportions of involved nodes were expressed as percentages ([np/nx] × 100%). The log odds of nodal involvement were computed using the empirical logistic transform: L = Log e ([np + 0.5]/ [nn + 0.5]) [ 30 ]. The transform, also called the sample logit, avoids singularities caused by null observations, and is the least biased estimator of the true log odds [ 31 ]. (Note that, with hindsight, Fig. 1 shows a logarithmic relationship.) Unadjusted mortality (the number of patients who died divided by the number of patients at risk) as a function of the ratios was used for descriptive purposes. The utilities of the percentage and log odds were evaluated in different multivariate Cox proportional hazards models [ 32 ]. The numbers np and nx, the percentage (np/nx) × 100%, and the L transform were entered as quantitative continuous variables in different combinations. The corresponding hazard ratios were each time computed within a Cox model that included tumor size, age at diagnosis, and year of diagnosis modeled as quantitative continuous variables; and the registry area, race, marital status, tumor topography, histologic type and grade, estrogen and progesterone receptor status, type of primary surgery, and administration of postoperative radiotherapy modeled as qualitative variables. The qualitative variables were converted or expanded as needed into dummy variables to allow binary coding ('married' versus 'not married', 'high grade' versus 'not high grade', for example, and so on). A first order interaction between type of surgery and postoperative radiotherapy was included for consistency with a previous analysis [ 33 ]. The models were computed in all cases irrespective of nodal status, and then as a function of positive or negative nodal status. The functional forms were assessed using the generalized additive model procedure [ 34 ]. The Nagelkerke R 2 index (R 2 N ) was used to score the different Cox models [ 35 ]. R 2 represents the proportion of variation explained by covariates in regression models [ 35 - 37 ]. R 2 N divides R 2 by its maximum attainable value to scale it to within the range 0–1. R 2 N is close to 1 for a perfectly predictive model, and close to 0 for a model that does not discriminate between short and long survival times. Statistical analyses were performed using Splus (Insightful Corporation, Seattle, WA, USA) statistical software. Results In the 2002 SEER release [ 29 ], 188,410 women were diagnosed with breast tumors from 1988 to 1997, of whom 132,457 had a hospital based histopathologic diagnosis of unilateral invasive carcinoma. A total of 83,686 cases matched the selection criteria; 58,070 were node-negative and 25,616 node-positive. The median follow-up time was 73 months (range 1–143 months) for patients still alive at the follow-up cut-off date (31 December 1999). Characteristics of the patients were presented elsewhere [ 33 ]. Except for some additional cases due to updated registration minus the exclusion of outliers resulting in 90 fewer cases, there were no noticeable differences in the distribution of the characteristics. The median number of nodes examined was 15 (range 1–50, mean ± standard deviation 15.4 ± 6.5). Among the node-positive patients, the median number of involved nodes was 2 (1–46, 4.1 ± 4.8). Table 1 shows the distribution of the percentages of involved nodes. Figure 2 is a plot of the corresponding breast cancer mortality, which appears to increase linearly with the np/nx percentage. Table 2 shows the distribution of the log odds for nodal involvement. Figure 3 plots the corresponding breast cancer mortality. There is an initial, almost flat segment for values of L ≤ -3, which is followed by a steeply sloping upward segment. The initial flat segment corresponds mostly to node-negative cases. The sloping upward segment corresponds to node-positive cases, with more positive L values indicating more involved nodes and/or fewer uninvolved nodes. There is an overlap between node-negative and node-positive cases for L values between -3.5 and -1. In multivariate analyses, np and nx exhibited marked nonlinearity and widely diverging confidence intervals (Fig. 4a,4b ). The linearity improved for the percentage (np/nx) × 100% and the L transform, which also showed more homogeneously distributed confidence intervals (Fig. 4c,4d ). The upper section of Table 3 shows a comparison between proportional hazards models that included different combinations of np, nx, np/nx, and L for all patients, irrespective of nodal status. Based on R 2 N , the best predictive covariate was L (model 6), with a small improvement contributed by nx (model 10). In the model with nx alone (model 3), nx was statistically significant but its contribution to global model fit appeared negligible because the R 2 N did not change from the baseline 0.069 (model 1). The contribution of np alone was substantial, with a change of R 2 N from baseline 0.069 to 0.093 (model 2). However, adding np and/or nx onto L or onto np/nx provided no improvement, except in the already mentioned model 10. The middle section of Table 3 shows multivariate analysis performed for node-positive cases only. Models based on separately expressed numbers provided the lowest R 2 N (models 2–4). The largest R 2 N values were all observed in models incorporating L or np/nx (models 5–11). The simplest model appeared to be based on np/nx alone (model 5; R 2 N = 0.108). A small improvement was contributed by np (model 7; R 2 N = 0.109). The lower section of Table 3 shows the analysis performed for node-negative cases only. Because np, by definition, equals 0, there are only four models. They show that nx (model 3) and L (model 6) are statistically significant, but these variables either alone or in combination did not improve the index R 2 N . The multivariate computations were also performed by considering death from any cause as the end-point. There were no notable discrepancies. Discussion Although many data were evaluated in the present study, there are weaknesses. The data are heterogeneous. Histopathologic characteristics such as grade could not be verified. Neoadjuvant systemic treatment might have modified the yield of nodes [ 38 ]. Important information such as how patients were selected for any particular treatment is missing. Undocumented comorbidity might have affected the extent of nodal dissection. An imbalance in the delivery of chemotherapy or hormone therapy could have affected the distribution of deaths. For all of these reasons, the present results should be considered explorative and must be validated independently. Since about 1999 a growing number of studies have investigated nodal ratios. In the studies that compared the numbers of involved nodes with ratios in multivariate models, the majority found that ratios were better than numbers as prognostic indicators [ 13 , 16 , 18 , 19 , 24 , 26 ]. Ratios (expressed as percentages or log odds) have a better prognostic impact than do isolated numbers and, unlike numbers, they are not associated with inconsistent findings. Part of the explanation might be that a ratio can be interpreted as a form of standardization in which the number of involved nodes found in a patient is standardized to the number of nodes examined in that same patient [ 20 ]. It is noteworthy that the hazard ratios for np/nx were almost unaffected by the model (column np/nx [%] in Table 3 ), whereas the hazard ratios for np and/or nx exhibited more variability (columns np and nx in Table 3 ). As a prognostic factor, np/nx appears the most convenient. Figure 2 shows that, for node-positive cases presenting with 0–10% involved nodes, the crude breast cancer mortality risk for an average follow up of 6 years is about 5%, and with 90–100% involved nodes the mortality is about 45%. For any intermediate value for the percentage of node involvement, the mortality risk is easily interpolated. The estimated log odds L provided results very similar to those with np/nx. Overall, L improves on np/nx when all cases are considered together (column L in Table 3 ). The log odds appears useful for integrating node-negative and node-positive cases while avoiding more complex modeling, which we performed previously [ 39 ]. However, there is a range of L values in which node-negative and node-positive cases overlap (Table 2 , Figure 3 ). In an analysis of all cases pooled, the overlap might blur the prognostic difference between node-negative status based on a very small number of excised nodes, and node-positive status based on a large number of excised nodes but with few involved nodes. The literature on the log odds of node involvement is scarce, and the utility of the L transform needs independent confirmation. The present findings indicate that the favorable survival attributed to higher numbers of nodes removed, as suggested by Krag and Single [ 40 ], might be due to different model specifications. The number of patients is huge and statistical significance can easily be demonstrated but without necessarily implying any major clinical impact. Undoubtedly, the uncertainty about node negativity increases when nx (the number of excised nodes) is small. However, the predictive utility attributable to nx is exceedingly small (Table 3 , lower section). This dissociation between statistical significance and predictive utility appears counterintuitive. Nevertheless, it is in keeping with findings from Fisher and coworkers [ 41 ], who noted that prognosis was unaffected by the number of excised nodes when nodal status was reported to be negative. This is also supported by a recent report based on 3800 patients [ 42 ] in which the number of excised nodes was predictive of the risk for recurrence in node-positive but not in node-negative patients. Sentinel node biopsy has gained wide acceptance since 1997 and it is used to determine the need for axillary dissection [ 43 ]. Because our selection of patients was from 1988 until 1997, it is unlikely that sentinel nodes could have represented any substantial part of the present study. The prognostic impact of one involved node in patients who had one node removed in this study cannot be extrapolated to the patient found with one involved node in a sentinel node procedure. However, in the prediction of nonsentinel node involvement when one or more sentinel nodes are found to be involved, Cserni and coworkers [ 44 ] reported that the number of sentinel nodes and the percentage of positive sentinel nodes were jointly significant predictors. A closely related finding that also highlights the predictive role of ratios was reported in a recent Australian study [ 45 ], in which the prediction model was determined by patient age, by the number of sentinel nodes, and by the proportion of involved sentinel nodes. Conclusion We found the percentage of involved nodes to be the most directly useful indicator of nodal involvement, but this is limited to node-positive cases. The log odds of nodal involvement performed equally well in node-positive and node-negative patients. The log odds might provide a unified approach to the modeling of nodal involvement. The present results and the growing literature argue that ratios should be considered in the staging of axillary dissection. Competing interests The author(s) declare that they have no completing interests. Abbreviations L = empirical logistic transform (estimated log odds); nn = number of axillary lymph nodes free from tumor involvement; np = number of pathologically involved axillary lymph nodes; nx = number of axillary lymph nodes examined (excised); R 2 N = Nagelkerke R 2 index. Supplementary Material Additional File 1 Text on deriving a ratio based prognostic index. Click here for file Additional File 2 Table evaluating nodal staging measures using breast cancer data from the San Jose–Monterey registry: all cases irrespective of nodal status. Click here for file Additional File 3 Table evaluating nodal staging measures using breast cancer data from the San Jose–Monterey registry: node-positive patients. Click here for file Additional File 4 Table providing a simulation of small datasets of 300 breast cancer patients, irrespective of nodal status. Click here for file Additional File 5 Table providing a simulation of small datasets of 300 node-positive breast cancer patients Click here for file Additional File 6 Figure showing Kaplan–Meier survival estimates for T1–T2 breast cancer abstracted from the San Jose–Monterey registry. Click here for file Additional File 7 Text on ratios and the TNM categorization. Click here for file

D:\keerthana\PMC001xxxxxx\PMC1064081.xml

1064082

Over-expression of lysophosphatidic acid receptor-2 in human invasive ductal carcinoma

Introduction Lysophosphatidic acid (LPA) is a bioactive phospholipid with diverse effects on various cells. It interacts with at least three G-protein-coupled transmembrane receptors, namely LPA1, LPA2 and LPA3, whose expression in various tumours has not been fully characterized. In the present study we characterized the expression profile of LPA receptors in human breast cancer tissue and assessed the possible roles of each receptor. Methods The relative expression levels of each receptor's mRNA against β-actin mRNA was examined in surgically resected invasive ductal carcinomas and normal gland tissue using real-time RT-PCR. LPA2 expression was also examined immunohistochemically using a rat anti-LPA2 monoclonal antibody. Results In 25 cases normal and cancer tissue contained LPA1 mRNA at similar levels, whereas the expression level of LPA2 mRNA was significantly increased in cancer tissue as compared with its normal counterpart (3479.0 ± 426.6 versus 1287.3 ± 466.8; P < 0.05). LPA3 was weakly expressed in both cancer and normal gland tissue. In 48 (57%) out of 84 cases, enhanced expression of LPA2 protein was confirmed in carcinoma cells as compared with normal mammary epithelium by immunohistochemistry. Over-expression of LPA2 was detected in 17 (45%) out of 38 premenopausal women, as compared with 31 (67%) out of 46 postmenopausal women, and the difference was statistically significant ( P < 0.05). Conclusion These findings suggest that upregulation of LPA2 may play a role in carcinogenesis, particularly in postmenopausal breast cancer.

Introduction Lysophosphatidic acid (LPA; 1- or 2-acyl- sn -glycero-3-phosphate), the simplest glycerophospholipid, mediates a broad range of cellular responses including smooth muscle cell contraction, platelet aggregation, neurite retraction/cell rounding, regulation of cell proliferation, protection from apoptosis, modulation of chemotaxis and transcellular migration [ 1 - 3 ]. Previous reports have also demonstrated that LPA stimulates proliferation, migration and production of matrix metalloproteinases and angiogenic factors in ovarian cancer cell lines [ 4 - 9 ]. Moreover, a high level of LPA has been detected in ascitic fluid from ovarian cancer patients [ 4 , 10 , 11 ]. These findings suggest that LPA may be a mediator of tumour development and progression [ 12 ]. At least three receptors – LPA1/Edg-2, LPA2/Edg-4 and LPA3/Edg-7 – have been identified as specific receptors for LPA [ 13 , 14 ]. These LPA receptors differ with respect to their distribution in various tissues. LPA1 is broadly expressed in various normal tissues, whereas expression levels of LPA2 and LPA3 are more restricted, which may account for the various biological effects of LPA [ 13 , 14 ]. In recent studies LPA1 and LPA2 were found to have redundant functions in mediating multiple endogenous LPA responses, including phospholipase C activation, calcium mobilization, cell proliferation and stress fibre formation in mouse embryonic fibroblast cells [ 15 , 16 ]. However, in the same study the different phenotypes of knockout mice for each receptor suggested distinct roles for LPA1 and LPA2 in vivo . Previous studies also found that malignant transformation resulted in aberrant expression of LPA2 or LPA3 in ovarian or thyroid cancers, suggesting that LPA may play a role in tumour biology and that shifts in LPA receptor expression are related to carcinogenesis [ 17 - 20 ]. However, patterns of expression of these LPA receptors in other tumours have not been adequately examined. In the present study we characterized the expression patterns of LPA receptors in human breast cancer tissue (another common form of cancer in females) and analyzed correlations with other clinical and pathological findings to determine the possible role played by each LPA receptor in the development of breast cancer. Methods Patients and materials In the first part of the study, mRNA expression for each LPA receptor was evaluated in 25 cases of invasive ductal carcinoma, which were surgically resected in the Department of Surgery, University of Tokyo, from 1998 to 2003, and in six samples of adjacent normal gland tissue. In the next part of the study, protein expression of LPA2 was evaluated by immunohistochemical staining in the 25 cases and an additional 59 cases of invasive ductal carcinoma, which were resected from 1992 to 1997, also in the Department of Surgery, University of Tokyo. All of the resected primary tumours and regional lymph nodes were histologically examined by haematoxylin–eosin staining, in accordance with the International Union Against Cancer TNM classification. Several discrete histological parameters and lymph node metastasis were also examined. Oestrogen receptor levels were evaluated using enzyme immunoassay of frozen tumour specimens with cutoff levels for positivity of 5 fmol/mg protein. Isolation of total RNA and reverse transcription The tumour tissue resected from the primary lesion and paired nontumour tissue (taken 10 cm away from the neoplasm) were immediately frozen in liquid nitrogen and kept at -80°C until extraction of RNA. Total RNA was extracted from each sample using the acid guanidine isothiocyanate/phenol/chloroform extraction method. One microgram of total RNA was reverse transcribed using a SuperScript First-Strand Synthesis System (Invitrogen Co., Carlsbad, CA, USA). The reverse transcription reaction was carried out in a total volume of 20 μl, in accordance with the manufacturer's instructions. The cDNA was stored at -20°C until use. Preparation of cDNA calibrators We used pcDNA3 vector (Invitrogen) containing LPA1, LPA2, or LPA3 as the standard cDNA for each LPA receptor [ 14 ]. cDNA for human β-actin was prepared using human colon cDNA as the template DNA and the following oligonucleotides: CCATC GAATTC ACCACCATGGATGATGATATCGCCGCGCTC and AAGGT GCGGCCGC CTAGAAGCATTTGCGGTGGACGAT. The resulting DNA fragments were digested by Eco RI/ Not I and ligated into pBlueScript (Strategene, La Jolla, CA, USA). The DNA sequence of cDNA prepared by RT-PCR was confirmed by DNA sequencing and used to calibrate for β-actin. Real-time fluorescence quantitative PCR The primers used in the analysis of LPA1, LPA2, LPA3 and β-actin gene expression are given in Table 1 . All the primers were designed using Primer Express software (Applied Biosystems, Foster, CA, USA). Real-time PCR reactions were conducted in an ABI PRISM 7000 (Perkin-Elmer/Applied Biosystems, Foster, CA, USA) using SYBR Green I (Perkin-Elmer) with the following profile: one step at 50°C for 2 min, one step at 95°C for 10 min, and 40 cycles at 95°C for 30 s and 60°C for 1 min. Thermocycling was done in a final volume of 20 μl containing 1 μl of cDNA sample. The ABI PRISM software constructed the calibration curve by plotting the crossing point against the logarithm of the number of copies for each calibrator. The number of copies in unknown samples was calculated by comparing their crossing points with the calibration curve. To correct for differences in both RNA quality and quantity between samples, data were normalized using the ratio of the target cDNA concentration to that of β-actin. Both RT and PCR reactions were performed in triplicate, and the mean values were calculated against against β-actin. Monoclonal antibodies against human LPA2 A peptide consisting of the carboxyl-terminal 17 amino acids of human LPA2 (335–351) was conjugated with keyhole limpet haemocyanin. The conjugate was injected into the hind foot pads of WKY/Izm rats using Freund's complete adjuvant. The enlarged medial iliac lymph nodes from the rats were used for cell fusion with mouse myeloma cell line PAI. The antibody-secreting hybridoma cells were selected by screening with enzyme-linked immunosorbent assay, immunofluorescence and Western blotting. Several clones were established. In this study, monoclonal antibody from clone 10D5 (rat IgG 1 ) was used for immunohistochemical staining. Specificity of the monoclonal antibody against human LPA 2 used in the present study (clone 10D5) was confirmed as follows. HeLa cells were transfected with LPA 1 -pcDNA3, LPA 2 -pcDNA3, LPA 3 -pcDNA3, or empty pcDNA3 vector using lipofectamine 2000 (Invitrogen). Twenty-four hours after transfection, the cells were rinsed with phosphate-buffered saline (PBS) and recovered using a cell scraper. Protein (10 μg) was separated on SDS-PAGE (12.5% polyacrylamide) and transferred onto nitrocellulose transfer membrane (Schleicher & Schuell, Einbeck, Germany). Using the enhanced chemiluminescence (ECL) method, the expression of each receptor was confirmed using anti-LPA1, anti-LPA2, or anti-LPA3 antibodies (Fig. 1 ). Immunohistochemical study of LPA2 Expression of LPA2 was examined by immunohistochemical staining using the anti-LPA2 antibody. Sections (3 μm thick) were deparaffinized in xylene, hydrated through a gradually diluted ethanol series, and heated in a microwave oven for two 7-min cycles (500 W). After being rinsed in PBS, endogenous peroxidase activity was inhibited by incubation with 0.3% hydrogen peroxide in 100% methanol for 30 min. After washing three times in PBS, nonspecific reaction was blocked by incubation with PBS containing 5% skimmed milk for 30 min at room temperature, and then the sections were incubated with normal rabbit serum for 30 min. The sections were incubated overnight at 4°C in humid chambers with the primary antibody to LPA2 at a dilution of 1/100. As negative control, anti-LPA1 monoclonal antibody (rat IgG 1 ), as well as control rat IgG (Nichirei, Tokyo, Japan), was used. After washing three times with PBS, the sections were incubated with biotinylated rabbit antirat immunoglobulin for 30 min. After washing again with PBS, the slides were treated with peroxidase-conjugated streptavidin for 30 min, and developed by immersion in 0.01% hydrogen peroxide and 0.05% diaminobenzidine tetrahydrochloride for 3 min, followed by light counterstaining with Mayer's haematoxylin. Immunoreactivity was assessed by two evaluators who had no knowledge of the background features. When more than half of the carcinoma cells in these fields exhibited significantly stronger staining intensity than the corresponding epithelium from normal mammary glands, these tumours were defined as having high expression of LPA2. In other tumours, most of the carcinoma cells exhibited only weak, if any, immunostaining, and, if present, the staining intensity was similar to that of normal epithelium; these tumours were thus defined as having low expression of LPA2. Statistical analysis All statistical calculations were conducted using StatView-J 5.0 statistical software (SAS Institute, Cary, NC, USA). mRNA levels were compared using unpaired Student's t-test, and relationships between the over-expression of LPA2 and clinicopathological features were examined by Fisher's exact test. P < 0.05 was considered statistically significant. Results Expression of LPA receptors at the mRNA level The relative expression level of each LPA receptor's mRNA was quantitatively evaluated by real-time RT-PCR against that of β-actin (Fig. 2 ). A significant level of LPA1 mRNA was detected in both normal and carcinoma tissues, and expression levels did not differ significantly between tissues (1438.5 ± 425.5 versus 1675.3 ± 299.8). However, the expression level of LPA2 in cancer tissue was significantly greater than that in normal tissue (3479.0 ± 426.6 versus 1287.3 ± 466.8; P < 0.05). In comparison, LPA3 was weakly expressed in both normal and cancer tissue, although one cancer contained a relatively high level of LPA3 transcript. When the ratio of LPA2 to LPA1 was calculated, cancer tissue exhibited a threefold higher ratio than did normal gland tissue (3.69 ± 0.96 versus 1.12 ± 0.43; P = 0.0016). Expression of LPA2 receptor in the immunohistochemical study The expression of LPA2 in the same tumours was also evaluated at the protein level by immunohistochemical staining (Fig. 3 ). In 15 out of 25 (60%) carcinomas, enhanced staining for LPA2 was clearly detected in comparison with normal tissue. In these carcinoma cells, LPA2 was detectable in the cytoplasm and in the cell membrane, but not in the nucleus. Some of the normal epithelium in the same specimens, as well as interstitial cells, also stained slightly, but the intensity was significantly weaker than that in carcinoma cells in the 15 tumours. Thus, these cases were categorized as tumours with high LPA2 expression. In the other 10 tumours, carcinoma cells exhibited apparently the same staining intensity as normal epithelium and were categorized as having low expression of LPA2. When the immunoreactivity was compared with the relative ratio of LPA2 mRNA against β-actin, 11 out of the 12 tumours (92%) with ratios greater than 3000 had high expression of LPA2, whereas 10 out of 13 tumours with a ratio below 3000 had low expression (Table 1 ). Thus, the results of the immunohistochemical study exhibited significant correlation with the results of quantitative mRNA study in breast cancer tissue ( P < 0.001). Relationship between LPA2 expression and clinical and pathological factors Because LPA2 expression showed a strong correlation between mRNA and protein levels in 25 cases, we performed immunostaining of LPA2 in an additional 59 cases. The relations between LPA2 expression and clinical or pathological findings in these 84 cases are presented in Table 2 . High expression of LPA2 was frequently detected in relatively old postmenopausal patients. High expression of LPA2 was detected in 17 out of 38 (45%) premenopausal patients and in 31 out of 46 (67%) postmenopausal women; this difference was statistically significant ( P < 0.05). LPA2 expression did not correlate with oestrogen receptor expression. The frequency of nodal or distant metastasis tended to be higher in tumours with low expression of LPA2, although the difference was not statistically significant. Discussion In the present study we found that breast cancer frequently exhibited enhanced expression of LPA2 mRNA as compared with normal breast gland tissue, although the expression level of LPA1 and that of LPA3 were not significantly different from those in normal tissue. Over-expression of LPA2 was confirmed at the protein level in some of these cases by immunohistochemical staining. Previous studies showed that LPA1 is widely expressed in various tissues, whereas LPA2 and LPA3 are known to be highly expressed in malignant cells, suggesting the potential role of these receptors in the pathophysiology of cancer [ 5 , 18 , 19 , 21 , 22 ]. Our findings regarding LPA2 are basically consistent with those results and suggest that upregulation of the LPA2 gene is a common feature of carcinogenesis in various organs. Protein expression of LPA2 in various organs has not been investigated thus far because of the lack of an appropriate antibody against LPA2. Our antibody clearly detected significantly enhanced expression of LPA2 in more than half of the ductal carcinoma cells, and the staining pattern exhibited a strong correlation with the mRNA data obtained with quantitative RT-PCR. Normal epithelium of mammary glands also showed slight immunoreactivity, but the staining intensity was markedly weaker than that of carcinoma cells in these cases. Positive staining was mostly detected in the cytosol and cellular membrane of carcinoma cells, although the distribution of immunoreactivity differed among the cells in each case, indicating that the cellular distribution of LPA2 is highly heterogeneous, even within the same tumour. The exact role played over-expression of LPA2 in breast carcinogenesis is not yet clear. However, in ovarian cancer cells the proliferative responses to LPA of ovarian cancer cells with high levels of LPA1 receptor were uniformly weaker than those of cells with low LPA1 receptor levels [ 22 ]. Moreover, LPA1 has been reported to transduce the inhibitory signal to LPA2-mediated proliferative responses and concurrently elicits apoptosis and anoikis [ 23 ]. On the other hand, LPA has been shown to have an antiapoptotic effect on various cells, although the LPA receptor that mediates this response has not been determined [ 24 , 25 ]. Deng and coworkers [ 26 ] reported that LPA prevents apoptosis of rat intestinal epithelial cells (IEC-6) induced by radiation and chemotherapeutic agents. In that study they used receptor specific inhibitors and concluded that the antiapoptotic effect was mediated through both LPA1 and LPA2. Thus, LPA1 can mediate both antiapoptotic and apoptotic effects, and this is apparently dependent on the cell type and circumstances. Taken together, these findings indicate that the stimulation of LPA1 receptor has suppressive effects, at least in part, on cell proliferation as compared with the LPA2 receptor. Thus, the enhanced expression of LPA2 with relatively reduced expression of LPA1 appears to promote mammary epithelial cell survival in LPA-rich circumstances and to favour development of breast cancer. Immunohistochemical finding showed that LPA2 is upregulated more frequently in postmenopausal than in premenopausal women, suggesting that over-expression of LPA2 is more strongly related to the carcinogenesis of postmenopausal breast cancer. It is well known that a high-fat diet has a positive association with development of breast cancer, especially postmenopausal breast cancer, although some conflicting findings have also been reported [ 27 - 32 ]. A high-fat diet is often associated with the insulin resistance syndrome, and hyperinsulinaemia and high level of insulin-like growth factor (IGF) have also been reported to be markers of increased breast cancer risk [ 33 - 36 ]. Recently, receptors for IGF-1 were shown to be functionally expressed in mammary epithelium [ 37 - 39 ]. Moreover, the IGF-mediated intracellular signal closely interacts with the LPA-mediated signal, although the specific mechanism is not yet known [ 40 - 43 ]. Therefore, our data raise the possibility that a shift in LPA receptor expression may alter the effects of the LPA on IGF-mediated signal in mammary epithelial cells, which may be closely associated with adiposity-related carcinogenesis of breast cancer. The crosstalk between the LPA1- or LPA2-mediated signal and the IGF- or insulin-mediated signal remains an interesting subject for future study. In summary, we found preferential expression of LPA2 in ductal carcinoma as compared with normal epithelium in human mammary gland tissue. Over-expression of LPA2 was more prominent in postmenopausal women. A shift in LPA receptor expression in mammary epethelium may be a key event, especially in adiposity-related breast carcinogenesis. LPA and its receptors could represent new chemopreventive targets in a novel therapeutic strategy in human breast cancer. Conclusion Human ductal carcinoma cells frequently exhibited enhanced expression of LPA2 as compared with normal epithelium in human mammary gland tissue, especially in postmenopausal women. This suggests that upregulation of LPA2 may be one of the key events in carcinogenesis, especially in adiposity-related breast cancer. Competing interests The authors declare that they have no competing interests. Abbreviations IGF = insulin-like growth factor; LPA = lysophosphatidic acid; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; RT = reverse transcription.

D:\keerthana\PMC001xxxxxx\PMC1064082.xml

1064084

Menopausal status dependence of the timing of breast cancer recurrence after surgical removal of the primary tumour

Introduction Information on the metastasis process in breast cancer patients undergoing primary tumour removal may be extracted from an analysis of the timing of clinical recurrence. Methods The hazard rate for local-regional and/or distant recurrence as the first event during the first 4 years after surgery was studied in 1173 patients undergoing mastectomy alone as primary treatment for operable breast cancer. Subset analyses were performed according to tumour size, axillary nodal status and menopausal status. Results A sharp two-peaked hazard function was observed for node-positive pre-menopausal patients, whereas results from node-positive post-menopausal women always displayed a single broad peak. The first narrow peak among pre-menopausal women showed a very steep rise to a maximum about 8–10 months after mastectomy. The second peak was considerably broader, reaching its maximum at 28–30 months. Post-menopausal patients displayed a wide, nearly symmetrical peak with maximum risk at about 18–20 months. Peaks displayed increasing height with increasing axillary lymph node involvement. No multi-peaked pattern was evident for either pre-menopausal or post-menopausal node-negative patients; however, this finding should be considered cautiously because of the limited number of events. Tumour size influenced recurrence risk but not its timing. Findings resulting from the different subsets of patients were remarkably coherent and each observed peak maintained the same position on the time axis in all analysed subsets. Conclusions The risk of early recurrence for node positive patients is dependent on menopausal status. The amount of axillary nodal involvement and the tumour size modulate the risk value at any given time. For pre-menopausal node-positive patients, the abrupt increase of the first narrow peak of the recurrence risk suggests a triggering event that synchronises early risk. We suggest that this event is the surgical removal of the primary tumour. The later, broader, more symmetrical risk peaks indicate that some features of the corresponding metastatic development may present stochastic traits. A metastasis development model incorporating tumour dormancy in specific micro-metastatic phases, stochastic transitions between them and sudden acceleration of the metastatic process by surgery can explain these risk dynamics.

Introduction The 1970s and 1980s witnessed a revolution in the conventional approach to the treatment of primary breast cancer. Early in the 1970s, the favourable results of postoperative systemic adjuvant therapy in women with positive axillary lymph nodes [ 1 , 2 ] started an avalanche of clinical trials that explored the role of several systemic treatments in different subsets of patients. The beneficial results were confirmed by a few overviews of randomised trials [ 3 , 4 ], and reports from individual studies proved that the benefit continued at 20 years of follow-up [ 5 ]. However, the significant, albeit moderate, improvement of disease-free survival and overall survival achieved by earlier adjuvant therapy trials has improved only slightly during subsequent years, despite a spate of new active drugs and the use of higher drug doses [ 6 ]. The benefits of adjuvant therapy have therefore apparently reached a plateau, and it is unlikely that further improvements will be obtained without a more complete and accurate understanding of the biology of the tumour–host system at the time of treatment. Surgical resection of primary tumour removal may either 'cure' a significant fraction of patients, or it may even change the 'natural' recurrence and death timing for some others, by accelerating the metastatic development [ 7 , 8 ]. Some specific biological mechanisms supporting this effect have been elucidated: in animals given surgery, a growth-stimulating factor was found in serum [ 9 ] and a switch of micro-metastatic foci to the angiogenic phenotype, due to withdrawal of an angiogenesis inhibitor from the primary tumour, was demonstrated [ 10 ]. Despite these provocative data, the residual tumour growth dynamics underlying the beneficial results of all adjuvant systemic treatments is virtually unknown in humans. Careful inspection of the timing of tumour recurrence after resection can be of considerable interest. The recurrence risk pattern in a given follow-up span, a useful estimate of which is the hazard function [ 11 ], provided information on the biological behaviour of metastases. The hazard functions for local-regional recurrences and distant metastases for breast cancer patients undergoing mastectomy alone [ 12 ] proved to be double-peaked, with an early peak at about 18 months after surgery, a second peak at about 60 months, and a plateau-like tail extending out to 15 years, the maximum period analysed. These findings were confirmed by a similar investigation on node-positive patients receiving adjuvant CMF (cyclophosphamide, methotrexate, fluorouracil) treatment [ 13 ], and a double-peaked hazard function was also demonstrated for the timing of death after primary tumour resection [ 14 - 16 ]. A reasonable hypothesis to explain these findings is that the early peak of the hazard function for recurrence is generated by the sudden acceleration of the metastatic process due to surgery. This hypothesis was the cornerstone for a biological model for the development of breast cancer metastasis, incorporating tumour dormancy in specific micro-metastatic phases and stochastic transitions between them [ 17 ]. A computer simulation of the model generated double-peaked relapse histograms reasonably similar to clinical data [ 18 ]. The model and the results of its computer simulation suggest that the first peak may result from the superimposition of metastatic recurrences with different dynamics. Here we report a detailed analysis of the first peak of the hazard rate curve, that is, of the recurrence timing during the first 4 years, for patients undergoing mastectomy without any adjuvant chemotherapy or radiotherapy. Our findings suggest that metastatic growth dynamics after primary tumour removal is markedly different depending on menopausal status and, perhaps, axillary nodal involvement. These results support the concept that surgery may have differential perturbing effects on tumour growth kinetics depending on both tumour and host traits. These results strengthen the leading lines of the proposed model for metastasis development and may be relevant to adjuvant treatment strategies. Methods All patients with mastectomy alone as primary treatment who entered into three different clinical trials for operable breast cancer were retrospectively evaluated for this analysis. The trials were performed at the Milan Cancer Institute between 1964 and 1980 to compare mastectomy with other surgical or combined therapeutic approaches. Before surgery, all patients underwent standard staging: complete physical examination, X-ray study of chest, skull, spine and pelvis, bilateral mammography, electrocardiogram, complete haemogram and routine biochemical tests. The primary tumour was treated by radical or modified radical mastectomy. No patient received postoperative radiotherapy or chemotherapy. After surgery, follow-up was performed as follows: physical examination, biochemical tests and chest X-ray every 6–8 months during the first 3 years and once a year thereafter; skeletal survey and mammography once a year. In the presence of controversial clinical findings, examinations were performed more often than originally planned. Appropriate radiological, radioisotopic and surgical investigations were performed whenever recurrence was suspected or clinically evident. Particular attention was paid to assessing the recurrence time by carefully reviewing clinical, radiological and laboratory documentation. More detailed characteristics of the studied series are reported in reference 12. Treatment failure (recurrence) was defined as the first clinically documented evidence of new disease manifestation(s) in either local-regional area(s) (namely chest wall, axilla and/or ipsilateral supra-clavicular region) or distant site(s) or any combination of these sites (classified as distant). Contra-lateral breast lesions may be either second primaries (confounding factors in the present investigation) or true metastases. Their annual incidence, together with the absence of a link with clinical prognostic factors of the primary breast cancer, suggest that most of them can be considered as second primary breast cancers [ 12 , 19 ]. Contra-lateral tumours were therefore considered second primaries. Recurrence-free survival was considered as the time elapsed from the date of surgery to the first documented evidence of treatment failure; clinical evaluation without recurrence, death without recurrence, and second primary cancer (including contra-lateral breast cancer) were considered censored events. Our application involved dividing the time axis into discrete intervals; the timing of the recurrence risk was studied by estimating the recurrence hazard rate, namely the conditional probability of manifesting recurrence in a time interval, given that the patient is clinically free of any recurrence at the beginning of the interval. The discrete hazards of recurrence and their standard deviations were calculated by means of the life-table method over 4 years. Because the event-specific rates display some instability arising from random variation, a kernel-like smoothing procedure [ 20 ] was adopted to aid the reader in visualising the underlying pattern, and the smoothed curves were represented graphically. Time intervals of 1, 2, 3, 6 and 12 months were used in a preliminary smoothing analysis that showed that a 3-month interval is a good compromise between smoothing data and displaying structure. We therefore used the hazard rate per 3-month interval. Menopausal status, which was collected and recorded at the time of primary tumour diagnosis, was defined as 'post-menopausal' if 1 year had elapsed since the last menstrual period. Cumulative survival curves were compared by two-sided log-rank test. Results The main clinical characteristics of the 1173 analysed patients, all undergoing regular follow-up examinations in accordance with protocol requirements, are reported in Table 1 . A total of 368 patients suffered from treatment failure within 4 years of surgery: the tumour recurrence was local-regional in 95 cases and distant in 273. A total of 749 patients were surviving and were at risk of recurrence at the end of the fourth year. Patients were lost to follow-up at a rate of 1.6% during the analysed years. As reported and discussed elsewhere [ 12 ], the hazard rate for recurrence of all patients during 10 years of follow-up (Fig. 1a ) displays a double-peaked pattern. In the present study we focused on the first peak (Fig. 1b ), which shows an asymmetrical pattern with an early steep rise, reaching the maximum recurrence risk at the end of the first year, followed by a slight decrease lasting about 1.5 years and then by a more distinct decline until the end of the fourth year. This shape suggests that the recurrence risk curve may result from the superimposition of differently timed peaks. As a first step, the hazard rate pattern by type of recurrence (local-regional versus distant) was studied and, because their timing proved to be remarkably similar, in the further analysis both recurrences were combined. Patients were then allocated to different subsets according to tumour size (T1 versus T2–T3) or to axillary nodal status (positive versus negative) or to menopausal status (pre-menopausal versus post-menopausal). The resulting hazard rate curves indicated that the recurrence risk pattern is definitely correlated to menopausal status, with a double-peaked curve for pre-menopausal patients (Fig. 2a ) and a single-peaked curve for post-menopausal women (Fig. 2b ). Tumour size did not show a noticeable influence on the recurrence risk pattern, and this finding was confirmed by the analysis of further subsets of patients (T1 and T2–T3 N-pre-menopausal, T1 and T2–T3 N+ pre-menopausal, T1 and T2–T3 N- post-menopausal, T1 and T2–T3 N+ post-menopausal), in whom tumour size (as well as type of recurrence) failed to change the hazard rate pattern (data not shown). Pre-menopausal patients were allocated to subsets of node-negative, node-positive, one to three nodes positive and more than three nodes positive, and the hazard rates were estimated (Fig. 3 ). The resulting curves show that for node-positive patients, the recurrence risk has a quite narrow early peak, with a very steep rise to a maximum at about 8–10 months after mastectomy, and a second wider increase peaking at about 28–30 months. The recurrence pattern and the peak position do not change for patients with different axillary involvement, whereas the peak height is positively correlated with the extent of nodal invasion. A similar analysis was performed in post-menopausal patients. All subsets of patients displayed a wide, nearly symmetrical recurrence risk peak, with a maximum at 18–24 months and a height similar to the corresponding estimates of pre-menopausal patients (Fig. 4 ). Even for these patients the extent of axillary node invasion was positively correlated to the peak height. The recurrence risk for node-negative patients displays an initial regular increase reaching a maximum at about 18–24 months after surgery, with a mild decrease afterwards. No multi-peaked pattern was evident or even suggested for either pre-menopausal (Fig. 3a ) or post-menopausal (Fig. 4a ) women. However, it should be noted that of the 598 node-negative patients at risk, only 74 recurrences were observed during the 4 years under investigation, 43 of which occurred within 2 years. The reported results, mainly those from the third and subsequent years, should therefore be very cautiously considered. Discussion First of all, it should be emphasised that findings resulting from the different subsets of patients are notably coherent. A remarkably stable two-peaked hazard function, with increasing peak height associated with the increasing axillary lymph node involvement, describes the risk of recurrence for pre-menopausal patients, whereas, quite differently, post-menopausal women always display a single peak. Moreover, each observed peak maintains the same position on the time axis in all analysed subsets of pre-menopausal and post-menopausal women. The occurrence of these peaks by chance, that is, resulting from random fluctuations of the recurrence timing, should therefore be considered very unlikely. Obviously, our findings must be confirmed by similar analyses of other comparable databases. We add here some important details to the previously reported investigation on the recurrence risk pattern of this same series [ 12 ]. Indeed, we found that even though pre-menopausal and post-menopausal patients display similar long time results (10-year recurrence-free survival: 56% versus 54%; not significant) and similar overall recurrence pattern (early surge during the first 4 years, a second minor increase at about 5 years and a tapering phase afterwards), the early recurrence dynamics may be very different between pre-menopausal and post-menopausal women. A very early significant difference in the recurrence risk between pre-menopausal and post-menopausal patients had already been noted [ 21 ]. In the present analysis we found that these differences cover all the first 4 years and persist in all patient subsets. Summarising, the recurrence risk for pre-menopausal patients displays an abrupt increase at the third 3-month period; it quickly reaches its prominent maximum level (about four times the level of the previous 6 months) and is followed by a second fairly symmetrical and considerably broader peak appearing at about 18 months of follow-up, with its maximum at 28–30 months after surgery. In comparison, the recurrence risk for post-menopausal patients is nearly symmetrical and, after a smooth increase, reaches its maximum level at about 18 months after primary tumour removal. These findings are clear in node-positive patients, and the higher the nodal involvement, the higher the amplitude of peaks. The limited number of events prevents us from drawing any conclusion for node-negative patients. Tumour size affects the recurrence risk value at a given time but does not affect the pattern of the hazard rate distribution (that is, the occurrence of peaks). It is difficult not to be impressed by the particular shape of the first recurrence peak of pre-menopausal node-positive patients and by the difference between it and the others of both menopausal subsets. It is also difficult, in the presence of a such an abrupt and prominent increase of the recurrence risk, to imagine something different from a triggering event resulting in recurrence synchronisation. As a first step, some possible non-biology-based explanations of this finding should be considered. The peak could be related to inadequate staging evaluation, because at the time of the study no bone scan, liver ultrasound or computed tomography scan staging was done. Patients with undiscovered metastatic disease at diagnosis would therefore have displayed recurrence at the initial post-treatment screening, thus contributing to the first peak. However, it should be considered that, for newly diagnosed otherwise asymptomatic operable patients, (1) very few cases prove to be metastatic at standard staging today (less than 5%) [ 22 ] and (2) bone scan ultrasound and computed tomography scan surveys have a very low detection rate (less than 1%) [ 23 , 24 ] and are not routinely recommended in these patients [ 25 ]. Even chest X-ray was considered unnecessary at staging of asymptomatic operable breast cancer, because of its very low diagnostic yield (0.1%) [ 26 ]. Therefore, even if a few missed metastatic patients at primary tumour diagnosis might not have been excluded, they would have had little influence on the observed first peak of recurrence risk involving, for pre-menopausal node-positive patients, 22% of eligible patients and 37% of recurrences within 10 years. Moreover, because up to 75% of breast cancer recurrences are detected by the patient and reported to the physician within a mean time of 1 month [ 27 , 28 ], we can reliably consider the observed recurrence timing as widely independent of the planned follow-up timing. Last but not least, post-menopausal patients, who underwent the same clinical management as pre-menopausal ones, did not display an early peak. A significant role of screening and diagnostics can therefore reasonably be ruled out and a biology-based explanation should be investigated. The first sharp peak, in our opinion, has a biological explanation and may result from some triggering event changing the unperturbed history of the disease. We suggest that primary tumour surgical removal is probably this event, selectively operating on some micro-metastases within similar biological conditions. This triggering effect could be modulated by the surgery extent [ 29 ]. However, in the present investigation data about the different surgical approaches (namely radical versus modified radical mastectomy) were not considered suitable for such an analysis, which would be more usefully performed for trials comparing mastectomy with breast-conserving surgery. The likely triggering effect of mastectomy is also supported by the peak position, implying a rapid activation and growth within 8–10 months. This is in quite good agreement with estimates of tumour growth after dormancy release (30 ± 8 weeks or less) obtained by very different methodology [ 30 ]. Assuming tumour dormancy release and the estimate of 8–10 months or less for tumour development from micro-metastasis to clinical metastasis, the hazard function should be considered as a description of the dormancy escape kinetics. A symmetrical, bell-shaped peak in the curve of hazard rate against time provides information about both the mean value (the peak position) and the variance (the peak width) of the timing of the constituent recurrences. The temporally stable position of the recurrence peak therefore suggests a common cause for recurrence development, a narrow peak indicates that the underlying event is most probably deterministic in nature, and a considerably wider peak suggests that some features of the transition from dormancy to growth may also present stochastic traits. We recently proposed a model [ 17 , 18 , 31 ] in which breast cancer metastasis development may include successive steps: first, single cells (or nests containing a few cells), where most of malignant cells are non-dividing; second, non-angiogenic micro-metastases (and angiogenic ones in the presence of anti-angiogenic factors) that cannot grow more than the size of avascular foci; and third, vascularised metastases that will reach the clinical level. This orderly process is apparently perturbed by surgical removal of the primary tumour, which can stimulate cells to proliferate and/or remove angiogenic inhibition, thus resulting, for some patients, in sudden acceleration of the metastatic process. The proposed model both reasonably explains and takes support from the findings of this study. The early sharp peak of the recurrence risk for pre-menopausal patients can be ascribed to a considerable switching of micro-metastatic foci to the angiogenic phenotype via the withdrawal of angiogenesis inhibitor(s) [ 10 ], and the sharpness of this peak supports the deterministic character of the triggering event. For post-menopausal patients, for whom the first sharp peak is absent, this effect may be much more modest. Other broader peaks may result from the induced proliferation of single cells through one or more growth-stimulating factors [ 9 ], resulting in successive avascular micrometastases, followed by the 'spontaneous' switching to the angiogenic phenotype, a process with stochastic characteristics, in keeping with peak shapes. As regards differences between pre-menopausal and post-menopausal patients, it should be taken into account that peculiar conditions relevant to breast cancer development, in particular conditions acting on angiogenetic traits, may be dependent on the host hormone milieu. Angiogenesis in invasive breast carcinoma, as evaluated by micro-vessel counts, is associated inversely with patient age, and higher vascular intensity is observed in younger patients [ 32 , 33 ]. Vascular endothelial growth factor, the first member of a family of glycoproteins with considerable stimulating activity for angiogenesis and lymphangiogenesis, waxes and wanes in normal breast tissue within each menstrual cycle [ 34 ]. Angiogenin, another potent angiogenic factor, displays a significant difference between serum levels drawn in the proliferative and secretory phases of the menstrual cycle [ 35 ]. Plasma levels of endostatin, a negative regulator of angiogenesis, are significantly higher in post-menopausal patients [ 36 ]. It is therefore reasonable to assume that some conditions related to menopausal status may select neoplastic cell populations with peculiar traits relative to angiogenesis. A further factor may have a role: it has been suggested that timing of surgery in the menstrual cycle might modulate the tumour–host relationship and ultimately the disease outcome [ 37 ]. This occurrence is restricted, obviously, to pre-menopausal patients. The findings of this investigation would be much more complete if other prognostic and predictive factors could be analysed. However, this database, like most databases of patients undergoing surgery alone without any adjuvant treatment, goes back to years when factors other than TNM (tumor, nodes, metastases) staging and menopausal status were poorly known or completely unknown. Conclusions Early metastatic growth dynamics shows a structured pattern that is correlated to menopausal status. Indeed, during the first 4 years, node-positive pre-menopausal patients display a split of the hazard function surge into two peaks, while post-menopausal patients always display a single peak. The timing and shape of peaks suggest the occurrence of some triggering event, correlated with primary tumour surgical removal, resulting in recurrence synchronisation, which seems to be especially important for pre-menopausal node-positive patients. In addition, findings support the hypothesis that some features of metastatic development may present stochastic traits. A metastasis development model incorporating tumour dormancy, stochastic transitions between specific micro-metastatic phases and sudden acceleration of the metastatic process due to surgery may be considered appropriate and to fit these findings well. The different post-resection relapse dynamics between pre-menopausal and post-menopausal patients may have important implications for adjuvant systemic treatments. For adjuvant chemotherapy the proposed picture makes it easier to understand retrospectively why adjuvant chemotherapy was found to be most effective in pre-menopausal patients, who during the first months after primary tumour removal would display an enhancement of actively proliferating, and hence more chemo-sensitive, micro-metastases [ 38 ]. Most importantly, the prominent role of angiogenesis in the surgery-driven dormancy escape for pre-menopausal patients suggests that early (preoperative) antiangiogenic treatment might favourably influence prognosis. Competing interests The authors declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064084.xml

1064093

The p53-dependent apoptotic pathway of breast cancer cells (BC-M1) induced by the bis-type bioreductive compound aziridinylnaphthoquinone

Introduction Several aziridinylbenzoquinone drugs have undergone clinical trials as potential antitumor drugs. These bioreductive compounds are designed to kill cells preferentially within the hypoxia tumor microenvironment. The bioreductive compound of bis-type naphthoquinone synthesized in our laboratory, 2-aziridin-1-yl-3-[(2-{2-[(3-aziridin-1-yl-1,4-dioxo-1,4-dihydronaphthalen-2-yl)thio]ethoxy}ethyl)thio]naphthoquinone (AZ-1), had the most potent death effect on the breast cancer cells BC-M1 in our previous screening. In the present study, we determined that the mechanism of the death effect of BC-M1 cells induced by AZ-1 was mediated by the apoptosis pathway. Methods We evaluated the cytotoxicity of AZ-1 and the anti-breast cancer drugs tamoxifen and paclitaxel to BC-M1 cells and MCF-7 cells by the MTT assay and measured the apoptosis phenomena by Hoechst 33258 staining for apoptotic bodies. We also quantified the sub-G 1 peak area and the ratio of the CH 2 /CH 3 peak area of the cell membrane in BC-M1 cells by flow cytometry and 1 H-NMR spectra, respectively. The apoptosis-related protein expressions, including p53, p21, the RNA-relating protein T-cell restricted intracellular antigen-related protein, cyclin-dependent kinase 2 (cell cycle regulating kinase) and pro-caspase 3, were detected by western blot, and the caspase-3 enzyme activity was also quantified by an assay kit. Results AZ-1 induced two of the breast cancer cell lines, with IC 50 = 0.51 μM in BC-M1 cells and with IC 50 = 0.57 μM in MCF-7 cells, and showed less cytotoxicity to normal fibroblast cells (skin fibroblasts) with IC 50 = 5.6 μM. There was a 10-fold difference between two breast cancer cell lines and normal fibroblasts. Of the two anti-breast cancer drugs, tamoxifen showed IC 50 = 0.12 μM to BC-M1 cells and paclitaxel had much less sensitivity than AZ-1. The expression of p53 protein increased from 0.5 to 1.0 μM AZ-1 and decreased at 2.0 μM AZ-1. The p21 protein increased from 0.5 μM AZ-1, with the highest at 2 μM AZ-1. Regarding the AZ-1 compound-induced BC-M1 cells mediating the apoptosis pathway, the apoptotic body formation, the sub-G 1 peak area, the ratio of CH 2 /CH 3 of phospholipids in the cell membrane and the enzyme activity of caspase-3 were all in direct proportion with the dose-dependent increase of the concentration of AZ-1. The death effect-related proteins, including T-cell restricted intracellular antigen-related protein, cyclin-dependent kinase 2, and pro-caspase-3, all dose-dependently decreased with AZ-1 concentration. Conclusions The AZ-1-induced cell death of BC-M1 cells mediating the apoptosis pathway might be associated with p53 protein expression, and AZ-1 could have the chance to be a candidate drug for anti-breast cancer following more experimental evidence, such as animal models.

Introduction The bioreductive drugs, aziridinylbenzoquinones, are a class of compounds designed to exploit one of the features of solid tumor biology caused by an inadequate blood supply to the solid tumor; namely, tumor hypoxia. Such regions generally are resistant to radiation and other oxygen-requiring treatment [ 1 - 4 ]. The ideal bioreductive drug should be administered as an inactive prodrug that is only activated under low-oxygen conditions by one-electron or two-electron reductase [ 5 ]. The aziridine-substituted benzoquinones such as mitomycin C, RH1, and E09 are three principal aziridinylquinone-class hypoxia-specific cytotoxins that are being developed for clinical use [ 6 - 8 ]. These agents are composed of an aziridinyl moiety on a quinone structure, and they convert on reductive metabolism to a bifunctional alkylating species that can cross-link DNA in the major grove that interacts predominantly on guanine-N7 [ 9 ]. These agents probably produce their major cytotoxic activities through the formation of DNA cross-links. Tumor tissue with lower oxidative reduction (redox) potential relative to most normal tissue could increase reductive activation of these quinone derivatives in the tumor [ 2 ]. The selectivity of bioreductive drugs is therefore governed not only by the difference in oxygen tension between the tumor and normal tissue, but also by levels of enzymes catalyzing bioreductive activation such as DT-diaphorase [ 4 , 10 , 11 ]. This fact led to the publication in 1990 of the concept of 'enzyme-directed bioreductive development' by Workman and Walton [ 12 ]. In many cases, the biological activity of quinone is attributed to the ability of accepted electrons to form the corresponding radical anion or dianion species. A quinone moiety substituted with aziridine has been shown to be a potent alkylating agent as a result of bioreduction by the one-electron reducing enzymes (e.g. NADPH cytochrome P450 reductase, cytochrome b5 reductase) or by two-electron reducing enzymes ([NADP]H oxidoreductase, NQO1) to form the corresponding aziridinyl hydroquinone [ 13 - 15 ]. The hydroquinone moiety in the corresponding aziridinyl hydroquinone effectively changes the p K value of the aziridine ring such that it is protonated and becomes activated toward nucleophilic attack under physiological pH. In previous work in our laboratory, we biosynthesized the bis-type of bioreductive aziridinylnaphthoquinone series compounds and evaluated their anti-cancer activity [ 16 , 17 ]. In the case of di-aziridinyl-substituted quinone, this highly cytotoxic bifunctional alkylating agent can cross-link DNA in cells that results in induction of a complex cellular mechanism leading to cell death by apoptosis or necrosis [ 18 ]. Based on our previous results, the compound 2-aziridin-1-yl-3-[(2-{2-[(3-aziridin-1-yl-1,4-dioxo-1,4-dihydronaphthalen-2-yl)thio]ethoxy}ethyl)thio]naphthoquinone (AZ-1) had the most potent death effect on breast cancer cells (BC-M1) and less cytotoxicity to normal fibroblast cells (skin fibroblasts). In this study, we were interested in determining the cytotoxicity of AZ-1 to our localized cell line (BC-M1 cells) and to determine the mechanism by which this occurs. We compared the death effect of two cell lines (BC-M1 and the estrogen-receptor-positive cell line MCF-7) induced by AZ-1, and also compared the cytotoxicity of AZ-1 with two clinical anti-breast cancer therapies, tamoxifen (aromatase inhibitors) and paclitaxel, in the cell viability of BC-M1 cells. Materials and methods Materials RPMI 1640 medium, DMEM medium, fetal bovine serum, 2 mM L-glutamine, MEM non-essential amino acid, trypsin-EDTA solution, PBS, Hank's balanced salt solution, pencillin-streptomycine and fungizone were purchased from Gibco Laboratories (Grand Island, NY, USA). NaHCO 3 - , MTT, trypan blue, EDTA, propidium iodide, Hoechst 33258, paclitaxel and tamoxifen were purchased from Sigma Chemical Co. (St Louis, MO, USA). The primary antibodies of T-cell restricted intracellular antigen-related protein (TIAR), pro-caspase and p53, cyclin-dependent kinase (cdk) 2 and cyclin B were purchased from BD Transduction Laboratories (BD Bioscience, Palo Alto, CA, USA). Peroxidase-conjugated affinity-purified goat anti-mouse IgG was purchased from Jackson Immuno Research Laboratories (West Grove, PA, USA). The chamber slide was purchased from NUNC (Roskilde, Denmark). All other chemicals were purchased from Merck (Darmstadt, Germany). AZ-1 was obtained from total synthesis in our laboratory, was dissolved in dimethsulfoxide before experimental use and was aliquoted to be stored at -20°C, with stability for several years. Methods Human cell lines cultured The BC-M1 (human breast adenocarcinoma) cell line was cultured in RPMI 1640 medium with 10% fetal bovine serum, 2 mM L-glutamine, and 25 mM Hepes. Skin fibroblasts and MCF-7 cells were cultured in DMEM medium with 10% fetal bovine serum, 2 mM L-glutamine, and MEM nonessential amino acid. The cell culture media for the two cell lines all contained pencillin-streptomycine and fungizone. All cells were incubated in a humidified atmosphere of 5% CO 2 at 37°C. Cell cultures were subcultured once or twice weekly using trypsin-EDTA to detach the cell from their culture flask. The numbers of cells were counted after trypsinization by a Neubauer hemocytometer (VWR Scientific Corp., Philadelphia, PA, USA). Cytotoxicity determined by MTT for cell viability The MTT assay was performed according to the method of Skehan and colleagues [ 19 ]. One day before drug application, cells were seeded in 96-well flat-bottomed microtiter plates (3000–5000 cells/well). Cells were incubated for 24 hours with various drugs, and applied as serial dilutions (100 μl/well) at various concentrations. Twenty microliters of MTT (5 mg/ml) were added to each well and incubated for 4 hours at 37°C. The formazan product was dissolved by adding 100 μl dimethylsulfoxide to each well, and the plates were read at 550 nm. All measurements were performed in triplicate and each experiment was repeated at least three times. The IC 50 value was calculated from the 50% formazan formation compared with a control without addition of drugs. Apoptoic body stained by Hoechst 33258 The cells were cultured in RPMI 1640 complete medium for BC-M1 cells and in DMEM complete medium for MCF-7 cells on a chamber slide (1 × 10 4 cells/ml). Various concentrations of AZ-1 compound were added and incubated at 37°C. After 24 hours of incubation, the cultured medium was removed and the cells were fixed by acetic acid/methanol (1:3) solution for 10 min. In the following step, the fixed solution was removed and cells were air-dried for another 10 min. The cell was stained by Hoechst 33258 stain solution (0.5 μg/ml in Hank's balanced salt solution) at room temperature for 30 min. After staining, the solution was removed, the cell was washed three times with distilled water and then one drop of mounting solution (0.1 M citric acid:0.2 M disodium phosphate:glycerol, 1:1:2) was added before being covered by a cover slide. Apoptotic cells showed blue, peripherally clumped or fragmented chromatin. Apoptosis analysis by flow cytometry The apoptotic nuclei of BC-M1 cells induced by AZ-1 were also identified by the flow cytometry analysis method as described by Dive and colleagues with minor modification [ 20 ]. The BC-M1 cells were treated with various concentrations of AZ-1 for 24 hours. Cells were harvested and DNA was stained with propidium iodide. The DNA content was measured by flow cytometry (Becton Dickinson FACScan, San Jose, CA, USA). Western blot analysis This analysis method was that according to Bacus and colleagues with slight modifications [ 21 ]. Briefly, cells were collected from a 100 mm cultured dish after challenge by various concentrations of AZ-1 compound at 24 hours. Cell pellets were spun-down by centrifugation (1000 × g × 20 min). Pellets were resuspended in cold buffer (10 mM HEPES [pH 7.9], 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 30 mg/ml leupeptin, 5 mg/ml aprotinin, and 5 mg/ml pepstatin A; all from Sigma) and were incubated on ice for 5 min and lysed by sonication. Cell lysate (25 μg) was separated by 12% SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Amersham, Little Chalfont, UK). Blots were incubated with blocking buffer (11 mM Tris-vase [pH 7.4], 154 mM NaCl, and 5% skim milk), washed by washing buffer (11 mM Tris-vase [pH 7.4], 154 mM NaCl, and 0.1% Tween-20) and incubated with specific antibodies to probe specific proteins. The primary antibodies were from mouse anti-human monoclonal antibodies (Imgenex Co., San Diego, CA, USA). The secondary antibody (Jackson ImmunoResearch Laboratories) was conjugated with horseradish peroxidase at an appropriate dilution by blocking buffer. The dilution factors for the primary and secondary antibodies were various (dependent on different proteins) and 1:5000, respectively. The primary antibodies used were a mouse monoclonal antibody to GAPDH and β-actin with dilution factors 1:2000 (Biogenesis, Poole, UK) and 1:10,000 (Calbiochem, San Diego, CA, USA), respectively. Immunodetection was carried out using enhanced chemiluminescence (NEN, Boston, MA, USA) detection system. Quantification of various protein expressions were achieved by measuring the intensity of chemiluminescence of the second antibody using a densitometer (BioRad Gel Doc 2000 software, analyzed using Gel Doc; BioRad Laboratories, Hercules, CA, USA) The values in the relative expression-quantifying table of various proteins represent the relative amounts of protein expression with respect to GAPDH or β-actin expression divided by its control. Caspase-3 activity assay The BC-M1 cells were treated with various concentrations of AZ-1 for 24 hours. The cells were harvested and washed with PBS buffer twice. The caspase-3 activity of BC-M1 cells challenged by AZ-1 was measured by the CPP32/caspase-3 protease colorimetric assay kit, with the assay method according to the procedure described by the manufacturer (Chemicon, Temecula, CA, USA). Briefly, the cells were first lysed by cold lysis buffer (supplied by the assay kit) for 10 min. The cell lysate was then centrifuged to remove the cell debris, and the protein concentration determined by Barford reagent (BioRad Laboratories). The supernatant was ready to detect the caspase-3 activity. Apoptosis analysis by 1 H-NMR spectra 1 H-NMR spectroscopy was performed in vitro using methods previously described by Francis and colleagues with minor modification [ 22 ]. Briefly, 5 × 10 7 cells were harvested and washed twice with PBS medium made with D 2 O (90% purity), suspended in a final volume of 500 μl, and were placed immediately on ice until data acquisition. Samples were analyzed on a 400 MHz high-resolution Brucker CSI Omega spectrometer (Brucker, Karlsruhe, Germany) at 18°C, with pulse acquisition, a 90° flip angle, a repetition time of 10 s, 64 or 128 excitations (depending on the desired signal-to-noise ratio), 8 k points, and a 5 kHz bandwidth. A coaxial tube filled with trimethylsialoproponic acid, 0.1% solution in D 2 O, was used as reference (0.0 ppm) for each experiment. The relative areas underneath the CH 2 and CH 3 resonance curves (at 1.3 ppm and 0.9 ppm, respectively) were calculated by integration of the proton spectrum using the trough between the CH 2 and CH 3 resonances as a baseline reference. Results The cytoxicity of AZ-1 to BC-M1 cells and MCF-7 cells The cytotoxicities of the IC 50 value in AZ-1 to BC-M1 cells and MCF-7 cells were 0.51 μM and 0.57 μM in a dose-dependent manner, respectively. The response of these two cell lines (MCF-7 and BC-M1) to AZ-1 was very similar to cell viability in a dose-dependent manner. In the normal fibroblast cell (skin fibroblast) there was still a 90% survival rate at 2 μM (Fig. 1 ). According to the time-dependent treatment of AZ-1 using an IC 50 concentration of 0.51 μM to BC-M1 cells, the cell viability was less than 10% after 36 hours of treatment in a time-dependent manner (Fig. 2 ). The two clinical therapies paclitaxel and taxmoxifen were compared regarding cytoxicity in BC-M1 cells with AZ-1, and the estrogen receptor antagonist taxmoxifen and AZ-1 were more potent than paclitaxel. The taxmoxifen had a low IC 50 value of 0.05 μM, which is lower than AZ-1, with a plateau concentration from 0.05 μM to 2 μM (Fig. 3 ). Apoptosis assay by flow cytometry and Hoechst staining Analysis of the DNA content of cells was used to determine whether cell apoptosis was induced by AZ-1. The data of the sub-G 1 area indicated that BC-M1 cells had a significant population of cell apoptosis in the sub-G 1 area on treatment with AZ-1 for 24 hours compared with dimethylsulfoxide alone (Fig. 4 ). The sub-G 1 area started at 1.0 μM AZ-1 and exhibited a large increase in apoptotic cells identified as subcellular populations with decreasing DNA. The BC-M1 cells treated with 2.0 μM AZ-1 exhibited the largest apoptotic area (29.2%). This is similar to the apoptotic bodies observed using the Hoechst stain (Fig. 5 ). The Hoechst staining method was used to identify the apoptotic nuclei in BC-M1 cells and MCF-7 cells. Apoptotic cells that contained the apoptotic bodies showed blue peripherally clumped or fragmented chromatin, as indicated by arrows in Figs 5 and 6 . From the visual observation of the Hoechst staining results, the BC-M1 cells treated with 2.0 μM had the highest number of apoptotic bodies (Fig. 5 ). The formation of apoptotic bodies was observed obviously at concentrations as low as 0.5 μM AZ-1 in MCF-7 cells (Fig. 6 ). Tumor cell apoptosis associated with protein expression and caspase-3 activity We next set out to determine whether the induction of BC-M1 cell apoptosis by AZ-1 was associated with expression of apoptosis-related proteins. We found that AZ-1 induced changes in BC-M1 cell expression of the checkpoint protein p53 and the cell arrest protein p21 in a dose-related manner (Fig. 7 ). The p53 protein showed increasing expression from 11% to 43% at 0.5 μM and 1 μM AZ-1, and then decreased to 31% at the 2 μM concentration of AZ-1, and p21 protein increased from 6% to 22% from the 0.5 μM to 2 μM concentrations of AZ-1 added to BC-M1 cells for 24 hours compared with control, respectively. The other proteins including TIAR, pro-caspase protein and cell cdk2 also showed a dose-dependent decreasing manner (Fig. 8 ). From the western blot results and the values in the relative protein expression-quantifying table (Figs 7 and 8 ) it was revealed that the expression of proteins was affected by various concentrations of AZ-1 in BC-M1 cells after 24 hours of treatment, and these relative protein expressions were compared with the control. The cdk2, pro-caspase protein and TIAR were reduced to about 62%, 85% and 65%, and to 40%, 67% and 57% when BC-M1 cells treated with 1 μM AZ-1 and 2 μM AZ-1 for 24 hours compared with control, respectively. Based on the results of western blot analysis in the expression of pro-caspase protein, we determined the enzyme activity of caspase-3 in BC-M1 cells after challenge by various concentrations of AZ-1 from 0.5 μM to 4 μM for 24 hours. The enzyme activity of caspase-3 dose-dependently increased with concentrations of AZ-1. The activity was more than twofold over the control in BC-M1 cells after 3 μM AZ-1 treatment for 24 hours (Fig. 9 ). Assay of the apoptosis signal by 1 H-NMR According to some previous reports, the ratio of the CH 2 and CH 3 peak area on the cell membrane was directly in proportion with the signal of apoptosis. From our results we also observed the same phenomena that the ratio of the CH 2 and CH 3 peak area was increasing according to the concentration of AZ-1 (Fig. 10 ). It was about 1.7-fold higher than the control at 2 μM AZ-1 treatment in BC-M1 cells. Discussion Breast cancer is the most common malignancy in women, and it is highly curable if diagnosed at early stage. It is now well established that adjuvant systemic therapy improves survival in patients with early-stage breast cancer [ 23 , 24 ]. Treatment options for early-stage breast cancer include chemotherapy (e.g. anthrocyclines, taxanes) and hormone therapy (e.g. tamoxifen, aromatase inhibitor). Estrogen mediates its functions through two specific intracellular receptors, estrogen receptor alpha and estrogen receptor beta, which act as hormone-dependent transcriptional regulators [ 25 , 26 ]. The estrogen receptor pathway plays a critical role in the pathophysiology of human breast cancer. Overexpression of estrogen receptor alpha is a well-established prognostic and predictive factor in breast cancer patients. The prognostic significance of estrogen receptor beta is not well defined [ 27 - 30 ]. Our previous study on the different series of bis-aziridinylnaphthoquinone compounds identified that they exhibit a more potent response toward the solid tumors than the circulation tumors [ 17 ]. This result was supported by other reports that there are differences in the reductive metabolism between the solid tumors and the circulation tumors [ 2 ]. Considering the importance of all the cellular reductases (e.g. NADPH cytochrome P450 reductase, cytochrome b5 reductase, [NADP]H oxidoreductase, NQO1) in response to the whole cellular reductive metabolism, these reductases are probably involved in bioactivation of AZ-1. The bioreductive drugs AZQ, mitomycin C and E09, however, have been developed to exploit the oxygen deficiency in the hypoxic fraction of solid tumors on the premise that hypoxic cells should show a greater propensity for reductive metabolism than well-oxygenated cells [ 2 , 31 - 33 ]. The major events involved in tissue homeostasis are proliferation, differentiation, and apoptosis. The processes of cell reproduction normally take place through an ordered process, generally known as the cell cycle. The tumor suppressor gene p53 is a multifunctional protein mainly responsible for maintaining genomic integrity, and it is the most frequently mutated gene in human tumors [ 34 ]. In response to DNA damage, aberrant growth signals, or chemotherapeutic drugs, p53 is stabilized and induces apoptosis and/or cell cycle arrest. While the mechanisms of p53-dependent apoptosis are not understood well, p53-dependent cycle arrest is primarily mediated by the cdk inhibitor p21 [ 34 - 36 ]. p21 regulates the cellular repair response to damaged DNA [ 37 ]. In the cytotoxicity results, AZ-1 induced the death effect of BC-M1 cells and MCF-7 cells in a dose-dependent and time-dependent manner in BC-M1 cells (Figs 1 and 2 ). From the results shown in Fig. 3 , the hormone antagonist taxmoxifen and AZ-1 were more potent than paclitaxel to our local cell line BC-M1. This indicates that the BC-M1 cell is more sensitive to the hormone antagonist drug and our bioreductive compound AZ-1 than to a nonhormone antagonist such as paclitaxel. We assumed that the BC-M1 cell is an estrogen receptor-positive cell line. According to the results of Figs 4 and 5 , the apoptosis phenomena were observed in BC-M1 cells induced by AZ-1 for 24 hours. We saw the apoptotic bodies increasing in direct proportion with the concentration of AZ-1 based on the sub-G 1 area measurement and Hoechst staining (Figs 4 and 5 ). The apoptotic bodies were increasing slightly in 0.5 μM AZ-1, and were most abundant at a concentration of 2.0 μM. The apoptotic bodies were also observed in MCF-7 cells after treatment by AZ-1 for 24 hours (Fig. 6 ). For the mechanism of the apoptotic pathway, we found that p53 might be involved in the death effect of BC-M1 cells induced by the bioreductive compound AZ-1 in this study. Figure 7 shows that the expression of the p53 protein of BC-M1 cells was upregulated from 0.5 μM, was highest at 1.0 μM and downregulated at 2 μM AZ-1 for a 24-hour challenge. The response of the expression sequence in p21 protein was upregulation from 1.0 μM to 2.0 μM AZ-1 treatment to BC-M1 cells, which was later than for p53 protein. The p21 protein is the inhibitor of cdk2, the results of Fig. 8 showing that it was decreasing to 40% of the cdk2 at 2.0 μM AZ-1 and was without any disturbance to cdk2 expression in 0.5 μM AZ-1. TIAR is a membrane of RNA recognition motif-type RNA-binding protein and also regulates the general translational arrest that accompanies environmental stress [ 38 ]. In the roles of the RNA-binding protein, TIAR involved the connections between the eukaryotic initiation factor kinase system, mRNA stability, and cellular chaperone levels [ 39 , 40 ]. Mutant mice lacking TIAR exhibit partial embryonic lethality and defective germ-cell maturation, implicating this protein in certain aspects of vertebrate development [ 41 ]. In DT40 cells, TIAR is also required for cell viability that involved the splicing of the exons [ 42 ]. In our results, the expression of TIAR protein was decreased with the concentrations of AZ-1 increasing in BC-M1 cells. We therefore proposed that TIAR was involved in the death effect of BC-M1 cells induced by AZ-1, which might play a role in cell viability, without direct proportion to apoptosis correlation. According to the report of Schmidt and colleagues, p53 expression was correlated with the resistance against paclitaxel [ 43 ]. From this report, none of the tumors with p53 expression (11 patients) responded to paclitaxel. In contrast, 10 of the 22 patients without p53 expression showed an objective response. This phenomena was also supported by the report of Takara and colleagues [ 44 ], whereby the doses of paclitaxel used in BC-M1 cells are much higher than the usual IC 50 values for most breast cancer cell lines reported in the literature (Fig. 3 ). This was also supported by the data from our results (Fig. 7 ) that the p53 protein really existed in BC-M1 cells, and with function. The report from Zhang and colleagues failed to detect the p53 protein in normal breast epithelial cells, and p53 positivity was 24% and 30% in intraductal and invasive cancer tissues, respectively [ 45 ]. The p53 protein could therefore be detected in tumor cells such as BC-M1 cells in untreated conditions, as in our results. Caspases are a large family of cysteine proteases, most of them playing central roles in the execution of apoptosis [ 46 ]. The pro-caspase protein was in its inactive form when the cell was in the rest state and the protein expression was decreasing to convert to the active form of caspase-3 in the process of apoptosis [ 47 ]. In our result, the expression of pro-caspase protein was in inverse proportion to the concentrations of AZ-1 (Fig. 8 , lane 2), and compared the enzyme activity assay of caspase-3 that was in direct proportion to AZ-1 concentration treatment to BC-M1 cells for 24 hours, as observed in Fig. 9 . The apoptosis pathway of BC-M1 cells induced by AZ-1 compound was mediated by caspase-3. In this study, we also approached the apoptosis phenomena by the NMR spectra method. Some reports provided evidence that NMR spectroscopy allows the early detection (after 6 hours) of apoptosis-induced cellular changes in cells by observing signals from small and mobile molecular species, particularly the ratio of intensities of the CH 2 and CH 3 resonance [ 22 , 48 ]. Figure 10 shows that we observed that the ratio of CH 2 and CH 3 in BC-M1 cells induced by AZ-1 for 24 hours was also in direct proportion with the concentration of AZ-1. This is also one of the pieces of evidence to prove the apoptosis phenomena in BC-M1 cells induced by AZ-1. Conclusion The apoptosis pathway in BC-M1 cells induced by various concentrations of AZ-1 was initially triggered by the activation of p53 protein and then upregulation of the p21 protein to inhibit the cyclin kinase cdk2 expression to arrest BC-M1 cell temporality and, finally, into the apoptosis process. In BC-M1 cells, the arrest and apoptosis processes, the activation of caspase-3 activity, apoptotic body formation, the ratio of CH 2 and CH 3 increasing and the expression of the RNA-binding protein TIAR decreasing were all involved in the death effect of BC-M1 cells induced by this bioreductive compound. From this study, AZ-1 could be used as a cocktail in combination with other anti-breast cancer drugs to cure the hypoxia tumor tissue of breast cancer patients to prevent cancer cell recurrence. Abbreviations AZ-1 = 2-aziridin-1-yl-3-[(2-{2-[(3-aziridin-1-yl-1,4-dioxo-1,4-dihydronaphthalen-2-yl)thio]ethoxy}ethyl)thio]naphthoquinone; cdk = cyclin-dependent kinase; DMEM = Dulbecco's modified Eagle's medium; IC 50 = inhibition concentration of 50% cell growth; NMR = nuclear magnetic resonance; PBS = phosphate-buffered saline; TIAR = T-cell restricted intracellular antigen-related protein. Competing interests The author(s) declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064093.xml

1064094

Presence of papillomavirus sequences in condylomatous lesions of the mamillae and in invasive carcinoma of the breast

Background Viruses including Epstein–Barr virus (EBV), a human equivalent of murine mammary tumour virus (MMTV) and human papillomavirus (HPV) have been implicated in the aetiology of human breast cancer. We report the presence of HPV DNA sequences in areolar tissue and tumour tissue samples from female patients with breast carcinoma. The presence of virus in the areolar–nipple complex suggests to us a potential pathogenic mechanism. Methods Polymerase chain reaction (PCR) was undertaken to amplify HPV types in areolar and tumour tissue from breast cancer cases. In situ hybridisation supported the PCR findings and localised the virus in nipple, areolar and tumour tissue. Results Papillomavirus DNA was present in 25 of 29 samples of breast carcinoma and in 20 of 29 samples from the corresponding mamilla. The most prevalent type in both carcinomas and nipples was HPV 11, followed by HPV 6. Other types detected were HPV 16, 23, 27 and 57 (nipples and carcinomas), HPV 20, 21, 32, 37, 38, 66 and GA3-1 (nipples only) and HPV 3, 15, 24, 87 and DL473 (carcinomas only). Multiple types were demonstrated in seven carcinomas and ten nipple samples. Conclusions The data demonstrate the occurrence of HPV in nipple and areolar tissues in patients with breast carcinoma. The authors postulate a retrograde ductular pattern of viral spread that may have pathogenic significance.

Introduction Breast cancer is the most frequently diagnosed cancer among women in the USA, and the incidence of breast carcinoma has increased by more than 40% over 25 years [ 1 ]. The aetiology of breast cancer remains unknown. Many risk factors have been associated with the pathogenesis of this disease, including family history, hormones, cigarette smoking and alcohol consumption [ 2 - 7 ]. Hormones, cigarette smoking and family history have also been demonstrated to enhance infections with papillomaviruses, mainly the high-risk human papillomavirus (HPV) types involved in the aetiology of cervical carcinoma [ 8 ]. Viral infection as an aetiological factor has been addressed in other studies. The data obtained from studies investigating the presence of viral sequences in breast cancer biopsies and cell lines have been controversial. A role for herpes viruses, and specifically Epstein–Barr virus (EBV), in the aetiology of this cancer has been questioned [ 9 , 10 ]. A human equivalent of the murine mammary tumour virus (MMTV) has been described in breast tumours, as well as in breast cancer cell lines [ 11 ]. These results have been questioned and negated by others (R Schmidt and H zur Hausen, unpublished results) [ 12 ]. Immortalisation of normal breast epithelial cells by the HPV types HPV 16 and HPV 18 [ 13 ] has been used to study the functions of the viral early genes E6 and E7 in different cellular pathways [ 14 - 18 ]. However, the presence of HPV in malignant tumours of the breast has been controversial. Several studies reported positive results [ 19 - 25 ], whereas others reported negative results [ 26 - 28 ]. In the present study we report the presence of HPV DNA in biopsy samples from mammary carcinomas, as well as these viral sequences in the tissue of the corresponding mamilla. Materials and methods Samples and DNA extraction Samples from breast tumour tissue for the HPV analysis were supplied by two of the authors (RES and CEB). The 29 cases used for the retrospective study and demonstration of papillomavirus sequences were selected from routine mastectomy specimens undertaken in a course of treatment for carcinoma of the breast. Case selection was based on the availability of adequate nipple and tumour tissue from the same patient and the presence of histological features suggestive of HPV infection in nipple and areolar sections. The patients, all female, ranged in age from 30 to 88 years, with an average age of 61.3 years. Twelve tumours arose in the right breast and 17 in the left. Five tumours were subareolar and 24 were located at a distance from the nipple and areola. Tumours ranged in size from 6 mm to 15 cm. One patient received cytoxan chemotherapy before surgery. In all other cases primary treatment was surgical as represented by the specimen. One patient had undergone contralateral mastectomy ten years earlier for a localised breast cancer, which was judged clinically and pathologically to represent a separate primary tumour. Among the 29 patients there were three with remote histories of malignancy without evidence of recurrence (one cutaneous malignant melanoma, one renal cell carcinoma and one soft-tissue malignancy). One patient had undergone hysterectomy for high-grade cervical dysplasia. There were no patients with documented cervical carcinoma. No patients were immune-impaired or immunosuppressed. Formalin-fixed paraffin-embedded samples from both the nipple ( n = 29) and the carcinoma of the breast ( n = 29) from the same patient were sectioned, and the total DNA was extracted from each sample. Great care was taken during sectioning to avoid any contamination between samples. Two different individuals sectioned the blocks in several small random groups at different time intervals over a period of one week. The microtome was cleaned thoroughly and exposed to ultraviolet between samples; in addition, new blades were used for each sample. Deparaffinisation was performed by rotation overnight in 1 ml of xylene, followed by centrifugation and subsequent removal of the supernatant. This step was repeated twice (each for one hour) with fresh xylene. The xylene was in turn removed by rotation in 1 ml of 100% ethanol for one hour followed by centrifugation and subsequent replacement of the supernatant with 90% ethanol for 45 min, and repeated by steps of 80% ethanol for 45 min and 70% ethanol for 45 min. Samples were freeze-dried, and total cellular DNA was extracted as described previously [ 29 ]. The DNA was digested with Proteinase K and subjected to extraction with phenol followed by extraction with chloroform/isoamyl alcohol. The extracted DNA was precipitated with ethanol and the pellet was resuspended in 10 mM Tris/HCl, pH 8.0. PCR analysis and cloning DNA (50–100 μg) of each sample was amplified by polymerase chain reaction (PCR). The quality of the DNA obtained from the fixed samples was controlled by amplification with primers to detect the β-actin gene [ 30 ]. The primers RS42 and KM29 were initially used to amplify the 536-base-pair fragment of the β-actin gene. If this size of fragment could not be amplified, the primers PC03 and PC04 were used to amplify the 110-base-pair (bp) fragment of the β-actin gene. Papillomavirus sequences were amplified by three different methods, each targeting highly conserved regions within the L1 open reading frame of papillomaviruses. These included the GP5 + /GP6 + primers [ 31 ], the CP primers using modified conditions as previously described [ 32 ] and the FAP primers [ 33 ]. All three primer combinations are routinely used on individual samples tested in our laboratory. We initially modified the PCR conditions for each described method to include the amplification of the largest number of individual HPV types possible. Cloned DNA of each known HPV type was used to optimise these conditions. Amplification with the GP primers was performed in 2 mM MgCl 2 using 40 cycles with an annealing temperature of 40°C. The expected size of the amplicon was 140–150 bp. The expected size with the FAP primers is 480 bp after 45 cycles in 3.5 mM MgCl 2 and an annealing temperature of 50°C. The nested amplification approach with the CP primers results in an amplicon size of 370–400 bp. The initial amplification was performed in 2 mM MgCl 2 and 40 cycles at 50°C annealing temperature, after which an aliquot was re-amplified with nested primers but with the same PCR conditions. All amplicons were eluted after gel electrophoresis, purified and cloned. At least ten fragments of each amplicon were sequenced. Sequencing was performed with either the Sequenase 2.0 DNA Sequencing Kit (USB, Cleveland, OH) or an ABI Model Sequencer with Big Dye Terminator chemistry (Perkin Elmer Applied Biosystems Division, Dreieich, Germany). All precautions were taken to avoid contamination before or during extraction of the DNA from the embedded formalin-fixed samples. After initial results had been obtained, a second sectioning was performed on the same embedded samples and DNA was again extracted to ensure that no contamination had taken place during the first round of experiments. This second round of analyses was performed in different rooms and by different individuals. Sequence analysis Sequences were compared with all available data banks with the aid of the Husar software package (Deutsches Krebsforschungszentrum). In situ hybridisation In situ hybridisation was performed as described by Kawase and colleagues [ 34 ]. Papillomavirus probes were digoxigenin-labelled by PCR in accordance with the manufacturer's instructions (PCR DIG Probe Synthesis Kit, catalogue no. 1636090; Roche, Mannheim, Germany). The FAP primers were used to label HPV 6 and HPV 11 DNA and the CP primers to label HPV 16 DNA. Denaturation was performed in a Biozym cycler for 5 min at 95°C followed by hybridisation overnight at 37°C under stringent conditions as described previously [ 35 ]. This was followed by washing with solutions in accordance with instructions in the Genpoint Kit (catalogue no. K0620; Dako, Hamburg, Germany). Anti-digoxigenin-AP antibodies (anti-digoxigenin-AP, Fab fragments, catalogue no. 1093274; Roche), blocking reagent (catalogue no. 1096176; Roche) and Nitro Blue Tetrazolium/5-bromo-4-chloroindol-3-yl phosphate (catalogue no. 1681451; Roche) were applied for visualisation of the signals. Sections from each sample were used for in situ hybridisation. Reading of the samples was blinded and performed by one of us (HzH). Results Nipple specimens were reviewed and selected by two of us (CEB, RES) for histological features consistent with human papillomavirus infections. Features identified included epithelial hyperplasia, hypergranulosis, parakeratosis, and distorted nuclear shape associated with nuclear clearing and pyknosis suggestive of koilocytosis. In addition, squamous metaplasia of lactiferous ducts and shedding of metaplastic cellular elements into lactiferous ducts and sinuses were noted (Fig. 1 ). In addition to the 29 invasive carcinomas, ductal carcinoma in situ was identified in lactiferous ducts in four samples and small benign papillomas were present in three samples. Most invasive carcinoma samples (18 of 29) displayed a duct cell carcinoma pattern (Table 1 ). Other types included medullary carcinoma (five cases), tubular carcinoma (three cases), lobular carcinoma (two cases) and mucinous carcinoma (one case). DNA extracted from the samples taken from the nipple and the corresponding carcinoma of each breast (29 pairs) was analysed by PCR for the presence of papillomavirus sequences. The quality of the DNA was controlled by amplification of the β-actin gene. The 536-base-pair fragment of this gene was amplified in 47 of 58 samples. The 110-base-pair fragment was amplified in the remaining 11 samples (Table 2 ). Intact papillomavirus particles are resistant to degeneration. Extraction of DNA from fixed tissue can harbour intact viral genomes even under conditions in which the cellular DNA has been degraded. The three different methods used for amplification of the papillomavirus DNA were chosen to ensure amplification of all known papillomavirus types, as well as putative new types. All three primer combinations (GP, CP and FAP) are routinely used on individual samples tested in our laboratory. We initially modified the PCR conditions for each described method to include the amplification of the largest number of individual HPV types possible. Cloned genomic DNA of each known HPV type was used to optimise these conditions. The GP primers have been designed to amplify the majority of papillomavirus types associated with mucosal lesions. The FAP primers, in contrast, were initially designed to amplify the majority of papillomavirus types associated with cutaneous lesions. The amplicon size described is 480 bp, but in our hands smaller fragments identified as papillomavirus sequences (after cloning and sequencing) were also generated, depending on the HPV type. The CP primers were originally designed to amplify all the epidermodysplasia verruciformis-associated HPV types, but under modified conditions a large spectrum of mucosal types are amplified as well. Combining the three methods with the modified PCR conditions allows us to amplify all known HPV types with more or less equal efficiency. The amplicons generated, using any of the described methods, often constitute cellular sequences depending on the viral load of the sample. Normal breast and areolar tissue was not available for use as control samples. The described methods of testing have been used routinely in our laboratory over several years in the analyses of tissues from a large variety of organs. We have analysed large numbers of samples from benign and malignant tumours of the oesophagus, head and neck and skin, as well as normal skin tissue for the presence of papillomavirus DNA with identical methods. HPV positivity varied depending on the type of tumour and normal tissue [ 29 , 36 - 38 ] (E-MdV, unpublished data). The present study describes the first series of tissue of the breast analysed in our laboratory for the presence of papillomavirus DNA. Sequence analyses showed that 25 (86%) of the breast carcinoma samples and 20 (69%) of the samples from the mamilla harboured papillomavirus sequences. The results for each sample using each of the primer combinations are presented in Table 2 . Only one pair of samples were both negative for HPV DNA. HPV DNA was present in 17 (59%) pairs of samples, namely nipple as well as carcinoma, whereas papillomavirus DNA was detected only in the carcinomatous tissue in eight cases (28%) and only in the nipple in three cases (10%) (Table 3 ). HPV 11 DNA was present in both carcinoma and nipple samples in seven cases, HPV 6 in one case and HPV 57 in two cases. The most prevalent HPV types were HPV 11 (19 carcinomas and 8 nipples) followed by HPV 6 (7 carcinomas and 6 nipples) (Table 4 ). Other HPV types detected were HPV 16 (2 carcinomas and 3 nipples), HPV 57 (2 carcinomas and 3 nipples), HPV 27 (2 carcinomas and 4 nipples), HPV 66 (2 nipples), candHPV 87 (4 carcinomas), HPV 37 (4 nipples) and the putative new HPV type DL250 (HPV 9-related) in two carcinomas. HPV types detected in single samples of the nipple were HPV 20, 21, 23, 32, 38 and GA3-1 (HPV 8-related) and HPV 3, 15, 23, 24 and DL473 (HPV 15-related) in carcinoma samples. Multiple HPV types were demonstrated in eight carcinomas (24%) and ten nipple (34.5%) samples. In situ hybridisation was performed on all samples that had previously been positive by PCR amplification for HPV 6, HPV 11 or HPV 16. Positive signals were clearly distinguishable from the surrounding negative tissue (Fig. 2 ). All carcinoma samples that had been negative for one of these three HPV types by PCR were also subjected to in situ hybridisation with each of the respective labelled HPV probes. No positive hybridisation signal was observed in any of these samples. In situ hybridisation supported the presence of HPV 6, 11 and 16 in nipple and areolar samples and in all tumour tissue in which these type-specific HPV sequences were detected by PCR analysis. Discussion The ubiquitous distribution of papillomavirus infections of the skin and genital and oral mucosa has been documented [ 39 - 41 ]. Infection with specific papillomavirus types has been shown to be a necessary but not sufficient cause in the pathogenesis of malignant genital tumours. The malignant tumours generally develop after long latency periods during which additional cellular modifications occur within the infected cell [ 8 ]. The E6 and E7 genes of the most prevalent high-risk HPV types, HPV 16 and HPV 18, modulate cellular pathways, thereby regulating proliferation and cell survival [ 8 , 42 ]. In contrast, the E6 and E7 proteins of the low-risk types, HPV 6 and HPV 11, do not influence these cellular pathways in the same manner, although they have occasionally been demonstrated in premalignant and malignant tumours. Additional cofactors are probably needed to modulate cellular proteins so as to immortalise and transform the infected cells [ 8 ]. The mechanism through which other HPV types, mainly associated with cutaneous lesions, might be involved in the pathogenesis of tumours has received little attention. However, preliminary data indicate that several molecular pathways are probably followed, depending on the HPV type involved [ 43 ]. The detection of papillomavirus infections in tissues largely depends on the method used. Described methods vary greatly in their sensitivity and specificity. The HPV types HPV 6, 11, 16, 18, 31, 33, 35, 39, 52 and 58 are frequently associated with genital lesions and are therefore most often targeted for HPV detection. Tests to demonstrate any of the 96 characterised HPV types in tissues require extensive analyses, including the sequencing of cloned amplimers. Most published studies therefore used methods restricted to the detection of specific, single, or combinations or groups of HPV types. The use of type-specific primers may increase the number of positive samples but is biased with regard to the HPV types involved, because other HPV types present cannot be detected. The present study investigated breast samples by using DNA PCR amplification of two conserved regions within the L1 open reading frame of the papillomavirus genome. Two of the primer combinations used (CP and GP primers) amplify an overlapping region within the L1 open reading frame. By using all three primer combinations to amplify the DNA of a single sample we are able to detect all known HPV types and also putative new HPV types. In addition, the cloning of the amplified product, in combination with sequencing of the cloned inserts, allows us to distinguish individual HPV types present in one sample, including multiple infections in one sample. Studies applying the techniques described here have demonstrated a larger spectrum of papillomavirus types associated with different types of benign and malignant tumours and argue that the historical grouping into 'mucosal' and 'cutaneous' HPV types can no longer be upheld [ 44 ]. We are not aware of any study outside our group that has analysed tissues for the presence of infection by any HPV types as extensively as the study described here. Other investigators have analysed nipple and areolar specimens for the presence of HPV DNA. HPV 6/11 was detected in one of 20 papillomas of the nipple, whereas ten nipple duct adenomas were negative [ 45 ] and seven low-risk and high-risk HPV types were not present in 20 cases of Paget's disease of the nipple [ 46 ]. Single cases of papillomas of the nipple resembling condylomata acuminata harboured HPV 6 DNA [ 47 ] and HPV 41 DNA [ 48 ]. Reports on the detection of HPV 16, HPV 18 and/or HPV 33 DNA in breast carcinoma samples range between 11% and 68% positivity [ 19 - 25 ]. The present study detected HPV 16 DNA in 7% of the carcinoma and in 10% of the nipple samples. The only other high-risk HPV type, HPV 66, was present in two samples from the nipple. The reasons for the differences in published reports are unclear, but may be attributed to numbers of samples tested, methodological differences or the demographics of the samples tested. In contrast, several other studies failed to demonstrate HPV sequences in tumours of the breast: all of 15 papillomas and 28 breast carcinoma samples were negative for HPV 6, 16 and 18 DNA [ 26 ], no HPV 16 and HPV 18 DNA could be detected in 26 fine needle aspirates and four breast carcinoma biopsies [ 28 ], and no HPV DNA was found in 80 samples of breast carcinomas [ 27 ]. HPV 16 and HPV 18 oncogenes immortalise normal human mammary cells under in vitro conditions [ 14 - 18 ]. It is unclear whether these data truly represent the in vivo situation. The present study detected papillomavirus sequences in 86% (25/29) of the breast carcinoma samples and 69% (20 of 29) of the nipple samples. HPV 6 and HPV 11 infections were most prevalent in our collection of samples. Either of the two HPV types was present in 20 of 29 (69%) of the breast carcinoma biopsies and in 12 of 29 (41%) of the samples of the mamilla. We detected the simultaneous presence of more than one HPV type in about one-third of the samples. HPV types associated with mucosal lesions, and also several associated with skin lesions, were detected. HPV 27 and HPV 57 are closely related HPV types. HPV 27 is very frequently found in cutaneous warts, whereas HPV 57 has been demonstrated in mucosal and cutaneous lesions [ 49 ]. Several other HPV types found in the breast samples have historically been associated with the rare hereditary disease epidermodysplasia verruciformis but have more recently been shown in non-melanoma skin cancers and their precursors in both immunosuppressed and immunocompetent patients [ 44 ]. The confirmation of PCR-derived evidence of HPV sequences in nipple, areolar and tumour tissue by in situ hybridisation in our hands lends substantive support to the conjecture that viral sequences are present and not an artefact in the tissues we have examined. The absence of HPV DNA in several samples, as well as the detection of different HPV types in the carcinoma versus the nipple of the same patient, might be attributed to the viral load at the time of sampling in combination with the sensitivity of the tests performed. The tissue sections also harboured unaffected tissue, leading to the dilution of virus-containing cells. In addition, the amplification of viral DNA by PCR with degenerate or consensus primers will detect replicating viral sequences and is not sensitive enough to detect all individual HPV types with equal sensitivity in a mixed population of cells. The in situ hybridisation technique is also less sensitive and will not detect a single viral genome copy per cell. HPV 6 and HPV 11 are regarded as low-risk HPV types because they have only rarely been demonstrated in premalignant or malignant tissue, and the early genes E6 and E7 do not immortalise primary keratinocytes in vitro [ 50 ]. HPV 6 DNA has been found in carcinomas [ 51 , 52 ] and giant condylomas (Buschke–Löwenstein tumours) [ 53 , 54 ]. The molecular mechanism by which these low-risk HPV types induce or participate in the transformation of cells has not been resolved. Duplications of the long control region [ 54 - 57 ], minimal sequence dissimilarities [ 58 , 59 ] and integration [ 60 ] of HPV 6 and/or HPV 11 genomes have been shown in malignant tissue. The mere presence of the low-risk HPV 6 and 11 in most of the mamillae and in the concomitant breast carcinoma samples does not prove its role in the aetiology of the disease, although it merits further investigation. Unfortunately, because of the previous fixation of our samples, RNA from these tumours was not available for the detection of viral transcripts as additional support of our results. Relatively few studies have been performed in an attempt to unravel the intracellular mechanisms through which HPV 6 or 11 may immortalise cells, and information is still fractional in comparison with data on the high-risk HPV types. Introduction of the HPV 6 E6 into normal mammary cells leads to immortalisation and reduced levels of the p53 protein [ 61 ]. Binding of the HPV 6 E6 protein modulates the function of the cellular proteins Bak [ 62 ], Gps2 [ 63 ] and Zyxin [ 64 ]. Additional in vitro studies investigating the influence of hormones, smoke adducts and genetic factors on the interaction of the HPV 6 and 11 proteins with cellular proteins should provide valuable information on a possible role for HPV 6 and 11 in the pathogenesis of breast carcinoma. The data presented in this study also indicate a strong need for epidemiological studies correlating HPV, cigarette smoking, hormone use and family history to substantiate in vitro findings. This study suggests that human papillomavirus may infect the epithelium of the nipple and areola. Human papillomavirus infection may be identified with recognisable histological features comparable to human papillomavirus infection at other sites. Furthermore, this study is consistent with a pathogenic mechanism involving transfer in a retrograde fashion via the nipple, areola, lactiferous ducts and sinuses of the human papillomavirus. If confirmed, the pathogenic mechanism proposed may have significant implications for the prevention and treatment of breast carcinoma and the identification of individuals at risk for carcinoma development. Examination of cellular samples of nipple and areola inclusive of cytological examination and methods for identifying the presence of human papillomavirus might aid early diagnosis and perhaps therapy. Conclusions The data demonstrate the occurrence of HPV in nipple and areolar tissues in patients with breast carcinoma. The presence of papillomavirus DNA in most mamillae and concomitant breast carcinoma samples merits further investigation. Competing interests The author(s) declare that they have no competing interests. Abbreviations EBV = Epstein–Barr virus; HPV = human papillomavirus; MMTV = murine mammary tumour virus; PCR = polymerase chain reaction.

D:\keerthana\PMC001xxxxxx\PMC1064094.xml

1064095

Polychlorinated biphenyls, cytochrome P450 1A1 (CYP1A1) polymorphisms, and breast cancer risk among African American women and white women in North Carolina: a population-based case-control study

Introduction Epidemiologic studies have not shown a strong relationship between blood levels of polychlorinated biphenyls (PCBs) and breast cancer risk. However, two recent studies showed a stronger association among postmenopausal white women with the inducible M2 polymorphism in the cytochrome P450 1A1 ( CYP1A1 ) gene. Methods In a population-based case-control study, we evaluated breast cancer risk in relation to PCBs and the CYP1A1 polymorphisms M1 (also known as CYP1A1 *2A), M2 ( CYP1A1 *2C), M3 ( CYP1A1 *3), and M4 ( CYP1A1 *4). The study population consisted of 612 patients (242 African American, 370 white) and 599 controls (242 African American, 357 white). Results There was no evidence of strong joint effects between CYP1A1 M1-containing genotypes and total PCBs in African American or white women. Statistically significant multiplicative interactions were observed between CYP1A1 M2-containing genotypes and elevated plasma total PCBs among white women ( P value for likelihood ratio test = 0.02). Multiplicative interactions were also observed between CYP1A1 M3-containing genotypes and elevated total PCBs among African American women ( P value for likelihood ratio test = 0.10). Conclusions Our results confirm previous reports that CYP1A1 M2-containing genotypes modify the association between PCB exposure and risk of breast cancer. We present additional evidence suggesting that CYP1A1 M3-containing genotypes modify the effects of PCB exposure among African American women. Additional studies are warranted, and meta-analyses combining results across studies will be needed to generate more precise estimates of the joint effects of PCBs and CYP1A1 genotypes.

Introduction The role of organochlorine compounds such as polychlorinated biphenyls (PCBs) as etiologic agents for breast cancer has been investigated in several epidemiologic studies, but findings are inconsistent [ 1 ]. Initial studies examining breast tissue or serum showed significantly higher concentrations of PCBs in patients with breast cancer compared with controls [ 2 - 6 ], but more recent studies observed no association [ 7 - 10 ] or an inverse association [ 11 ] with breast cancer. Several mechanisms have been proposed to explain how PCB exposure might increase the risk of breast cancer [ 12 - 15 ]. Owing to their lipophilic nature, organochlorine compounds undergo lifelong sequestration in animal and human adipose tissues. Adipose tissue within the human breast contains measurable levels of PCBs [ 16 ]. Metabolism of PCBs within breast tissue can generate reactive intermediates that cause oxidative damage and react with DNA. The proposed mechanism involves cytochrome P450 1A1 ( CYP1A1 ), a gene that is inducible by PCBs [ 17 , 18 ]. CYP1A1 encodes aryl hydrocarbon hydrolase (AHH), an enzyme involved in the production of reactive epoxide intermediates from polycyclic aromatic hydrocarbons, steroid hormones, and other aromatic compounds [ 19 ]. On the basis of this model, individuals with higher CYP1A1 activity would be at increased risk of breast cancer when exposed to high levels of PCBs. Several polymorphisms have been identified in CYP1A1 , some of which lead to more highly inducible AHH activity [ 19 - 22 ]. CYP1A1 polymorphisms include M1 (T→C substitution at nucleotide 3801 in the 3'-noncoding region), M2 (A→G substitution at nucleotide 2455 leading to an amino acid change of isoleucine to valine at codon 462), M3 (T→C substitution at nucleotide 3205 in the 3'-noncoding region), and M4 (C→A substitution at nucleotide 2453 leading to an amino acid change of threonine to asparagine at codon 461). Studies of CYP1A1 in cultured human lymphocytes showed significantly elevated levels of inducible enzyme activity among persons with M2 genotypes [ 20 , 22 ]. Crofts and colleagues [ 21 ] reported that M2 alleles were associated with increased CYP1A1 inducibility at the mRNA level, and a threefold elevation in AHH enzyme activity. The M1 allele also encodes an inducible form of CYP1A1 [ 19 , 22 , 23 ]. Two recent epidemiologic studies showed an increased risk of breast cancer among postmenopausal white women with at least one M2 (valine) variant allele of CYP1A1 and a high serum level of PCBs [ 24 , 25 ]. These findings suggest that interactions between CYP1A1 polymorphisms and PCBs could be involved in the development of breast cancer, and genotyping for CYP1A1 could help to resolve inconsistencies in past epidemiologic studies. We used data from the Carolina Breast Cancer Study (CBCS), a population-based case-control study of African American and white women in North Carolina, to evaluate the effects of PCBs and CYP1A1 genotypes in relation to breast cancer risk. Materials and methods Study population CBCS study participants were women aged 20–74 years residing in 24 contiguous counties in central and eastern North Carolina [ 26 ]. Patients were women with a first diagnosis of invasive breast cancer identified through a rapid ascertainment system implemented in cooperation with the North Carolina Central Cancer registry. Controls were selected from records of the North Carolina Division of Motor Vehicles and US Health Care Financing Administration, and were frequency-matched to patients on the basis of race and age. Information was collected through in-person interviews and included reproductive history, diet and lifestyle factors, a detailed family history of cancer, and occupational history. About 98% of participants who were interviewed agreed to give a 30 ml blood sample at the time of interview. Informed consent to obtain DNA and measurement of plasma concentration of PCBs was sought with the use of a form approved by the Institutional Review Board of the UNC School of Medicine in compliance with the Helsinki Declaration. The CBCS was conducted in two phases. Phase I of the study was conducted between May 1993 and December 1996 [ 26 ]. Phase II of the study was conducted between 1996 and 2001. The total numbers of participants enrolled in Phase I of the CBCS were 861 patients (335 African American, 526 white) and 790 controls (332 African American and 526 white). The response rates were 76% for patients and 55% for controls. Genotyping for CYP1A1 was performed on the first 688 breast cancer patients and 702 controls enrolled in Phase I of the CBCS. Laboratory assays for organochlorines were conducted on blood samples from 748 patients and 659 controls in Phase I of the CBCS [ 26 ]. The present analyses were based on Phase I participants for whom both CYP1A1 genotyping and organochlorine measurements were available: 612 patients (242 African American, 370 white) and 599 controls (242 African American, 357 white). Laboratory methods Methods for identification of plasma PCBs on CBCS plasma samples using gas chromatography/electron capture detection have been described previously [ 26 ]. Methods for genotyping of CYP1A1 on CBCS samples using restriction fragment length polymorphism–polymerase chain reaction have been described previously [ 27 ]. In a newly proposed nomenclature for CYP1A1 alleles, the M1 allele corresponds to CYP1A1 *2A, M2 to CYP1A1 *2C, M3 to CYP1A1 *3, and M4 to CYP1A1 *4 [ 28 ]. Data were missing on the following CYP1A1 genotypes due to unreadable gels or failure to amplify: M1 (10 patients and 7 controls), M2 (none missing), M3 (10 patients and 7 controls), and M4 (none missing). To correct for differences in plasma concentrations of chlorinated hydrocarbons attributable to blood lipids, the quantity of different lipid components in the plasma of each subject was measured with an automated enzymic assay, as described previously [ 26 ]. Statistical methods Imputed total PCB values were used in the data analyses. Imputation was performed by setting zero values and values below the detection limit (0.0125 ng/ml) to the detection limit divided by the square root of 2 (26). Hornung and Reed [ 29 ] recommended this method of imputation as an accurate approach for computing means and standard deviations for variables that include nondetectable values. In the present study, imputation did not affect the estimation of odds ratios (ORs) for breast cancer and total PCBs, because individuals in the lowest category (that is, below the median) remained in the lowest category after imputation. Lipid-adjusted total PCBs levels were calculated with Equation 2 of Phillips and colleagues [ 30 ]. Univariate analyses were performed for total PCBs, lipid-adjusted total PCBs, low to moderate chlorinated PCBs, highly chlorinated PCBs, and individual PCB congeners 99, 105, 118, and 153. Among controls in the CBCS, levels of individual PCB congeners and congener groups were highly intercorrelated [ 31 ]. ORs for the joint effects of CYP1A1 genotypes and PCB exposure did not differ substantially according to specific PCB congeners or congener groups, and ORs were similar to results obtained with total PCBs. We did not observe differences in ORs for breast cancer and PCB exposure, or the joint effects of CYP1A1 genotypes and PCBs, when we analyzed PCB congeners 99, 105, 118, or 153 alone, as suggested by Demers and colleagues [ 32 ]. Therefore, only results for lipid-adjusted total PCBs are presented. Adjusted ORs for breast cancer and 95% confidence intervals (CIs) were calculated from unconditional logistic regression models. Total PCBs were categorized as greater than or equal to versus below the median, on the basis of the distribution in African American or white controls separately. PROC GENMOD of software package SAS (version 8.1; SAS Institute, Cary, NC) was used to incorporate offsets derived from the sampling probabilities used to identify eligible participants. Covariates included age, race (whites/African Americans), parity (nulliparous, 1 or ≥ 2), use of hormone replacement therapy (never or ever), oral contraceptive use (never or ever), breast feeding (never or ever), smoking (never, current, or former smoker), alcohol consumption (yes/no), family history of breast cancer (yes/no), benign breast biopsy (yes/no), income (less than $30,000/year or at least $30,000/year), and education (lower than high school, or high school and above). Continuous covariates, which included height, waist/hip ratio, and body mass index, were categorized on the basis of the median values of each variable in all controls. ORs did not differ after adjusting for additional covariates, so results are presented for African American and white women adjusting for sampling fractions and age. Menopausal status was defined as described previously [ 27 ] and used to stratify analyses of CYP1A1 genotypes and PCB levels. Stratified analyses were also conducted on the basis of smoking history (ever or never). In addition, we estimated the increase in odds of breast cancer per 0.10 ng/ml increase in total PCB levels by coding lipid-adjusted total PCBs as a continuous variable, and stratifying on CYP1A1 genotypes. Linearity in the logit was tested using the Box–Tidwell transformation test as implemented in SAS (version 8.1). The assumption of linearity was not violated in premenopausal or postmenopausal African American or white women. To assess the interaction on the additive scale between PCB levels and CYP1A1 genotypes, indicator variables were created for each category of joint exposure of PCBs and CYP1A1 genotype. Women with the homozygous common (non-M1, non-M2, non-M3, or non-M4) genotypes and the lowest level of exposure to PCBs were used as a common reference group. Interaction contrast ratios (ICRs) and 95% CIs were calculated for joint effects of the M1 genotype and PCBs [ 33 ]. ICR values greater than zero indicate greater than additive effects (synergy), and 95% CIs for the ICR that exclude zero can be used as a test for statistical significance at an alpha level of 0.05. To test for interaction on a multiplicative scale, likelihood ratio tests (LRTs) were conducted comparing logistic regression models with main effect terms for lipid-adjusted total PCBs and CYP1A1 genotypes and a product interaction term compared with models with main effects only. P values for LRTs less than 0.20 were considered statistically significant [ 34 ]. To test for independence of environmental exposure and genotype, means, medians, and distributions of total PCBs were determined in relation to CYP1A1 genotypes among African American and white controls. Medians for lipid-adjusted total PCBs were compared by using the Wilcoxon rank sum test. There were no statistically significant differences in medians for total PCBs comparing participants with CYP1A1 M1-containing versus non-M1-containing genotypes, M2 versus non-M2, M3 versus non-M3, or M4 versus non-M4 genotypes. Results Genotype frequencies for CYP1A1 , and ORs for breast cancer, have previously been reported for the CBCS [ 27 ]. In brief, genotype frequencies for CYP1A1 M1- and M3-containing genotypes were higher among African Americans than among whites, whereas M2- and M4-containing genotypes were more prevalent among whites. ORs for CYP1A1 genotypes and breast cancer were close to the null in African Americans and whites [ 27 ]. The association between plasma levels of total PCBs and breast cancer in the CBCS was reported previously [ 26 ]. A weak positive association for high levels of PCBs and breast cancer was found among African American women. ORs were close to the null for PCBs and breast cancer in white women. ORs for breast cancer estimating the joint effects of total PCBs and CYP1A1 M1 genotypes on an additive scale are presented in Table 1 . ORs were slightly elevated for PCB exposure greater than or equal to the median among premenopausal African American women regardless of CYP1A1 M1 genotype. There was no evidence for strong joint effects of PCBs and CYP1A1 M1-containing genotypes. ICRs were close to zero in African American and white women, and LRTs were not statistically significant. Joint effects for PCBs and CYP1A1 M2-containing genotypes among white women are presented in Table 2 . Greater than additive joint effects were observed among all participants, with an ICR greater than zero and an associated P value of 0.03. Although imprecise, joint effects seemed to be stronger among premenopausal than among postmenopausal women. P values for LRTs were 0.02 among all participants, 0.007 among premenopausal women, and 0.66 among postmenopausal women. ORs estimating the joint effects of PCBs and CYP1A1 M3 genotype among African American women are presented in Table 3 . ORs were slightly elevated for women with elevated PCB levels and CYP1A1 M3-containing genotypes. An ICR greater than zero was observed in postmenopausal women. P values for LRTs were 0.10 in all participants, 0.71 among premenopausal women, and 0.11 in postmenopausal women. Joint effects of PCBs and CYP1A1 M4 genotypes among white women are presented in Table 4 . An ICR greater than zero ( P = 0.03) was observed among postmenopausal women. The P value for the LRT in postmenopausal women was 0.07. ICRs greater than zero for the joint effects of exposures with ORs less than 1.0 can be interpreted as antagonism [ 33 ]. Results were similar among smokers and non-smokers (data not shown). ORs for breast cancer per 0.10 ng/ml increase in lipid-adjusted total PCBs showed only slight differences according to CYP1A1 M2 genotypes in white women. ORs for premenopausal women were 1.3 (95% CI 0.7–2.3) for those with CYP1A1 M2-containing genotypes and 1.0 (95% CI 0.9–1.1) for non-M2 genotypes. The corresponding ORs for postmenopausal women were 1.2 (95% CI 0.8–1.8) and 1.1 (1.0–1.2). Discussion We estimated the joint effects of plasma levels of total PCBs and CYP1A1 genotypes in association with breast cancer, using previously collected data from a population-based case-control study of African American and white women in North Carolina. ORs were slightly elevated for premenopausal white women with CYP1A1 M2-containing genotypes and lipid-adjusted total PCBs greater than the median, and for postmenopausal African American women with M3-containing genotypes and total PCBs greater than the median. ORs and ICRs were imprecise because of small sample size, and many of our results could be due to chance. However, biologic evidence and previous epidemiologic studies support the possibility of causal interaction between CYP1A1 genotypes (in particular, M2-containing genotypes) and PCB exposure in the etiology of breast cancer. PCBs are metabolized by cytochrome P450 enzymes, activate CYP1A1 , and produce free-radical-induced oxidative DNA damage in breast tissue [ 35 ]. With regard to CYP1A1 M1-containing genotypes, a previous study by Laden and colleagues [ 25 ] did not report a positive association for M1-containing genotypes and high levels of PCBs in postmenopausal women: the OR for women in the highest one-third of lipid-adjusted total PCBs (at least 0.67 μg per g lipid) and M1-containing genotypes was 1.1 (95% CI 0.5–2.5) compared with women in the lowest one-third of PCBs with non-M1-containing genotypes. In our data set, the corresponding age-adjusted OR for lipid-adjusted total PCBs of 0.67 ng/ml or more and CYP1A1 M1-containing genotypes among postmenopausal white women was 1.8 (95% CI 0.5–6.9). With regard to CYP1A1 M2-containing genotypes, Moysich and colleagues [ 24 ] reported an OR for breast cancer of 2.9 (95% CI 1.2–7.5) in postmenopausal women with total PCBs between 3.72 and 19.04 ng/g and CYP1A1 M2-containing genotypes compared with women with low PCBs and non-M2-containing genotypes. In our data set, the corresponding age-adjusted OR for total PCBs of 3.73 ng/ml or more (not lipid-adjusted) and CYP1A1 M2-containing genotypes was 6.3 (95% CI 0.7–55.0). Laden and colleagues [ 25 ] reported an OR of 2.8 (95% CI 1.0–7.8) for women with the highest one-third of lipid-adjusted total PCBs (0.67 μg per g lipid) and CYP1A1 M2-containing genotypes, compared with women in the lowest one-third of PCBs and non-M2-containing genotypes. The corresponding OR in our data set for lipid-adjusted total PCBs ≥ 0.67 ng/ml and CYP1A1 M2-containing genotypes was 4.7 (95% CI 0.5–43.1). It would be helpful to combine individual-level results across these three studies to obtain more precise estimates of the joint effects of CYP1A1 genotypes and high levels of PCB exposure. A strength of our study is the fact that we included both African American and white women, and we examined the effects of the four known CYP1A1 alleles: M1, M2, M3, and M4. Previous studies [ 24 , 25 ] included only white women and did not estimate ORs for CYP1A1 M3- or M4-containing genotypes. In previous analyses of this data set [ 26 ], African American women showed higher plasma levels of PCBs and a stronger relationship between total PCBs and breast cancer than that in white women. Our results suggest that further study of CYP1A1 M3-containing genotypes in African American women is warranted, particularly in combination with PCB exposure. Another strength of our study is that fact that we estimated joint effects for specific PCB congener groups and CYP1A1 genotypes. PCB congener groups may differ in biologic activity in ways that are relevant to breast cancer etiology [ 35 - 39 ]. However, because of the strong intercorrelation of specific congeners with total PCBs, we were unable to distinguish strong congener-specific effects. A weakness of our study was the fact that, because of the small sample size, we were unable to estimate ORs with precision in all of the subgroups of interest. Moysich and colleagues [ 24 ] reported that breast cancer risk was significantly increased among women with elevated PCB body burden and CYP1A1 M2 genotypes who had ever smoked cigarettes. Laden and colleagues [ 25 ] observed similar results. We did not observe differences in ORs for the joint effects of total PCBs and CYP1A1 genotypes according to smoking status. However, we lacked power to address the effects of smoking dose or duration. ORs were imprecise when we subdivided study participants on the basis of menopausal status, and any differences could be due to chance. Owing to the case-control study design, patients were enrolled after diagnosis of breast cancer. As described previously [ 26 ], we did not observe evidence for disease-related changes in plasma organochlorine levels when we adjusted for weight loss or gain and stage at diagnosis. Limited information on diet was collected in this study. We did not observe correlations between total PCB levels and the consumption of fruits, vegetables, or fish (data not shown), but the confounding effects of other dietary exposures cannot be ruled out. Conclusions Data from the CBCS provides evidence that subgroups of women with CYP1A1 M2 and M3 polymorphisms and high levels of PCB exposure might have a modestly elevated risk of breast cancer. Our results confirm findings from previous studies with respect to the highly inducible CYP1A1 M2 genotype, and suggest that M3-containing genotypes might also modify risk of breast cancer associated with PCB exposure in African American women. Additional studies with a large sample size are warranted to confirm or refute these findings. In addition, meta-analyses combining individual-level data from epidemiologic studies of CYP1A1 and breast cancer will be needed to generate more precise estimates of the joint effects of PCBs and CYP1A1 genotypes. Abbreviations AHH = aryl hydrocarbon hydrolase; CBCS = Carolina Breast Cancer Study; CI = confidence interval; CYP1A1 = cytochrome P450 1A1; ICR = interaction contrast ratio; LRT = likelihood ratio test; OR = odds ratio; PCBs = polychlorinated biphenyls. Competing interests The author(s) declare that they have no competing interests. Authors' contributions YL and LC conducted the laboratory analyses; YL, RM, and C-KT conducted the statistical analyses; YL, RM, DB, LC, C-KT, BN and KC participated in writing the manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064095.xml

1064097

Imaging in situ breast carcinoma (with or without an invasive component) with technetium-99m pentavalent dimercaptosuccinic acid and technetium-99m 2-methoxy isobutyl isonitrile scintimammography

Introduction The aim of the study was to retrospectively define specific features of the technetium-99m pentavalent dimercaptosuccinic acid ( 99m Tc-(V)DMSA) and technetium-99m 2-methoxy isobutyl isonitrile ( 99m Tc-Sestamibi [ 99m Tc-MIBI]) distribution in ductal breast carcinoma in situ and lobular breast carcinoma in situ (DCIS/LCIS), in relation to mammographic, histological and immunohistochemical parameters. Materials and methods One hundred and two patients with suspicious palpation or mammographic findings were submitted preoperatively to scintimammography (a total of 72 patients with 99m Tc-(V)DMSA and a total of 75 patients with 99m Tc-Sestamibi, 45 patients receiving both radiotracers). Images were acquired at 10 min and 60 min, and were evaluated for a pattern of diffuse radiotracer accumulation. The tumor-to-background ratios were correlated (T-pair test) with mammographic, histological and immunohistochemical characteristics. Results Histology confirmed malignancy in 46/102 patients: 20/46 patients had DCIS/LCIS, with or without coexistent invasive lesions, and 26/46 patients had isolated invasive carcinomas. Diffuse 99m Tc-(V)DMSA accumulation was noticed in 18/19 cases and 99m Tc-Sestamibi in 6/13 DCIS/LCIS cases. Epithelial hyperplasia demonstrated a similar accumulation pattern. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value for each tracer were calculated. Solely for 99m Tc-(V)DMSA, the tumor-to-background ratio was significantly higher at 60 min than at 10 min and the diffuse uptake was significantly associated with suspicious microcalcifications, with the cell proliferation index ≥ 40% and with c- erbB-2 ≥ 10%. Conclusion 99m Tc-(V)DMSA showed high sensitivity and 99m Tc-Sestamibi showed high specificity in detecting in situ breast carcinoma ( 99m Tc-(V)DMSA especially in cases with increased cell proliferation), and these radiotracers could provide clinicians with preoperative information not always obtainable by mammography.

Introduction Ductal carcinoma in situ (DCIS) of the breast, alone or in the presence of an invasive tumor (invasive ductal carcinoma [IDC]), presents a problem in its detection and treatment. DCIS represents a number of biologically different processes that exhibit variable frequencies of occult invasion and variable risks for local recurrence, after attempts at excision biopsy or lumpectomy. The term 'extensive intraductal component' or 'extensive intraductal carcinoma' (EIC) is defined as DCIS within and around an invasive tumor, comprising at least 25% of the neoplasm. Of all IDC cases, 15–30% are considered to have EIC [ 1 ]. Tumors that are predominantly DCIS and appear with focal invasion are also classified as EIC. DCIS not within the confines of the invasive tumor, but frequently in the adjacent breast tissue, makes breast-conserving surgical treatment less safe with regard to the recurrence rate. The great majority of recurrences are observed at or near the site of the primary tumor. On pathological and radiological evaluation of mastectomy specimens, some studies [ 1 ] have shown that EIC tumors commonly had DCIS not only in their immediate vicinity, but also as far as 4–6 cm from the invasive tumor. The risk of local recurrence after a breast-conserving procedure can be evaluated according to the histological type, the extent of the lesion and the state of the surgical margins. It has been reported that the local recurrence rate after a limited resection of the tumor, followed by radiotherapy, is significantly greater for EIC-positive patients than for EIC-negative patients [ 2 , 3 ]. If the real extent of EIC was known preoperatively, it might be possible to safely limit the breast resection, since an initial wide excision would constitute overtreatment for the majority of patients not having a substantial amount of DCIS outside the invasive tumor. The fact that younger patients are more likely than older ones to have EIC, and that the association between EIC and breast cancer recurrence is stronger in premenopausal women than in postmenopausal women, renders the detection of such lesions imperative [ 1 ]. Mammographic evidence of microcalcifications with a certain form and distribution is presently the only method of detecting the presence of DCIS. The failure of mammography to recognize DCIS or EIC in radiologically dense breasts in young populations raises a significant problem. Scintimammography with technetium-99m 2-methoxy isobutyl isonitrile ( 99m Tc-Sestamibi [ 99m Tc-MIBI]) has been used to detect primary breast cancer [ 4 - 8 ]. It concentrates in cancer cells by an energy-requiring transport mechanism and a transmembrane electronegative potential, in addition to nonspecific mechanisms, and it is stored within the mitochondria. Only a few reports [ 9 , 10 ] have discussed the efficacy of 99m Tc-Sestamibi in detecting DCIS. Technetium-99m pentavalent dimercaptosuccinic acid ( 99m Tc-(V)DMSA) is a tumor-seeking agent already known for its efficacy in detecting medullary thyroid carcinoma. Its mechanism of accumulation has been thought to be related to the structural similarity between the (V)DMSA core and the phosphate anion (PO 4 3- ), which is avidly taken up by some cancer cells [ 11 , 12 ]. Denoyer and colleagues recently reported that the radiotracer uptake is specifically mediated by NaPi cotransporter type III in cancer cells, similar to phosphate ions, which enter cells via NaPi cotransporters [ 13 ], while other reports suggest that its uptake in tumors is related to glucose metabolism-mediated acidosis [ 14 ]. 99m Tc-(V)DMSA has been tried out in identifying invasive and preinvasive breast cancer [ 15 - 17 ]. The present study retrospectively assessed the capability of 99m Tc-(V)DMSA and 99m Tc-Sestamibi to image DCIS and EIC. The study also investigated whether this was related to the presence of suspicious microcalcifications on mammography, stromal reaction, lymphocytic infiltration and immunohistochemical parameters such as the cell proliferation index (Ki-67), c- erbB-2 , and p53 expression. Materials and methods Patients and scintimammography A total of 102 women (mean age ± standard deviation [SD], 62.5 ± 12.49 years) referred with suspicious breast lesions on physical examination and/or an abnormal mammogram underwent 99m Tc-Sestamibi and/or 99m Tc-(V)DMSA scintimammography prior to any surgical intervention. All patients were intended to have studies performed with both agents. The first agent to be used was selected on a random basis. Not all patients managed to have the second study performed preoperatively, for various reasons. Forty-five patients underwent both 99m Tc-(V)DMSA and 99m Tc-Sestamibi breast scintigraphy at separate sessions, in a head-to-head, double-phase study with a 48-hour time interval. Twenty-seven patients underwent only 99m Tc-(V)DMSA and 30 patients underwent only 99m Tc-Sestamibi scintimammography. A total of 72/102 patients therefore underwent 99m Tc-(V)DMSA and a total of 75/102 patients underwent 99m Tc-Sestamibi scintigraphy. Inclusion and exclusion criteria for entry into the study are summarized in Table 1 . The study was in accordance with the ethical principles of the Declaration of Helsinki. Early and late planar images (at approximately 10–20 min and 60–70 min, respectively) were acquired, in the lateral prone and anterior supine positions, after intravenous administration of 925–1100 MBq radiotracer. Acquisitions were obtained using a special positioning pad (PBI-2 Scintimammography Pad Set ® ; Pinestar Technology Inc., Greenville, PA, USA). Scintimammography was performed using a single-head γ camera (Sophycamera DS7 ® ; Sopha Medical Vision International, Buc Cedex, France), equipped with a high-resolution parallel hole collimator connected to a dedicated computer (Sophy NxT ® ; Sopha Medical Vision International). The matrix was 256 × 256 pixels and the photopeak was centered at 140 keV, with a symmetrical 10% window. The acquisition time for images with both radiotracers varied between 7 min and 10 min per view (the time required to acquire 2,000,000–2,500,000 counts per image). Tomographic imaging (single-photon emission computerized tomography) was not performed since it has not been found to be superior to planar imaging, regarding the sensitivity and the specificity, for invasive breast tumors [ 18 ] and since no consensus has been reached regarding its utility. It was considered that no additional information would be obtained (except for axillary lymph node involvement detection), and tomographic imaging is time consuming and relatively inconvenient for the patient. 99m Tc-(V)DMSA was prepared using a domestically available kit (DMS(V)/Demoscan ® ; National Center of Physical Sciences, Institute of Radioisotopes and Radiodiagnostics 'DEMOKRITOS', Athens, Greece). The kit for the preparation of 99m Tc-Sestamibi (Cardiolite ® ) was obtained from Bristol-Myers Squibb GmbH (Regensburg, Germany). Both pharmaceuticals were labeled with technetium-99m within the Nuclear Medicine Department. Diagnosis was made by histopathology of the specimens obtained surgically. The scintigraphic results and the mammograms were compared with histologic and immunohistochemical findings. Image analysis Scintigrams were retrospectively evaluated, regarding the site, shape, pattern, extent, homogeneity and dispersion of radiotracer distribution. The background count rate of normal breast tissue depends on various factors (breast size, injected radioactivity dose, etc.) and is readily distinguishable from tracer accumulation in pathological tissue. Any increased focal radiotracer accumulation was assessed as suggestive of invasive breast cancer, whereas any other pattern of more widespread diffuse uptake was assessed as representing noninvasive lesions [ 10 , 15 , 16 ]. We set and evaluated as the criterion for the scintigraphic detection of noninvasive breast carcinoma the presence of any pattern of increased radiotracer accumulation other than focal activity (i.e. diffuse heterogeneous and diffuse homogeneous), independently of the coexistence of any focal increased activity. Accordingly, two experienced nuclear medicine physicians blinded to any clinical and pathologic data characterized the scintimammograms as positive or negative. Disagreement between them was resolved by consensus or by obtaining a third opinion. Radiotracer accumulation in breast tumors was evaluated in early and delayed images (at 10–20 min and 60–70 min post injection, respectively). This was performed visually and semiquantitatively. The tumor-to-background (T/B) ratio was calculated by drawing regions of interest of standardized size and shape over the site of the greatest radioactivity within the areas of diffusely increased tracer uptake – assumed to represent noninvasive in situ lesions – and over the surrounding normal breast tissue. The same ratio was calculated for the areas of focal increased tracer uptake (presumably representing invasive tumor). In the late (60 min) 99m Tc-(V)DMSA images of some cases with extensive DCIS and associated IDC, the margins of the focal uptake area – corresponding to invasive disease – were not very clearly demarcated from the diffuse uptake. Nevertheless, the invasive component was usually clearly visible on the early (10 min) image. Thus, by comparing early and late images, that area was avoided when creating the region of interest for diffuse uptake. Mammography All women underwent conventional X-ray mammography using cranio-caudal and medio-lateral projections. Images were interpreted by two experienced radiologists and were characterized as positive or negative with regard to the presence or the absence of suspicious branching and clustered coarse granular-type microcalcifications, suggestive of in situ carcinoma [ 19 ]. Disagreement was resolved by consensus or by obtaining a third opinion. Histopathology and immunostaining The assessment of the extent of the in situ carcinoma was based on the number of histologic slices containing the lesion. Whole specimens were examined in serial sections and the final major diameter of the lesion was estimated by multiplying the number of slices containing the lesion by 0.3 cm, which is the medium size of a section. The EIC was defined as DCIS within and around an invasive tumor, comprising at least 25% of the neoplasm. In order to investigate whether any of these factors is associated with the diffuse pattern of tracer distribution, an immunohistochemical method (avidin biotin peroxidase-horseradish peroxidase complex) was performed on paraffin-embedded breast tissue sections for in situ carcinomas, for the demonstration of Ki-67, c- erbB-2 and p53 expression. A semiquantitative estimation of these levels was performed, based on the staining intensity and the percentage of positive cells. Regarding Ki-67, a threshold of 40–45% of cells with positive staining is generally used for separating 'moderate' staining from 'intense' staining. Overexpression for c- erbB-2 is defined, in cases of invasive carcinoma, by dense membrane staining in ≥ 10% of the epithelial cells. Similarly, p53 overexpression is defined by staining in ≥ 10% of the epithelial cell nuclei. Although there is no consensus for defining c- erbB-2 and p53 overexpression in noninvasive breast carcinoma, the latter generally has the same pattern as invasive breast carcinoma and thus the same thresholds could be used [ 20 ]. Data analysis The T-pair test was applied to the T/B ratios of diffuse lesions between early and late acquisitions. The T/B ratios in the late (60 min) images for both tracers were also correlated with the presence of or the absence of suspicious mammographic microcalcifications, with the stromal reaction, with lymphocytic infiltration, with Ki-67 values < 40% and ≥ 40%, with c- erbB-2 values < 10% and ≥ 10%, and with p53 values < 10% and ≥ 10%, in order to assess any significant difference in diffuse tracer uptake between these populations. Finally, linear regression univariate analysis was performed to reveal any possible correlation between the T/B ratio at 60 min and the tumor size. For all tests, P < 0.05 was considered statistically significant. In vitro assay of 99m Tc-(V)DMSA uptake and micro-autoradiographic study An in vivo autoradiographic study performed on tumor specimens excised immediately after scintimammography could provide precise data about 99m Tc-(V)DMSA localization on DCIS areas. Such a study was attempted, but the radioactivity of the tumor specimen was not sufficient. This was not surprising, since, according to the literature [ 21 ], in vivo autoradiographic studies in mice require approximately 37 MBq (1 mCi or 1221 MBq/kg body weight), a dose that is approximately 66 times higher than the injected dose per kilogram of body weight in humans. Given the impossibility of implementing the study as described, the tumor specimen was thus used for an in vitro autoradiographic study. The frozen tumor tissue of one patient with DCIS was sliced into serial sections in the cryostat microtome chamber (Microm HM 505 N ® ; Microm Laborgerate GmbH, Walldorf, Germany), was mounted onto gelatin-coated slides, was dried at 37°C for 1 hour and was then incubated in a solution containing 99m Tc-(V)DMSA. The sections were then exposed to Kodac X-OMATT XAR ® film (Eastman Kodak Co., Rochester, NY, USA) in an autoradiographic cassette for 24 hours. Results Breast cancer was histologically confirmed in 46/102 women. Of these women, 26/46 had invasive carcinomas, mainly ductal (IDC) and lobular (invasive lobular carcinoma [ILC]). This was found to be the prominent histological lesion in this group of IDC patients, and these patients were therefore classified as EIC-negative. The other 20/46 cases were diagnosed as having carcinoma in situ . Of these, 18/20 presented DCIS: in four cases (patients 9, 12, 17, and 18; see Table 2 ) DCIS was the sole histological finding, while the majority (14 patients) had invasive carcinoma (IDC) associated with extensive DCIS clearly outside the confines of the invasive tumor, and thus these patients were characterized as EIC-positive. Both the remaining 2/20 cases had extensive lobular carcinoma in situ (LCIS), associated with an invasive component (ILC). EIC was found more frequently in younger women (mean age ± SD, 57.1 ± 13.7 years for EIC-positive patients versus 65.8 ± 8.0 years for EIC-negative patients; P < 0.05). The size of the invasive tumors (IDC/ILC) associated with extensive DCIS/LCIS ranged from 0.7 to 6.0 cm (mean ± SD, 2.75 ± 1.50 cm), and the size of in situ carcinomas ranged from 0.8 to 9.0 cm (mean ± SD, 4.92 ± 1.93 cm). The basic data of these 20 DCIS/LCIS cases are presented on an individual basis in Table 2 . Benign breast lesions were found in the remaining 56/102 women. Among them, epithelial hyperplasia (usual type hyperplasia and/or atypical type hyperplasia) was present in 14/56 patients, with or without fibrosis, adenosis and ductal dilatation. All scintimammograms were characterized as positive or negative for in situ carcinoma, with reference to the diagnostic criteria set for the detection of DCIS/LCIS (i.e. presence of the pattern of diffusely increased tracer uptake, regardless of the presence of any focal increased activity). Any study displaying only focal uptake was therefore a (true or false) negative for the in situ criterion (Fig. 1 ). Comparative statistics (sensitivity, specificity, accuracy, positive predictive value and negative predictive value) for each radiotracer, in the total patient groups studied with each one and in the subgroup studied with both (head-to-head study), are presented in Table 3 . The interobserver agreement rate between the two physicians assessing the breast scans was 92%. Of the 20/46 breast cancer patients with DCIS/LCIS, 19/20 underwent 99m Tc-(V)DMSA and 13/20 underwent 99m Tc-Sestamibi scintimammography (12/20 were part of the head-to-head subgroup of 45 patients that underwent both). Locally diffuse, heterogeneous (patchy), poorly circumscribed increased 99m Tc-(V)DMSA accumulation, sometimes covering and surrounding focal increased accumulation (if present), was noticed in 16/17 of DCIS cases (Figs 2 , 3 , 4 ) and in 2/2 LCIS cases (Fig. 5 ), a total of 18/19 (95%) patients. The one lesion that was not detected had a size of 0.8 cm (patient 9). A similar pattern of distribution was found with 99m Tc-Sestamibi, but only in 6/13 (46%) patients. Among those 12/20 DCIS/LCIS patients that underwent both examinations, this pattern of tracer distribution was noticed in 11/12 patients (92%) with 99m Tc-(V)DMSA and in 5/12 patients (42%) with 99m Tc-Sestamibi (Tables 3 and 4 ). In women with histologically confirmed usual type hyperplasia and/or atypical type hyperplasia, a similar pattern of intense diffuse, not patchy, but more homogeneous, 99m Tc-(V)DMSA distribution was observed in 10/14 patients (71%). Diffuse accumulation of 99m Tc-Sestamibi in sites of epithelial hyperplasia was observed in 6/14 patients (43%), yet it was less intense in comparison with 99m Tc-(V)DMSA (Fig. 6 ). In the absence of epithelial hyperplasia, the benign histological lesions of fibrosis, adenosis and ductal dilatation alone did not accumulate either of the two radiotracers. These results are also presented in Table 4 . In the series studied, the T/B ratio for 99m Tc-(V)DMSA uptake in the sites of diffuse accumulation pattern demonstrated a tendency to increase over time (Figs 2 , 3 , 4 , 5 , 6 ). Semiquantitative analysis (T/B ratio) performed separately in DCIS and in IDC revealed a statistically significant increase in late images only of the diffuse 99m Tc-(V)DMSA accumulation in DCIS (mean ± SD, 1.27 ± 0.22 at 10 min versus 1.76 ± 0.25 at 60 min; P < 0.01). On the contrary, T/B ratios for the focal 99m Tc-(V)DMSA uptake in invasive tumors were not significantly increased in the late acquisitions (mean ± SD, 1.76 ± 0.28 at 10 min versus 1.94 ± 0.28 at 60 min; P > 0.05). The pattern of diffuse 99m Tc-(V)DMSA distribution observed in epithelial hyperplasia demonstrated a similar tendency to increase over time. The T/B ratios for 99m Tc-Sestamibi did not demonstrate significant variability with time, either in DCIS or in IDC (mean ± SD for DCIS, 1.34 ± 0.29 at 10 min versus 1.33 ± 0.33 at 60 min; P > 0.05; and mean ± SD for IDC, 2.06 ± 0.97 at 10 min versus 1.98 ± 0.84 at 60 min; P > 0.05). Similarly, the diffuse 99m Tc-Sestamibi distribution observed in epithelial hyperplasia did not appear to increase significantly over time. Patients with LCIS and associated ILC gave parallel results, but due to the limited number of cases ( n = 2) they could not be statistically evaluated. Linear regression univariate analysis revealed no statistically significant correlation between the 99m Tc-(V)DMSA T/B ratio at 60 min and tumor size ( r = 0.28, P > 0.1). Mammography depicted suspicious microcalcifications in 10/20 (50%) patients with DCIS/LCIS (Table 2 ). The diffuse uptake of 99m Tc-(V)DMSA in DCIS/LCIS at 60 min was significantly higher in patients with suspicious microcalcifications on mammography as compared with patients without (mean ± SD T/B ratio, 1.81 ± 0.05 versus 1.4 ± 0.07, respectively; P = 0.003). 99m Tc-Sestamibi diffuse uptake at 60 min was not significantly different between the patient groups with and without microcalcifications (mean ± SD T/B ratio, 1.49 ± 0.19 versus 1.31 ± 0.09, respectively; P = 0.49). An intense stromal reaction was found in 1/20 patient, a moderate stromal reaction in 4/20 patients, and a mild stromal reaction in 3/20 patients with DCIS/LCIS; a total of 8/20 patients (40%). Microcalcifications combined with stromal reaction appeared in 6/20 cases. Mild or moderate lymphocytic infiltration was noted in 4/20 patients (20%). Pure comedo was the most prominent type (10/18) among DCIS (Table 2 ). No statistically significant difference was found in 99m Tc-(V)DMSA or 99m Tc-Sestamibi uptake between patients with stromal reaction and/or lymphocytic infiltration and those patients without. With reference to the studied immunohistochemical parameters in DCIS/LCIS patients, the Ki-67 value ranged from 10% to 55% (mean ± SD, 33.4 ± 13.4%), p53 ranged from 0% to 27% (mean ± SD, 6.4 ± 8.8%) and c- erbB-2 ranged from 0% to 80% (mean ± SD, 21.7 ± 26.4%) (Table 2 ). The uptake of 99m Tc-(V)DMSA was significantly higher in patients with Ki-67 overexpression (≥ 40%) as compared with patients with Ki-67 levels < 40% (mean ± SD T/B ratio at 60 min, 1.86 ± 0.027 versus 1.52 ± 0.047, respectively; P = 0.025]. No significant difference was found in 99m Tc-Sestamibi scans in relation to Ki-67 levels ≥ 40% and < 40% (mean ± SD T/B ratio at 60 min, 1.51 ± 0.27 versus 1.24 ± 0.08, respectively; P = 0.46). Overexpression of c- erbB-2 (≥ 10%) was associated with higher 99m Tc-(V)DMSA uptake in DCIS/LCIS (mean ± SD T/B ratio at 60 min, 1.89 ± 0.01 versus 1.52 ± 0.09 for values ≥ 10% and < 10%, respectively; P = 0.004). 99m Tc-Sestamibi uptake was not significantly different between women with c- erbB-2 values ≥ 10% and < 10% (mean ± SD T/B ratio at 60 min, 1.43 ± 0.06 versus 1.3 ± 0.16, respectively; P = 0.64). Diffuse uptake of both 99m Tc-(V)DMSA and 99m Tc-Sestamibi was not significantly different with regard to p53 values < 10% and ≥ 10% (mean ± SD T/B ratio at 60 min, 1.71 ± 0.068 versus 1.72 ± 0.06, respectively, for 99m Tc-(V)DMSA; P = 0.93; and 1.60 ± 0.11 versus 1.29 ± 0.17, respectively, for 99m Tc-Sestamibi; P = 0.78). The c- erbB-2 levels were significantly higher in DCIS cases with coexistent IDC (EIC-positive) than in EIC-negative patients ( P < 0.05), whereas p53 expression did not demonstrate such a statistically significant difference. In vitro autoradiography of the breast tumor of one patient with DCIS revealed a very good correlation between the regional distribution of 99m Tc-(V)DMSA radioactivity and the distribution of DCIS cell clumps, as observed after hematoxylin & eosin staining of the same sections (Fig. 7 ). Discussion In situ breast carcinoma tumors are very difficult to identify, and the only available method to date for their detection is the evaluation of the shape and distribution of microcalcifications on mammography, with all the limitations inherent in this method (dense breasts, scars, implants, etc.). Scintigraphic imaging of EIC, DCIS or epithelial hyperplasia with 99m Tc-(V)DMSA has already been reported [ 15 , 16 ]. Few reports have been published concerning the ability of 99m Tc-Sestamibi not actually to detect DCIS, but rather to improve invasive breast cancer detection in the presence of DCIS [ 9 , 10 ], and these studies seem to be contradictory. Howarth and colleagues [ 10 ] reported no change in sensitivity in the presence or absence of carcinoma in situ , and furthermore that there is no relationship between 99m Tc-Sestamibi uptake and DCIS, except in the presence of an invasive tumor. Several other authors, however, have suggested that diffuse 99m Tc-Sestamibi uptake in benign breast disorders was associated with proliferative changes [ 15 , 22 , 23 ], demonstrating an increased risk of developing into breast cancer, while other workers described hormonal influence to be the cause of diffuse breast uptake on scintimammography [ 24 ]. In the present study, hormonal influence or the menstrual cycle could not be responsible for this kind of tracer distribution, since all but two women were postmenopausal and they were not on any hormone replacement therapy. In the group of patients with breast malignancy, the detected invasive tumors (IDC/ILC) presented mostly with focal increased uptake of both radiotracers. On the contrary, the pattern of diffuse heterogeneous (patchy) uptake was mainly observed in cases with DCIS, independently of the presence of an invasive tumor. Among the 26 invasive tumors not associated with extensive amounts of in situ carcinoma, only a couple of cases gave false positive results (for DCIS) with 99m Tc-(V)DMSA (Table 4 ): one consisted of multiple microscopic foci of IDC, and the other case presented IDC with intense lymphocytic infiltration. LCIS was found in two cases, and due to this limited number specific scintigraphic results could not be statistically evaluated. They could have been considered as false positive findings for DCIS, yet their scintigraphic appearance was the same; they both revealed a similar diffuse inhomogeneous pattern of 99m Tc-(V)DMSA uptake, apart from the focal uptake corresponding to the coexistent ILC. They could therefore not be separated from DCIS, except after histological examination of the biopsied tissue, and thus they are studied together in this series. If this diffuse uptake was attributable to the lobular invasive component, known for its peculiar pattern of growth and local invasion, then it should have been clearly visible from the early (10 min) images, simultaneously with the focal uptake. Yet the extensive diffuse 99m Tc-(V)DMSA uptake was prominent in the late (60 min) image, whereas the focal tracer uptake was also visible in the early image (Fig. 4 ). 99m Tc-Sestamibi, known for its efficacy in imaging invasive carcinoma, in this case managed to image only the focal uptake in both early and late images. The diffuse pattern of radiotracer uptake (not patchy, but more homogeneous) was also noticed in some cases of epithelial hyperplasia, more prominently with 99m Tc-(V)DMSA. Epithelial hyperplasia was therefore the major source of false positive results. The discrimination between atypical epithelial hyperplasia and in situ carcinoma is certainly not easy, even on frozen section. Yet any false positive scan attributable to hyperplasia should not be considered a major disadvantage, since it may lead to a search but not to an incorrect final decision concerning excision. This is because scintigraphic data (in addition to suspicious microcalcifications, if present) can serve as a guide and provide the surgeon with useful information preoperatively, in order to decide the extent of the search and to facilitate a guided biopsy, rather than directly determine whether an excision has to be performed. Increased cell proliferation activity (Ki-67 ≥ 40%) was found to be significantly correlated with the diffuse 99m Tc-(V)DMSA uptake in breast carcinoma in situ . Papantoniou and colleagues [ 25 ] recently reported this factor to be independently correlated with focal 99m Tc-(V)DMSA accumulation in invasive breast cancer, while their preliminary reports [ 26 ] revealed that cases of usual type hyperplasia imaged with 99m Tc-(V)DMSA tend to be related to elevated Ki-67 values and are therefore at risk of developing breast malignancy, according to the literature [ 27 - 31 ]. Cutrone and colleagues claimed that increased proliferative activity is found not only in invasive cancers, but also in precancerous lesions [ 27 ]. Shaaban and colleagues [ 28 ], along with several other reports [ 29 , 30 ], suggested that benign breast lesions such as epithelial hyperplasia, when associated with increased levels of Ki-67 and estrogen receptors type A, define a subset of hyperplastic lesions with high risk of subsequent breast cancer development. Such patients could be selected for prophylactic antiestrogen therapy to diminish the proliferative activity [ 28 , 31 ]. The observed tendency of the 99m Tc-(V)DMSA diffuse uptake pattern to increase over time (being more notable on the delayed 60-min images), as well as its relation to elevated proliferative cellular activity (Ki-67), could be considered indicative of the existence of a distinct primary pathway for the mechanism of its accumulation in tumor tissue. This is considered to reflect the tracer's uptake in structures that participate in mitotic activity of cancerous and precancerous cell populations [ 25 ]. The possibility for this finding to be due to locally increased vascularity, angiogenesis or vessel permeability is therefore highly unlikely; if that was the case, then this diffuse pattern would have been equally clearly visible in the early 10-min images. On the contrary, diffuse 99m Tc-Sestamibi distribution was not found to be significantly correlated with Ki-67 expression. Palmedo and colleagues [ 32 ] reported that the 99m Tc-Sestamibi concentration was one-half that of 99m Tc-(V)DMSA in an experimental model of rats bearing poorly differentiated breast tumours. Cutrone and colleagues found a moderate correlation between 99m Tc-Sestamibi uptake and the degree of cellular proliferation [ 27 ], while Cwikla and colleagues did not [ 33 ]. Angiogenesis and an oxidative metabolism seem to be favorable factors for 99m Tc-Sestamibi tumor uptake, rather than proliferative activity [ 25 , 29 ]. This could provide an explanation for the lower sensitivity of this radiotracer, as compared with 99m Tc-(V)DMSA, in imaging such diffuse lesions. Overexpression of c- erbB-2 (≥ 10%) is found in almost 60% of cases of high-grade comedo-type DCIS, in 10–40% of IDC and in only a few cases of ILC [ 34 , 35 ]. An increased c- erbB-2 level was a factor significantly correlated with diffuse 99m Tc-(V)DMSA uptake (but not with 99m Tc-Sestamibi uptake). This appears reasonable, since oncoprotein overexpression is usually found in aggressive in situ carcinomas with increased levels of cell proliferative activity. The measurement of Ki-67 and c- erbB-2 on histological specimens is not related to the size of the tumor. In the present series, the T/B ratio for 99m Tc-(V)DMSA in the late image was not found to be significantly correlated with tumor size. It is therefore considered that the correlation found for the 99m Tc-(V)DMSA T/B ratio to Ki-67 and/or c- erbB-2 expression is independent of tumor size. p53 overexpression is known to reflect tumor aggressiveness and a decreased disease-free interval following therapy. An intriguing finding requiring further investigation, however, is that p53 overexpression, unlike that of Ki-67 and c- erbB-2 , was not significantly correlated with this pattern of uptake of either tracer in the present series. DCIS of the comedo type is usually accompanied by a desmoplastic stromal reaction with pronounced neovascularization. Elastosis is also more common in DCIS [ 36 ]. A variable inflammatory infiltrate is present in the periductal stroma. This consists of lymphocytes and histiocytes, in amounts ranging from sparse to abundant [ 37 ]. The suggestion that stromal reaction and/or lymphocytic infiltration could be considered mediators for the diffuse radiotracer uptake was not confirmed in the studied population. Suspicious mammographic microcalcifications were observed in only one-half of the DCIS/LCIS cases, whereas 99m Tc-(V)DMSA revealed a diffuse uptake pattern in the vast majority (18/19 patients). Nevertheless, diffuse uptake for 99m Tc-(V)DMSA (but not for 99m Tc-Sestamibi) was found to be significantly increased in women with suspicious microcalcifications as compared with those without. A possible explanation for this could be that clustered microcalcifications are usually found in more aggressive types of DCIS (such as comedo type), which tend to be related to increased Ki-67 levels. The majority of DCIS found in this series were of comedo type. Although it has been reported for the comedo type that suspicious microcalcifications approximately correspond to the histologically confirmed size, their extent in the present study was substantially smaller in comparison with the spread of diffuse 99m Tc-(V)DMSA uptake, which in turn was found to correlate very well with the histologically confirmed size of DCIS. These findings imply that calcium deposits may not represent a 99m Tc-(V)DMSA uptake mechanism, as was previously presumed [ 16 ]. Examining scintimammograms in combination with mammography could help limit false positive results. In the presence of suspicious microcalcifications, a diffuse 99m Tc-(V)DMSA uptake could be considered more suggestive of in situ carcinoma. In the absence of microcalcifications, however, a diffuse 99m Tc-(V)DMSA uptake could imply the possibility of a false positive study, probably due to epithelial hyperplasia. In vitro 99m Tc-(V)DMSA autoradiography of one tumor specimen demonstrated a very good correlation between radioactivity uptake and the distribution of foci of DCIS cells. This is considered a strong indication for the tracer localization in these cell clumps. Autoradiography nevertheless needs to be performed in a larger series of tumor specimens to verify the findings. Conclusion 99m Tc-(V)DMSA displayed an excellent sensitivity and negative predictive value in detecting DCIS/LCIS, especially in cases associated with increased cell proliferation (Ki-67) and c- erbB-2 overexpression, independent of the presence of suspicious microcalcifications. Its relatively lower specificity and positive predictive value, caused by false positive scans mainly attributable to epithelial hyperplasia, should not be considered a major disadvantage, since to date there is no diagnostic technique other than suspicious mammographic microcalcifications to define high-risk breast areas that should be biopsied. 99m Tc-Sestamibi appeared to be less sensitive, yet it demonstrated a very good specificity and could also be useful in preinvasive lesion imaging. In our opinion, any diffuse tracer accumulation, either as an isolated finding or extending beyond the margins of a focal, well-circumscribed accumulation, could be considered to probably correspond with highly proliferating tissue and to possibly represent DCIS/LCIS or epithelial hyperplasia. Since the difference between in situ carcinoma and atypical type hyperplasia cannot always be discriminated, even histologically, we note the potential usefulness of scintimammography in imaging these lesions and making them accessible to biopsy. There is, however, a need for a larger series of patients to verify these preliminary observations, as well as prospective studies to estimate its reliability in affecting treatment decisions. Abbreviations DCIS = ductal carcinoma in situ ; EIC = extensive intraductal carcinoma; IDC = invasive ductal carcinoma; ILC = invasive lobular carcinoma; Ki-67 = cell proliferation index; LCIS = lobular carcinoma in situ ; SD = standard deviation; T/B, tumor-to-background; 99m Tc-(V)DMSA = technetium-99m pentavalent dimercaptosuccinic acid; 99m Tc-Sestamibi ( 99m Tc-MIBI) = technetium-99m 2-methoxy isobutyl isonitrile. Competing interests The author(s) declare that they have no competing interests. Authors' contributions VP is the guarantor of integrity of the entire study and was responsible for study concepts, study design, literature research, data acquisition, data analysis/interpretation, statistical analysis, manuscript preparation, manuscript definition of intellectual content, manuscript editing, and manuscript revision/review. ST participated in data acquisition, data analysis/interpretation, statistical analysis, manuscript preparation, manuscript editing, and manuscript revision/review. EM, VV, MSou, LN, and AT participated in data acquisition, data analysis/interpretation, and manuscript preparation. MSot and IP participated in experimental studies (autoradiography in vitro ), data acquisition, data analysis/interpretation, and manuscript preparation. DL, AL, and MM participated in clinical studies, data acquisition, data analysis/interpretation, and manuscript preparation. JK participated in data acquisition, data analysis/interpretation, statistical analysis, and manuscript preparation. CZ participated in data analysis/interpretation, manuscript preparation, manuscript definition of intellectual content, manuscript editing, and manuscript revision/review. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064097.xml

1064098

Breast fibroblasts modulate epithelial cell proliferation in three-dimensional in vitro co-culture

Background Stromal fibroblasts associated with in situ and invasive breast carcinoma differ phenotypically from fibroblasts associated with normal breast epithelium, and these alterations in carcinoma-associated fibroblasts (CAF) may promote breast carcinogenesis and cancer progression. A better understanding of the changes that occur in fibroblasts during carcinogenesis and their influence on epithelial cell growth and behavior could lead to novel strategies for the prevention and treatment of breast cancer. To this end, the effect of CAF and normal breast-associated fibroblasts (NAF) on the growth of epithelial cells representative of pre-neoplastic breast disease was assessed. Methods NAF and CAF were grown with the nontumorigenic MCF10A epithelial cells and their more transformed, tumorigenic derivative, MCF10AT cells, in direct three-dimensional co-cultures on basement membrane material. The proliferation and apoptosis of MCF10A cells and MCF10AT cells were assessed by 5-bromo-2'-deoxyuridine labeling and TUNEL assay, respectively. Additionally, NAF and CAF were compared for expression of insulin-like growth factor II as a potential mediator of their effects on epithelial cell growth, by ELISA and by quantitative, real-time PCR. Results In relatively low numbers, both NAF and CAF suppressed proliferation of MCF10A cells. However, only NAF and not CAF significantly inhibited proliferation of the more transformed MCF10AT cells. The degree of growth inhibition varied among NAF or CAF from different individuals. In greater numbers, NAF and CAF have less inhibitory effect on epithelial cell growth. The rate of epithelial cell apoptosis was not affected by NAF or CAF. Mean insulin-like growth factor II levels were not significantly different in NAF versus CAF and did not correlate with the fibroblast effect on epithelial cell proliferation. Conclusion Both NAF and CAF have the ability to inhibit the growth of pre-cancerous breast epithelial cells. NAF have greater inhibitory capacity than CAF, suggesting that the ability of fibroblasts to inhibit epithelial cell proliferation is lost during breast carcinogenesis. Furthermore, as the degree of transformation of the epithelial cells increased they became resistant to the growth-inhibitory effects of CAF. Insulin-like growth factor II could not be implicated as a contributor to this differential effect of NAF and CAF on epithelial cell growth.

Introduction The structure and homeostasis of normal breast parenchyma is maintained by dynamic interactions between breast epithelial cells and their associated stroma. These stromal elements include the vasculature, adipocytes, resident immune cells, and fibroblasts with their numerous cellular products, including various growth factors and extracellular matrix (ECM) components. In breast cancers, the stroma differs from stroma found in normal breast. The stromal alterations that accompany most invasive breast carcinomas are morphologically characterized by an enhanced accumulation of fibroblasts and a modified, collagenized extracellular matrix. Breast carcinoma-associated fibroblasts (CAF) have been reported to express increased amounts of specific ECM molecules, various molecules that modulate the ECM, and several peptide growth factors, including insulin-like growth factor (IGF) II, in comparison with fibroblasts in histologically normal breast (i.e. normal breast-associated fibroblasts [NAF]) [ 1 , 2 ]. Fibroblasts surrounding ductal carcinoma in situ (DCIS), prior to the development of invasive carcinoma, also differ from those in histologically normal breast tissue. Some of the molecular alterations found in CAF also have been documented, by immunohistochemistry and by in situ hybridization, in the fibroblasts surrounding the DCIS [ 3 - 6 ]. This suggests an accumulation of alterations in stromal fibroblasts (i.e. a progression from NAF to CAF) surrounding the breast epithelium as it progresses from normal to hyperplasia to DCIS and invasive cancer. The role that these stromal changes play in the development and progression of breast cancer and their effect on fibroblast–epithelial cell interactions is a current topic of much interest. It is theorized that CAF act to enhance breast cancer progression [ 2 ], and much of the experimental evidence to date supports this contention [ 7 - 14 ]. However, assessment of the effect of CAF has concentrated on established breast cancers. The focus in the present article is on the effect of fibroblasts on the growth of epithelial cells derived from benign breast disease, specifically proliferative breast disease (i.e. MCF10A cells and MCF10AT cells), and the potential role of fibroblast–epithelial cell interactions to promote the development of breast cancer. If CAF promote epithelial cell growth to a greater degree than NAF, could preventing the alterations that occur in fibroblasts surrounding epithelial lesions during carcinogenesis inhibit the progression of the epithelial lesion? To address this possibility, the key signaling and regulatory pathways mediating the effects of fibroblast–epithelial cell interactions, as well as the way in which these interactions are altered during carcinogenesis, must be identified. Studies to date indicate that the IGF system may play a role in the stromal–epithelial interactions that affect breast cancer progression. IGF I and IGF II both function in cellular growth, in differentiation, and in survival in all tissues. Signaling by IGF I and IGF II through their principal receptor, the insulin-like growth factor receptor, can promote cell cycle progression and can inhibit apoptosis [ 15 ]. In breast cancer, the normal regulation and functioning of the IGF system is altered, and the insulin-like growth factor receptor is expressed in 39–93% of breast cancers [ 16 ]. By in situ mRNA analysis and immunohistochemistry, IGF II expression is reported to be increased in the stromal cells within some breast cancers in comparison with the stromal cells adjacent to normal breast epithelium [ 17 ]. In the present study, the effect of NAF and CAF on the growth of pre-cancerous breast epithelial cells was compared. To explore a potential role for IGF II in the growth modulation of breast epithelial cells by NAF and CAF, the level of expression of IGF II in CAF versus that in NAF was assessed. NAF and CAF were grown in direct contact co-cultures with MCF10A breast epithelial cells and MCF10AT breast epithelial cells, both of which are considered representative of pre-invasive breast disease. MCF10A cells were derived from benign proliferative breast disease. These cells carry a deletion of the chromosomal locus containing p16 and p14ARF, and amplification of MYC [ 18 , 19 ]. MCF10A cells were transfected with mutated T24 H-ras to yield the MCF10AT cells. When suspended in the basement membrane material Matrigel ® , the MCF10AT cells persist as xenografts in nude mice. The cells initially form structures that resemble normal breast epithelium, and then gradually undergo transition to structures resembling proliferative breast disease and DCIS. Approximately 25% of the MCF10AT xenografts develop invasive carcinoma. The MCF10AT model thus reflects temporally and morphologically high-risk human proliferative breast disease [ 20 , 21 ]. Our in vitro model consists of a three-dimensional (3D) direct co-culture system in Matrigel ® similar to that utilized by Debnath and colleagues and by Shekhar and colleagues [ 19 , 22 ]. A 3D system was selected over standard monolayer cultures because it more closely simulates in vivo growth [ 19 ]. When grown in Matrigel ® , human luminal breast epithelial cells, primary or immortalized, form spherical polarized structures that resemble normal lobular acini. This spatial organization of cells determines how cells perceive and respond to signals from the stromal microenvironment [ 23 ]. Importantly, it has been demonstrated that intracellular signaling directing the proliferation of breast epithelial cells differs in cells grown in two dimensions versus those grown in three dimensions [ 23 - 25 ]. Incorporation of NAF and CAF in this 3D culture system allowed assessment of soluble and insoluble secreted factors and of direct contact factors in the fibroblast–epithelial interactions influencing epithelial cell growth. Methods Maintenance of epithelial cell lines MCF10A cells (American Type Culture Collection, Manassas, VA, USA) and MCF10AT cells (Karmanos Cancer Institute, Detroit, MI, USA) were cultivated in DMEM/Ham's F-12 (Cambrex, Walkersville, MD, USA) supplemented with 0.1 μg/ml cholera toxin (Calbiochem, San Diego, CA, USA), 10 μg/ml insulin (Sigma, St Louis, MO, USA), 0.5 μg/ml hydrocortisone (Sigma), 0.02 μg/ml epidermal growth factor (Upstate Biotechnology, Lake Placid, NY, USA) and 5% horse serum (Invitrogen, Carlsbad, CA, USA). Subconfluent cultures (80–90% confluence) were utilized in experiments. Isolation, characterization and maintenance of fibroblast cultures Fibroblasts were derived from mammary reduction specimens (NAF) and from primary breast cancers (CAF). The tissues were remnants of diagnostic surgical specimens and were obtained from The University of Alabama at Birmingham Tissue Procurement Facility after Institutional Review Board approval. H&E-stained, frozen histologic sections were prepared from each tissue sample to confirm benignity or malignancy. The tissue samples from the breast reduction specimens consisted predominantly of adipose tissue, but interspersed fibrous areas were selected for fibroblast isolation. The tissue was minced and digested for 18–24 hours at 37°C in DMEM (Vitacell, Manassas, VA, USA) with 10% fetal bovine serum (Invitrogen) supplemented with 100 U/ml streptomycin, 100 μg/ml penicillin, 2.5 μg/ml Fungizone (GibcoBRL, Life Technologies, Grand Island, NY, USA), 150 U/ml hyaluronidase (Sigma) and 200 U/ml collagenase type III (GibcoBRL). The digested tissue was centrifuged at 100 relative centrifugal force and plated in T25 tissue culture flasks with DMEM and 10% fetal bovine serum. Differential trypsinization was applied during subculturing to select for the growth of fibroblasts [ 26 ]. Early passages (below passage 9) of all fibroblasts were subjected to immunocytochemical evaluation with anti-vimentin (mouse IgG 1 , clone V9; Neomarkers, Fremont, CA, USA), anti-epithelial membrane antigen (mouse IgG 2a , clone ZCE113; Zymed, San Francisco, CA, USA), and anti-cytokeratin (CK) 5/CK 8 (mouse IgG 1 , clone C-50; Neomarkers) as confirmation of their stromal origin (i.e. strong vimentin expression, and absence of epithelial membrane antigen and CK 5/CK 8). Epithelial membrane antigen and CK 8 are expressed primarily in luminal breast epithelial cells, whereas CK 5 is found in myoepithelial cells [ 27 ]. For immunocytochemical evaluation, fibroblasts were grown on glass coverslips, fixed in 70% ethanol and were permeabilized with acetone. Fibroblasts were incubated in 3% hydrogen peroxide followed by 3% goat serum at room temperature. Anti-vimentin (0.5 μg/ml), anti-epithelial membrane antigen (1 μg/ml), or anti-CK 5/CK 8 (0.3 μg/ml) were applied for 1 hour at room temperature. Secondary detection was accomplished with a streptavidin/horseradish peroxidase secondary detection system (Signet Laboratories, Dedham, MA, USA) and diaminobenzidine (BioGenex, San Ramon, CA, USA). Harris hematoxylin was used as a counterstain. The fibroblasts were routinely maintained in DMEM and 10% fetal bovine serum. Subconfluent cultures (70–90% confluence) of lower passages (below passage 9) were utilized for the experiments described. Only early-passage fibroblast cultures are used in experiments to more closely simulate their in vivo phenotype. It has been shown that early-passage (below passage 9) colonic primary fibroblasts maintain many of their in vivo characteristics, including expression of alpha-smooth muscle actin, collagen IV and laminin 1 [ 28 ]. Multiple NAF and CAF cultures were utilized because of potential variation among fibroblasts from different individuals and different breast cancers. Preparation of the 3D cultures The effect of NAF and CAF on MCF10A breast epithelial cells and MCF10AT breast epithelial cells was studied using a 3D in vitro model. In co-cultures, epithelial cells and fibroblasts were mixed in Human Endothelial–SFM Basal Growth Media (Invitrogen) supplemented with 10 ng/ml epidermal growth factor and 20 ng/ml basic fibroblast growth factor (Invitrogen). This was followed by dispersal on 100 μl basement membrane material (Growth Factor Reduced Matrigel ® ; BD Biosciences, Bedford, MA, USA) in each well of eight-well chamber slides (Lab-Tek ® Chamber Slide™ System; Nalge Nunc International, Naperville, IL, USA). The ratio of epithelial cells to fibroblasts (E:F) ranged from 2:1 to 1:3 by varying the number of fibroblasts while keeping the number of epithelial cells constant (100,000 cells/well). Controls consisted of 3D monocultures of MCF10A cells and MCF10AT cells (100,000 cells/well). All cultures were incubated in a 37°C, 5% CO 2 humidified incubator for 14 days with supplementation of fresh Human Endothelial-SFM Basal Growth Media supplemented with 10 ng/ml epidermal growth factor and 20 ng/ml basic fibroblast growth factor at 4-day intervals. Morphologic development was observed by phase contrast microscopy. 5-bromo-2'-deoxyuridine (BrdU) (0.2 mg/ml; Calbiochem) was applied to all cultures for 24 hours. The cultures were removed from the chamber slides, were embedded in HistoGel specimen processing gel (Richard-Allan, Kalamazoo, MI, USA), were fixed in 10% neutral-buffered formalin, and were embedded in paraffin. Histologic sections were prepared for H&E staining and immunocytochemistry. BrdU detection by immunocytochemistry Distinction of epithelial cells from fibroblasts in the 3D cultures was accomplished by examination of the cell morphology and location within the culture. To ensure that only epithelial cells were counted, however, immunostaining with anti-BrdU (mouse IgG 1 , clone Bu20a; DAKO, Carpinteria, CA, USA) was followed by immunostaining with anti-CK 5/CK 8, expressed only in MCF10A cells and MCF10AT cells in 3D cultures. The staining entailed pretreatment with low-temperature antigen retrieval (i.e. incubation in 0.01 M citric acid monohydrate, pH 6.0, for 2 hours in an 80°C water bath), followed by sequential incubation in 1 N HCl, 3% hydrogen peroxide, 1% goat serum, and anti-BrdU (3 μg/ml). Secondary detection was as previously described. This was followed by incubation with anti-CK 5/CK 8 (4 μg/ml) and secondary detection using a streptavidin/alkaline phosphatase reagent (Signet Laboratories) and New Fuchsin (BioGenex) as the chromogen. The slides were lightly counterstained with Harris hematoxylin. The resulting dual coloration of anti-BrdU (brown nucleus) and anti-CK 5/CK 8 (pink cytoplasm) enabled identification of proliferating epithelial cells. A BrdU-labeling index in the epithelial cells was determined by calculating the percentage of epithelial cells with nuclear staining for anti-BrdU in complete cross-sections of the 3D cultures. A minimum of 500 epithelial cells was counted. Negative controls consisted of histologic sections of each 3D culture processed without the addition of primary antibodies. BrdU detection by flow cytometry Cells were gently removed from 3D cultures after treatment with Dispase (BD Biosciences Discovery Labware, Bedford, MA, USA), followed by 5–10 mM EDTA. Recovered cells were washed with cold PBS. To distinguish co-cultured epithelial cells from fibroblasts, allophycocyanin-conjugated anti-EpCAM (mouse IgG 1 , clone EBA-1; BD Biosciences Immunocytometry Systems, San Jose, CA, USA) was used to label MCF10A cells and MCF10AT cells. The cells were permeabilized and fixed (BD Cytofix/Cytoperm and Perm/Wash kit; BD Biosciences Pharmingen, San Diego, CA, USA). Prior to staining with FITC-conjugated anti-BrdU (mouse IgG 1 , clone B44; BD Biosciences Immunocytometry Systems), cells were treated with DNase I (Roche Diagnostics GmbH, Penzberg, Germany). Samples were analyzed on a BD FACS Calibur™ flow cytometer (BD Biosciences). The percentage of BrdU-labeled epithelial cells (positive for anti-EpCAM and anti-BrdU) was calculated from the total acquired events. Assessment of apoptosis by TUNEL assay Apoptosis was quantified in epithelial cells in 3D cultures by the TUNEL Assay (ApopTag Peroxidase In Situ Apoptosis Kit; Integren Co., Purchase, NY, USA) as per the manufacturer's instructions. In the TUNEL assay, terminal deoxynucleotide transferase was used to label fragmented DNA with digoxigenin-linked nucleotides. These nucleotides were then detected using an anti-digoxigenin antibody. Negative controls consisted of histologic sections of each 3D culture processed without the addition of terminal deoxynucleotide transferase. Sequential sections of each 3D culture were stained with anti-CK 5/CK 8, as previously described, and were used to aid in identification of epithelial cells in co-cultures. The percentage of epithelial cells with nuclear staining was determined in complete cross-sections of the 3D cultures. A minimum of 500 epithelial cells was counted. Preparation of cell lysates Cells were gently removed from Matrigel ® in 3D co-cultures of MCF10AT cells (100,000 cells/well) with fibroblasts (50,000 cells/well) and monocultures of MCF10AT cells (100,000 cells/well) or fibroblasts (50,000 cells/well) using dispase, as previously described. The released cells were washed with cold PBS to remove residual Matrigel ® . Cell lysates were prepared from 3D cultures and monolayer cultures with NEB lysis buffer (Cell Signaling Technology, Beverly, MA, USA) containing 20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 1 mM Na 2 EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na 3 VO 4 , 1 μg/ml leupeptin, 10 mM sodium fluoride, a Complete Mini Protease Inhibitor Cocktail tablet (Roche, Indianapolis, IN, USA) and 1 mM phenylmethylsulfonylfluoride, and were cleared by centrifugation. All lysates were concentrated with Millipore Microcon ® Centrifugal Filter Devices YM-3 (Millipore Corporation, Bedford, MA, USA). The protein concentration of lysates was measured with the BCA Protein Assay Kit (Pierce, Rockford, IL, USA). Enzyme-linked immunosorbent assay The protein levels of IGF II in cell lysates were measured by ELISA for human IGF II (DSL-10-9100 Active™ IGF-II ELISA; Diagnostic System Laboratories, Webster, TX, USA) as per the manufacturer's instructions. The kit includes a modified version of the standard acid–ethanol extraction prior to ELISA. The ELISA for IGF II shows no detectable cross-reactivity with IGF I, insulin, IGF binding protein 1, and IGF binding protein 3. The minimum detection limit of the assay is 0.25 ng/ml. The amount of IGF II present in lysates from cell lines and 3D cultures was normalized to the total protein concentration. Quantitative real-time PCR RNA isolation (Trizol reagent; GibcoBRL) was followed by RNA clean-up with RNeasy columns (Qiagen, Valencia, CA, USA). Spectrophotometric ratios of A 260 to A 280 were greater than 1.8. The forward, reverse and probe oligonucleotides were synthesized and purified by HPLC (Applied Biosystems, Foster City, CA, USA) after complete evaluation of the IGF II and ribosomal S9 gene sequences (GenBank database) (IGF II forward, 5'-GTCGATGCTGGTGCTTCTCA-3' and reverse, 5'-GGGCGGTAAGCAGCAATG-3' ; probe, 5'-6FAMCTTCTTGGCCTTCGCCTCGTGCTTAMRA-3' ; ribosomal S9 forward, 5'-ATCCGCCAGCGCCATA-3' and reverse, 5'-TCAATGTGCTTCTGGGAATCC-3' ; probe, 5'-6FAMAGCAGGTGGTGAACATCCCGTCCTTTAMRA-3') using the Primer Express software (Applied Biosystems). Fluorescent signal data were collected by the ABI Prism 7700 Sequence Detection System (Applied Biosystems). The log-linear phase of amplification was monitored to obtain the threshold cycle (defined as the fractional cycle number at which the amount of the amplified target reaches a fixed threshold) values for each RNA sample. Ribosomal S9 was used as the internal reference and was selected because it exhibits minimal variability in tissues of different origins [ 29 ]. The comparative threshold cycle method was employed to determine IGF II expression levels in each sample relative to a calibrator, in this case MCF10AT cells [ 29 , 30 ]. Each sample was run in triplicate. Statistical analysis A relationship between BrdU labeling or apoptotic indexes and co-culture with NAF or with CAF was assessed by multiple linear regression analysis to allow combining replicate experiments performed on different days. Results of ELISA and quantitative real-time PCR were compared by t test. Outliers were eliminated prior to analysis using a basic outlier test, where a high outlier was defined as a number greater than quartile 3 + 1.5 (interquartile range) and a low outlier was less than quartile 1 – 1.5 (interquartile range). Results Fibroblasts alter the 3D morphology of MCF10A cells and MCF10AT cells The immunocytochemical characterization of fibroblasts used in the described experiments confirmed their stromal nature. Immunostaining for vimentin was strongly positive and staining for epithelial membrane antigen and CK 5/CK 8 was negative. In 3D monocultures, both MCF10A cells and MCF10AT cells initially form a lattice-like network of duct-like structures. After several days, the lattice-like network is replaced by a predominance of rounded epithelial cell groups (spheroids) (Fig. 1a,1b,1c,1d ). MCF10A cells form small, rounded spheroids (Figs 1b and 2a ). MCF10AT cells aggregate into larger solid groups or sheets with extensive squamous metaplasia (Figs 1d and 2c ). The formation of larger three-dimensional structures and the abnormal differentiation (i.e. squamous metaplasia) of MCF10AT cells supports their greater degree of transformation. In 3D co-cultures with NAF or with CAF, the epithelial cells and fibroblasts form large rounded structures (Fig. 1e ). In histologic sections of co-cultures with NAF or with CAF, MCF10A cells and MCF10AT cells form spheroids or sheets within Matrigel ® , as seen in monocultures, and these epithelial groups surround an aggregate of fibroblasts (Fig. 2b,2d ). These 3D co-cultures resemble a terminal duct-lobular unit in the normal breast or in proliferative breast disease/DCIS (Fig. 3a,3b,3c ), in that epithelial cells are arranged in groups (similar to an acinus or a terminal duct) surrounded by a laminin-rich basement membrane with fibroblasts located outside the basement membrane and separating epithelial groups. The 3D in vitro model differs from normal breast or in situ breast disease in vivo , however, in that the number of fibroblasts centrally located and between epithelial cell groups is greater than that typically seen in vivo . Additionally, the fibroblasts in the in vitro model are surrounded by an ECM rich in laminin and collagen IV [ 31 ], whereas in vivo collagen I would typically predominate [ 32 ]. In 3D co-cultures, both NAF and CAF markedly suppressed squamous metaplasia of MCF10AT cells as observed on H&E-stained sections of monocultures, thereby normalizing the morphology to a glandular phenotype (Fig. 2c,2d ). No obvious differences between NAF and CAF in morphology, growth pattern or adhesion to substrate, in either a monolayer or the 3D culture, were observed. NAF and CAF affect the rate of proliferation of MCF10A cells and MCF10AT cells in 3D co-culture In replicate co-cultures of MCF10A cells with three different NAF and CAF grown in an E:F of 2:1, both types of fibroblasts significantly reduced proliferation of MCF10A cells. The mean BrdU-labeling index of MCF10A cells, when measured by immunocytochemistry, was decreased by 47% in co-culture with NAF ( n = 19, P = 0.009) and by 39% in co-culture with CAF ( n = 19, P = 0.024) relative to the MCF10A monoculture (Table 1 and Fig. 4 ). The BrdU-labeling index of MCF10AT cells was reduced by 49% in the presence of NAF ( n = 20, P = 0.013), relative to the MCF10AT monoculture, whereas co-culture with CAF failed to significantly lower the MCF10AT BrdU-labeling index ( n = 22, P = 0.935) (Table 2 and Fig. 4 ). The effect of NAF versus CAF on the rate of proliferation of MCF10AT cells was significantly different ( P < 0.001). The effect was further confirmed by repeating the co-cultures to measure the BrdU-labeling index by flow cytometry, rather than by immunocytochemistry (Fig. 5 ). There was variability among NAF cultures and among CAF cultures in their ability to suppress proliferation of MCF10A cells and MCF10AT cells (Tables 1 and 2 ) in this 3D culture system, potentially reflecting heterogeneity among the individuals from which the fibroblasts were derived. Because of this variability, detection of a significant difference in the function of NAF and CAF required many replicates and multiple fibroblast cultures derived from different individuals. In a prior report, CAF was found to promote, rather than inhibit, the growth of MCF10A cells in a similar 3D co-culture system [ 22 ]. One of several possible explanations for this discrepancy between the prior result and the present result is a difference in E:F. Shekhar and colleagues used an E:F of 1:1 rather than the E:F of 2:1 we initially used [ 22 ]. The number of fibroblasts has been shown to have an effect on the response of epithelial cells [ 7 , 9 , 14 ]. We therefore repeated the 3D co-cultures of MCF10A cells using NAF-2 and CAF-1 with increasing numbers of fibroblasts (i.e. a decreasing E:F) (Fig. 6 ). BrdU labeling was assessed by immunocytochemistry of histologic sections of 3D cultures. As previously, NAF-2 at an E:F of 2:1 suppressed proliferation of MCF10A cells. However, with increasing numbers of NAF-2, this suppression effect was gradually weakened ( P = 0.043). Although we found no significant difference in the suppressive effect of NAF-2 in an E:F of 2:1 versus an E:F of 1:1 or of 1:2, there was a significantly greater rate of proliferation of MCF10A cells with NAF-2 in an E:F of 1:3 compared with in an E:F of 2:1 ( P = 0.028). More importantly, CAF-1 at an E:F of 1:1 did not significantly suppress proliferation, whereas our original ratio of 2:1 did ( P = 0.025). CAF-1 at an E:F of 1:2 also conferred a higher rate of proliferation of MCF10A cells than the E:F of 2:1, but this did not reach statistical significance ( P = 0.054). At an E:F of 1:3, however, CAF-1 caused a decrease in proliferation of MCF10A cells and enhanced cell death, as assessed by microscopic morphology. At an E:F of 1:3, the total number of viable MCF10A cells was reduced in co-culture with both NAF-2 and CAF-1; however, this reduction was more marked with CAF-1. Neither NAF nor CAF had a significant effect on the rate of apoptosis of MCF10A cells or MCF10AT cells when grown at an E:F of 2:1 after 2 weeks of co-culture, as assessed by TUNEL assay (Fig. 7 ). Quantities of IGF II are no different in NAF versus CAF As an initial attempt to identify differences between NAF and CAF that explain our observed results, expression of IGF II in NAF and in CAF was assessed. A higher level of expression of IGF II in CAF than in NAF may provide an explanation for the higher rate of proliferation of MCF10AT cells allowed by CAF in comparison with NAF. ELISA performed on cell lysates of NAF and CAF cultures demonstrated variability in expression of IGF II among cultures, but no significant difference was observed in the mean IGF II quantity between NAF and CAF in monolayer cultures (Table 3 ) or in 3D monocultures (Table 4 ). Although in monolayer cultures more CAF than NAF had IGF II levels > 10 ng/μg protein, the mean levels of IGF II were similar (NAF versus CAF, P > 0.05) (Table 3 ). This comparison included five different NAF cultures and six different CAF cultures. ELISA was also performed on NAF-conditioned media and CAF-conditioned media but, despite the concentration of samples, the levels were too low for reliable quantification by ELISA (data not shown). Additionally, because IGF II mRNA was previously reported to be expressed at a higher level in CAF than in NAF, IGF II mRNA was assessed by quantitative real-time PCR in monolayer cultures of NAF and CAF. The relative expression levels for each fibroblast culture are provided in Fig. 8 . Although more CAF cultures than NAF cultures expressed IGF II mRNA at relatively high levels, the mean relative expression level of IGF II mRNA for NAF (4.2, n = 5) and for CAF (7.7, n = 5) did not differ significantly ( P = 0.390). Co-culture of MCF10A cells or MCF10AT cells may enhance expression of IGF II in fibroblasts, as has been reported in fibroblasts co-cultured with MCF-7 cells [ 33 ]. Furthermore, expression of IGF II may be enhanced to a greater degree in CAF versus NAF, thus possibly explaining the greater rate of proliferation of MCF10AT cells in co-culture with CAF than with NAF. 3D culture of NAF and CAF also could alter expression of IGF II in comparison of the monolayer culture. To investigate these possibilities, IGF II levels were assessed by ELISA in 3D monocultures of the same three NAF cultures and three CAF cultures used in previous co-cultures, in 3D monocultures of MCF10AT cells, and in 3D co-cultures of MCF10AT cells with NAF and CAF in an E:F of 2:1, identical to those co-cultures assessed for BrdU labeling previously (Table 4 ). In the 3D cultures, fibroblasts expressed IGF II at a significantly higher level than the epithelial cells ( P < 0.01), and there was no difference in IGF II between NAF and CAF ( P > 0.05). In co-cultures, the overall expression of IGF II was lower than in fibroblast monocultures. This latter result is expected because of the addition of a relatively large number of MCF10AT cells (E:F of 2:1) that express IGF II at a lower level than fibroblasts. Furthermore, there was no difference in mean IGF II levels in co-cultures with NAF versus CAF. The IGF II levels in 3D cultures were generally higher than in monolayer cultures, and this was particularly true when comparing 3D monocultures of fibroblasts versus two-dimensional monocultures of fibroblasts. The potential reasons for this include an effect of 3D growth, or the presence of Growth Factor Reduced Matrigel ® on the expression of IGF II, and/or deviations between the ELISA runs. When the protein levels of IGF II in 3D monocultures of fibroblasts or in 3D co-cultures were correlated with the rate of proliferation of MCF10AT cells in co-culture with the matching NAF or CAF, no significant correlation was observed ( r = 0.030 or r = 0.258, respectively; P > 0.05, Pearson Product Moment Correlation). Discussion Our results indicate that both NAF and CAF have the ability to inhibit epithelial cell proliferation and to induce glandular differentiation to a more normal phenotype. However, CAF have less inhibitory capacity than NAF. At relatively low concentrations of fibroblasts (E:F of 2:1) NAF can suppress proliferation of both MCF10A cells and MCF10AT cells, whereas CAF can suppress proliferation of MCF10A cells but not the more transformed MCF10AT cells. In vivo NAF may thus have an inhibitory and regulatory effect on the proliferation of normal epithelial cells. This suppressive ability may be lost or reduced as epithelial lesions gradually progress from hyperplasia to DCIS and invasive cancer, and correspondingly NAF become CAF. Differences in gene expression between CAF and NAF have been documented by examination of human breast cancers by immunohistochemistry and in situ hybridization, and also by analysis of cultures of fibroblasts isolated from breast cancers. These documented characteristics of breast-derived CAF are thoroughly reviewed by Kunz-Schughart and Knuechel [ 1 , 2 ], and include an increased expression of several growth factors [ 3 , 34 ], of ECM molecules [ 35 - 37 ], and of proteases and protease inhibitors involved in modulating the ECM [ 5 , 6 ]. Many of these differences in the expression profile of CAF in comparison with NAF have the potential to enhance the development, growth and progression of breast carcinoma [ 1 ]. Furthermore, a subset of the phenotypic alterations documented in CAF have been identified in fibroblasts surrounding DCIS by examining these lesions using immunohistochemistry or in situ hybridization [ 3 - 6 ]. However, many of the changes observed in CAF have not yet been examined in DCIS, and it is therefore quite possible that fibroblasts surrounding DCIS share more features with CAF than are currently documented. Attempts to actually demonstrate a promotional effect of CAF on the growth of epithelial cell lines derived from breast cancer are limited in number and have met with somewhat conflicting results. A variety of different methods were used in these studies, complicating comparison among them. In most in vitro analyses, however, direct and indirect co-culture with CAF increased the growth of MCF-7 breast cancer cells compared with MCF-7 cells alone [ 7 , 12 , 14 , 38 ]. Co-culture of NAF with MCF-7 cells caused both growth promotion [ 7 , 10 , 11 , 13 , 39 , 40 ] and inhibition [ 8 ] of MCF-7 cells. In contrast, co-culture of NAF and CAF with the breast cancer cell line MDA-MB-231 had no significant effect on epithelial cell growth [ 10 , 11 ]. In vivo , NAF and CAF increased growth of the MCF-7 xenografts or NAF had no effect [ 7 , 38 ]. Only a few reports have addressed the effects of fibroblast–epithelial interactions on the growth of nontransformed or nontumorigenic breast epithelial cells, rather than on the growth of breast carcinoma cells. NAF and CAF have been reported to stimulate the proliferation of normal human breast epithelial cells [ 7 , 39 ] or to have no effect on the rate of proliferation of normal breast epithelial cells immortalized by SV40 large-T antigen [ 8 ]. The overriding observations from these previous studies of fibroblast–epithelial interactions are that co-cultured fibroblasts affect the growth of epithelial cells, but this growth is dependent on the source of the fibroblasts, the characteristics of the epithelial cells, and the culture conditions utilized. In a co-culture system similar to that presented here, NAF (two different cultures) inhibited the growth, measured by direct counting of total viable cells in co-cultures, of MCF10A cells and MCF10AT-EIII8 cells – whereas CAF induced the growth of both epithelial cell lines [ 22 ]. While our findings for NAF are similar to the previous results, we did not find a promotional effect of CAF on epithelial cell growth. This discrepancy may be a result of the interindividual variation found in fibroblast cultures and/or differences in epithelial cells (MCF10AT cells versus MCF10AT-EIII8 cells, an estrogen-induced derivative of the MCF10AT cells). In addition, the total number of viable cells present (fibroblasts and epithelial cells) in co-cultures were counted, whereas in the current study only proliferation of epithelial cells was measured. In the previous study, an E:F ratio of 1:1 was utilized compared with the current E:F of 2:1. In support of the latter, we found that CAF at lower E:F values of 1:1 and 1:2 no longer inhibited the growth of MCF10A cells. Although not identical, this result is more in keeping with that of Shekhar and colleagues [ 22 ]. Varying E:F ratios have been utilized in a few prior studies of the effect of fibroblasts on the growth of breast carcinoma cells. Ratios have varied from a great predominance of epithelial cells [ 9 ] to a predominance of fibroblasts [ 7 , 14 ]. In these previous studies, an increasing proportion of breast fibroblasts, either NAF or CAF, correlates with an increase in growth of co-cultured cancerous breast epithelial cells to a plateau where no further enhancement of growth is seen. This is in general concordance with our results, where the inhibitory effect of both NAF and CAF on proliferation of MCF10A cells was less with increasing numbers of fibroblasts, particularly for NAF. This suggests the presence of fibroblast-derived factors that both inhibit and promote the proliferation of epithelial cells; at higher concentrations of fibroblasts, the promotional effect predominates. To be biologically relevant, the most meaningful ratio of epithelial cells to fibroblasts depends on the lesion or tissue being modeled. In the present study, the intent was to model proliferative breast disease and DCIS, the putative precursors of invasive carcinoma, which are believed to have their origins in terminal ducts within terminal duct-lobular units [ 41 ]. Microscopic examination of such intraductal lesions in the human breast reveals a range of E:F, depending on the extent of fibrosis of the lesion. In the normal terminal duct-lobular unit and in situ lesions depicted in Fig. 3 , the E:F varies from 3:1 to 2:1. These ratios were determined by counting epithelial cells (both luminal and myoepithelial cells) and stromal cells identified as fibroblasts and located within or in close proximity to the terminal duct-lobular unit involved. It is probable that those fibroblasts in proximity to the epithelial structures have the greatest influence on epithelial cell behavior. Our choice to use an initial E:F of 2:1 is therefore appropriate, whereas lower ratios with many more fibroblasts are less common within terminal duct-lobular units in vivo . Prior attempts to identify fibroblast-derived factors that are mediating the effect of NAF and CAF on breast epithelial cell growth have identified IGF I and/or IGF II as partially contributing to the mitogenic effect of fibroblasts [ 11 ]. Previous studies have reported that IGF II was expressed at a moderate to high level in 43–57% of breast cancers by in situ hybridization and by immunohistochemistry [ 17 , 42 , 43 ], with localization primarily to stromal fibroblasts or vessel walls [ 17 , 42 ], making IGF II a potential candidate to mediate fibroblast–epithelial interactions. Additionally, in cultures of CAF and NAF, IGF II mRNA was detected at higher levels more frequently in CAF than in NAF in some studies [ 33 , 44 ], but not in other studies [ 45 , 46 ]. In the current study, IGF II levels in fibroblasts were assessed in both monolayer cultures and 3D cultures by ELISA and quantitative real-time PCR. While more CAF cultures than NAF cultures had relatively higher levels of IGF II mRNA or protein, no significant differences in mean quantities of IGF II mRNA or protein between these CAF and NAF were observed. We also found no difference in IGF II expression in the 3D co-cultures of MCF10AT cells prepared with NAF versus CAF, suggesting that MCF10AT cells do not alter IGF II expression to a different degree in co-cultured NAF and CAF. Furthermore, we found no correlation between proliferation of MCF10AT cells in co-culture and the level of IGF II protein in 3D fibroblast monocultures or 3D co-cultures. Differences in IGF II expression in NAF and CAF are therefore unlikely to explain the difference in the effect of CAF versus that of NAF on proliferation of the epithelial cells described in this study. Our results do not eliminate a role for the IGF system in these fibroblast–epithelial cell interactions as other family members, such as IGF II receptor and a multitude of IGF binding proteins, may be mediating these interactions in other ways. Prior studies by other workers [ 33 , 44 - 46 ] and the current work underscore the variability in expression of IGF II and in the growth inhibitory effect of NAF and CAF, and emphasize the importance of including several NAF and CAF cultures from different individuals in studies of fibroblast–epithelial cell interactions. The interindividual heterogeneity observed among NAF and CAF complicates assessment of mechanisms underlying fibroblast–epithelial interactions. Broad generalizations based on the results of experiments using only one or two fibroblast cultures should be avoided. In conclusion, both NAF and CAF have the ability to suppress breast epithelial cell proliferation; however, the capacity of CAF to inhibit proliferation is less than that of NAF. This suggests that there are differences in either secreted factors or intercellular interactions between NAF and CAF that render CAF less effective in inhibiting proliferation, particularly of more transformed epithelial cells. Furthermore, differences between the phenotypes of the H-ras overexpressing MCF10AT cells and the parental MCF10A cells cause MCF10AT cells to be more resistant to the suppressive effect of fibroblasts. Future work to identify the key fibroblast–epithelial interactions mediating these effects may reveal mechanisms to allow restoration of the inhibitory of effect of fibroblasts during the carcinogenic process. Abbreviations 3D = three-dimensional; BrdU = 5-bromo-2'-deoxyuridine; CAF = carcinoma-associated fibroblasts; CK = cytokeratin; DCIS = ductal carcinoma in situ ; DMEM = Dulbecco's modified Eagle's medium; ECM = extracellular matrix; E:F = ratio of epithelial cells to fibroblasts; ELISA = enzyme-linked immunosorbent assay; FITC = fluorescein isothiocyanate; H&E = hematoxylin and eosin; HPLC = high-performance liquid chromatography; IGF = insulin-like growth factor; NAF = normal breast-associated fibroblasts; PBS = phosphate-buffered saline; PCR = polymerase chain reaction. Competing interests The author(s) declare that they have no competing interests. Authors' contributions AS carried out experimental design and conduct, and manuscript preparation. ZN provided technical assistance with flow cytometry. MRJ provided technical assistance with quantitative real-time PCR, and carried out manuscript preparation. DBB provided technical assistance with the conduction of experiments. SRG provided assistance with the conduction of experiments. GPP provided assistance with the statistical analysis. JVT provided technical assistance and carried out manuscript preparation. DRW carried out experimental design and manuscript preparation. ARF was the principal investigator responsible for overall experimental design and conduct, and manuscript preparation.

D:\keerthana\PMC001xxxxxx\PMC1064098.xml

1064099

Systemic chemotherapy induces microsatellite instability in the peripheral blood mononuclear cells of breast cancer patients

Introduction Systemic chemotherapy is an important part of treatment for breast cancer. We conducted the present study to evaluate whether systemic chemotherapy could produce microsatellite instability (MSI) in the peripheral blood mononuclear cell fraction of breast cancer patients. Methods We studied 119 sequential blood samples from 30 previously untreated breast cancer patients before, during and after chemotherapy. For comparison, we also evaluated 20 women who had no relevant medical history (control group). Results In 27 out of 30 patients we observed MSI in at least one sample, and six patients had loss of heterozygosity. We found a significant correlation between the number of MSI events per sample and chemotherapy with alkylating agents ( P < 0.0001). We also observed an inverse correlation between the percentage of cells positive for hMSH2 and the number of MSI events per sample ( P = 0.00019) and use of alkylating agents ( P = 0.019). Conclusion We conclude that systemic chemotherapy may induce MSI and loss of heterozygosity in peripheral blood mononuclear cells from breast cancer patients receiving alkylating agents, possibly mediated by a chemotherapy-induced decrease in the expression of hMSH2. These effects may be related to the generation of secondary leukaemia in some patients, and may also intensify the genetic instability of tumours and increase resistance to treatment.

Introduction Systemic chemotherapy is an important part of treatment for breast cancer in both the adjuvant and palliative settings. Despite the consistent improvement in overall survival afforded by systemic adjuvant chemotherapy, about 1% of patients develop secondary leukaemia and/or myelodysplasia, probably as a consequence of the genotoxic effects of this type of treatment [ 1 ]. Deficiencies in the DNA mismatch repair system leading to microsatellite instability (MSI) may produce a syndrome of familial predisposition to colon, endometrial and upper gastrointestinal cancers, known as hereditary nonpolyposis colon cancer [ 2 ]. The DNA mismatch repair system depends on the coordinated interplay of several proteins encoded by various genes, including hMLH1, hMSH2, hPMS1 and hPMS2 [ 2 ]. Several groups have reported a significant association between MSI and treatment-related secondary acute myeloid leukaemia (AML) and myelodysplasia [ 3 - 5 ]. In some studies MSI in treatment-related secondary AML/myelodysplasia cases was accompanied by hMLH1 hypermethylation [ 3 ] and MSH2 polymorphisms [ 6 ]. In order to evaluate whether systemic chemotherapy could produce MSI in normal peripheral blood mononuclear cells (PBMCs), we analyzed PBMCs from breast cancer patients collected before, during and after systemic chemotherapy. Methods Collection and preparation of samples This protocol was approved by our institutional review board. We obtained blood samples from 33 patients with histologically confirmed breast cancer after informed consent had been obtained. We had only the initial sample in three patients, and so we could not include them in the study. We therefore studied 119 sequential blood samples from 30 previously untreated breast cancer patients, collected at 3-month intervals before, during and after receiving systemic treatment (13 adjuvant, 12 neoadjuvant and 5 palliative). Three patients initially received hormones (two adjuvant and one palliative). Chemotherapy combinations containing cyclophosphamide were classified as alkylating regimens (FAC [5-fluorouracil, doxorubicin and cyclophosphamide], AC [doxorubicin and cyclophosphamide], CMF [cyclophosphamide, methotrexate, and 5-fluorouracil]). We also studied PBMCs from 20 normal control women, who had no relevant previous medical history, by immunocytochemistry. Afterward, we collected peripheral blood from these 20 normal control women on two occasions with a 3-month interval to evaluate MSI using the TP53Alu and PCR15.1 markers (described below). Each sample consisted of 20 ml venous blood, from which we separated the PBMCs by Ficoll gradient (Ficoll Hypaque; Organon Teknica ® , Durham, NC, USA), yielding a final concentration of 1.0 × 10 6 cells/ml; we sent part of the sample for cytospin for immunocytochemical studies and part for DNA extraction. DNA was extracted from mononuclear fraction by the use of Trizol (Invitrogen, Carlsbad, CA, USA), in accordance with the manufacturer's instructions. For immunocytochemistry analysis, assay slides were prepared (described in detail elsewhere [ 7 ]). Microsatellite analysis The sequences of all primers used for six microsatellite loci (BAT-26, BAT-40, MFD-28, MFD-41, TP53, PCR15.1, TP53ALU) and the PCR protocols used to study each of these markers were described previously [ 8 ]. Briefly, PCR products were denatured and subjected to electrophoresis in Gene Gel Clean 15/24 (Amershan Pharmacia Biotech AB, Uppsala, Sweden) for 90 min at 600 V and 8°C, and then silver stained using a Hoefer Automed Gel Stainer (Amershan Pharmacia Biotech AB). MSI was defined by the appearance of novel bands and loss of heterozygosity (LOH) whenever disappearance of previously visualized bands occurred, as described previously [ 8 ]. Three independent observers analyzed the results of each gel for LOH and MSI. Immunocytochemistry For immunostaining for proliferating nuclear antigen (PCNA), hMLH1, hPMS1, hPMS2, hMSH2 and TP-53 proteins, we used an avidin–biotin–peroxidase complex and 3,3'-diaminobenzidine as chromogen. All antibodies were supplied by Santa Cruz Biotechnology (Santa Cruz, CA, USA), and for each clone we employed the following dilutions: PC-10 (anti-PCNA), 1:1000; C-20 (anti-hMLH1), 1:25; K-20 (anti-hPMS1), 1:25; C-20 (anti-hPMS2), 1:200; N-20 (anti-hMSH2), 1:200; and Pab1801 (anti-P53), 1:400. Endogenous peroxidase activity was blocked by incubation with H 2 O 2 and washed in tap water. The cytospin slides were then treated with 7% skimmed milk for 60 min to block nonspecific protein binding. Each antibody was applied separately, and cytospins were incubated for 18 hours at 3°C. After 3 washes in phosphate-buffered saline, antigen-bound primary antibody was detected using a standard streptavidin–biotin complex (Strep AB Complex; Dako, Carpinteria, CA, USA). After brief washing, slides were incubated with diaminobenzidine and H 2 O 2 for 10 min and then counterstained with haematoxylin, dehydrated in graded alcohols and cleared in xylene. Two independent observers scored 300 cells/slide as positive or negative according to the presence of nuclear staining for each of the aforementioned antibodies. Then, the results from these two observers were averaged to obtain a percentage of positive cells per sample. Statistical methods We analyzed correlations between categorical variables using the χ 2 or Fisher's exact test. We used analysis of variance to study correlations between continuous and categorical variables, and simple regression to analyze correlations between continuous variables. Results We studied 30 patients with median age 52 years (range 25–80 years). Twelve patients had stage II, 13 had stage III and five had stage IV breast cancer. We evaluated samples from 20 normal control women by immunocytochemistry for PCNA, hMLH1, hPMS2, hMSH2 and TP-53 (Table 1 ); the raw data for each of the included patients appear in Additional file 1 . We observed MSI in 40 out of the 119 samples collected, and MSI was observed in 27 out of 30 patients in at least one sample. We observed LOH at the PCR15.1 locus in six patients. Whereas MSI was usually transient, disappearing in the next collected sample, LOH was persistent in all subsequent samples in five of the six observed cases (Fig. 1 ). In order to ensure that our findings in the patients studied were not due to chance, we studied 20 normal blood donors (women who had no relevant medical history; see Additional file 2 ). In those women blood samples were collected twice, with a 3-month interval between samples. We studied both samples from each of these normal women for MSI using TP53ALU and PCR15.1 markers. In only one normal woman was any MSI identified (one additional band was detected in the PCR15-1 marker after 3 months). None of them had LOH (data not shown). Therefore, in only one blood sample out of 40 drawn from these 20 normal women was there any evidence of MSI, as compared with 28 out of 119 samples from the breast cancer patients studied using the same markers ( P = 0.018). We observed a significant correlation between the number of MSI events per sample and the use of chemotherapy ( P = 0.005), especially chemotherapy containing alkylating agents ( P < 0.0001; Fig. 2 ). We also observed an inverse correlation between the number of MSI events per sample and the percentage of cells positive for MSH2 by immunocytochemistry ( P = 0.00019) and especially for the MFD41 marker ( P = 0.000638). The percentage of cells positive for MSH2 by immunocytochemistry was inversely correlated with the use of chemotherapy regimens containing alkylating agents ( P = 0.019). We found no significant correlations between the number of MSI events or the presence of LOH and the frequency of cells positive for TP53, hMLH1, hPMS1 and hPMS2. In the present study, the percentage of cells positive for PCNA, as determined by immunocytochemistry, was directly correlated with use of radiation therapy ( P = 0.046) and with the percentage of MSH2-positive cells ( P = 0.000096), and inversely correlated with the presence of LOH ( P = 0.043), but it was not correlated with the number of MSI events per sample. Discussion Systemic chemotherapy administered to breast cancer patients induces secondary leukaemia and myelodysplasia in about 1% [ 1 ]. In our study, in 90% of patients MSI was noted in at least one of the samples. This finding does not appear due to chance because only one of the sequential blood samples drawn from 20 normal donor women exhibited MSI. Furthermore, we observed a significant and direct correlation between the number of MSI events per sample and the use of chemotherapy, especially with regimens containing alkylating agents. We believe that our data reflect the genotoxic effects of chemotherapy on normal cells because in only one normal control woman out of 20 was there evidence of MSI (after 3 months) and with only one of the two surveyed MSI markers. Furthermore, no LOH was observed within this cohort of normal women. Therefore, because MSI and LOH occurred at significantly greater frequency among patients with breast cancer, it is unlikely that our findings could be due to normal variations in MSI occurrence or to technical artifacts. We also noted a significant, inverse correlation between the number of MSI events per sample and the expression of hMSH2 by immunocytochemistry. Because we noted fluctuation in hMSH2 levels throughout the treatment of several patients, it is possible that hMSH2 downregulation occurs at the post-transcriptional level and could be related to the use of chemotherapy based on alkylating agents. Interestingly, Watanabe and coworkers [ 9 ] described induction of MSI in ovarian tumours from patients studied before and after they received cisplatin-based chemotherapy, but in their study the cisplatin-induced MSI correlated with a significant decrease in hMLH1 expression. We believe that the transient decrease in hMSH2 expression that we noted might have predisposed to an increased number of MSI events. In fact, Zhu and coworkers [ 10 ] also described abnormal hMSH-2 expression in about one-third of cases of AML, mainly in those with treatment-related secondary AML and in patients who were elderly. PCNA is a 36 kDa nuclear protein that functions as a cofactor for δ-DNA polymerase, which is regulated in a cell cycle dependent manner. PCNA expression also increases when cells are actively engaged in DNA repair [ 11 ]. The correlation of PCNA with LOH but not with the number of MSI events found in the present study may indicate that PCNA may have greater expression in situations of more intense DNA damage that tend to be long-lasting. Interestingly, our group previously showed that PCNA over-expression was associated with resistance to chemotherapy AML and chronic lymphocytic leukaemia [ 12 , 13 ]. Conclusion We conclude that systemic chemotherapy may induce MSI and LOH in PBMCs from breast cancer patients receiving chemotherapy regimens, especially chemotherapy regimens containing alkylating agents. Furthermore, it is possible that these alterations are associated with a chemotherapy-induced decrease in the expression of hMSH2. These effects may be related to the generation of secondary leukaemia in some patients, and may also intensify the genetic instability of tumours and increase resistance to treatment. Abbreviations AML = acute myeloid leukaemia; LOH = loss of heterozygosity; MSI = microsatellite instability; PBMC = peripheral blood mononuclear cell; PCNA = proliferating nuclear antigen; PCR = polymerase chain reaction. Competing interests The author(s) declare that they have no competing interests. Supplementary Material Additional File 1 An Excel file showing raw data for each patient studied. Click here for file Additional File 2 A figure showing the results of testing in 20 normal women. The 20 normal women were tested for MSI using the PCR15.1 and TP53Alu markers. Each normal woman's sample corresponds to one letter (e.g. 'A' represents the first sample of the first control woman and 'A1' her second sample, collected 3 months latter). We noted only one case (S and S1) in which the second sample showed an extra band with the marker PCR15.1 (indicated by an arrow). Click here for file

D:\keerthana\PMC001xxxxxx\PMC1064099.xml

1064100

Variants in estrogen-biosynthesis genes CYP17 and CYP19 and breast cancer risk: a family-based genetic association study

Background Case-control studies have reported inconsistent results concerning breast cancer risk and polymorphisms in genes that control endogenous estrogen biosynthesis. We report findings from the first family-based association study examining associations between female breast cancer risk and polymorphisms in two key estrogen-biosynthesis genes CYP17 (T→C promoter polymorphism) and CYP19 (TTTA repeat polymorphism). Methods We conducted the study among 278 nuclear families containing one or more daughters with breast cancer, with a total of 1123 family members (702 with available constitutional DNA and questionnaire data and 421 without them). These nuclear families were selected from breast cancer families participating in the Metropolitan New York Registry, one of the six centers of the National Cancer Institute's Breast Cancer Family Registry. We used likelihood-based statistical methods to examine allelic associations. Results We found the CYP19 allele with 11 TTTA repeats to be associated with breast cancer risk in these families. We also found that maternal (but not paternal) carrier status of CYP19 alleles with 11 repeats tended to be associated with breast cancer risk in daughters (independently of the daughters' own genotype), suggesting a possible in utero effect of CYP19 . We found no association of a woman's breast cancer risk either with her own or with her mother's CYP17 genotype. Conclusion This family-based study indicates that a woman's personal and maternal carrier status of CYP19 11 TTTA repeat allele might be related to increased breast cancer risk. However, because this is the first study to report an association between CYP19 11 TTTA repeat allele and breast cancer, and because multiple comparisons have been made, the associations should be interpreted with caution and need confirmation in future family-based studies.

Introduction Cumulative exposure to circulating estrogen is considered to be of primary importance in breast cancer etiology. Estrogen biosynthesis, cellular binding and metabolism involve many steps, and the genes controlling these steps may contribute to inherent variability in breast cancer susceptibility. Endogenous estrogen is produced predominantly in the ovarian theca cells in premenopausal women and in the breast stromal adipose cells in postmenopausal women. The present study focuses on CYP17 and CYP19 , two key genes that control the biosynthesis of estradiol and estrones from their lipid precursors and are expressed in these cells. CYP17 controls two successive early steps of endogenous estrogen biosynthesis by converting pregnenolone and progesterone to precursors of androgen and estrogen. CYP19, also known as aromatase, controls the terminal step of estrogen biosynthesis by converting 19-carbon steroids (testosterone and androstenedione) to 18-carbon estrogens (estradiol and estrone). A T→C single-nucleotide polymorphism in the 5' promoter region of the CYP17 gene and a TTTA repeat polymorphism in the exon 4–intron 5 boundary region of the CYP19 gene have been investigated in breast cancer by several studies, with inconsistent results [ 1 , 2 ]. For both polymorphisms the variant alleles are considered to be related to an increased biosynthesis of endogenous estrogen. The CYP17 T→C polymorphism is thought to create an Sp1-type (CCACC) promoter site (although one study did not confirm this [ 3 ]) and is associated with an increased serum estrogen level [ 4 , 5 ]. After Feigelson and colleagues first published their study [ 6 ] showing a higher risk of breast cancer in relation to the CYP17 C allele among non-Caucasian women, many other authors attempted to replicate this in other populations. Although some studies confirmed this initial finding, others did not. All studies reporting an increased risk, including the original study, found the increased risk in one or more certain subgroups of women studied, for example women with advanced disease [ 6 ], women aged less than 40 years [ 7 ], women aged less than 40 years with family history [ 8 ], women aged more than 55 years [ 9 ], and women also carrying other genetic polymorphisms [ 10 ]. Two studies found that women carrying CYP17 C allele are less likely to use hormone replacement therapy [ 5 , 11 ] and three studies found that the protective effect of later age at menarche is stronger among women who do not carry the C allele [ 5 , 6 , 12 ]. A recent meta-analysis concluded that the CYP17 T→C polymorphism is not a significant independent risk factor for breast cancer [ 2 ]. The CYP19 gene contains a variable number (range 7–13) of TTTA repeats in the exon 4–intron 5 boundary region, creating polymorphisms that have been examined in five studies [ 13 - 17 ]. Kristensen and colleagues [ 13 ] and subsequently others found a roughly twofold to fourfold elevated risk in relation to certain numbers of CYP19 TTTA repeat polymorphisms. Although one small study found a higher risk in relation to the TTTA seven repeats allele, (TTTA) 7 [ 14 ], most studies reporting an association found elevated risks in relation to one of the higher number of TTTA repeat alleles: 10 repeats, (TTTA) 12 [ 13 ]; 12 repeats, (TTTA) 10 [ 15 , 16 ]; or 10 or more repeats, (TTTA) ≥10 [ 17 ]. A meta-analysis published in 1999 based on some of the earlier studies found that women carrying the CYP19 (TTTA) 10 allele were at higher risk of breast cancer [ 1 ]. All published studies of association between the CYP17 and CYP19 polymorphisms and breast cancer discussed above used a classical case-control design. A recent meta-analysis of CYP17 T→C polymorphism indicates substantial differences in genotype frequencies in case-control studies conducted in different populations [ 2 ], with proportions of carriers ranging from 0.46 in the UK [ 18 ] to 0.79 in Japan [ 19 ] and proportions of homozygotes ranging from 11% in Finland [ 12 ] to 36% in Taiwan [ 10 ]. Similarly, the allele frequency of the CYP19 (TTTA) 10 allele ranges from 0.5% [ 15 ] to 1.8% [ 14 ]. Given that case-control studies can be susceptible to population stratification bias, it is important to examine these potentially important biologically plausible hypotheses in family-based studies that are free from such bias. In this study we examine the association between the CYP17 promoter T→C and CYP19 TTTA repeat polymorphisms and female breast cancer by using a family-based design among nuclear families participating in the Metropolitan New York Registry (MNYR), one of the six international centers of the National Cancer Institute's Breast Cancer Family Registry project. Although other polymorphisms in the CYP17 and CYP19 genes have been reported, we focused on these two polymorphisms because they have been studied most extensively both in relation to their potential associations with breast cancer and also in relation to their influence on circulating estrogens. All published studies focused on the relationship between a woman's own constitutional genotype and her breast cancer risk. A body of recent literature has provided limited data suggesting that a woman's breast cancer risk might be related not only to her own endogenous estrogens during adolescence and adulthood, but also to her prenatal exposure; that is, her exposure in utero to her maternal circulating estrogens [ 20 - 25 ]. In addition to the main association between a woman's own genotype and her breast cancer status, the family-based design of the present study allows us to address this hypothesis indirectly, by examining the association between maternal carrier status of CYP17 or CYP19 gene variants (that is, exposure in utero to an altered level of maternal estrogens) and breast cancer status in daughters. Methods Selection of study participants Since 1995 the MNYR has been recruiting families with breast and/or ovarian cancers in clinical and community settings within the metropolitan New York area. Families meeting one or more of the following criteria are invited to participate: a female less than 45 years of age at diagnosis of breast cancer; a female with both breast and ovarian cancer; three or more relatives with breast or ovarian cancer diagnosed at age 45 years or more, or any male with breast cancer. After identification of a proband he/she is invited to participate in the registry and his/her family's eligibility is assessed. If the family is eligible and the proband agrees to participate, after appropriate informed consent, he/she is interviewed either in person or by phone with an epidemiology questionnaire and a family-history questionnaire. The proband is also asked to provide permission to contact family members. Blood or buccal samples are also collected and participants are provided with a self-administered dietary questionnaire to be returned by mail. Once family members consent to participate, data and blood or buccal samples from the family members are also collected in a similar manner. For members affected with cancer, tumor tissue samples are collected and reviewed pathologically. Genomic DNA from white blood cells or buccal samples has been collected for participants who donated biological samples. So far, the MNYR has enrolled 1158 families and more than 3900 total participants. For this study we restricted attention to nuclear families with at least one affected daughter and at least one parent and/or sibling for whom DNA samples were available. Of the 1158 families enrolled in the MNYR so far, 278 families met these eligibility criteria. Subjects can participate in the MNYR with or without completion of the full epidemiology questionnaire and/or blood samples. There were 1123 family members in the 278 eligible nuclear families, of whom 702 completed the full epidemiology questionnaire and provided blood samples. However, accurate data on relevant variables for the statistical method used in this study (see below) for the remaining 421 members were available from the family-history questionnaire completed by the 702 members. There was 99% concordance in data on age and affected status between women who completed the full epidemiology questionnaire and women who did not. Laboratory analysis We evaluated association between the T→C single-nucleotide polymorphism in the promoter region of the CYP17 gene and the tetranucleotide (TTTA) repeat polymorphism in the exon 4–intron 5 boundary of the CYP19 gene. A total of 23 subjects could not be genotyped for CYP17 , and 26 subjects could not be genotyped for CYP19 . Genotype data were available on a total of 679 members (from 277 nuclear families) for CYP17 and 676 members (from 278 nuclear families) for CYP19 . The CYP17 promoter polymorphism was determined with template-directed primer extension and detection by fluorescence polarization in a 96-microwell-based format [ 26 , 27 ]. In brief, DNA isolated from blood cells by salting out was used for genotyping subjects. First, the target DNA was amplified by polymerase chain reaction (PCR; using forward primer 5'-TTTAAAAGGCCTCCTTGTGC-3' and reverse primer 5'-TTGGGCCAAAACAAATAAGC-3') to generate products in the range 100–200 base pairs. After amplification by PCR, the primers were digested with shrimp alkaline phosphatase and Escherichia coli exonuclease I. Then single-nucleotide extension was performed in the presence of the appropriate allele-specific ddNTPs differentially fluorescence-labeled with either R110 or tetramethylrhodamine purchased from NEN Life Sciences (Boston, MA). For the single-nucleotide extension reaction both forward and reverse probes were tested to select the optimum (the forward probe 5'-GCCACAGCTCTTCTACTCCAC-3') on the basis of clear signal differences. The incorporation resulted in diminished rotation of the fluor compared with the ddNTP. Finally, the fluorescence polarization was read on a fluorescence polarization microplate reader (Tecan Polarion, Research Triangle Park, NC). The reader generates the genotype data on the basis of the distinct separations (with appropriate cut-offs) of the fluorescent intensity values for different alleles in comparison with internal controls. The CYP19 TTTA repeats were determined by PCR amplification (using the forward primer 5'-GTCTATGAATGTGCCTTTTT-3' and the reverse primer 5'-GTTTGACTCCGTGTGTTTGA-3') followed by analysis on an ABI 377 system with GenScan software on the basis of the separations on gel according to the differences in the number of TTTA repeats. All laboratory assays were performed with laboratory personnel blinded to the subject's disease status or family relationships. In addition to assay-specific quality-control samples, 10% of samples were reassayed after relabeling to keep laboratory staff blinded to its identity. Statistical analysis We used the Family Genetic Analysis Program (FGAP [ 28 ], freely available at ) to test the null hypothesis of no association between genotype and breast cancer risk in nuclear families. The FGAP computes two test statistics: the nonfounder statistic (NFS), a generalization of the transmission disequilibrium test (TDT) [ 29 , 30 ], which evaluates transmission disequilibrium from parents to offspring, and the founder statistic (FS), which compares the distribution of parental genotypes with that expected under the null hypothesis of no association. The FGAP statistics fully exploit data from families with variable numbers of affected/unaffected members with variable (known/unknown) patterns of parental genotypes. They are similar to, but can be more powerful than, those available in the software FBAT [ 31 ]. (See [ 32 ] for a comparison of the methods.) On the basis of the previous evidence [ 6 , 13 , 15 , 17 ], we hypothesized that breast cancer risk is elevated among carriers of the CYP17 C allele and the CYP19 variant alleles with 10 or more TTTA repeats, namely the (TTTA) 10 , (TTTA) 11 , (TTTA) 12 , and (TTTA) 13 alleles. The data analysis was focused on two specific components of the study hypotheses: first, whether a woman's carrier status of the hypothesized alleles is associated with her breast cancer status, and second, whether a mother's carrier status of the hypothesized alleles is associated with her daughter's breast cancer risk. For testing the first component of a hypothesis, we applied the FS and NFS to assess whether specific genotypes of each of the studied genes are related to breast cancer. Because FS and NFS follow a normal Gaussian distribution under the null hypothesis, the assessment of statistical significance of the association can be done on the basis of the deviation of these statistics from the standard critical values under normal distribution. For simplicity, we describe these analyses for the CYP17 gene as applied to nuclear families consisting of two parents and at least one daughter. Parents may be untyped and the mother's breast cancer status may be unknown. The test statistics, which are likelihood-based score statistics, are obtained by summing the score contributions from each family. These family-specific scores are obtained in three steps. In the first step we imputed a probability distribution for the genotypes of each pair of parents, conditional on the observed genotypes of all family members. To do this, we obtained maximum-likelihood estimates of the genotypes TT, TC and CC for each of a pair of parents, given the observed genotypes in the family. These estimates do not require the assumption of Hardy–Weinberg frequencies for parental genotypes. If, for example, both parents' genotypes were known, then the probabilities are degenerate at the observed genotypes. Similarly, if both parents' genotypes were unknown but two offspring had observed CYP17 genotypes TT and CC, then the parental distributions are degenerate at TC because both parents must be heterozygotic. In the second step we used the inferred parental genotype distribution and the offspring's observed genotypes to test whether heterozygous parents were equally likely to transmit T and C alleles to affected daughters. This evaluation is based on the NFS. Under the null hypothesis of equal transmission of T and C alleles from parents to affected daughters, the NFS has an asymptotic standard Gaussian distribution. The NFS generalizes the TDT to families with untyped parents and to families with both affected and unaffected daughters. It can be considerably more powerful than the sibling TDT test [ 33 ] when applied to families without unaffected daughters. In the final step we used the inferred parental genotypes (and the mothers' breast cancer phenotypes) in the FS to compare the parental genotype distribution with the expected distribution under the null hypothesis of no association. This statistic treats the affected and unaffected mothers like cases and controls in a case-control study. However, each parent's contribution is weighted in proportion to his/her number of affected and unaffected daughters, so that parents of many affected daughters receive higher weights than do those of few affected daughters. To test the second component of our hypothesis, namely the association between maternal carrier status and daughter's breast cancer status, we evaluated whether the genotypes of mothers with more affected daughters differ from those of mothers with less affected daughters. Such deviation might be expected if some aspect of a daughter's environment in utero , governed by the mother's genotype, influences the daughter's risk of subsequent breast cancer development. The FS was adapted to evaluate this question by comparing the observed or imputed genotypes of mothers of affected daughters with the genotypes expected in the parental population. It is a weighted sum of differences between each mother's observed (or inferred) C allele count and the average C count in the population. In symbols, each family's contribution to this sum is proportional to the quantity ( n A - n U )( C obs - C exp ), where n A and n U are,, respectively, the numbers of affected and unaffected daughters in the family, and C obs and C exp are the observed and expected C-allele counts for the mother. Under the null hypothesis of no association between maternal genotype and daughters' breast cancer risks, C obs has a mean value C exp , so C obs - C exp = 0 in expectation for all families. Thus the FS has expectation zero and the correct type I error rate regardless of the actual numbers of affected and unaffected daughters in each family. Under the alternative hypothesis that maternal C-allele count is associated with daughters' breast cancer risks, one expects that C obs - C exp > 0, and thus families with many affected daughters and few unaffected daughters (that is, n A - n U >> 0) contribute larger values to the FS than those with few affected daughters or those with many unaffected daughters. A statistically significant value of the FS when restricted to the mothers (with an insignificant value when restricted to the fathers) would provide evidence for this association. When the null hypothesis is rejected, it is useful to estimate a measure of association between genotype and risk, such as the odds ratio, and to evaluate the effects of potential confounding by hormonal factors. To do so, we also performed conditional logistic regression analyses [ 34 , 35 ] on all the available sibships containing at least one affected sibling and at least one unaffected sibling who had provided blood samples and relevant epidemiology data for statistical adjustment (165 sibships for CYP17 and 169 sibships for CYP19 ). Results Of the 277 nuclear families eligible for CYP17 analyses, 229 were Caucasian, 4 were African American, 41 were Hispanic, and 3 were Asian American. Of the 278 nuclear families eligible for CYP19 analyses, 229 were Caucasian, 4 were African American, 42 were Hispanic, and 3 were Asian American. Table 1 shows the distribution of the study subjects according to CYP17 and CYP19 genotypes, by family position and breast cancer status. The numbers in each cell represent the number of specific type of family members in our study population carrying a particular genotype. The number of TTTA repeats in intron 4 of the CYP19 gene ranged between 7 and 13 in our study population, with the (TTTA) 7 and (TTTA) 11 alleles being the most frequent (allele frequencies 53.9% and 28.8%, respectively). These frequencies are consistent with those found in Caucasian populations in other studies in the USA [ 15 ]. The frequency of the CYP17 variant C allele was 42.8% in this study population, which is similar to that found in other studies conducted in Caucasians [ 4 ]. The distribution of the nuclear families according to mother's and father's carrier status and mother's and daughter's affected status is presented in Table 2 . A majority (about 55%) of the nuclear families contained one affected and one unaffected daughter. The majority of the nuclear families had one or more parents who did not have the genotyping information available. Table 3 presents the FS and NFS for testing the associations between the a priori hypothesized CYP17 and CYP19 variant alleles and breast cancer. Each test statistic has an approximately standard Gaussian distribution under the null hypothesis of no association between genotype and breast cancer risk. A positive value of a NFS reflects excess transmission of the variant allele to affected daughters, and a negative value represents fewer such transmissions than expected under the null. Thus a test statistic that is negative but large in absolute value would suggest that the variant allele is associated with reduced risk. We computed the FS and NFS under recessive, dominant, and additive models. For the dominant models, the number of affected daughters carrying one or more copies of the variant alleles was compared with that expected from the parental genotypes in accordance with Mendelian expectation. Similarly, for the recessive models, the number of affected daughters homozygous for the variant allele was compared with that expected under Mendelian expectation. For the additive models, the total variant allele count in affected daughters was compared with that expected from the parental genotypes in accordance with Mendelian expectation. On the basis of the literature, we hypothesized a priori that CYP19 alleles with 10 or more TTTA repeats would be associated with breast cancer. In addition, we examined the association between the CYP19 genotype and breast cancer by defining the variant allele(s) by treating each of the 10 or more repeat alleles, (TTTA) 10 , (TTTA) 11 , (TTTA) 12 and (TTTA) 13 , separately as the variant allele under each of the three models (realizing that this might have increased the chance of our finding of a statistically significant association; see the Discussion section). As seen in Table 3 , the NFS for association between the (TTTA) 11 allele and breast cancer under the dominant model is 1.83, which is higher than the critical value (1.65) for a one-tailed test statistic, suggesting that affected daughters were more likely to receive the (TTTA) 11 allele from their parents (irrespective of their ethnic distribution) than unaffected daughters. Like the NFS, the FS was also statistically significant under the dominant model, supporting an association between the CYP19 (TTTA) 11 allele and breast cancer among the parents in these families. The results for CYP19 TTTA ≥10 alleles did not show a consistent association, because only the FS was statistically significant under the dominant model. None of the other specific CYP19 alleles showed a consistent association with breast cancer on the basis of the NFS and FS (results not shown). Although the FS found an association between the CYP19 (TTTA) 13 allele and breast cancer, this was not supported by the more robust NFS (results not shown). Neither the FS nor the NFS suggested any significant association between the CYP17 variant C allele and breast cancer, under any of the models of FGAP analyses (see Table 3 ). Table 4 presents the results of conditional logistic regression analysis comparing the CYP17 and CYP19 genotypes between affected and unaffected sisters. These results, adjusted for age (in years), hormone replacement use (ever/never), oral contraceptive use (ever/never), age at menarche (in years) and term pregnancies (yes/no), are similar to the FGAP results although because of the smaller number of available sibships the associations did not achieve statistical significance. As seen in Table 4 , carriers of the CYP19 (TTTA) 11 allele had an increased risk of breast cancer (odds ratio 1.8; 95% confidence interval 0.9–3.5). Table 5 presents results relating maternal and paternal carrier statuses for the variants of estrogen-biosynthesis genes CYP17 and CYP19 to breast cancer risk in daughters. Mothers of affected daughters were more likely to carry the CYP19 (TTTA) 11 allele than expected in the parental population. There were no such associations between the paternal carrier status of (TTTA) 11 and any of the other CYP19 alleles and breast cancer in daughters. For this hypothesis, the findings for analysis involving CYP19 (TTTA) ≥10 corroborated that for (TTTA) 11 alleles. Although maternal carrier status of the CYP17 C allele tended to be positively associated with daughter's breast cancer, this association was not specific to the mothers but was also present among the fathers. Discussion Despite a sound biological basis for the role of estrogen-biosynthesis genes in breast cancer, the findings of studies investigating the relationship between these genes and breast cancer have not been consistent. Employing a case-control design, many of these prior studies, especially those examining the CYP17 gene–breast cancer relationships, produced conflicting results. Although in comparison with CYP17 a smaller number of studies investigated the association of breast cancer with CYP19 , findings for CYP19 have been more consistent, with most studies showing a positive association between CYP19 alleles with a higher number (10, 12, or 10 or more) of TTTA repeats and breast cancer [ 13 , 15 - 17 ]. Using a family-based design we investigated the relationships between the CYP17 and CYP19 gene variants and breast cancer in families participating in the MNYR. Like many of the previous case-control studies, the present study did not find any association between the CYP17 C (variant) allele and breast cancer. However, our findings support an association between certain alleles of the CYP19 intron 4 TTTA repeat polymorphism and breast cancer. On the basis of the previous studies we defined each of the CYP19 alleles with 10, 11, 12, or 13 TTTA repeats as the 'variant' allele and examined each association with breast cancer. Unlike some of the previous case-control studies we did not find the CYP19 (TTTA) 10 or (TTTA) 12 alleles to be associated with breast cancer. However, we found the CYP19 (TTTA) 11 allele to be significantly associated with breast cancer in these nuclear families, under a dominant model. Although we also observed a significantly positive association between the CYP19 (TTTA) 13 allele and breast cancer among the parents in these families, we did not observe excess transmission from parents to affected daughters, suggesting that the association might be due to chance or bias. The evidence of an increased risk in relation to the CYP19 (TTTA) 11 allele was also observed in the conditional logistic regression analysis adjusting for potential confounding variables among the subset of families containing discordant sibships. However, because of the reduced power of these analyses among only a subset of families [ 36 ], results of these discordant sibship analyses did not achieve statistical significance. In addition to evaluating associations between a woman's breast cancer risk and her own constitutional genotype, we also evaluated whether maternal genotypes are associated with the breast cancer risk in the daughters (independent of the daughter's own genotype). We found that the maternal (but not the paternal) genotypes of the CYP19 (TTTA) 11 allele conferred a non-significantly elevated breast cancer risk to daughters. This effect was also observed when all (TTTA) ≥10 alleles were treated as the variant allele. This association is consistent with evidence from the previous literature on the association between exposure to hormonal factors in utero and breast cancer risk in adulthood [ 20 ]. Although the association might be due to chance, if confirmed in subsequent studies it will have important implications in advancing our understanding of the breast cancer etiology. Some limitations of the present study merit consideration. The major limitation concerns statistical power. The analysis, which is based on 287 nuclear families, might not have had enough power to detect small increases in risk associated with certain of the CYP17 genotypes. For example, we lacked power to evaluate interactions between genotypes for CYP17 and CYP19 and both endogenous and exogenous hormonal characteristics, such as age at menarche, timing and number of pregnancies and the use of exogenous hormones. In addition, although there is evidence for variations in the allele frequencies of the studied polymorphisms across ethnic groups, we lacked statistical power to conduct ethnicity-specific analyses. The evaluation of such analyses will be the subject of a separate future analysis, based on additional numbers of Breast Cancer Family Registry families. Although the hypotheses examined in this study are not novel, the study design (which is free from population stratification bias) and the analytical approach have not been applied to these hypotheses in previous studies. Several limitations of this study require caution when interpreting the findings. First, the selection of nuclear families participating in this study from the MNYR was not population-based. Although this might limit the generalizability of the findings it should not affect the validity of the observed associations. Second, although it is possible for variations in the number of nucleotide repeats in hormone-related genes to be associated with cancer risk, such an association is less plausible biologically for the TTTA repeat numbers in the CYP19 gene. This is because the TTTA polymorphism is in the intronic region of the gene and so it is less likely that the variant alleles of the gene are directly associated with the functional status of endogenous estrogens in the body. Nevertheless, it is possible that one or more of the CYP19 TTTA alleles, including the (TTTA) 11 allele, are in linkage disequilibrium with other functionally relevant alleles, as suggested by other studies [ 16 ]. Third, the present study compared multiple CYP19 TTTA alleles with breast cancer under different models. Although it is possible that multiple comparisons might have led to the observed associations, the consistency of the associations involving the CYP19 (TTTA) 11 allele across both parents and transmission to offspring as well as the similarity between the associations with both constitutional and maternal genotypes suggest that these findings might have a biological basis. Further, the fact that the association was observed under specific susceptibility models and was consistent with conditional logistic regression analysis might be suggestive of the specificity of the finding. Conclusion This family-based study found that the CYP19 (TTTA) 11 allele is associated with breast cancer risk among families participating in a breast cancer family registry. The study also suggests that maternal carrier status of the CYP19 (TTTA) 11 allele might be associated with breast cancer in daughters in these families. These associations might have important implications for understanding the etiology and risk prediction of breast cancer. However, because this is the first study to report an association with the CYP19 (TTTA) 11 allele, and because multiple comparisons have been made, the associations reported in this study should be interpreted with caution and need to be confirmed in future family-based studies. Abbreviations FGAP = Family Genetic Analysis Program; FS = founder statistic; MNYR = Metropolitan New York Registry; NFS = nonfounder statistic; PCR = polymerase chain reaction; TDT = transmission disequilibrium test. Competing interests The author(s) declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064100.xml

1064101

Histopathological features of breast tumours in BRCA1, BRCA2 and mutation-negative breast cancer families

Introduction Histopathological features of BRCA1 and BRCA2 tumours have previously been characterised and compared with unselected breast tumours; however, familial non- BRCA1/2 tumours are less well known. The aim of this study was to characterise familial non- BRCA1/2 tumours and to evaluate routine immunohistochemical and pathological markers that could help us to further distinguish families carrying BRCA1/2 mutations from other breast cancer families. Methods Breast cancer tissue specimens ( n = 262) from 25 BRCA1 , 20 BRCA2 and 74 non- BRCA1/2 families were studied on a tumour tissue microarray. Immunohistochemical staining of oestrogen receptor (ER), progesterone receptor (PgR) and p53 as well as the histology and grade of these three groups were compared with each other and with the respective information on 862 unselected control patients from the archives of the Pathology Department of Helsinki University Central Hospital. Immunohistochemical staining of erbB2 was also performed among familial cases. Results BRCA1 -associated cancers were diagnosed younger and were more ER-negative and PgR-negative, p53-positive and of higher grade than the other tumours. However, in multivariate analysis the independent factors compared with non- BRCA1/2 tumours were age, grade and PgR negativity. BRCA2 cases did not have such distinctive features compared with non- BRCA1/2 tumours or with unselected control tumours. Familial cases without BRCA1/2 mutations had tumours of lower grade than the other groups. Conclusions BRCA1 families differed from mutation-negative families by age, grade and PgR status, whereas ER status was not an independent marker.

Introduction Women predisposed to hereditary or familial breast cancer form a heterogeneous group. It would be useful if we could identify carriers of the high-risk BRCA1 and BRCA2 genes and target the expensive and time-consuming genetic testing to individuals who most probably carry those mutations. Besides family history, histopathological markers could also be useful in distinguishing patients and families likely to carry a BRCA1/2 germline mutation from mutation-negative families and breast cancer patients in general. Several studies have compared the characteristics of breast cancers in BRCA1 carriers and in sporadic controls. Distinct features between BRCA1 -associated tumours have been found, such as high tumour grade, oestrogen receptor (ER) negativity, and overexpression of p53 [ 1 - 3 ]. In addition, negativity for progesterone receptor (PgR) [ 3 , 4 ], a higher proportion of medullary and atypical medullary carcinomas [ 5 , 6 ], and tumours with a low expression of c-erbB-2 [ 1 , 4 , 6 ] have been detected. Besides the higher proportion of medullary histology, a higher frequency of ductal carcinoma has also been reported [ 7 , 8 ]. Recently, cDNA expression analyses have suggested a basal epithelial phenotype for BRCA1 tumours [ 9 ] and expression of cytokeratins 5/6 has been associated with BRCA1 tumours [ 10 ]. Among BRCA2 -associated tumours, a slight increase has been observed in the incidence of lobular or tubulolobular carcinomas [ 11 , 12 ]. However, results are inconsistent, and in most cases no significant difference has been found between BRCA2 -associated tumours and sporadic cancers [ 1 , 4 , 6 , 13 ]. Complementary DNA (cDNA) expression analysis has suggested distinct expression profiles for BRCA2 tumours as well [ 14 ]. It has been clear for some time that in many families hereditary susceptibility is not due to the BRCA1 or BRCA2 genes [ 15 , 16 ]. Only a few studies have evaluated the features of this large group of families. It would be crucial for genetic counselling and for our understanding of tumour development to learn more about these patients. Lakhani and colleagues [ 17 ] studied the pathology of 82 familial breast cancers not attributable to BRCA1 or BRCA2 mutations, and they found these non- BRCA1/2 cancers to be of lower grade, to show less pleomorphism and to have a lower mitotic count than sporadic cancers or BRCA mutation-positive cancers. No other features differed significantly in their study. However, they did not examine immunohistochemical characteristics. There has been only one study so far that has characterised the immunohistochemical features of familial non- BRCA1/2 cancers: Palacios and colleagues [ 18 ] studied immunohistochemical staining and histopathology and compared 37 non- BRCA1/2 cancers with 20 BRCA1 -associated and 18 BRCA2 -associated cancers, and also with unselected control cancers. They similarly found those to be of lower grade than in all the other three groups. In comparison with sporadic cancers they were also more frequently p53-negative and erbB2-negative, and expressed reduced E-cadherin and β-catenin. However, the number of patients in this study was small and it was restricted to only a univariate analysis of the studied parameters. Most of the previous studies on BRCA1 -associated and BRCA2 -associated cancers have studied highly selected patient groups, but in the present study families were collected with a simple criterion of at least three first-degree or second-degree relatives with breast or ovarian cancer with no restriction on age. We studied an extensive material of 152 non- BRCA1/2 tumours, and also 110 tumours from BRCA1/2 families for histopathological features as well as for the immunohistochemical expression of ER, PgR, p53 and erbB2. We describe here the histopathological profile of the tumours originating from non- BRCA1/2 breast cancer families and also present a multivariate analysis to find the independent markers that can further help in distinguishing especially BRCA1 mutation-positive families from other familial cases. Patients and methods Familial breast cancer patients were identified and collected by a systematic screening for family history at the Department of Oncology, Helsinki University Central Hospital, as described previously [ 19 ]. We defined breast cancer families by the selection criterion of at least three first-degree or second-degree relatives with breast or ovarian cancer (including the proband). We confirmed the genealogy of the families through population registries, and cancer diagnoses through the Finnish Cancer Registry. In this study we included 25 BRCA1 families, 20 BRCA2 families and 74 families not associated with either of these genes (non- BRCA1/2 families) (Table 1 ). All families had previously been tested for BRCA1 and BRCA2 mutations by mutation analysis of the whole coding sequences and exon/intron boundaries of the genes as described [ 16 , 20 ] or tested for all previously reported 18 Finnish BRCA1 and BRCA2 mutations [ 16 , 20 - 22 ]. We collected all the paraffin blocks of all the primary breast cancers that were available ( n = 262) from these families. However, cases tested to be non-carriers in the mutation-positive families were excluded. In total, 51 cancers from the 25 BRCA1 families, 59 cancers from the 20 BRCA2 families and 152 cancers from the 74 non- BRCA1/2 families were included in this study. We studied the haematoxylin and eosin sections of the original blocks to achieve histological diagnosis and grading. Grading was performed according to Scarff-Bloom-Richardson modified by Elston and Ellis [ 23 ]. The most representative area of the tumour was punched to produce a hereditary breast cancer tissue microarray including two cores (diameter 0.6 mm) from all of the original blocks. The array block of non- BRCA1/2 cases was described previously by Vahteristo and colleagues [ 24 ]. All of the microarray slides were stained with routine methods by antibodies against ER, PgR, p53 and erbB2 in the same pathology laboratory as our controls. In brief, 5 μm sections were cut from paraffin-embedded blocks, deparaffinated in xylene, and dehydrated in a series of graded alcohols. The sections were pretreated in a microwave oven and incubated overnight with antibody. Antibodies for ER (dilution 1:50) and c-erbB-2 (NCL-CB11; dilution 1:400) were purchased from Novocastra (Newcastle upon Tyne, UK), and those for PgR (dilution 1:250) and p53 (dilution 1:100) were from Dako (Copenhagen, Denmark). The evaluation of the staining results was similar to that used in routine diagnostics and samples were considered positive when 10%, 10% and 20% of the cells were stained for ER, PgR and p53, respectively. Samples having a moderate or intense staining of the entire membrane in more than 10% of the tumour cells (2+ and 3+) were considered to be c-erbB-2-positive. Other staining patterns were considered to be negative (0 and 1+). As a control group we drew from the archives of the Pathology Department at Helsinki University 862 unselected breast cancer tumours from the years 1997–2001 that were scored by the same pathologist (PH) as the tumours of the hereditary breast cancer patients. Statistical analysis was conducted with SPSS version 8.0 for Windows. We tested the differences in continuous variables by the Mann–Whitney test and in dichotomous variables by the χ 2 test or Fisher's exact test. In multivariate analysis, we used logistic regression analysis (stepwise backwards logistic regression, 99%). All P values are two-sided. Permissions for this study were obtained from the ethics committees of the Department of Oncology and the Department of Obstetrics and Gynaecology, Helsinki University Central Hospital, and of the Ministry of Social Affairs and Health in Finland. Blood and tumour samples were used in this study with written informed consent from probands and family members. Results BRCA1 -associated cancers were diagnosed at a younger age than unselected breast cancers (median ages 44 and 56 years, respectively; P ≤ 0.0005) and they were more often ER-negative and PgR-negative, p53-positive and of higher grade (Table 2 ). The frequency of medullary histology was also higher. In logistic regression analysis, taking into account all of these factors, the independent factors were age ( P ≤ 0.0005), ER status ( P = 0.0597), PgR status ( P = 0.0170) and medullary histology ( P = 0.0636). In comparison with familial non- BRCA1/2 cancers (median age 55 years), a univariate analysis of BRCA1 tumours showed similar differences from unselected cancers (Table 2 ). However, in multivariate analysis, taking into account the same factors, the independent factors were age ( P = 0.0012), grade ( P = 0.0014) and PgR negativity ( P = 0.0196) (Table 3 ). We did not find any significant differences between groups of tumours from carriers with different mutations or when comparing tumours from the carriers of the mutations at the 5' end or the 3' end of the gene. BRCA2 -associated cancers were diagnosed at younger age (median age 47 years) than unselected breast cancers (median age 56 years, P ≤ 0.0005). They were also more often ER-negative and PgR-negative (Table 2 ). In multivariate analysis, the independent factors were age ( P ≤ 0.0005), PgR status ( P = 0.0365) and p53 status ( P = 0.0318). When compared with familial non- BRCA1/2 cancers, the BRCA2 -associated cancers were diagnosed at a younger age (median age 47 years; among non- BRCA1/2 patients the median age was 55 years; P ≤ 0.0005). No other variable differed significantly from non- BRCA1/2 cancers (Table 2 ). We did not find any significant differences between tumours originating from different mutation carriers or when comparing tumours from the carriers of mutations from the OCCR (Ovarian Cancer Cluster region) ( n = 7) and the end of the gene. In a logistic regression analysis comparing BRCA2 -associated cancers with non- BRCA1/2 cancers and taking into account age, grade and all tested histological and immunohistochemical factors, the only independently significant factor was age ( P = 0.0001) (Table 2 ). Among familial non- BRCA1/2 patients, the median age of onset was marginally younger (55.0 years) than among the unselected controls (56.0 years, P = 0.060). The frequency of grade I and II tumours was much higher among the non- BRCA1/2 group than in the unselected control group (odds ratio 1.8, P = 0.009) (Table 2 ) or in the BRCA1 and BRCA2 groups. In a multiple regression analysis comparing non- BRCA1/2 tumours with unselected controls, grade (odds ratio 0.54, P ≤ 0.00005) and ER status (odds ratio 2.4 for negative ER status, P = 0.0006) were the independent significant factors. The erbB2 results were very similar in the three groups of familial cases, with 18.6%, 15.1% and 17.4% of the BRCA1 , BRCA2 and non- BRCA1/2 tumours, respectively, expressing the erbB2 antigen (Table 2 ). Discussion Although genetic testing for BRCA1 and BRCA2 mutations is available, it is expensive, time-consuming and stressful to patients. Several models have therefore been developed for evaluating the probability of carrying a BRCA1 or BRCA2 mutation [ 25 - 30 ]. In our previous study of Finnish breast cancer families [ 15 , 29 ], multivariate analysis suggested simple family history criteria for breast cancer onset under the age of 40 years and the presence of ovarian cancer to be most strongly associated with BRCA1/2 mutation status [ 29 ]. However, in addition to family history, it is important to find other markers that could help to identify mutation carriers. In many countries, markers that are already routinely used, such as ER, PgR, p53 and the grade of the tumour, could serve as an excellent tool to aid distinguishing families because this information is easily available. In this study, we found that BRCA1 -associated cancers have a different histological profile and can be distinguished from other familial cancers. Specifically, multivariate analysis revealed age, grade and PgR negativity as the independent factors distinguishing BRCA1 tumours from familial non- BRCA1/2 tumours. BRCA2 -associated cancers did not differ greatly, but those were also significantly younger and there was a trend towards higher grade than among familial non- BRCA1/2 cancers. ER negativity has previously been highlighted to be linked to BRCA1 tumours. It was obvious in this study too. Our results are consistent with, for example, the study of Lakhani and colleagues [ 4 ], in which BRCA1 cancers were clearly more ER-negative (90%) than unselected control cancers (35%). However, in our study the overall percentages of ER-negative cancers were much lower for both groups (67% among BRCA1 and 19% among unselected controls). This might be due to a different age distribution because the prevalence of ER-negative cases is higher among young breast cancer patients [ 31 , 32 ] and our familial cases were not selected on the basis of young age of diagnosis. Furthermore, there is a very high coverage mammography screening (in the general age group 50–60 years) in Finland, aiding in finding asymptomatic cancers, which are more often ER-positive as well as being of lower grade and PgR-positive [ 33 ]. In this study, we specifically sought to compare BRCA1 cancers with familial non- BRCA1/2 cancers as well, which is the relevant question in a clinical and genetic counselling setting. Surprisingly, in this analysis the ER status was not an independent marker in multiple regression. More important markers were age, PgR status and grade. ER status was dependent on age and grade. Palacios and colleagues [ 18 ] also compared BRCA1 cases with non- BRCA1/2 cases (cases from families with at least three affected with breast cancer, one of them diagnosed under 50 years of age). In a univariate analysis, the results showing ER and PgR negativity, p53 overexpression and high grade were very similar to ours. However, multivariate analysis was not used for evaluating independent markers. Recently, cDNA expression analyses have suggested a basal epithelial phenotype for BRCA1 tumours [ 9 ]. Epithelial phenotype is associated with breast cancers that express neither ER nor erbB2, a feature that also occurs in BRCA1 -mutation carriers [ 10 ]. A large majority (20 of 22 tumours) of BRCA1 -associated tumours were also ER-negative and erbB2-negative in our study, in accordance with the high frequency of ER-negative tumours among BRCA1 carriers. The frequency of erbB2 expression was similar in all three groups of familial tumours. BRCA2 -associated tumours did not differ significantly from familial non- BRCA1/2 tumours, although they were diagnosed at an earlier age. Our results on BRCA2 cancers were quite similar to those of Palacios and colleagues [ 18 ] and Lakhani and colleagues [ 17 ], although in the former study the BRCA2 cancers were more ER-positive and PgR-positive (on the basis of smaller sample set of 14 cases). BRCA1 and BRCA2 mutations have been detected in quite a low proportion of breast cancer families. In Finland, among families with three affected first-degree or second-degree relatives without age restrictions, the proportions were 10% and 11% for BRCA1 and BRCA2 , respectively [ 16 , 20 ]. Thus an important and large group of families is not due to mutations in the BRCA1 and BRCA2 genes. In this study, the familial non- BRCA1/2 cancers were diagnosed at a marginally younger age than those among unselected cases, and were more often of lower grade than the control cancers or BRCA1 and BRCA2 cancers. These factors might be influenced by recall bias because patients having affected relatives undergo diagnostic procedures earlier than women with no family history of breast cancer. However, no such influence is seen for BRCA1 or BRCA2 cases, which often have a much stronger family background of cancer. Furthermore, a lower grade among familial non- BRCA1/2 cases has been detected elsewhere as well [ 17 , 18 ]. In comparison with BRCA1 or BRCA2 cancers, familial non- BRCA1/2 cancers represented a much greater number of ER-positive and PgR-positive cases; however, in comparison with the unselected control group they were more receptor-negative. The frequency of ER-positive cancers among our control cancers is quite high, which might be due to a later year of diagnosis than for non- BRCA1/2 cancers and in general because of age distribution [ 34 ], mammography screening [ 33 ] or ethnic differences [ 35 ] in different populations. Conclusions In this report we have characterised familial non- BRCA1/2 tumours and evaluated routine immunohistochemical and pathological markers that could help us to further distinguish families carrying BRCA1/2 mutations from other breast cancer families. It is noteworthy here that, although ER negativity has been considered a hallmark of BRCA1 tumours, logistic regression analysis indicated that this was not an independent marker but was dependent on the age of diagnosis and tumour grade. When considering the possibility of mutation testing in the context of genetic counselling, for instance, it would be important to consider the tumour characteristics specifically in comparison with those from other breast cancer families. It also seems crucial to consider the histopathological features with regard to the age of the patients. In this study, the independent markers that distinguished BRCA1 carrier tumours from familial non- BRCA1/2 tumours were earlier age of diagnosis, negative PgR status and higher grade. These immunohistochemical and pathological characteristics of the tumours, which are available in routine pathological diagnostics, should be of value in evaluating the possibility of mutation and in targeting mutation screening in such families, especially when considering the characteristics of several tumours in the family and combined with the family history of cancer. Abbreviations cDNA = complementary DNA; ER = oestrogen receptor; PgR = progesterone receptor. Competing interests The author(s) declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064101.xml

1064102

Breast cancer oestrogen independence mediated by BCAR1 or BCAR3 genes is transmitted through mechanisms distinct from the oestrogen receptor signalling pathway or the epidermal growth factor receptor signalling pathway

Introduction Tamoxifen is effective for endocrine treatment of oestrogen receptor-positive breast cancers but ultimately fails due to the development of resistance. A functional screen in human breast cancer cells identified two BCAR genes causing oestrogen-independent proliferation. The BCAR1 and BCAR3 genes both encode components of intracellular signal transduction, but their direct effect on breast cancer cell proliferation is not known. The aim of this study was to investigate the growth control mediated by these BCAR genes by gene expression profiling. Methods We have measured the expression changes induced by overexpression of the BCAR1 or BCAR3 gene in ZR-75-1 cells and have made direct comparisons with the expression changes after cell stimulation with oestrogen or epidermal growth factor (EGF). A comparison with published gene expression data of cell models and breast tumours is made. Results Relatively few changes in gene expression were detected in the BCAR-transfected cells, in comparison with the extensive and distinct differences in gene expression induced by oestrogen or EGF. Both BCAR1 and BCAR3 regulate discrete sets of genes in these ZR-75-1-derived cells, indicating that the proliferation signalling proceeds along distinct pathways. Oestrogen-regulated genes in our cell model showed general concordance with reported data of cell models and gene expression association with oestrogen receptor status of breast tumours. Conclusions The direct comparison of the expression profiles of BCAR transfectants and oestrogen or EGF-stimulated cells strongly suggests that anti-oestrogen-resistant cell proliferation is not caused by alternative activation of the oestrogen receptor or by the epidermal growth factor receptor signalling pathway.

Introduction The development and progression of breast cancer is dependent on steroid sex hormones and polypeptide growth factors. Oestrogen action has been implicated in the development of breast cancers and frequently contributes to tumour growth. The action of oestrogen is relayed through its specific nuclear oestrogen receptor (ER), which belongs to the family of ligand-inducible transcription factors [ 1 , 2 ]. Two distinct genes for ER ( ERα and ERβ ) have been identified. The role of ERα in breast cancer has been studied extensively and this receptor has been the subject of targeted therapies. Less information is available for ERβ [ 3 , 4 ], which exhibits differential tissue distribution and alternative responses to selective ER modulators [ 5 , 6 ]. Epidermal growth factor receptor (EGFR) expression is mainly present in ERα-negative breast tumours and is a marker of poor prognosis [ 7 , 8 ]. The frequent occurrence of ERα in breast tumours (about 75%) has been used as a guide for treatment. Endocrine treatment regimens, which either reduce the endogenous oestrogen levels (for example aromatase inhibitors) or interfere with ERα activation (anti-oestrogen such as tamoxifen), have been shown to block tumour growth and in some cases cause tumour reduction or disappearance [ 9 , 10 ]. However, the resistance of ERα-positive breast tumours is a severe limitation of endocrine treatment. About half of the ERα-positive breast tumours completely fail to respond (intrinsic resistance), whereas all responsive breast cancers ultimately progress and become resistant to the treatment (acquired resistance). Despite much effort, the basis for the resistance of breast cancer to endocrine treatments is still poorly understood [ 11 ]. In general, tamoxifen resistance is not accompanied by loss of ERα expression [ 12 , 13 ]. Previous work has shown that the treatment outcome might be the result of a delicate balance between positive and negative regulators acting in concert with the hormone receptor [ 2 , 14 ]. In addition, alternative growth regulatory pathways might be available to tumour cells, permitting escape from the treatment [ 15 ]. We have searched for specific Breast Cancer Anti-oestrogen Resistance ( BCAR ) genes involved in the progression of oestrogen-dependent breast cancer cells to anti-oestrogen resistance [ 16 ]. The BCAR3 gene was shown to control anti-oestrogen-resistant cell growth in two different oestrogen-dependent cell lines and its product exhibits features of a cytoplasmic signalling molecule [ 17 ]. The BCAR1 gene causes anti-oestrogen resistance in our cell model and is the human homologue of the rat Crk-associated substrate ( p130Cas ) gene [ 18 ]. This docking protein has been implicated in many types of intracellular signalling processes [ 19 , 20 ]. Moreover, studies of human breast cancer specimens have shown that high BCAR1 expression is associated with poor prognosis and also predicts a poor response of recurrent disease to treatment with tamoxifen [ 21 - 24 ]. Recent developments in global gene expression profiling have elegantly shown their applicability in tumour classification and in predicting the prognosis of the patient [ 25 - 30 ]. Furthermore, studies in model systems have highlighted the use of gene expression profiling for unravelling delicate cellular processes. The aim of our study was to use gene expression profiling to investigate the anti-oestrogen-resistant growth regulatory process induced by overexpression of the BCAR genes and to establish whether oestrogen or epidermal growth factor (EGF) signalling are involved. Materials and methods Breast cancer cell lines cultures and RNA preparation The oestrogen-dependent human breast cancer cell line ZR-75-1 was maintained in RPMI 1640 medium supplemented with 10% bovine calf serum and 1 nM 17 β-oestradiol (R/BCS/E2) as described previously [ 16 ]. The derived EGFR-transfectant cell line ZR/HERc(1A) [ 31 ], hereafter referred to as ZR/EGFR, a BCAR1-transfectant cell line (4A12) [ 18 ] and a BCAR3-transfectant cell line (B3-10) [ 17 ] were also maintained in R/BCS/E2 medium. For short-term induction experiments, cells were cultured for 4 days in regular medium lacking added oestrogen (R/BCS) in 162 cm 2 flasks, given fresh R/BCS medium 24 hours before manipulation, and cultured for 6 hours in the presence of 100 nM oestradiol or 1 μM ICI 164,384 (or ethanol vehicle alone) in R/BCS medium. Hormones and anti-hormones were provided by N.V. Organon (Oss, The Netherlands). For long-term induction experiments, cells were grown for 7 days in R/BCS/E2 medium or R/BCS medium containing 10 ng/ml EGF (Roche Diagnostics Nederland B.V., Almere, The Netherlands). Medium was replaced after 3 days and at 24 hours before harvest. After completion of the culture, the medium was removed and cells were lysed directly with 16–20 ml of RNAzol B solution (Campro Scientific, Veenendaal, the Netherlands). RNA was prepared as described by the manufacturer, quantified and checked for integrity on agarose gels. Poly(A) + mRNA was prepared from pooled total RNA samples of two independent cultures by two cycles of binding to oligo(dT) with the use of OligoTex (Qiagen/Westburg B.V., Leusden, The Netherlands), and checked for integrity and contamination with ribosomal RNA by capillary electrophoresis (Lab-on-a-Chip, Agilent Technologies 2100 bioanalyzer, Amstelveen, The Netherlands). Expression analysis Production of Cy5-labelled and Cy3-labelled cDNA from the purified mRNA, hybridisation of the UniGEMV cDNA microarrays and quantification of the signals were performed by Incyte Genomics (Mountain View, CA, USA) as described previously [ 32 ]. Two batches of UniGEMV2 microarrays were used for these experiments. All hybridisations (Table 1 ) were performed in duplicate with a fluor reversal to minimise possible bias caused by the molecular structure of the Cy3 and Cy5 dyes. Data analysis was performed with the Rosetta Resolver software package (v 3.2) with an Incyte/UniGEM microarray error model (Rosetta Inpharmatics Inc., Kirkland, WA, USA). Genes exhibiting at least once a significant difference ( P ≤ 0.01) in expression in these experiments were used for further analysis ( n = 2373). In addition to the actual measured gene expression ratios we calculated the expression ratios between different experimental conditions from two measurements that contained a common sample [ 33 ]. Because the calculated gene expression ratios were in good agreement with available actual measurements, we used these calculations as 'virtual experiments'. For hierarchical clustering of the measured and calculated expression ratios, we used Resolver software (average linkage agglomerative clustering using Euclidean distance and weighted by error) and Spotfire Decision Site 7.1 analysis package (Spotfire Inc., Somerville, MA, USA) using the Unweighted Paired-Group Method with Arithmetic mean (UPGMA) and Pearson's correlation as a similarity measure. Information on the function of genes has been retrieved from various public databases (for example PubMed, OMIN, GENECARD, KEGG and GO) and from the LifeSeq Gold database (Incyte Genomics). Expression data publicly available from prostate cancer, breast cancer and cell lines were linked to our data by means of the Unigene cluster number. The strength of the relations between oestrogen and EGF-induced gene expression (log [ratio]) data was tested by Spearman rank correlation. The relations between categorised expression data (differential expression [DE] ≥ 1.60; 1.60 > DE > – 1.60; DE ≤ – 1.60) and gene association with tumour ER status (positive or negative) were tested by Spearman rank correlation. All computations were done with the STATA statistical package, release 8.0 (STATA Corp., College Station, TX, USA). All P values are two-sided. Quantitative RT–PCR For quantitative reverse transcriptase-mediated polymerase chain reaction (RT–PCR), 2.5 μg of total RNA (or 100 ng of poly(A) + mRNA), 0.8 μg of oligo(dT) 12–18 (Invitrogen Corporation) and 0.5 μg of random hexamer (Pharmacia) in 20 μl of RNase-free water were heated for 5 min at 65°C and cooled on ice. The final reaction of 40 μl contained 0.4 mM dNTPs (Pharmacia), 60 units of RNAseOUT and 300 units of SuperscriptII Reverse Transcriptase (Invitrogen). Incubations were for 2 min at 0°C, 10 min at 25°C, 50 min at 42°C and 10 min at 55°C; the reaction was stopped by heating for 15 min at 75°C. RNA was destroyed by treatment with 2 units of RNAseH (Promega) for 30 min at 37°C. cDNA products were diluted to 100 μl with 10 mM Tris/HCl pH 7.5; these stocks were stored at -80°C. Forty cycles of amplification of 5 μl of cDNA stocks in distilled water (Invitrogen) diluted 1:19 were performed with an SYBR green PCR mix (Applied Biosystems or Stratagene) and 0.33 μM forward and reverse primers in a volume of 25 μl on a ABI Prism 7700 (Applied Biosystems) in accordance with the recommended protocol. Primer annealing was performed at 60 or 62°C. A dilution series (1:4 to about 1:10,000) of a reference cDNA pool (mixture of cDNA preparations of RNA derived from six different cell lines) was used for normalising gene expression. The intron-spanning gene primers used are listed in Table 2 . The cycling conditions were as follows: denaturation for 10 min at 95°C; 40 cycles (15 s at 95°C, 30 s at 60°C (CTSD, TFF1 and MYC) or 62°C, 10 s ramping to 72°C, 20 s at 72°C, 10 s ramping to 79°C, 20 s at 79°C). Data were collected at 72 and 79°C and were analysed at 79°C. At the end of the amplification, the melting curve of the products was determined. PCR products showed discrete melting curves and specific bands of correct lengths on agarose gels after 40 cycles of amplification. We also used Assays-on-Demand™ (Applied Biosystems) for various genes, in accordance with the manufacturer's protocol. cDNA (5 μl, diluted 1:19 or 1:39) was measured in 25 μl reactions with the TaqMan Universal PCR master Mix. All cDNA samples were normalised for HPRT1 levels (four independent measurements) and are presented relative to the gene level in ZR-75-1 cells maintained in R/BCS medium. Results Overall gene expression To evaluate the effects of oestrogen, of EGF and of previously identified BCAR genes involved in oestrogen-independent growth of the human breast cancer cell line ZR-75-1, we determined the global gene expression in these cells by using UniGEMV2 cDNA microarrays. We performed direct comparisons of mRNA samples without the use of a general reference RNA sample (Table 1 ). Of the approximately 9000 sequences (about 88% represent known genes according to UNIGENE build no. 160) present on the microarray, the expression of 2373 genes (i.e. 26% of total sequences) was significantly ( P ≤ 0.01) affected by either the oestrogen treatment or the EGF treatment or the BCAR transfections. Hierarchical clustering distributes the expression profiles according to the experimental culture conditions; that is, profiles of long-term oestrogen-treated cells were separated from short-term oestrogen-treated cells or BCAR-transfected cells ( Additional file 1 ). The majority of the large changes in gene expression were observed in the oestrogen-stimulated or EGF-stimulated cultures. Hybridisation profiles of short-term oestrogen stimulation of ZR-75-1 cells and BCAR3-transfected cells are very similar and distinct from hybridisation profiles of long-term oestrogen-stimulated cell lines ( Additional file 1 ). The EGF-stimulated ZR/EGFR cells as well as the BCAR1- and BCAR3-transfected cells exhibit clearly different profiles in comparison with the oestrogen-stimulated cultures. The comparison of BCAR-transfected cell lines with each other or with non-stimulated parental cells revealed modest changes in gene expression, indicating that gene expression differences in these BCAR cell lines are generally subtle. Below we discuss the expression profiles (average gene expression log(ratios) of the independent experiments) of a selection of 1006 genes exhibiting a |DE| ≥ 1.60 in at least one of the actual or virtual experiments. Effects of oestrogen and EGF on gene expression of ZR-75-1 human breast cancer cells ZR-75-1 breast cancer cells are completely dependent on oestrogen for growth. In standard medium without added oestrogen, growth is strongly reduced. Addition of anti-oestrogen completely abolishes the growth of these cells [ 16 ]. Oestrogen-induced cell proliferation is mediated by the transcription activation function of the ERα. To identify the early effects (that is, the transcription targets) of oestrogen stimulation, the expression profile was analysed after a 6-hour high-dose pulse of 100 nM oestrogen. From Fig. 1 and Additional file 1 it is clear that limited changes in gene expression (115 genes with |DE| ≥ 1.60) have occurred during this short treatment compared with mock-stimulated cultures. Over 75% of these genes seemed to be induced. Among these genes are well-known oestrogen targets such as TFF1 (PS2), CSTD (cathepsin D), CCND1 (cyclin D1) and PGR (progesterone receptor). These and several novel genes were rapidly induced by oestrogen both in the parental ZR-75-1 cells and in the BCAR3-transfected cells (Fig. 1 and Additional files 1 2 3 ). An extended picture emerges after continuous exposure to the regular dose of 1 nM oestrogen. About 400 genes exhibit consistent changes (at least 1.6-fold) in expression, of which about 60% of the sequences exhibit a significant decrease of gene expression (up to sevenfold) and 40% are increased (up to more than 10-fold). The genes specifically modulated by oestrogen comprise members of all functional compartments and processes in the cell. One-quarter of the early-induced genes remain expressed (DE > 1.60) during continuous exposure to oestrogen (see Fig. 1 ), whereas the expression of others is turned off (for example AMD1 , BCL2 , CCND1 , GJA1 , MEIS3 , RIP140 , RUNX1 and STC1 ) or even downregulated (for example HIF1A , IL6ST , MYB , PC4 and UGT2B7 ). Definitions of these and other and other gene names can be found in Additional file 3 . Statistical analysis of all 1006 genes shows a clear positive correlation between early-induced gene expression and genes expressed after 7 days of oestrogen treatment ( r s = 0.36, P < 0.0001). We have previously shown that the addition of EGF to the culture medium does not support growth of ZR-75-1 cells because of the absence of EGF receptors [ 31 , 34 ]. The introduction of EGFR into ZR-75-1 cells (ZR/EGFR) permits a response to EGF and can support proliferation independently of oestrogen [ 31 ]. Gene expression of ZR/EGFR cells stimulated with EGF for 7 days was directly compared with that of ZR-75-1 cells stimulated with oestrogen continuously. It is clear from this direct comparison that 247 genes are specifically altered more than 1.6-fold ( Additional file 1 ). From the virtual experiment, which compares the EGF stimulation of ZR/EGFR cells with unstimulated ZR-75-1 cells, we can conclude that EGF modulates a large cohort of genes (707) at least 1.6-fold (Fig. 1 ). Statistical analysis of all 1006 genes shows a strong positive correlation for gene expression regulated by EGF and long-term oestrogen ( r s = 0.67, P < 0.0001), but no significant association with genes induced early by oestrogen ( r s = 0.16). As expected, EGFR is one of the most prominently changed genes in our analysis as a consequence of the transgene expression. In addition, the expression of genes implicated in signalling processes, cell adhesion and structure, protein modification, transport and metabolic processes is specifically regulated by treatment with EGF or shows a pronounced alteration in comparison with that in oestrogen-treated cultures (Fig. 1 ). Effects of overexpression of BCAR1 or BCAR3 in ZR-75-1 cells We have previously shown that stable overexpression of BCAR1 or BCAR3 induces cell proliferation independently of oestrogen and anti-oestrogen [ 17 , 18 ]. In an attempt to pinpoint the effects of these cytoplasmic signalling molecules on global gene expression in cultures without added oestrogen, we compared BCAR3 and BCAR1, and BCAR3 and ZR-75-1, directly on microarrays, and calculated the gene expression relation between BCAR1 and ZR-75-1 as a virtual experiment. A total of 79 genes exhibited consistent differences in expression of at least 1.6-fold (Fig. 1 ). Few genes are modulated solely by the overexpression of a BCAR gene; most are also a target for hormonal and/or EGF stimulation (Fig. 1 and Additional file 1 ). As expected, the largest observed difference (up to 15-fold) in these comparisons was derived from the BCAR3 transgene expression. No cDNA sequence corresponding to the BCAR1 gene was present on this microarray. Changes in gene expression specifically caused by BCAR3 overexpression in ZR-75-1 cells were seen for CSTA , DLG7 , FMOD , FOLR1 , FOXJ1 , HSPC242 , IGFBP5 , LGALS1 , LIV1 , NEDD4L , NELL2 , PCDH7 and UBE2C . In addition, a clear induction of several genes involved in glucose metabolism ( PGK1 , LDHA , TPI1 , and moderate induction levels of ALDOC , ENO1 , ENO3 and PFKP ; Fig. 1 ) was observed. This coherent change in gene expression is unlikely to represent a culture artefact because no expression change was observed in these genes in BCAR3-transfected cells after 6 hours of induction with oestrogen ( Additional file 1 ). Specifically altered genes in BCAR1-transfected ZR-75-1 cells include APOD , ASS , CRIP1 , ELL2 , FOXM1 , HSPG2 , ID1 , IL1R1 , IL6ST , LGALS8 , PDZK1 , TK1 , PPP3CA , TOMM20 , VIPR1 and various genes encoding ribosomal proteins (Fig. 1 ). A moderately opposing direction of gene expression change in these two transfectant cell lines compared with ZR-75-1 cells was indicated by TFF3 and MGP (Fig. 1 ). A set of genes (including BF , BCL2 , BIRC5 , CDKN1A , INADL , KIAA0101 , MAPKAPK2 , MYB , P2RY10 , PC4 , PRCP , OCLN , RAB6KIFL (= KIF20A ), SERPINA5 , SLC7A2 and UGT2B11 ) exhibited differential expression in both BCAR-transfected cell lines when compared with the parental cell line ZR-75-1. Verification of expression differences by quantitative PCR To establish that the measured differences in gene expression on the microarrays did indeed reflect the concentration of the respective mRNAs, we performed quantitative RT–PCR (Q-PCR) on a selection of 22 genes on the same RNA samples and on additional RNA preparations from different culture conditions. Standard quantities of intact total RNA or mRNA were reverse transcribed and subjected to Q-PCR. To normalise the cDNA samples we used HPRT1 (not present on the microarray) as a reference. Results have been presented relative to non-stimulated ZR-75-1 cells to facilitate direct comparison with microarray data (Table 3 ). In general, we found good agreement between the levels of HPRT1 and another housekeeping gene ( HMBS ) in our experimental samples. The expression of MYC was fairly constant in our series, with the exception of a slight decrease in EGF-stimulated ZR/EGFR cells (Table 3 ). This is in agreement with the results of our microarray hybridisations, which showed MYC expression to be significantly reduced only by EGF (Fig. 1 ). Oestrogen targets such as TFF1 , PGR , PDZK1 and CTSD are indeed increased by treatment of our ZR-75-1 cells and BCAR1-transfected and BCAR3-transfected cells with oestrogen (Table 3 ). These genes are already elevated after 6 hours of treatment with oestrogen (not by the pure oestrogen antagonist ICI 164,384), but their levels increase further ( TFF1 up to 30-fold) after prolonged oestrogen treatment, in agreement with our microarray data. After stimulation of ZR/EGFR cells with EGF, the expression of PGR and PDZK1 was completely abolished, whereas TFF1 and CTSD levels were induced under EGF (Table 3 ). TFF3 and MGP levels were clearly induced after long-term treatment with oestrogen and reduced after stimulation with EGF. The levels of ERα (not present on the array) show some decrease after treatment of ZR-75-1-derived cell lines with oestrogen (Table 3 ). A much stronger decrease in ER α levels (10-fold) is achieved after 7 days of treatment of ZR/EGFR cells with EGF. ERβ levels were found to be reduced about 1000-fold compared with ERα in our cells and not strongly affected by the culture conditions. The levels of HER2/Neu ( ERBB2 ) were decreased after treatment with both oestrogen and EGF. The expression levels of BCAR3 , EGFR and BCAR1 were not strongly affected in the various cultures, except for the cells containing the introduced transgene. The observed expression modulation of IGFBP5 , IL1R1 and CSTA in our microarray experiments is very well reproduced by Q-PCR (Fig. 1 and Table 3 ). Although oestrogen treatment causes a moderate decrease in these genes in ZR-75-1 cells, treatment of ZR/EGFR cells with EGF causes a marked effect (100-fold decrease in IGFBP5 after 7 days). The Q-PCR data also support the microarray data that BCAR3 cells have decreased levels of IGFBP5 mRNA and increased levels of CSTA mRNA in comparison with the BCAR1 and parental cells. IL1R1 and APOD levels are slightly modulated in BCAR1 cells, in agreement with the hybridisation data. PSAT1 levels were found to be further decreased in BCAR1 cells than suggested by the array experiments. Discussion The expression of a large proportion of genes investigated on this cDNA microarray does not alter significantly after stimulation of ZR-75-1 cells with oestrogen, or ZR/EGFR cells with EGF, or transfection of BCAR1 or BCAR3 genes. Furthermore, different ZR-75-1-derived cell clones showed very similar expression profiles after growth manipulation with oestrogen, indicating the stability of the parental cell line and the absence of extensive variation between cell clones. This result is in agreement with previous observations that this human breast cancer cell line is extremely stable and is a suitable target for in vitro insertion mutagenesis with retrovirus [ 16 ]. The growth of this cell line is completely dependent on oestrogen, and the proliferation signal is mediated primarily through ERα because ERβ mRNA levels were very low. The ERα mRNA is readily detected in our cells by Q-PCR and is moderately decreased by oestrogen treatment (Table 3 ). In contrast, ERα mRNA is strongly decreased (about 10-fold) in EGF-treated ZR/EGFR cells, which might relate to the observation of growth interference between signalling by oestrogen and by EGF in these cells [ 31 ]. Of the 1006 significantly affected genes in our series of experiments, only few are immediate/early targets of oestrogen-activated ERα (Fig. 1 and Table 3 ). Most changes in gene expression observed in our cell model are the result of long-term culture with either oestrogen or EGF. Many genes here identified as oestrogen targets have previously been reported to be directly regulated by oestrogen using conventional northern blotting, serial analysis of gene expression (SAGE) [ 35 ] or gene expression profiling [ 36 - 40 ] (see also Fig. 1 , 'cell lines' column). Good concordance with literature data was observed for the immediate and late targets of oestrogen in our cells (categorised data, r s = 0.42 and 0.41, respectively; P < 0.01). As expected, no association between EGF-regulated gene expression and reported oestrogen targets was seen (categorised data, r s = - 0.07). Undisputed early targets are TFF1 , CTSD , CCND1 , PGR , PDZK1 and MYB , whereas induction of IGFBP4 by oestrogen was reported to be dependent on the presence of serum in MCF7 cells [ 37 ]. Mostly overlapping results for oestrogen-regulated genes in MCF7 cells have been presented recently [ 40 ]. Differences between the various cell line models might explain individual differences (see Fig. 1 , for example MYC , XBP1 and MGP ). STC1 is rapidly induced in our experiments but has been reported not to be regulated in MCF-7 cells with the use of SAGE and microarrays [ 35 , 40 ]. In contrast, its family member STC2 was strongly increased by oestrogen treatment of MCF7 cells [ 35 , 37 ]. On our microarrays we did observe a moderate induction (1.7-fold) in STC2 transcript levels after induction with oestrogen for 6 hours, essentially parallel to STC1 (Fig. 1 ). Linking expression databases derived from cell line models and clinical samples provides the opportunity to extract additional information. We have connected our in vitro gene expression data with the public results of gene expression profiling of clinical breast cancer specimens [ 28 , 41 ] through the Unigene cluster number. About one-fifth of the 1006 genes in our selection were reported to be associated with ER status, BRCA mutation status and/or breast cancer prognosis (Fig. 1 ). Comparison of the categorised data of early oestrogen-regulated gene expression and expression association with ER status reveals a positive correlation ( r s = 0.21, P < 0.002). Clear examples of genes showing a positive correlation with ER status in breast carcinoma and oestrogen-induced gene expression in ZR-75-1 cells are CCND1 , GFRA1 , GJA1 , IGFBP4 , IL6ST , MEIS3 , MYB , PDZK1 , PGR , SERPINA5 and TFF1 (Fig. 1 ). Examination of the genes regulated in our cells by long-term treatment with oestrogen reveals an unexpected inverse relation ( r s = - 0.23; P < 0.002) with expression association to ER status of the tumour. About half of the genes regulated by long-term oestrogen and not regulated by EGF exhibit concordance with ER status, whereas most of the genes regulated similarly by both treatments show a discordant relation with ER expression (Fig. 1 ). A much stronger negative relation exists between EGF-regulated gene expression and expression association with ER status ( r s = - 0.43; P < 0.0001), which concurs with the established inverse relation between ER and EGFR in breast cancer [ 7 , 8 ]. The results of this comparison of cell line expression data with profiles of clinical samples indicate that part of the molecular markers for ER status in primary breast tumours might indeed represent genuine ER targets. Various other markers are not linked to oestrogen action but might reflect activation of the EGFR pathway in ER-negative tumour cells. Partial overlap was also reported for genes associated with breast cancer ER status and oestrogen-responsive genes in MCF7 cells [ 40 ]. Some genes reported to be associated with the prognosis of node-negative breast cancer ( TK1 , FBP1 and IGFBP5 ) or with BRCA mutation-induced disease ( CPE and P2RY10 ) seem to be regulated by BCAR1 and/or BCAR3 (see Fig. 1 ). In addition, the expression of IL1R1 , FOXM1 and PC4 changes markedly during the progression of normal prostate to metastasised prostate cancer [ 27 ]. These observed relations of genes regulated by BCAR1 and/or BCAR3 with clinical features of malignancies remain interesting and are targets for further study. The overall results show that BCAR1-transfected or BCAR3-transfected cells in unsupplemented cultures exhibit only modest changes in gene expression compared with unstimulated ZR-75-1 cells (Fig. 1 and Table 3 ). The prominent changes in gene expression induced by BCAR3 are upregulation of CSTA , HSPC242 and LGALS1 and downregulation of IGFBP5 , NEDD4L and PCDH7 . These genes are involved in protein degradation, cell–cell adhesion, the assembly of extracellular matrix and control of cell growth and metabolism. In BCAR1-transfected cells, upregulation of BIRC5 , ELL2 , FOXM1 , ID1 , IL1R1 , MYB and TK1 and downregulation of APOD , IL6ST and LGALS8 was observed. Several of these genes modulated by BCAR1 have been shown to be important in cell signalling and in the regulation of cell proliferation or possibly in increasing cell survival. These clearly different patterns of gene expression in BCAR1 and BCAR3 transfectants indicate that signalling proceeds along alternative pathways. This contrasts with the co-occurrence of BCAR1 and BCAR3 in a protein complex in some of our cell models (Ton van Agthoven, Arend Brinkman, Lambert CJ Dorssers, unpublished results) and the functional association of BCAR1/p130Cas and BCAR3/AND-34 in cell migration [ 42 ]. From the profiles in Fig. 1 and Additional file 1 it is clear that partial overlap exists in the expression profiles of the BCAR1 and BCAR3 transfectants with both oestrogen-induced and EGF-induced cells. Most of these genes are modulated in most experiments and thus might represent expression features of proliferating ZR-75-1 cells. The remaining overlap with either oestrogen-modulated or EGF-modulated genes is limited, making it highly unlikely that the BCAR genes use major parts of these signalling pathways. This observation agrees with previous results showing that BCAR1 and BCAR3 cell lines generated by retroviral insertion mutagenesis had all lost ERα protein expression and did not acquire responsiveness to EGF [ 16 , 17 ]. Furthermore, growth of BCAR1 and BCAR3 transfectants (which are fully responsive to oestrogen; Table 3 and Additional file 1 ) was not stimulated by anti-oestrogen [ 17 , 18 ], suggesting that there is no role for ERα in the anti-oestrogen-resistant proliferation of these cell models. In clinical specimens, BCAR1 was found to be an independent marker (multivariate analyses also including ERα) for early recurrence of breast cancer and for failure of tamoxifen treatment of recurrent disease [ 21 - 24 ]. Because not all genes are present on this microarray and only a limited set of experimental conditions have currently been analysed in our cell model, we cannot exclude from these microarray experiments the possibility that the BCAR transfectants selectively use components of the hormonal receptor and/or growth factor receptor signalling pathways for proliferation control. In addition, growth control mediated by BCAR1 and BCAR3 might not be reflected in gene expression but could also be supervised at the level of regulatory protein activation. Our results present an overview of gene expression changes after perturbation of ZR-75-1 breast cancer cells with treatment with oestrogen or EGF or by the overexpression of BCAR genes. The data suggest that oestrogen-independent cell proliferation induced by overexpression of BCAR1 or BCAR3 does not depend merely on the oestrogen-signalling or EGFR-signalling pathways. Because BCAR1 protein levels have been associated with clinical outcome for breast cancer patients, further studies are needed to resolve the underlying mechanism. This study also shows that important cell biological properties such as oestrogen-independent proliferation can be regulated in several subtle ways and thus might be hidden in the excess of gene expression differences observed in profiles of specimens from patients. The combination of expression profiles of relevant cell models, which can be manipulated in vitro , and high-quality specimens from patients might permit the identification and understanding of the important cellular pathways contributing to major clinical features of malignant diseases [ 43 ]. This information could ultimately lead to the development of improved or novel treatment strategies for breast cancer. Abbreviations BCAR = Breast Cancer Anti-oestrogen Resistance; DE = differential expression; E2 = oestradiol; EGF = epidermal growth factor; EGFR = epidermal growth factor receptor; ER = oestrogen receptor; P130Cas = Crk-associated substrate; Q-PCR = quantitative RT–PCR; R/BCS = RPMI 1640 medium supplemented with 10% bovine calf serum; r s = Spearman rank correlation; RT–PCR = reverse transcriptase-mediated polymerase chain reaction; SAGE = serial analysis of gene expression; ZR/EGFR = EGFR-transfectant cell line ZR/HERc(1A). Competing interests The author(s) declare that they have no competing interests. Supplementary Material Additional File 1 A TIFF file showing hierarchical clustering of gene ratios. Signal intensity tables were imported into Rosetta Resolver, combined for dye swapping and evaluated using the Incyte/UniGEM error model. A total of 2373 genes displaying at least once a P value of 0.01 or less were used for hierarchical clustering. In this colour picture, increased expression is shown in red and decreased expression is shown in green. Black represents no change and grey indicates missing data. The compared RNA samples are indicated. The gene names and log(ratios) are provided in Additional file 2. Click here for file Additional File 2 An Excel file containing gene expression data, namely Incyte spot ID, accession number, sequence name and description, and the Rosetta Resolver output data columns: log(ratio), P value and log(error). Click here for file Additional File 3 An Excel file containing all data from the actual and virtual experiments meeting the selection criteria (see Fig. 1), Unigene cluster number, gene name and description, and cited literature data with regard to oestrogen-regulated gene expression in cell lines and associations with tumour phenotypes. Ordering is in accordance with the hierarchical clustering of all genes, using the columns depicted in Fig. 1. Click here for file

D:\keerthana\PMC001xxxxxx\PMC1064102.xml

1064103

Prenylation inhibitors stimulate both estrogen receptor α transcriptional activity through AF-1 and AF-2 and estrogen receptor β transcriptional activity

Introduction We showed in a previous study that prenylated proteins play a role in estradiol stimulation of proliferation. However, these proteins antagonize the ability of estrogen receptor (ER) α to stimulate estrogen response element (ERE)-dependent transcriptional activity, potentially through the formation of a co-regulator complex. The present study investigates, in further detail, how prenylated proteins modulate the transcriptional activities mediated by ERα and by ERβ. Methods The ERE-β-globin-Luc-SV-Neo plasmid was either stably transfected into MCF-7 cells or HeLa cells (MELN cells and HELN cells, respectively) or transiently transfected into MCF-7 cells using polyethylenimine. Cells deprived of estradiol were analyzed for ERE-dependent luciferase activity 16 hours after estradiol stimulation and treatment with FTI-277 (a farnesyltransferase inhibitor) or with GGTI-298 (a geranylgeranyltransferase I inhibitor). In HELN cells, the effect of prenyltransferase inhibitors on luciferase activity was compared after transient transfection of plasmids coding either the full-length ERα, the full-length ERβ, the AF-1-deleted ERα or the AF-2-deleted ERα. The presence of ERα was then detected by immunocytochemistry in either the nuclei or the cytoplasms of MCF-7 cells. Finally, Clostridium botulinum C3 exoenzyme treatment was used to determine the involvement of Rho proteins in ERE-dependent luciferase activity. Results FTI-277 and GGTI-298 only stimulate ERE-dependent luciferase activity in stably transfected MCF-7 cells. They stimulate both ERα-mediated and ERβ-mediated ERE-dependent luciferase activity in HELN cells, in the presence of and in the absence of estradiol. The roles of both AF-1 and AF-2 are significant in this effect. Nuclear ERα is decreased in the presence of prenyltransferase inhibitors in MCF-7 cells, again in the presence of and in the absence of estradiol. By contrast, cytoplasmic ERα is mainly decreased after treatment with FTI-277, in the presence of and in the absence of estradiol. The involvement of Rho proteins in ERE-dependent luciferase activity in MELN cells is clearly established. Conclusions Together, these results demonstrate that prenylated proteins (at least RhoA, RhoB and/or RhoC) antagonize the ability of ERα and ERβ to stimulate ERE-dependent transcriptional activity, potentially acting through both AF-1 and AF-2 transcriptional activities.

Introduction Both estrogen receptor (ER) subtypes, ERα and ERβ, are ligand-activated transcription factors. ERα is the major ER in mammary epithelium and is an important regulator of cell growth, differentiation and malignant transformation. After binding to estrogen, the receptors associate with specific estrogen response elements (EREs) within the promoters of estrogen-regulated genes or the receptors affect the activity of other transcription factor complexes such as AP-1 (Jun–Fos). The two ER subtypes share affinity for the same ligands and DNA response elements [ 1 ]. These nuclear receptors consist of six domains including the A/B domain containing the AF-1 autonomous transcription activation domain, the C domain containing the DNA binding domain, the E domain containing the ligand binding domain, and the AF-2 ligand transcription activation domain located in the C terminus of the receptor. Transcriptional activation by ERα is mediated by the synergistic action of the two distinct activation functions; although AF-1 is constitutively active, it is usually weaker than the AF-2 activity. In contrast, ERβ appears to have no significant AF-1 activity and thus depends entirely on the ligand-dependent AF-2 activity [ 2 ]. The current model for ER action suggests that the ER modulates the rate of transcription through interactions with the basal transcription machinery and by altering the recruitment of co-activators that modify chromatin organization at the promoter level of target genes [ 3 - 5 ]. In addition, tissue-specific nuclear receptor co-activators and co-repressors have been described that can modify the transcriptional activity of the ER [ 6 - 8 ]. There is increasing evidence, however, that not all the biological effects of estrogens are mediated by direct control of target gene expression; indeed, some effects are attributed to estrogenic regulation of signaling cascades [ 9 - 11 ]. Several rapid effects suggest that estrogens can interact with receptors that are located in close proximity to the plasma membrane [ 12 , 13 ]. These receptors, which appear to form a subpopulation of the classical ER, are associated with the cell membrane and are responsible for several manifestations of estrogenic signaling [ 14 , 15 ]. Recent data explain how the coordinate interactions between a newly identified scaffold protein, MNAR, the ER and Src lead to Src activation, demonstrating the integration of ER action in Src-mediated signaling [ 11 , 13 ]. These data highlight new evidence for a cross-talk between estradiol (E2) and growth-factor-induced cytoplasmic signaling. Several components of these signaling pathways are low molecular weight GTPases, such as Ras, that require prenylation to function. Ras belongs to the Ras superfamily of low molecular weight proteins. The activity of such proteins is controlled by a GDP/GTP cycle. Members of the Ras superfamily include the Ras, Rho and Rab subfamilies. The Ras and Rho proteins of this superfamily are modified post-translationally by the isoprenoid lipids farnesylpyrophosphate and geranylgeranylpyrophosphate. Farnesyltransferase and geranylgeranyltransferase I respectively catalyze the covalent attachment of the farnesyl group (C 15 ) and the geranylgeranyl group (C 20 ) to the carboxyl-terminal cysteine of prenylated proteins. Prenylation appears to be essential not only for membrane association, but also for biological activity [ 16 ]. In a previous report, we demonstrated the implication of both farnesylated and geranylgeranylated proteins in E2 actions, as prenylation inhibitors block the cell-cycle progression driven by E2 and stimulate the transcriptional activity of the ER [ 17 ]. Our data strongly suggest that the transcriptional stimulation of the ER by prenylation inhibitors is due to a shift in the association of a transcription co-regulator with the ERα. Among the numerous co-activators identified to date, one of the best characterized is the p160 family of proteins including SRC-1, GRIP1 and RAC3, which enhance ER transactivation by recruiting other transcriptional regulatory factors such as CREB-binding protein and p300 (reviewed in [ 18 ]). We demonstrated that both FTI-277 (a farnesyltransferase inhibitor) and GGTI-298 (a geranylgeranyltransferase I inhibitor) increase the association of the SRC-1 co-activator with ERα and that FTI-277 decreases the association of HDAC1 with ERα, which is essential for transcriptional repression. In the present report, we assess in further detail the role of prenylated proteins in the estrogenic regulation of transcription in MCF-7 cells and in HeLa cells transiently expressing ERα, ERβ or ERα mutants. Our data clearly outline that prenyltransferase inhibitors (FTI-277 and GGTI-298) only stimulate ERE-dependent luciferase activity in stably transfected MCF-7 cells. They stimulate both ERα-mediated and ERβ-mediated ERE-dependent luciferase activity, without any obvious relocation of the ER from the cytoplasm to the nuclei. By contrast, the roles of both AF-1 and AF-2 of ERα appear to be determining in this effect. These results further establish the involvement of prenylated proteins, and more specifically the Rho-mediated signaling pathway, in the negative regulation of ER transcriptional activity. Materials and methods Materials Materials used for cell culture were from InVitrogen Life Technology (Cergy Pontoise, France). Polyethylenimine was purchased from Sigma-Aldrich S.a.r.l. (St Quentin Fallavier, France). FTI-277 and GGTI-298 were a generous gift from S Sebti (University of South Florida, Tampa, FL, USA). Both FTI-277 and GGTI-298 peptidomimetics were dissolved in a solution of 10 mM dithiothreitol in dimethylsulfoxide. Cell culture The MCF-7 human breast adenocarcinoma cell line was obtained from the American Tissue Culture Collection (Rockville, MD, USA). The development of stable transfectants of MCF-7 cells (MELN cells) or HeLa cells (HELN cells) has been described previously [ 19 ]. These cells were established by transfecting MCF-7 cells or HeLa cells with the ERE-β-globin-luc-SV-Neo plasmid and therefore expressing luciferase in an estrogen-dependent manner. MCF-7 cells were routinely cultured in RPMI 1640, and MELN cells and HELN cells were cultured in DMEM growth media, supplemented with 5% FCS. Cells were incubated at 37°C in a humidified 5% CO 2 incubator. Cell transfection Cells were grown for 3 days in phenol red-free medium, containing 5% dextran-coated charcoal-treated fetal calf serum (DCC-FCS). Then 6 × 10 4 cells were seeded per well in 12-well plates and grown for 1 day in phenol red-free medium, containing 5% DCC-FCS. Cells were then transfected (0.4 μg/well DNA, 1 μl/well polyethylenimine in a final volume of 500 μl). Five hours after transfection, and for optimal FTI-277 treatment, cells were pretreated for 24 hours with the farnesyltransferase inhibitor prior to receiving E2 (5 nM) and FTI-277 [ 17 ]. For GGTI-298 treatment, no preincubation was performed and the inhibitor was added simultaneously with E2. Control experiments were also carried out; on the one hand in the absence of E2, and on the other hand in the presence of E2 and in the absence of inhibitors. The plasmids used were the β-glob-Luciferase and the corresponding PGL3 empty vector for MCF-7 transient transfection, pSG5-REβ coding the full-length ERβ, HGO coding the full-length ERα, HG19 coding the AF-1-deleted ERα, HG7 coding the AF-2-deleted ERα, and the pSG5 empty vector for transfection of HELN cells. The HG0, HG19 and HG7 plasmids [ 20 ] were provided by Prof Chambon (Institut de Génétique et de Biologie Moléculaire et Cellulaire, Centre National de la Recherche Scientifique, Strasbourg, France). The pRL-CMV Vector (Promega, Charbonnières, France) coding the Renilla Luciferase was co-transfected as a reporter gene. Luciferase assay in cellular homogenates Cells were seeded in 12-well plates as already described, and were treated or not with prenyltransferase inhibitors and E2 for 16 hours in a final volume of 500 μl. At the end of the treatment, cells were washed with PBS and lysed in 100 μl lysis buffer (Promega). The luciferase activity from both Luciferase genes was measured using the Dual Luciferase Reporter (Promega), according to the manufacturer's instructions. For MELN cells that were not transfected with the Renilla Luciferase gene, protein concentrations were measured using the Bradford technique to normalize the luciferase activity data. A minimum of 15,000 relative light units were generated for the control conditions, used as the reference onefold induction. For each condition, average luciferase activity was calculated from the data obtained from three independent wells. ER immunocytochemical detection For each condition, 5 × 10 5 MCF-7 cells were seeded onto glass slides in 100 mm Petri dishes and were grown for 6 days in phenol red-free medium, containing 5% DCC-FCS. Cells were then preincubated or not with farnesyltransferase inhibitor and received E2 (5 × 10 -9 M) and/or inhibitors 24 hours later. At the end of the treatment period (24 hours), cells were fixed in paraformaldehyde (4% in PBS) for 1 min. The staining was performed by a Techmate Horizon™ slide processor using a one-step peroxidase-conjugated polymer backbone visualization system (En Vision™ DAKO, Glostrup, Denmark) according to the manufacturer's instructions. The chromogenic substrate was 2,3'-diaminobenzidine. The primary antibody used was a monoclonal anti-ERα antibody (clone 1D5; DAKO). Negative controls were performed using the same staining technique without incubation with the primary antibody. Data were quantified using the ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA, USA). Results are expressed in arbitrary units for either nuclei or cytoplasms of analyzed cells, in each experimental condition. Clostridium botulinum C3 exoenzyme treatment MELN cells were grown for 3 days in phenol red-free medium, containing 5% DCC-FCS. For each condition, 10 5 cells were seeded per well in 12-well plates and were grown for 1 day in phenol red-free medium, containing 5% DCC-FC. Cells were then treated with 20 μg/ml C3 exoenzyme (produced from the pGEX2T-C3 plasmid; kindly provided by Dr LA Feig, Tufts University School of Medicine, Boston, MA, USA) in a final volume of 0.5 ml [ 21 ]. Cells received E2 (5 × 10 -9 M) 24 hours later and lysates were obtained after a further 16 hours for the luciferase assay as already described. Statistical analysis Statistical analysis of the data was conducted using an unpaired two-sample t test. Significance was defined as P < 0.02 or P < 0.05, as indicated in the text. Results Prenyltransferase inhibitors stimulate ERE-dependent luciferase activity in stably transfected MCF-7 cells but not in transiently transfected MCF-7 cells We previously showed that both FTI-277 and GGTI-298 markedly enhance ER-mediated transcription in MELN cells [ 17 ], a clone of MCF-7 cells stably transfected with the ERE-β globin-luciferase reporter gene (Fig. 1a ). After 5 days of E2 deprivation, cells were treated or not with 5 nM E2, in the presence of or in the absence of prenyltransferase inhibitors, and the luciferase activity was quantified 16 hours later. For optimal FTI-277 treatments, a 24-hour preincubation was systematically performed before E2 addition, whereas GGTI-298 was added simultaneously with E2 [ 17 ]. We confirmed the efficiency and specificity of FTI-277 and GGTI-298 to inhibit protein prenylation in MELN cells in these experimental conditions. Indeed, for this purpose, we checked whether HDJ2, a farnesylated protein, and Rap1A, a geranylgeranylated protein, were indeed not prenylated in the presence of 10 μM FTI-277 and 5 μM GGTI-298, respectively (data not shown). E2 alone induced a 14.7-fold increase in the luciferase activity in MELN cells. In the absence of E2 (Fig. 1a , white bars), FTI-277 and GGTI-298 stimulated the basal transcriptional activity by 8.4-fold and 3.9-fold, respectively. In the presence of E2 (Fig. 1a , grey bars), FTI-277 and GGTI-298 dramatically enhanced the ability of E2 to stimulate transcription by an additional 4.1-fold for FTI-277 and 2.5-fold for GGTI-298. The ability of E2 to stimulate transcription was statistically enhanced by both FTI-277 and GGTI-298 ( P < 0.02). In parallel to this cell model, in which the Luciferase gene is stably integrated into the genome, we determined the effects of prenylation inhibitors in MCF-7 cells that were transiently transfected with the ERE-β-glob-Luciferase plasmid (Fig. 1b ). In vehicle-treated cells, E2 still induced a 6.4-fold induction of luciferase activity. In the absence of E2, FTI-277 and GGTI-298 only stimulated the basal transcriptional activity by 3.2-fold and twofold, respectively. In the presence of E2, the transcription was only enhanced by an additional 1.1-fold for FTI-277 and 1.4-fold for GGTI-298. The ability of E2 to stimulate transcription was not statistically enhanced by FTI-277 or by GGTI-298 ( P < 0.05). This result outlines the necessity of having a stably integrated reporter gene in order to elicit the effects of prenylation inhibitors. This necessity was confirmed by checking the effects of FTI-277 in two other cell lines having a stably integrated reporter system, beside the MELN cells. We used the MVLN cells (MCF-7 cells stably transfected with an ERE-vit-tk-luc plasmid [ 22 ]; data not shown) and HELN cells (HeLa cells stably transfected with the ERE-β-globin-luc-SV-Neo plasmid and transiently transfected with the ERα, as described in Fig. 2 ). The data clearly indicate that in the three models the farnesyltransferase inhibitor does stimulate the ERE-dependent luciferase activity, in the absence of and in the presence of E2. Prenyltransferase inhibitors stimulate both ERα-mediated or ERβ-mediated ERE-dependent luciferase activity in HELN cells We then examined whether the effects of prenylation inhibitors were preserved in cells that stably expressed the gene coding the ERE-β glob-Luciferase but only transiently expressed the ER. For this purpose we used HELN cells, which are derived from the HeLa human cervical cancer cell line. HELN cells have no endogenous ER but stably express the gene coding the ERE-β glob-Luciferase [ 19 ]. As expected, we observed no induction of luciferase activity by E2 in cells transfected with the empty vector (Fig. 2 ). For all experiments, luciferase activity was normalized according to the expression of the co-transfected Renilla Luciferase gene so as to take into account the variations due to transfection efficiency. HELN cells were then transfected with either ERα or ERβ. In both cases, E2 induced the luciferase activity in vehicle-treated cells (3.2-fold and 2.2-fold, for ERα-transfected and ERβ-transfected cells, respectively), which was lower than the activity obtained in MELN cells or MCF-7 cells (14.7-fold and 6.4-fold, respectively). In the absence of E2 (Fig. 2 , white bars), FTI-277 and GGTI-298 only weakly stimulated the basal transcriptional activity by 3.6-fold and 2.4-fold, respectively, for ERα and by 1.5-fold and 1.7-fold, respectively, for ERβ. These stimulations were statistically significant ( P < 0.05). In the presence of E2 (Fig. 2 , grey bars), FTI-277 and GGTI-298 enhanced the basal transcriptional level by an additional 2.2-fold and 1.6-fold, respectively, for ERα and by an additional 2.7-fold and 1.6-fold, respectively, for ERβ. Altogether, the effect of E2 was significantly enhanced by both prenylation inhibitors with either ERα or ERβ ( P < 0.05). As previously described for MELN cells [ 17 ], these effects were strongly inhibited by ICI 182,780 (data not shown). Role of AF-1 and AF-2 in the effect of prenyltransferase inhibitors on ERE-dependent luciferase activity in HELN cells In order to understand how prenylated proteins regulate ER transcription, we investigated whether the effects of prenylation inhibitors on transcription involve the AF-1 and/or AF-2 functions of the ERα. For this purpose, we used HELN cells that stably expressed the ERE-β glob-Luciferase reporter gene and were transiently transfected with plasmids coding either the full-length human ERα (HEG0) or the ERα mutants with defective AF-1 (HEG19) or AF-2 (HEG7) activities [ 20 ]. Cells, deprived of E2 for 4 days, were co-transfected with the Renilla luciferase plasmid and the plasmid coding full-length ERα or its mutants (Fig. 3 ). As expected, E2 alone could induce the luciferase activity only in cells transfected with plasmids coding either the full-length human ERα (HEG0) or the AF-1-defective mutant (HEG19). Indeed, AF-1 is constitutively active, is ligand independent and is mainly induced by growth factors. The absence of induction in cells transfected with the plasmid coding the AF-2 defective mutant (HEG7) is in agreement with the ligand dependence of AF-2. In the presence of E2 (Fig. 3 , grey bars), FTI-277 and GGTI-298 induced a statistically significant increase in the stimulation of transcription by E2 by 2.3-fold and 1.6-fold, respectively, in cells transfected with the full-length ERα (HEG0) and by 1.9-fold and 2.3-fold, respectively, in cells transfected with the AF-1 defective mutant (HEG19) ( P < 0.02). In the absence of E2 (Fig. 3 , white bars), FTI-277 and GTI-298 statistically increased the transcription of the luciferase gene in cells transfected with plasmids coding either the full-length human ERα or the two defective mutants by twofold to threefold. Cells transfected with the plasmid coding the AF-2 defective mutant (HEG7) exhibited no significant induction of luciferase activity in the presence of the prenylation inhibitors compared with E2 alone. The induction observed in cells transfected with the empty vector in the presence of FTI-277 had no biological significance as the final induction level was very low (1.2-fold). Decreased cytoplasmic and nuclear presence of ERα due to prenyltransferase inhibitors in MCF-7 cells To investigate the effects of prenylation inhibitors on ERα localization within the cell, we detected the presence of ERα by immunocytochemistry in MCF-7 cells, treated or not with E2 and prenylation inhibitors for 24 hours. The result of this experiment is shown in Fig. 4a , with the relative quantification presented in Fig. 4b . As expected, ERα is highly concentrated in the nuclei of the untreated control with a significant staining of the corresponding cytoplasms (indicated by arrows in Fig. 4a ). In the presence of E2, the staining intensity of both nuclei and cytoplasms was clearly decreased, with no distinct staining of the cytoplasms. In the absence of E2, similar decreases in labeling intensity were observed in the cytoplasms and nuclei of FTI-277-treated cells. Indeed, the presence of FTI-277 dramatically enhanced the ability of E2 to decrease nuclear staining, with the persistent absence of cytoplasmic staining. GGTI-298 similarly induced a decrease of ERα staining intensity in the cell nuclei, although to a lesser extent than did FTI-277, both in the presence of and in the absence of E2. Unexpectedly, no significant decrease in the staining intensity was observed in the cytoplasms of GGTI-298 treated cells, either in the presence of or in the absence of E2. It must be outlined that these effects were observed after 24 hours of treatment with the prenylation inhibitors, whereas no effect was observed after short-term incubation (2 hours; data not shown). Involvement of Rho proteins in ERE-dependent luciferase activity in MELN cells C. botulinum C3 exoenzyme, a specific inhibitor of RhoA, RhoB and RhoC proteins, was used to determine whether the effects of prenylation inhibitors on ERE-mediated transcription were due to Rho proteins. After 4 days of E2 deprivation, cells were treated or not with C3 exoenzyme. E2 was added after 24 hours, and the luciferase activity was quantified 16 hours later (Fig. 5 ). The inhibition of ADP-ribosylation had already been examined under these conditions, demonstrating that the C3 exoenzyme does enter the treated cells (data not shown). The results clearly show an induction of luciferase activity by the C3 exoenzyme, in the presence of or in the absence of E2. This demonstrates the involvement of Rho proteins in the negative control of the ERE-dependent luciferase activity. Discussion Activation of ERs by estrogens triggers both ER nuclear transcriptional activity and the Src/Ras/Erks pathway-dependent mitogenic activity [ 9 , 13 ]. Prenylated proteins have been implicated in both estrogenic actions [ 17 , 23 ], and the prenylated Rho GTPases are now considered important modulators of ER transcriptional activity [ 24 , 25 ]. We previously demonstrated that prenylated proteins antagonize the ability of ERα to stimulate ERE-dependent transcriptional activity, potentially through the recruitment of a co-regulator [ 17 ]. In the present report, we proved that the stable integration of a plasmid coding the ERE-dependent luciferase in MCF-7 cells, or in HeLa cells transfected with the ERα gene, is necessary for FTI-277 and GGTI-298 effects. We did, however, observe a significant but expected stimulation of luciferase activity by E2 alone in the transiently transfected cells. The need for a stable integration of the reporter system suggests that integration in the cell genome, an optimal chromatin environment of the ERE, and a precise low number of copies of the gene of interest are relevant conditions for the effects of the prenylation inhibitor, and consequently for the prenylated protein role. More insight in the interactions between the co-regulator complex in chromatin and prenylated proteins will be possible by performing chromatin immunoprecipitation experiments. We chose HELN cells as a model of ER-free cells. These cells are stably transfected with the plasmid coding the ERE-dependent luciferase. Neither E2 nor prenylation inhibitors had any effect on the luciferase activity in these cells. In contrast, the ERE-dependent activation of transcription induced in the presence of both prenyltransferase inhibitors and reversed by the antiestrogen ICI 182,780 (data not shown) could be observed after transient transfection of HELN cells with the ERα gene. The classical consensus ERE is directly bound to ERα and ERβ, and at this site ERβ is a weaker transactivator [ 3 - 5 ]. In the presence of E2 and either ERα or ERβ, the comparison of the effects of prenylation inhibitors revealed that prenylated proteins had similar effects on the transcriptional activities of both ERs. These results are in agreement with the fact that Rho GDIα, a Rho guanine dissociation inhibitor that represses the activity of Rho GTPases including RhoA, Rac1 and Cdc42, has been shown to be a positive regulator of both ERα and ERβ transcriptional activities in human osteosarcoma cells [ 24 ] It must be outlined, however, that there are important DNA binding sites for ERs other than ERE [ 26 , 27 ] where ERβ exerts different, or sometimes even opposite, effects to ERα (e.g. at AP-1 and Sp-1 sites) [ 28 , 29 ]. Such differential signaling between ERα and ERβ may exist with estrogen and prenylated proteins at the AP-1 response element. As it has been shown that AP1-mediated transcription is downregulated by RhoB, then RhoB may to some extent act on E2-mediated effects through this pathway [ 30 ]. Moreover, the activity of a number of transcription factors that may determine the overall promoter activity of ERE-containing genes is known to be altered by prenylated proteins (e.g. SRF, NF-κB) [ 31 - 34 ]. Our data strongly suggest the involvement of AF-2 in the transcriptional activation induced by prenylation inhibitors. Indeed, significant luciferase activity in cells expressing the AF-1 deleted mutant is observed in the presence of inhibitors and in both the presence of and the absence of E2, whereas the rates of induction observed in the AF-2-deleted mutant in the presence of inhibitors are similar in the presence of and the absence of E2. This is confirmed by the fact that the effects of prenylation inhibitors are similar on ERα and ERβ, whereas ERβ seems to have no significant AF-1 activity and thus depends entirely on the ligand-dependent AF-2 [ 2 ]. Interestingly, prenylated proteins appear to also regulate ER in an AF-2-independent manner, as the AF-2-deleted mutant did not completely abolish the increase in ER transcriptional activity induced by either FTI-277 or GGTI-298, with a significant stimulation in the absence of E2. This is in agreement with the data suggesting that induction of ER transactivation by RhoGDI is mediated largely by an ER AF-2-dependent mechanism, and to a lesser extent by an AF-2-independent mechanism [ 25 ]. The process by which a portion of the classical ER relocates to the cell membrane remains undetermined. Steroid receptors rapidly shuttle between the nucleus and cytoplasm by active nuclear import and export mechanisms. In the absence of identified post-translational modifications that could be involved in the attachment of the ER to the cell membrane, it is probable that membrane translocation of the ER is facilitated by another protein such as caveolin-1, which has been reported to interact with the ER [ 35 ]. Indeed, proteins of the Rho family (RhoA and Rac1) are targeted to caveolae in fibroblasts and are retained there by an unknown mechanism [ 36 ]. Moreover, p122, a GTPase activating protein is localized in the caveolae of both fibroblastic and epithelial cells, and plays an important role in caveolin distribution through the reorganization of the actin cytoskeleton [ 37 ]. The intracellular localization of MNAR and the way it traffics around the cell with the ER is still to be determined. Prenylated proteins may then modulate ER signaling through each key step in the various pathways: membrane attachment, translocation to the nucleus, stability, and genomic or nongenomic activities. The mechanisms that are responsible for the effects of prenylation inhibitors could involve modifications in the intracellular localization, or the level of ER protein expression, which would facilitate transcription. As an example, RhoA is a geranylgeranylated protein known to stimulate the degradation of the cyclin-dependent kinase inhibitor, P27 kip . We previously demonstrated that the ER protein and RNA expression levels in total cell lysates are decreased by both FTI-277 and GGTI-298 treatment [ 17 ], and that ICI 182,780, a pure estrogen antagonist, induces the rapid relocation of ERα to an insoluble nuclear fraction, followed by its degradation [ 38 ]. Furthermore, the association of the SRC-1 co-activator to the ER is increased by FTI-277 and GGTI-298 [ 17 ], and slows the cytoplasm to nucleus mobility of ligand–ER complexes [ 39 ]. Immunocytochemical analysis showed a clear decrease in the intensity of ER staining in the nuclei of MCF-7 cells treated with either E2 or with FTI-277, or both, with a concomitant decrease in cytoplasmic ER staining. It is important to note that GGTI-298 has a much weaker effect, mainly in the cytoplasm. This suggests that farnesylated proteins predominantly maintain a high level of ER expression within the whole cell (either by repressing the degradation of ER or by increasing its expression or stability), whereas geranylgeranylated proteins could only stimulate the nuclear expression of the ER. Such a mechanism must be elucidated; indeed, the degradation of ERα induced by E2 is dependent on the proteasome [ 40 ], as has already been suggested for RhoB [ 41 ]. Since prenylation inhibitors had no effect on ER localization after a short incubation period in MCF-7 cells (2 hours; data not shown), we speculate that prenylated inhibitors act mainly by inhibiting active regulatory prenylated protein rather than by activating inhibited proteins. Our data strongly suggest that both farnesylated proteins and geranylgeranylated proteins modulate ER-mediated activities, through either a common pathway or a distinct pathway. A special emphasis has already been placed on the role of Rho GTPases, and especially RhoA, Rac1 and cdc42, as ER-negative modulators in human osteosarcoma and breast cancer cells [ 24 ]. RhoA may therefore be the target of GGTI-298, explaining part of the described effects. We evaluated the role of the RhoA, RhoB, or RhoC proteins in the negative regulation of ER transcriptional activity, using the C3 exoenzyme. This toxin inhibits the three small GTPases through ADP-ribosylation. We could therefore implicate Rho proteins, which are generally geranylgeranylated, as negative modulators of ER transcriptional activity in MCF-7 cells. As both farnesylated proteins and geranylgeranylated proteins play a role in the negative control of ER-mediated transcriptional activity, many farnesylated proteins are potential candidates for the stimulatory effects of FTI-277. Due to the absence of specific inhibitors of the GTPase activity of farnesylated proteins, a more progressive screening of the numerous farnesylated proteins is required to identify those that are involved in the negative regulation of ER-mediated transcriptional activity. It is important to note that RhoB can be both farnesylated and geranylgeranylated, which modifies the cellular localization of the protein. Indeed, farnesylated RhoB is located at the plasma membrane and geranylgeranylated RhoB is endosomal [ 42 ]. An increase in RhoB mRNA is observed in some breast cancer cell lines [ 43 ]. In rat fibroblasts, it has been shown that in the presence of farnesyltransferase inhibitors, the antiproliferative effect is driven by the geranylgeranylated form of RhoB [ 44 ]. Rho proteins and other prenylated proteins may then modulate ER-mediated transcription. Apart from the involvement of RhoGDI as a positive regulator of ER transactivation, it has been shown that Brx, a guanine nucleotide exchange factor for Rho proteins and member of the Dbl oncogene family, contains a nuclear hormone receptor binding region. Brx affects ER-mediated gene activation by a mechanism that is dependent on the Cdc42Hs signaling pathway [ 45 ]. Moreover, constitutively active forms of c-Raf and Rac synergistically enhance the CREB-binding protein/p300-mediated increase of transcription in T-cell activation signals [ 46 ]. Finally, a constitutively active form of Cdc-42 induces H4 hyperacetylation in chromatin [ 47 ]. Altogether, our data strongly suggest that several prenylated proteins, either farnesylated or geranylgeranylated, modulate ER signaling by repressing its transcriptional activity and/or increasing its stability. The determination of the exact Rho protein(s) involved in the effect we described is a major step of the research we are currently performing in the laboratory, in addition to the identification of the potential ER co-regulators. Conclusion Targeting the estrogen signaling pathway has proved to be of great value in the treatment of human breast cancer, and current evidence suggests that such targets hold great potential in the prevention of human breast cancer. Evaluation of the multiple pathways that can cross-talk with estrogen signaling pathways should help improve our understanding of some of the possible mechanisms of de novo and acquired tamoxifen resistance. Farnesyltransferase inhibitors are a new class of anticancer drugs that are at present in phase III clinical evaluation [ 48 - 50 ]. We previously showed that the combination of a farnesyltransferase inhibitor and tamoxifen may increase the antitumor effect of either drug alone in breast cancer [ 23 ]. Gaining further insight into the involvement of prenylated proteins in the estrogen signaling pathways should allow a more rational approach to treating and/or preventing hormone-resistant phenotypes. Abbreviations DCC-FCS = dextran-coated charcoal treated fetal calf serum; DMEM = Dulbecco's modified Eagle's medium; E2 = estradiol; ER = estrogen receptor; ERE = estrogen response element; FCS = fetal calf serum; FTI-277 = farnesyltransferase inhibitor; GGTI-298 = geranylgeranyltransferase inhibitor; MNAR = modulator of non-genomic activity of estrogen receptor; PBS = phosphate-buffered saline. Competing interests The author(s) declare that they have no competing interests. Authors' contributions The authors' contributions to this research are reflected in the order shown, with the exception of SDS who supervised the research and the preparation of this report. PC, GS, CMG and SDS carried out the experiments and PC drafted the manuscript. PB provided the MELN cells and the HELN cells, and provided technical support to initiate the study. PR carried out the immunocytochemical labeling. SDS and JCF conceived of the study, and participated in its design and coordination. GF, head of the department, gave helpful advice during the whole process. SDS performed the statistical analysis. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064103.xml

1064104

Membrane estrogen receptor-α levels in MCF-7 breast cancer cells predict cAMP and proliferation responses

Introduction 17β-estradiol (E 2 ) can rapidly induce cAMP production, but the conditions under which these cAMP levels are best measured and the signaling pathways responsible for the consequent proliferative effects on breast cancer cells are not fully understood. To help resolve these issues, we compared cAMP mechanistic responses in MCF-7 cell lines selected for low (mER low ) and high (mER high ) expression of the membrane form of estrogen receptor (mER)-α, and thus addressed the receptor subform involved in cAMP signaling. Methods MCF-7 cells were immunopanned and subsequently separated by fluorescence activated cell sorting into mER high (mER-α-enriched) and mER low (mER-α-depleted) populations. Unique (compared with previously reported) incubation conditions at 4°C were found to be optimal for demonstrating E 2 -induced cAMP production. Time-dependent and dose-dependent effects of E 2 on cAMP production were determined for both cell subpopulations. The effects of forskolin, 8-CPT cAMP, protein kinase A inhibitor (H-89), and adenylyl cyclase inhibitor (SQ 22,536) on E 2 -induced cell proliferation were assessed using the crystal violet assay. Results We demonstrated a rapid and transient cAMP increase after 1 pmol/l E 2 stimulation in mER high cells; at 4°C these responses were much more reliable and robust than at 37°C (the condition most often used). The loss of cAMP at 37°C was not due to export. 3-Isobutyl-1-methylxanthine (IBMX; 1 mmol/l) only partially preserved cAMP, suggesting that multiple phosphodiesterases modulate its level. The accumulated cAMP was consistently much higher in mER high cells than in mER low cells, implicating mER-α levels in the process. ICI172,780 blocked the E 2 -induced response and 17α-estradiol did not elicit the response, also suggesting activity through an estrogen receptor. E 2 dose-dependent cAMP production, although biphasic in both cell types, was responsive to 50-fold higher E 2 concentrations in mER high cells. Proliferation of mER low cells was stimulated over the whole range of E 2 concentrations, whereas the number of mER high cells was greatly decreased at concentrations above 1 nmol/l, suggesting that estrogen over-stimulation can lead to cell death, as has previously been reported, and that mER-α participates. E 2 -mediated activation of adenylyl cyclase and downstream participation of protein kinase A were shown to be involved in these responses. Conclusion Rapid mER-α-mediated nongenomic signaling cascades generate cAMP and downstream signaling events, which contribute to the regulation of breast cancer cell number.

Introduction It has long been recognized that cAMP is an important intracellular messenger that is capable of regulating diverse functions such as steroidogenesis in the ovary, sugar metabolism in the liver, and contractility of the cardiac muscle [ 1 - 5 ]. In addition to these crucial and specific physiological functions, the cAMP pathway plays an important regulatory role in the development of other complex cellular states such as differentiation, proliferation, and synaptic plasticity. Elevated levels of intracellular cAMP can inhibit the growth of some cell types but stimulate the growth of others. Among the cell types whose growth is stimulated by cAMP are thyroid [ 6 ], pituitary [ 7 ], and normal human breast epithelial cells in culture [ 8 ]. Several studies have demonstrated that cAMP inhibits the growth of established breast cancer cell lines and breast cancer cells in primary culture [ 8 - 10 ]. However, such studies have not completely resolved issues such as the effective levels of cAMP in different cell types, the roles played by other participants in the signaling web that may cooperate to bring about different outcomes, and any subpopulations of response-initiating receptors. The intracellular cAMP level is determined by the rates of its synthesis and clearance. It is synthesized from ATP via a transmembrane adenylyl cyclase (AC), which is activated by stimulatory G s proteins coupled to cell surface receptors (G-protein-coupled receptors). Clearance is regulated either by cyclic nucleotide phosphodiesterases (PDEs) or by an efflux mechanism that secretes or transports cAMP out of cells [ 11 ]. The multiplicity of PDEs was recently established [ 12 ], but the regulation and utilization of different isoforms are experimental issues that remain unclear. The influence of 17β-estradiol (E 2 ) on MCF-7 breast cancer cell proliferation was previously assigned exclusively to genomic mechanisms. However, it has been known for some time that E 2 can rapidly (within a few minutes) induce cAMP production, and that this time frame is not compatible with the multiple macromolecular synthetic events necessary for genomic responses [ 13 ]. We have an incomplete understanding of the conditions under which cAMP responses are best measured, and of the participation of multiple enzymes in generating and degrading cAMP. There are also reports that some MCF-7 cell sublines either do not respond or do not reliably generate this response (personal communications from many laboratories). Therefore, estrogen receptor (ER)-mediated mechanisms that increase cAMP in MCF-7 and other responsive cell types have not fully been explained. Here, we compare cAMP responses in cell lines selected for low and high expression of the membrane form of ER (mER)-α and thus address the receptor subform that is involved in these mechanisms. Additionally, we further probe the details of mER-α-mediated activation of AC and the contributions of PDEs to final signal levels, as well as subsequent cAMP-dependent protein kinase A (PKA) activation and the regulation of cell proliferation. Methods Cell immunoseparation and subculturing MCF-7 cells originated from the Michigan Cancer Center. We separated them into two subpopulations by immunopanning [ 14 , 15 ] using C-542 carboxyl-terminal ER-α antibody, provided by Drs Dean Edwards and Nancy Weigel. This antibody is now commercially available from Stressgen Biotechnologies (Victoria, Canada). Briefly, sterile antibody on the surface of a petri plate bound cells at 4°C over a 1-hour time period, and cells that attached to the plate (mER + ) were propagated separately from those that did not bind (mER low ). Using the same antibody, mER + cells were then subjected to further separation via a fluorescence-activated cell sorter [ 16 ]. Cells selected by both types of sequential separation are highly enriched for mER-α (mER high ). Upon propagation, the separated cells were not exclusively mER positive or negative, and consequently they were named mER high and mER low . We previously observed this heterogeneity in similarly separated GH3/B6 cells [ 15 , 17 ]. All subpopulations of MCF-7 cells were routinely cultured in phenol-red free Dulbecco's modified eagle medium (Gibco-BRL, Grand Island, NY, USA) supplemented with 10% heat-inactivated DSCS (defined/supplemented bovine calf serum; HyClone Laboratories, Inc., Logan, UT, USA) and 1% antibiotic–antimycotic (Gibco Invitrogen Corporation, #15240-062). Cells used in the experiments were between passage 8 and 14 after separation. Three days before each experiment, the cell growth medium was replaced with medium containing 4 × dextran-coated charcoal stripped serum (DCSS medium). Stripped serum was produced by incubation with an equivalent volume of packed 0.25% weight/vol charcoal Norit A for 2 hours on a shaker at 4°C. The charcoal preparation had previously been suspended in 0.0025% (weight/vol) dextran T-70 (Sigma, St Louis, MO, USA) in 0.25 mol/l sucrose, 1.5 mmol/l MgCl 2 , and 10 mmol/l HEPES (pH 7.6). The charcoal was pelleted at 500 g for 10 min and the supernatant (stripped serum) decanted, and the process was repeated four times. In other experiments we deprived cells of serum 2–3 days before the experiment by placing them in a defined medium (DM) consisting of phenol-red free Dulbecco's modified eagle medium with 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenium (using the stock preparation from Sigma), and 0.1% bovine serum albumin. cAMP response measurement Cells were plated at a density of 0.25 × 10 6 per well in six-well plates and then incubated in either DM or DCSS for 3 days before the experiment. In experiments in which hormone treatments were performed at 37°C, on the day of the experiment the cells were pretreated for 10 min with 1 mmol/l IBMX (3-isobutyl-1-methylxanthine; Sigma) and then treated for different time intervals (3, 6, 10, 20, 30, 60 or 120 min) with 1 pmol/l E 2 or ethanol vehicle (0.1%) in the presence of IBMX. In experiments in which the hormone treatment (as described above) was performed at 4°C, cells were treated either in their growth plates or in suspension. For cell suspensions the cells were scraped from two 150 mm plates, pelleted at 1000 g , and resuspended in 3.5 ml DM. The incubation was then performed in 100 μl of cell suspension with permanent agitation and the reaction was terminated by spinning down the cells and washing them in cold DM. For attached cells, similar incubations were performed in the growth dishes. The specificity of this response for ER-α was tested with ICI172,780 (Tocris Cookson Inc., Ellisville, MO, USA). The cells were pretreated with ICI172,780 for 30 min (1 μmol/l) and incubated for an additional 15 min after adding 1 pmol/l E 2 , or they were simultaneously treated for 15 min with E 2 and ICI172,780. As a negative control, we also tested the time-dependent production of cAMP with 10 nmol/l 17α-estradiol (Sigma Aldrich, St Louis, MO, USA) – an inactive E 2 stereoisomer. Preparation of cell lysates and quantification of intracellular cAMP were performed according to the protocol provided with the cAMP detection kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). Each value for cAMP was normalized against protein concentration determined using a Bradford (Bio Rad, Hercules, CA, USA) protein detection kit. Cell proliferation Cells were plated at a density of 1000 or 2000 cells per well in 96-well plates. The next day the growth medium was replaced with DCSS containing the hormone (E 2 ) or other treatments (forskolin, 8-CPT cAMP, H-89 plus E 2 , SQ22 536 plus E 2 ). Controls for E 2 treatment were done on a separate plate, because we had previously determined that low amounts of volatilized estrogens can affect responses mediated via nongenomic signaling pathways [ 18 ]; control treatments were with the vehicle in which test compounds were solubilized (ethanol or dimethyl sulfoxide). In some experiments the growth medium was replenished every second day, and after 5 days the cells were fixed with 2% paraformaldehyde/0.1% glutaraldehyde in PBS. The number of the cells in each well was determined using the crystal violet (CV) assay [ 19 ], which we modified previously [ 17 ]. Briefly, the cells were incubated in 0.1% CV for 30 min at room temperature, excess dye was removed by three brief rinses with ddH 2 O, the plates were air dried, and the dye was extracted with 10% acetic acid, which was then read in a plate reader (Wallac 1420; Perkin Elmer, Boston, MA, USA) at 590 nm. The utility of the CV assay in measuring cell number was verified both in MCF-7 cells and in combination with immunodetection plate assays for GH3/B6 cells [ 17 ]. We compared CV results with DNA content measurements and with cell number counts by hemocytometer and both assays correlated very well (data not shown). Additionally, we compared the CV method with the MTT assay (ATCC, Manassas, VA, USA), which is often used to determine cell number and viability. Different numbers of mER high and mER low cells (1000–7000) were plated in 96-well plates, and fixed and treated with CV as described. Both sets of data were approximated with a single linear regression line (Fig. 1a ). The number of cells determined by CV versus MTT assays were linearly correlated (Fig. 1b ). Statistical analysis Statistical differences between two sets of data were determined using two-way analysis of variance. The dose–response curves were fitted with a four-parameter Gaussian distribution (Sigma Plot 8.0; Systat Software Inc., Point Richmond, CA, USA). The differences between the entire curves were tested by comparing the sum of squares of the residuals from each individual curve with the sum of squares of the residuals of the combined curves by applying a Microsoft Excel F test. P < 0.05 was considered statistically significant. Results Optimal conditions for E 2 -induced cAMP accumulation and measurement in MCF-7 cells enriched and depleted for membrane ER-α MCF-7 cell line enriched for membrane ER-α We observed that the level and kinetics of cAMP accumulation varied depending on the type of incubation medium, the incubation temperature, and pretreatment with the PDE inhibitor IBMX. In mER high MCF-7 cells, 1 pmol/l E 2 induced rapid and transient production of cAMP (Fig. 2a,2b,2c ). At the reduced temperature of 4°C in a completely defined medium (Fig. 2a ), a substantial increase in accumulated cAMP was seen as compared with the levels achieved at 37°C (Fig. 2b ). The response peak at 4°C was prolonged, as the accumulated cAMP decreased gradually, even in the absence of the PDE inhibitor IBMX, which is usually included to inhibit the decay of cAMP and enhance the response. The same level of accumulated cAMP was obtained at 5, 15 and 30 min at 4°C, regardless of whether the treatment was performed with cells attached to a plate (Fig. 2a , open circles) or in suspension (Fig 2a , closed circles). However, after 30 min the cAMP level declined abruptly in attached cells. The DCSS medium did not increase cAMP production at 4°C (data not shown). At 37°C, a low but significant degree of cAMP elevation was achieved by 5 min in attached cells treated in DM (Fig. 2b , triangles); a greater cAMP response was achieved in attached cells pre-incubated in DCSS (Fig. 2b , squares) as compared with DM. All of the experiments conducted at 37°C were performed in the presence of 1 mmol/l IBMX, as in most published protocols. The impeded ligand E 2 -peroxidase also triggered production of cAMP in suspended cells treated at 4°C in DM (Fig. 2c ). The level of E 2 presented by the conjugate in these studies approximated the level applied as free steroid in the experiments shown in the other panels. The resulting maximal level of accumulated cAMP was lower than that with the equivalent amount of free ligand, but the time required for peak accumulation was the same. MCF-7 cell line depleted for membrane ER-α When treated at 4°C in suspension, mER low MCF-7 cells exhibited only a small but significant rise in cAMP (Fig. 2d ) in response to E 2 . Hormone treatment caused a maximal accumulation of cAMP at 6–15 min, but the decay in this second messenger was very rapid. Insignificant changes were observed at 37°C in attached cells, both in serum-containing and in defined media (Fig. 2e ). The statistical comparison of cAMP accumulated in mER high versus mER low cells revealed that those differences were significant. cAMP dynamics Because our 4°C protocol without any added IBMX is relatively unusual, we wondered about the causes of the cAMP level dynamics. We next asked why accumulated cAMP did not plateau at 37°C. That is, what are the circumstances of cAMP decay in these cells, and does cAMP leave the cells or is it metabolized? To approach these questions we measured cAMP levels at 37°C after 1 pmol/l E 2 stimulation, in cells and in their extracellular medium in parallel. We also measured the kinetics of IBMX protection of cAMP by measuring the decrease in cAMP in the cytosolic fraction prepared from cells that had been treated with forskolin. The increase in cAMP in the medium did not match the decrease in intracellular cAMP content of the cell; although the intracellular cAMP of this population of cells decreased by about 80 pmol (from 95 pmol down to 15 pmol), the cAMP in the medium increased only by about 3 pmol (Fig. 3a ). As expected, in the absence of IBMX the decrease in cytosolic cAMP was rapid (3.8%/min), whereas in the presence of IBMX it decreased at a slower rate (1.2%/min). Therefore, the addition of IBMX protected cAMP from the action of some PDEs (Fig. 3b ), but there was still a significant fraction of cAMP that seemed refractory to this protection. ICI172,780 and 17α-estradiol effects on cAMP production ICI172,780 was effective in preventing the 15 min maximal cAMP accumulation (Fig. 4a ). Simultaneous application of ICI172,780 (1 μmol/l) and E 2 (1 pmol/l) was as effective as the 30 min pretreatment with ICI172,780. The usually inactive E 2 stereoisomer 17α-estradiol (10 nmol/l) was not capable of triggering cAMP production in mER high MCF-7 cells (Fig. 4b ). These studies are consistent with this effect being mediated by known ER proteins. E 2 -induced dose-responses of cAMP levels and cell number changes differ between MCF-7 cells enriched and depleted for membrane ER-α In Fig. 2 we observed that the characteristics of hormonally induced cAMP accumulation differed between mER high and mER low cells. Figure 5a shows that the E 2 dose–response curve for this response was biphasic in both cell types, but the sensitivity and upper limit of the responses differed. The mER high cells exhibited a higher maximal response level in cAMP levels. In mER low cells the response peak was achieved with a 50-fold lower E 2 concentration; the higher maximal stimulation of cAMP production in mER high cells was achieved with 100 pmol/l E 2 . Therefore, it can be assumed that a larger number of receptors had to be filled to achieve this higher response level in cells that had a larger receptor pool. E 2 also caused different dose effects on the proliferation of mER high versus mER low cells. Cells with low mER levels responded to E 2 with stimulated proliferation in the whole range of tested concentrations (from 0.1 fmol/l up to 0.1 mmol/l; Fig. 5b , open circles). However, mER high cells showed a biphasic proliferation pattern, with growth stimulation in the lowest range of concentrations of E 2 (up to 1 pmol/l) and inhibited proliferation at higher E 2 concentrations (from 1 nmol/l up to 1 mmol/l; Fig. 5b , closed circles). Therefore, the higher levels of cAMP production achieved with mER high cells result in growth inhibition. Because we had previously observed an effect of cell density on the level of mER expression and corresponding functional consequences in pituitary tumor cells, we wondered whether cell density would affect functional responses to E 2 in this breast cancer model system. Indeed, proliferation of mER high cells in the presence of 10 nmol/l E 2 was influenced by cell density (Fig. 6 ). At a low density (such as that used in the preceding experiments) proliferation was inhibited, whereas at higher cell densities this effect was reversed. When the cells were plated at the highest density studied (6000/well), their growth was stimulated at the same level as were the mER low cells. In contrast, proliferation of mER low cells was stimulated regardless of the cell plating density. E 2 -induced increase in adenylyl cyclase activity correlates with decreased proliferation of MCF-7 cells enriched for membrane ER-α cells via a PKA-activated pathway An activator of AC (forskolin), as well as the cell-permeable cAMP analog 8-CPT cAMP, inhibited the growth of both mER high and mER low cells (Figs 7 and 8 ) when provided in the same concentrations to both cell types. In mER high cells, both the AC inhibitor SQ22,536 and the PKA inhibitor H-89 abrogated E 2 -induced inhibition of cell proliferation (Figs 7 and 8 ). Both inhibitors increased the stimulatory effect of E 2 on mER low cell proliferation (Fig. 8 ). Therefore, bypassing receptor-mediated signaling mechanisms and controlling cAMP levels with other compounds (which either provided or acted to generate cAMP) made both cell types behave similarly with respect to changes in cell number. Likewise, artificially decreasing cAMP levels or their downstream effects (AC and PKA inhibitors) made cells respond similarly with respect to growth, regardless of their mER levels. Discussion Our studies provide evidence that membrane-associated ERs participate in the estrogen-regulated control of MCF-7 breast cancer cell number by changing cAMP levels and downstream cAMP-activated PKA. E 2 differentially influenced cell proliferation in mER high versus mER low cells. Measurements of and responses to cAMP manipulated via different culture conditions, assay incubation conditions, and specific response mimetics or inhibitors allowed us to implicate pathways and signal levels in the mechanism of control of cell number, and to suggest possible resolutions between previously reported conflicting results about cAMP and the control of breast cancer cell growth. Breast cancer cell lines (especially MCF-7 cells) are heterogeneous populations of cells; other investigators have succeeded in separating them into different constituent subpopulations [ 20 , 21 ]. We exploited this characteristic and separated MCF-7 cells into populations enriched and depleted for the membrane subform of ER-α by capturing live cells that bound ER-α-specific antibody to their membranes. These two cell subtypes express the same level of total (membrane plus nuclear) ER [ 22 ]. Thus, we have a system in which the consequences of mER levels can be assessed in the same cell line, without other transgenetic manipulations. Because our mER high and mER low cells were not clonally derived (from a single cell), the differences between these populations that we observed can not be attributed to clonal variations. Attempts to study E 2 -stimulated cAMP levels using the conditions reported by others for MCF-7 cells [ 23 - 25 ] resulted in variable and marginal responses in our hands. We tested different incubation times from 10 min up to 3 hours at 37°C, and all were relatively ineffective, even in the presence of a PDE inhibitor. When we decreased the incubation temperature to 4°C, the amplitude of maximally accumulated cAMP increased and the slope of its decline decreased considerably. Although this temperature is not physiologic, it is often applied for biochemical assessment of biologic responses. This is done to avoid competition by other biologic responses that could rapidly remove the molecule under study at the physiologic temperature. We also tested a variety of other conditions, including much shorter incubation times than 10 min and various media, as well as comparing attached versus suspended cell preparations. Both we and others [ 24 ] have found an influence of charcoal-stripped serum in the medium on cAMP production. However, whereas Fortunati and coworkers [ 24 ] could not identify any cAMP production in serum-free medium, we were able to detect a low but significant E 2 -induced increase. It is possible that this discrepancy could be due to different incubation times, as we found a very rapid effect (3 min) but the other study tested 15 min as the earliest time point. In more recent analyses it has become common to observe nongenomic responses that are complete and return to baseline by 15 min [ 26 - 28 ]. In our cells 1 mmol/l IBMX was only partially effective, and loss of cAMP was not the consequence of its transport out of the cells. We can assume that some other PDEs that are resistant to IBMX were present, unless our mER high cells possess other cAMP effectors (receptors or binding proteins) apart from PKA [ 29 ] that could capture cyclic nucleotides and render them undetectable in our cytosol assay. The possibility of multiple PDEs with their own individual patterns of regulation provide significant alternative opportunities for control of this important cell function, and add another level of complexity to its analysis. The impeded ligand that we used in these studies, E 2 -peroxidase, was also capable of triggering cAMP accumulation. The attachment of E 2 to a large molecule that cannot easily enter cells has previously been used to demonstrate action at the membrane [ 30 - 33 ]. Although this compound stimulated lower but significant levels of cAMP accumulation than did approximately equivalent levels of free steroid, it is likely that steric considerations (such as only partial contact of the large molecule with the cell surface; bound protein obscuring adjacent receptors) may prevent ligand–receptor contacts, or that some other aspect of the ligand–receptor complex configuration is disrupted by ligand tethering. Others have also reported that E 2 -peroxidase has lower affinity for the receptor compared with free ligand [ 34 ]. Although it has been reported that steroid can escape conjugates and elicit a response, we removed free steroid in our studies by incubating an aliquot with dextran-coated charcoal under conditions that remove more than 99% of free hormone [ 35 ] just prior to application of the impeded ligand. The release of steroid from conjugates happens over a much longer period of time than a rapid response [ 36 ]. We have shown that E 2 can play a dual role in breast cancer cells, both stimulating and inhibiting cell proliferation, depending on its concentration, which is consistent with the findings reported by Lippert and coworkers [ 37 ]. We have also shown that the same E 2 concentration (10 nmol/l) could either prevent or enhance mER high MCF-7 cell growth, depending on cell density. We previously showed that increased cell density can dramatically decrease the fraction of ER-α expressed in the cell membrane (see accompanying reports [ 18 ] and [ 22 ]), and others have observed similar effects on other membrane receptors [ 18 , 38 ]. The estrogenic responses of mER low cells were not affected by cell density, and their mER levels were already low. We therefore suggest that this type of control can be attributed to effects on the level of mER and thus the signaling cascades elicited by it. It remains to be determined whether E 2 -induced apoptosis, necrosis, or cell growth arrest could be part of the negative growth effects caused by high E 2 levels in mER high cells. Others have shown that very low concentrations of E 2 (as low as 1 pmol/l) induce apoptosis in MCF-7 LTED cells [ 39 ], selected by long-term growth in the absence of estrogens. The resulting increase in ER-α expression levels in these cells caused hypersensitivity to E 2 [ 40 ]. In addition, cells in which ER-α has been over-expressed by stable transfection also die in response to physiologic estrogen concentrations [ 41 ]. However, measurements in both of these studies failed to distinguish between nuclear and membrane forms of ER-α. Because our mER high cells were selected on the basis of their high mER expression levels (see accompanying report [ 22 ]), we can assume that their higher sensitivity to E 2 is a consequence of their mER levels. Our mER high cells exhibited a biphasic pattern of proliferation, with the maximal growth at approximately 10 -11 mol/l E 2 and the decline in cell number beginning in response to approximately 1 nmol/l E 2 . Maximal stimulation of LTED cells was achieved at 10 -14 mol/l, while stimulation of wild-type MCF-7 cells was maximal at 10 -10 mol/l E 2 . Hence, our mER high cells exhibit an intermediary proliferation pattern between wild-type and MCF-7 LTED cells with respect to the E 2 dose–response curves reported by Santen and coworkers [ 21 ]. Our mER low cells responded to E 2 by dose-dependent enhancement of their cell numbers. These cells have much lower levels of mER but some is present, and they express the intracellular ER at comparable levels to mER high cells (see accompanying report [ 22 ]). MCF-7 breast cancer cells have also been shown to respond to E 2 as an apoptosis survival factor after taxol or ultraviolet exposure [ 42 ]. This protective effect was directly dependent on the concentration of E 2 , and the highest effect was achieved with 10 nmol/l E 2 . This inhibition of apoptosis was also achieved with E 2 –bovine serum albumin, which indicated probable mediation through a plasma membrane ER. However, these studies did not directly assess the cellular level of mER (the populations studied were probably mixed, comprising cells with both high and low levels of mER). Therefore, it is hard to correlate their heterogeneous cell population results with our selected mER high or mER low cells. Whether protection from apoptosis plays any role in the survival of our mER low cells remains to be determined. Our findings also demonstrated that cAMP-activated PKA activity is associated with inhibition of cell proliferation. In mER high cells H-89 blocked the decrease in cell number caused by 10 nmol/l E 2 , and in mER low cells it further enhanced the already existing stimulation of proliferation. It is still not clear how cAMP inhibits cell proliferation. Lowe and coworkers [ 43 ] suggested that the inhibitory effect of cAMP (and forskolin, which induced it) occurs distal to extracellular signal-regulated kinase (ERK) activation, possibly by the inhibition of an ERK-independent pathway. However, other groups have reported that the cAMP/PKA pathway and mitogen-activated protein kinase (MAPK) pathway are connected in fibroblasts [ 43 , 44 ] and in MCF-7 cells [ 45 , 46 ]. Activated PKA phosphorylates Raf and inhibits phosphorylation of downstream MEK1/2 and Erk1/2, thus preventing proliferation, a counterbalance to E 2 -stimulated pathways that stimulate proliferation (Fig. 9 ). Our findings are consistent with the participation of PKA in inhibition of cell proliferation. In mER low cells, at both low (1 pmol/l) and high (10 nmol/l) E 2 concentrations, a very low amount of cAMP accumulated. Therefore, in these cells cAMP's inhibitory concentration is never achieved, although further inhibition of cAMP and its downstream effects (via inhibitors) could still enhance proliferation. In mER high cells, however, both low and high E 2 concentrations caused about the same amount of cAMP accumulation, but only at 10 nmol/l E 2 did the inhibitory pathway prevail. This finding suggests that the model depicted in Fig. 9 is simplified, and that other pathways participate and integrate at other levels, resulting in modulation of the cell's decision to proliferate, arrest, or die. It was recently shown that E 2 simultaneously activates the proliferative MAPK-Erk (via Src-Shc-Ras-Raf-MEK-Erk) and phosphatidylinositol-3 kinase pathways in MCF-7 cells, and that both pathways are required to trigger S-phase entry of cells. In addition, the temporal pattern of their activation is extremely important; a necessary immediate and transient consequence of hormone treatment is MAPK activation, whereas phosphatidylinositol-3 kinase activation must be sustained for at least 3 hours [ 47 ]. Our results represent a unique example of the extremely complex effects of steroid on target tissues, and emphasize the importance of a global approach when studying the proliferative effects of estrogens. Reports from other groups have indicated that there may be other membrane estrogen receptor types, such as GPR30 [ 48 ] and SHBG-R [ 24 , 49 ], which participate in E 2 signaling from the membrane. Clearly, a detailed analysis of different mER levels on the same cell backgrounds from a variety of responsive tissue types would be a valuable addition to these investigations. Finally, estrogen-dependent breast cancer cells synthesize and secrete growth factors in response to E 2 which can stimulate the epidermal growth factor receptor signaling pathway to induce proliferation [ 50 ]. Therefore, comparative studies of the contributions of each of these possible players are needed if we are to gain a complete understanding of E 2 -induced proliferation of breast cancer cells. Conclusion Our results indicate that a membrane-associated ER-α is responsible for rapid E 2 -induced cAMP accumulation and subsequent activation of the downstream PKA pathway. A high level of stimulation through the mER-α results in an interruption to breast cancer cell proliferation. Thus, knowledge of tumor mER-α levels and manipulation of estrogenic compound doses and resulting signaling mechanisms may offer an opportunity to refine treatment strategies for some patients with tumors that have these characteristics. Abbreviations AC = adenylyl cyclase; CV = crystal violet; DCSS = medium with 4 × dextran-coated charcoal-stripped serum; DM = defined medium; E 2 = 17β-estradiol; ER = estrogen receptor; ERK = extracellular signal-regulated kinase; IBMX = 3-isobutyl-1-methylxanthine; MAPK = mitogen-activated protein kinase; mER = membrane estrogen receptor; PDE = phosphodiesterase; PKA = protein kinase A. Competing interests The author(s) declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064104.xml

1064105

Membrane estrogen receptor-α levels predict estrogen-induced ERK1/2 activation in MCF-7 cells

Introduction We examined the participation of a membrane form of estrogen receptor (mER)-α in the activation of mitogen-activated protein kinases (extracellular signal-regulated kinase [ERK]1 and ERK2) related to cell growth responses in MCF-7 cells. Methods We immunopanned and subsequently separated MCF-7 cells (using fluorescence-activated cell sorting) into mER-α-enriched (mER high ) and mER-α-depleted (mER low ) populations. We then measured the expression levels of mER-α on the surface of these separated cell populations by immunocytochemical analysis and by a quantitative 96-well plate immunoassay that distinguished between mER-α and intracellular ER-α. Western analysis was used to determine colocalized estrogen receptor (ER)-α and caveolins in membrane subfractions. The levels of activated ERK1 and ERK2 were determined using a fixed cell-based enzyme-linked immunosorbent assay developed in our laboratory. Results Immunocytochemical studies revealed punctate ER-α antibody staining of the surface of nonpermeabilized mER high cells, whereas the majority of mER low cells exhibited little or no staining. Western analysis demonstrated that mER high cells expressed caveolin-1 and caveolin-2, and that ER-α was contained in the same gradient-separated membrane fractions. The quantitative immunoassay for ER-α detected a significant difference in mER-α levels between mER high and mER low cells when cells were grown at a sufficiently low cell density, but equivalent levels of total ER-α (membrane plus intracellular receptors). These two separated cell subpopulations also exhibited different kinetics of ERK1/2 activation with 1 pmol/l 17β-estradiol (E 2 ), as well as different patterns of E 2 dose-dependent responsiveness. The maximal kinase activation was achieved after 10 min versus 6 min in mER high versus mER low cells, respectively. After a decline in the level of phosphorylated ERKs, a reactivation was seen at 60 min in mER high cells but not in mER low cells. Both 1A and 2B protein phosphatases participated in dephosphorylation of ERKs, as demonstrated by efficient reversal of ERK1/2 inactivation with okadaic acid and cyclosporin A. Conclusion Our results suggest that the levels of mER-α play a role in the temporal coordination of phosphorylation/dephosphorylation events for the ERKs in breast cancer cells, and that these signaling differences can be correlated to previously demonstrated differences in E 2 -induced cell proliferation outcomes in these cell types.

Introduction Estrogen receptor (ER)-α has traditionally been defined as a ligand-dependent transcription factor that regulates its target genes by binding to estrogen response elements present in the promoters of many responsive genes [ 1 ]. However, an ever-increasing number of reports indicate that the cellular actions of estrogens can be initiated at the plasma membrane, through membrane versions of estrogen receptors (mERs) [ 2 - 4 ] or via other membrane-resident 17β-estradiol (E 2 )-binding proteins [ 5 ]. There is also evidence that mER-α from vascular endothelium and human MCF-7 breast cancer cells is localized in specialized cholesterol-rich membrane microstructures (caveolae), where it can associate with different signaling molecules and participate in various nongenomic actions [ 6 , 7 ]. A variety of rapid E 2 -induced signal transduction events can lead to stimulation of calcium flux, cAMP production, phospholipase C activation, and inositol phosphate production [ 8 ]. Mitogen-activated protein kinases (MAPKs) such as extracellular signal-regulated kinase (ERK)1 and ERK2 are also rapidly stimulated by estrogens in various cell types (e.g. endothelial [ 9 ], osteoblastic [ 10 ], neuroblastoma [ 11 ], and breast cancer cells [ 12 ]). However, the specific relationship of these responses to the levels of antibody-identified ER-α in the membrane has rarely been investigated [ 13 , 14 ]; other rapid estrogen-induced actions were specifically linked to mER-α in pituitary tumor cells in our previous studies [ 15 - 18 ]. The two isoforms of ERK (p42 and p44) play critical roles in the control of cell proliferation, differentiation, homeostasis, and survival. Traditionally, autophosphorylation of receptor tyrosine kinases after ligand binding initiates the cascade of phosphorylation steps that result in dual ERK phosphorylation (on Thr202 and Tyr204 in the human enzyme, or Thr183 and Tyr185 in the rat enzyme [ 19 ]). The signaling pathway initiated by E 2 at the level of the plasma membrane is not yet completely understood, although recent studies have implicated a cascade of intermediary proteins and signaling steps involving mER-α, G-proteins, Src-induced matrix metalloproteinases that liberate heparin-binding epidermal growth factor (EGF), and EGF receptor [ 13 ]; the involvement of many other signaling pathways remains unexamined. Whether different levels of mER can influence signaling parameters (such as kinetics and final levels of second messengers) that lead to physiological responses remains to be investigated. To address this question we separated MCF-7 cells into two subpopulations based on outer membrane-exposed mER-α levels and confirmed their differential mER-α expression by several methods. We investigated the association of mER-α with caveolin-rich membrane fractions in cells enriched for membrane display of these receptors. We then linked the level of mER-α to the magnitude and patterns of E 2 -induced ERK1/2 activation. Although activation of kinases was previously demonstrated, those other studies did not address the accompanying inactivation mechanisms for ERKs involving several specific cellular phosphatases. Methods Cell immunoseparation and subculturing Our MCF-7 cells originated from the Michigan Cancer Center. We separated them into two subpopulations by immunopanning [ 16 , 20 ] using C-542 carboxyl-terminal ER-α antibody provided by Drs Dean Edwards and Nancy Weigel; this antibody is now commercially available from Stressgen Biotechnologies (Victoria, Canada). Briefly, sterile antibody on the surface of a petri plate bound cells over a 1-hour time period at 4°C, and cells that attached to the plate (mER + ) were propagated separately from those that did not bind (mER low ). The mER + cells were then subjected to further selection via fluorescence-activated cell sorting (FACS) using the same antibody [ 21 ]. Cells undergoing sequential separation were highly enriched for mER-α (mER high ). All cell subpopulations were routinely cultured in phenol-red free Dulbecco's modified eagle medium (Gibco-BRL, Grand Island, NY, USA) supplemented with 10% heat-inactivated DSCS (defined/supplemented bovine calf serum; HyClone Laboratories Inc., Logan, UT, USA) and 1% of an antibiotic–antimycotic (Gibco Invitrogen Corporation, #15240-062). Cells used in the experiments were between passage 8 and 14 after separation. Three days before each experiment, the cell growth medium was replaced with medium containing 4 × dextran-coated charcoal stripped serum (DCSS medium) or completely defined medium (DM; phenol-red free Dulbecco's modified eagle medium with 5 μg/ml each of insulin and transferrin, 5 ng/ml selenium, and 0.1% bovine serum albumin). Stripped serum was produced by incubation with an equivalent volume of packed 0.25% weight/vol charcoal Norit A for 2 hours on a shaker at 4°C. The charcoal preparation had previously been suspended in 0.0025% (weight/vol) dextran T-70 (Sigma-Aldrich, St. Louis, MO, USA) in 0.25 mol/l sucrose, 1.5 mmol/l MgCl 2 , and 10 mmol/l HEPES (pH 7.6). The charcoal was pelleted at 500 g for 10 min, the supernatant (stripped serum) decanted, and the process repeated a total of four times. Fluorescence immunocytochemical detection of membrane ER-α The protocol published for immunocytohcemical detection of mER-α in the GH 3 rat pituitary cell line [ 22 ] was optimized for MCF-7 cells. Briefly, the cells were fixed in 2% paraformaldehyde/0.1% glutaraldehyde in phosphate-buffered saline for 25 min at room temperature; those conditions preserve the integrity of the cell membrane and enable detection of the specific ER-α antibody bound to the cell surface without interference from the usually more intense nuclear signal. Autofluorescence was reduced by blocking the aldehyde groups for 15 min with 1% Na 2 HPO 4 and 0.05% NaBH 4 . Nonspecific binding sites were blocked with 0.1% fish gelatin in phosphate-buffered saline for 45 min at room temperature, which we successfully used previously in GH6/B6/F10 cells [ 23 , 24 ]. For the present studies with MCF-7 cells, we compared the nonspecific antibody signal blocking abilities of fish gelatin, horse serum, bovine serum albumin and nonfat dry milk, all commonly used methods, and found them to be equally useful. The plates were incubated with C-542 antibody (5–10 μg/ml) overnight at 4°C in the blocking solution. To detect the bound primary antibody and amplify the signal, we used a biotinylated secondary antibody from the biotin/avidin Vectastain ABC-Alkaline Phosphatase kit and Vector red as a substrate, together with 40 μl of 125 mmol/l levamisol (which inhibits the endogenous alkaline phosphatases). All the components for detection of bound primary antibody were obtained from Vector Laboratories Inc. (Burlingame, CA, USA). Fluorescence photography was performed as previously described [ 15 , 22 ]. The images were photographed with Kodak HC4000 color film and a camera (Model C-35AD-4) with Olympus AHBT microscope and fluorescence attachment (model AH2-RFL) using the FITC filter, under 100 × magnification. Digital deconvolution and pseudo-coloring were performed with Image-Pro Plus software applied to images captured through the FITC filter with a CoolSNAP-Pro digital monochrome camera and a ProScan motorized stage (Meyer Instruments, Houston, TX, USA) attached to a Olympus AHBT microscope equipped with fluorescence attachments (model AH2-RFL). Quantification of membrane ER-α The mER was quantified with a protocol modified from one previously developed in our laboratory for GH 3 cells [ 23 ]. Briefly, cells plated and treated in 96-well plates were fixed as described for immunocytochemistry (see above), and the integrity of the membrane was verified by lack of staining with the anti-clathrin antibody (ICN Biomedicals, Aurora, OH, USA), as clathrin is localized just inside the plasma membrane. Different concentrations of C-542 ER-α antibody were tested (in the range 1–12 μg/ml), and the tagging enzymatic reaction (alkaline phosphatase) was monitored for different time intervals (5, 15 and 30 min) in order to determine optimal conditions for measurement. The specificity of the C-542 antibody was checked by comparing its binding with the nonspecific binding of mouse IgG 1k (mIgG 1k , of the same immunoglobulin isotype) and by the ability of the peptide representing the C-542 epitope (Genosys, Woodlands, TX, USA) to decrease C-542 binding. Other controls included incubation without any antibody to detect endogenous phosphatase not inhibited by levamisol, and without primary antibody to detect nonspecific binding of secondary antibody. The total cellular ER-α was measured by applying the same procedure to cells permeabilized by including 0.1% of the non-ionic detergent IGEPAL CA-630 (Sigma-Aldrich) during the fixation procedure. The absorbance of the alkaline phosphatase product paranitrophenol (pNp) in each well was measured at 405 nm and normalized against the number of cells determined by the absorbance of crystal violet (CV) at 590 nm, as previously described [ 23 ]. Caveolae preparation and Western analysis To concentrate caveolin-rich membranes, we extended a previously published protocol [ 25 ] by introducing a dialysis step to remove sucrose, and a vacuum spin at a low drying rate to concentrate the samples. Specifically, cells were seeded in three 150 mm diameter plates and grown in serum-supplemented medium until 60% confluent. The growth medium was replaced with DCSS medium devoid of antimycotic compound and cultured for an additional 3 days. Cells from all three plates were collected in 1 ml lysis buffer consisting of 50 mmol/l Tris/HCl (pH 7.5), 5 mmol/l EDTA, 100 nmol/l NaCl, 50 mmol/l NaF, 1 mmol/l PMSF, 0.2% TritonX-100 and protease inhibitor cocktail P8340 (100 × diluted; Sigma-Aldrich). Cells in solution were passed through a 25-g syringe needle, then homogenized with 25 strokes using a Dounce B-type pestle. The homogenate was adjusted to 45% sucrose by addition of an equivalent volume of 90% sucrose. A discontinuous sucrose gradient consisting of the sample, 35% sucrose, and a top layer of 5% sucrose was centrifuged for 18 hours at 200,000 g . Fractions of 1 ml were collected and checked for the presence of caveolin-1 and caveolin-2 by Western analysis using antibodies from Biosciences (San Jose, CA, USA). Fractions 5 and 6 (the interface between 5% and 35% sucrose) contained the highest amount of caveolin-1. Those two fractions were pooled and dialyzed overnight against the lysis buffer. The sample was then concentrated by vacuum spin, and 20 μg of these proteins separated by 4–20% SDS-PAGE. The proteins were transferred to nitrocellulose membranes and the mER-α bands were probed with C-542 ER-α antibody. After incubation with secondary antibody (conjugated with horse radish peroxidase) the bands were visualized with an ECL kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). Fixed cell-based enzyme-linked immunosorbent assay detection of activated ERK1/2 in 96-well plates A protocol previously developed in our laboratory for other cells [ 26 ] was optimized for MCF-7 cells. Cells were plated at a density of 4000/well in 96-well plates, and after 24 hours the growth medium was replaced with DCSS medium. After 3 more days of culture the cells were treated with E 2 (1 pmol/l) for different time intervals (3, 6, 10, 20, 30 and 60 min), or with different E 2 concentrations (from 10 -16 to 10 -7 mol/l) for 10 and 6 min for mER high and mER low cells, respectively. After treatments the cells were fixed in 2% paraphormaldehyde/0.21% picric acid for 2 days at 4°C. The cells were subjected to a 60 min blocking step with 0.1% fish gelatin and 0.1% Triton X-100 at room temperature. Incubation with the antibody raised against the phosphorylated forms of ERK1/2 (Cell Signaling Technology, Beverly, MA, USA) was performed overnight at 4°C (1:400 dilution). To quantify active ERKs, biotinylated secondary antibody (anti-mouse/anti-rabbit) conjugated to alkaline phosphatase was applied. Substrate pNp phosphate was added and incubated for 25 min at 37°C, and the absorbance of the pNp product was determined at 405 nm in a plate reader (Wallac 1420; Perkin Elmer, Boston, MA, USA). The levels of phosphorylate (pERK1/2) were normalized to the cell number in each well (measured using the CV assay). To confirm the activation of ERK1/2 (after 3, 6 and 10 min in mER high or 6 min in mER low ), we pretreated the cells for 15 min with 40 μmol/l U0126 MEK1/2 inhibitor (Cell Signaling Technology). The ER antagonist ICI182,780 (Tocris Cookson Inc., Ellisville, MO, USA) at a concentration of 1 μmol/l was tested with or without E 2 by preincubating the cells with ICI182,780 for 30 min followed by the addition of 1 pmol/l E 2 or by simultaneous addition of ICI182,780 and E 2 . MDA-MB-231 cells used to test the requirement of ER-α for these responses were obtained from ATCC (Manassas, VA, USA). We confirmed the absence of ER-α mRNA in these cells by multiple reverse transcription polymerase chain reaction primer pairs representing the ER-α sequence (data not shown). Cell proliferation Cells were plated at a density of 1000 cells/well in 96-well plates. The next day the growth medium was replaced with DCSS containing different treatments. The 1 pmol/l E 2 was present either for the duration of the experiment or as a short pulse treatment (10 min), whereas 1 pmol/l E 2 -peroxidase (Sigma-Aldrich; concentration based on the E 2 content of the conjugate) was used only as short pulse treatment. The effect of MEK inhibitor (40 μmol/l U0126) on the pulsed E 2 -induced proliferation of mER high cells was tested by a pretreatment for 15 min and an additional treatment for 10 min with E 2 together with the inhibitor. To block the effect of E 2 in the pulse treatment, we used ER-α antibody (AER315; NeoMarkers Inc., Fremont, CA, USA) raised against the ligand-binding domain. The cells were pretreated with 1 μg/ml AER315 antibody for 1 hour at room temperature, followed by a 10 min incubation at 37°C with 1 pmol/l E 2 in the presence of the antibody. Controls for E 2 treatment were done on a separate plate, because we previously determined that low amounts of volatilized estrogens can affect responses mediated via nongenomic signaling pathways [ 24 ]. After 5 days the cells were fixed with 2% paraformaldehyde/0.1% glutaraldehyde in phosphate-buffered saline in preparation for the CV assay (see below). Crystal violet assay The number of the cells in each well was determined with the CV assay [ 27 ], which we modified previously [ 23 ]. Fixed cells were incubated in 0.1% filtered CV solution for 30 min at room temperature, and excess dye was removed by three brief rinses with ddH 2 O. The plates were then air dried, the dye was extracted with 10% acetic acid, and the extract (in the same wells) was then read in a plate reader (Wallac 1420, Perkin Elmer) at 590 nm. The utility of this assay was previously verified for GH3B6 cells by comparison with other assays of cell number in combination with the immunoplate assays [ 23 ]. In addition, for MCF-7 cells we verified the utility of this assay for measuring cell number by comparison with DNA content measurements and with cell counts by hemocytometer (data not shown). We also compared the CV assay with the MTT assay, which is often used to determine viable cell number (ATCC), and we obtained a linear correlation for the two assays. These latter results are presented in the accompanying paper [ 28 ]. Statistical analysis Statistical differences between two sets of data were determined using two-way analysis of variance. The differences between the entire curves were tested by comparing the sum of squares of the residuals from each individual curve with the sum of squares of the residuals of the combined curve by applying a Microsoft Excel F test. P < 0.05 was considered statistically significant. Results Immunoseparated cell characterization of mER-α Immunopanning and subsequent FACS successfully separated MCF-7 cells into two populations according to the expression of mER-α observed in immunocytochemistry experiments. Punctate staining can be seen on the surface of unpermeabilized mER high cells (Fig. 1a ), whereas the majority of mER low cells did not exhibit this staining (Fig. 1b ). Whenever occasional staining was present on cells in the mER low population, its appearance was similar to that seen on mER high cells (data not shown). Secondary antibody staining alone was at levels similar to that shown for the mER low cells in Fig. 1b (not shown). When permeabilized, both subpopulations of cells exhibited plentiful cytoplasmic and nuclear staining (not shown) at similar levels. Digital deconvolution (in 15 separate cell planes) performed on a grouping of three unpermeabilized mER high cells clearly demonstrated punctuate staining all along the periphery of these cells (Fig. 2 ). To determine whether mER-α is in a submembrane location in our mER-α-enriched cells, we colocalized mER-α with caveolin proteins in gradient-separated membrane fractions. Our mER high cells express both caveolin-1 (form α and β represented by doublet bands) and caveolin-2 (Fig. 3a ; two different concentrations of protein were loaded in adjacent lanes). In the same density gradient fractions that contained the majority of these caveolar structural proteins (fractions 5 and 6), mER-α was found (Fig. 3b ). We detected several prominent bands from both the sucrose gradient fractions (fractions 5 and 6) and the whole cell lysates (Fig. 3b , panel 1). To determine those bands whose signal was specific for C-542 ER-α antibody, we performed an epitope competition (preblocking of antibody with peptide representing the epitope). We identified two ER-α bands competable with peptide: the 67 kDa (the size of classical ER-α) and a prominent lower band at 52 kDa (Fig. 3b , compare panels 1 and 2). To quantify mER-α with our plate assay, we first determined the lowest optimal C-542 antibody concentration that would saturate binding to the membrane form of the receptor, and then optimized assay conditions to give a reliably detectable signal in the linear range of the fluorescent product assay. All three signal generation incubation times (5, 15 and 30 min) gave acceptable data, but 15 min assays allowed clear delineation of antigen-saturating concentrations of C-542 antibody utilizing lower antibody concentrations (8 μg/ml; Fig. 4a ). Those conditions were used for subsequent assays. The same level of mER-α could be detected in cells kept for 3 days in a completely defined medium lacking serum or in the same medium supplemented with 10% DCSS (Fig. 4a , open circles and closed circles, respectively). The background values (obtained in the absence of primary and secondary antibody, or with no primary antibody) were very low (Fig. 4b ). At 8 μg/ml the isotype control mIgG 1k yielded a value only 10% of that for C-542 antibody; peptide competition resulted in an approximate 50% decrease in C-542 antibody binding, again verifying the specificity of the antibody for this epitope. This verified that the assay was specific, and not subject to interference by endogenous alkaline phosphatases in the presence of levamasole. The fixation conditions that we used preserved the integrity of the cell membrane, as demonstrated by the negligible anti-clathrin antibody binding in nonpermeabilized versus the very high binding in permeabilized cells. Therefore, equivalent protocols differing only in the inclusion of a permeabilization agent during fixation can be used to quantify the total ER (tER) versus the mER. Because we had previously observed an effect of cell density on the level of ER expression in GH 3 /B6/F10 cells, which occurred at a density when cells had just begun to make contact by cell processes [ 24 ], we wondered whether the same regulation was applicable to breast cancer cells. Considering this effect, plating a single density may not demonstrate the ability of cells to express mER-α. Therefore, we performed a cell density study for mER high versus mER low breast cancer cell types (Fig. 5 ). The mER-α signal decreased exponentially with increasing cell number (Fig. 5a ). At low plating densities, mER high cells clearly showed much higher levels of mER than did mER low cells (Fig. 5 , closed circles and open circles, respectively). The two curves approximating the level of mER were significantly different ( P = 0.0003). The ER-α-negative cell line MDA-MB-231 did not have mER-α (Fig. 5 , diamond), even at relatively low cell density, because their value was at the level of mIgG 1k isotype control antibody levels (horizontal hatched line). Increased cell density also decreased tER (Fig. 5b ); however, mER high and mER low MCF-7 cells exhibited the same level of tER, because all of these data could be approximated with the same curve. Because mER high cells had higher mER-α levels but the same tER-α levels as mER low cells, then, by subtraction, mER high cells have lower intracellular ER-α levels than do mER low cells. Kinetics of ERK1/2 activation by 17β-estradiol in MCF-7 cells enriched and MCF-7 cells depleted for membrane ER-α During the optimization of the fixed cell-based enzyme-linked immunosorbent assay for ERK activation, we established the optimal cell density for a 10 min EGF treatment, which is known to result in substantial phosphorylation of ERK1/2. Both the controls and EGF-treated cells exhibited increased pERK1/2 with increased cell number (Fig. 6a , main graph). As we expected, normalizing pERK1/2 values to the number of cells (CV absorbance at 590 nm) did not significantly change the ratio values except in the case of the highest number of cells plated, verifying that cells plated in the density range of CV values 0.2–0.6 could be used. Partly because others have shown that ERKs can be activated by mechanical stimulus in MCF-7 cells [ 29 ], we tested ethanol-treated controls over the same time course (3–60 min). A pronounced decrease in ERK1/2 phosphorylation was seen with time (Fig. 6b ), and so appropriate controls were performed for each time point in all subsequent experiments. In mER high MCF-7 cells, ERK1/2 activation with 1 pmol/l E 2 was fast and transient (Fig. 7a ). The maximal activation was achieved after 10 min, followed by a rapid decline in phosphorylated ERK1/2. However, continued incubation with E 2 (1 hour) resulted in the reactivation of ERK1/2. To test whether this signal was initiated at the membrane, the impeded ligand E 2 -peroxidase was applied to the cells at steroid concentrations that approximated the levels applied as free steroid in the previous experiments (Fig. 7b ). To eliminate any free steroid present, just before use we pretreated the E 2 -peroxidase with dextran-coated charcoal under conditions that remove more than 99% of free hormone [ 9 ]. The resulting maximal level of ERK1/2 activation was slightly higher than for treatment with the free ligand, but the peak time of activation was the same. Again, a recurrent later ERK activation was observed. Cells with lower levels of mER-α (Fig. 7c ) also had the capacity for fast and transient activation of ERK1/2, but this smaller activation peak appeared at 6 min after 1 pmol/l E 2 treatment. The levels of phosphorylated ERK declined between 10 and 30 min of E 2 treatment as they had with mER high cells; however, at longer incubations (1 hour) no reactivation was seen but rather a further ERK1/2 apparent dephosphorylation was observed. This implies that higher levels of mER-α associated with more robust early ERK activation are also responsible for the sustained ERK reactivation at the later stage. The inhibitor of the upstream MEK1/2, namely U0126, was effective in inhibiting ERK1/2 activation in both types of MCF-7 cells (Fig. 6a and 6c , insets), verifying that the values we measured in our plate assay were from MEK-phosphorylated ERK. In the MDA-MB-231 ER-α-negative cell line (Fig. 7d ), E 2 could not significantly activate ERK1/2, confirming that ER-α is necessary for ERK activation during this 60 min time period. Dose-dependent activation of ERK1/2 by 17β-estradiol is influenced by the level of membrane ER-α expression In mER high cells, the ability of E 2 to induce ERK activation was biphasic with respect to dose at the 10 min time point (early response peak). ERK phosphorylation was stimulated at a wide range of concentrations from 0.1 pmol/l to 100 nmol/l E 2 , although the highest E 2 concentrations resulted in less phosphorylation (Fig. 8a ). In mER low cells a biphasic response was also seen, but the only effective concentrations were 0.1 and 1 pmol/l for the 6 min (early and only) response peak (Fig. 8b ). Physiologic significance of early ERK1/2 activation Long-term treatment of mER high MCF-7 cells with 1 pmol/l E 2 resulted in substantial stimulation of proliferation (Fig. 9 ). A 10 min short pulse treatment also resulted in significant although lower stimulation of proliferation. The same level of stimulation was achieved with both E 2 and E 2 -peroxidase presented for a short pulse. E 2 -induced proliferation was prevented with MEK inhibitor (U0126) as well as with a specific ER-α antibody recognizing the ligand binding domain. These results are consistent with the participation of mER-α and ERK1/2 in the cell proliferation response. Phosphatase inhibitors differentially affect ERK activation in MCF-7 cells enriched and depleted for membrane ER-α We next asked whether the observed decrease in phosphorylated ERK1/2 after 20 min in both subpopulations of cells, and the continued low phosphorylation levels after 60 min in mER low cells, could successfully be abrogated with specific phosphatase inhibitors (Fig. 10 ). We tested inhibitors of protein phosphatase (PP)1, PP1A, and PP2B. These phosphatases can be considered principal enzymes of this class, based on their general abundance and broad specificity [ 30 ]. Okadaic acid (OA), an inhibitor of PP1 and PP1A, and cyclosporin A, an inhibitor of PP2B, were both able to reverse the ERK inactivation in mER high cells (Fig. 10a ). In mER low cells, both the 20 min and continued 60 min dephosphorylation were abrogated only with the PP2B inhibitor (Fig. 10b and 10c ). Because of the known apoptotic effect of OA at some concentrations [ 31 ], it is important to stress that we applied it at a low concentration (50 nmol/l). In addition, OA does not have toxic effects in short-term incubations [ 32 ]. These results suggest that dephosphorylation of ERKs is an important component of their process of action and that coordinated phosphorylation/dephosphorylation is required for strong signaling through this pathway. Rapid effects of ICI182,780 and 17α-estradiol on ERK1/2 activation To characterize the pharmacologic properties of ERK activation, we used mER high cells to study the effectiveness of the potent antiestrogen ICI182,780 (1 μmol/l) and the inactive E 2 stereoisomer 17α-estradiol (10 nmol/l). We used concentrations of these compounds shown by others to be effective in inhibiting the transcriptional activity of E 2 . We had also previously shown that a 10 nmol/l 17α-estradiol concentration could elicit another type of nongenomic estrogenic effect in our GH 3 /B6/F10 pituitary tumor cell model – rapid prolactin release [ 15 ]. ICI182,780 alone induced an activation pattern very similar to that seen with E 2 but with an earlier first peak (Fig. 11a ). A 30 min ICI182,780 preincubation before a subsequent 1 pmol/l E 2 challenge did not significantly change the E 2 activation pattern, although the first peak again appeared at 6 min (the ICI182,780 pattern) and there was a much more pronounced inactivation at 10 and 20 min (compare Fig. 11b with Figs 7a and 11a ). However, simultaneous application of ICI182,780 (1 μmol/l) and E 2 (1 pmol/l) blunted the response and altered the kinetics of ERK phosphorylation, shifting the now single weak activation to later times (20–60 min; Fig. 11c ). The E 2 stereoisomer (17α-estradiol) provoked a slightly delayed and blunted response also, but with some other interesting features (Fig. 11d ). A significant dephosphorylation occurred before the major activation peak, a return to baseline phosphorylation levels followed the 20 min activation peak, and no second activation peak occurred at 60 min. Discussion In the late 1970s, Pietras and Szego [ 33 ] reported the presence of high-affinity binding sites for E 2 associated with the plasma membranes of the MCF-7 human breast cancer cell line. Since that time few laboratories have followed up on this finding [ 34 , 35 ] until very recently, when more detailed studies started to emerge. However, none of these studies established a correlation between the level of identified mER-α expression and its functions. To address this issue we used immunopanning followed by FACS to separate MCF-7 cells into two subpopulations that were enriched and depleted for mER-α expression. We then used several approaches to assess the appearance and levels of mER-α in these two subpopulations. These studies demonstrated that MCF-7 cells are heterogeneous with respect to mER-α expression, and that difficulties in reproducibly demonstrating nongenomic estrogenic effects in these cells could in part be due to the dilution of responding cells in the largely nonresponsive total cell population. We were able to obtain two distinct cell subpopulations without applying plasmid-based transfection manipulations to over-express receptor (with probable accompanying altered regulatory parameters). Our experimental model avoids changes in stoichiometry of the multiple interacting proteins that are involved in steroid actions. Because our mER high and mER low cells were not clonally derived (from a single cell), the observed differences between these subpopulations cannot be attributed to clonal variations. A characteristic punctate staining for mER (earlier reported for GH 3 /B6/F10 cells [ 22 ]) was also detected on the surface of most of our mER high MCF-7 cells. MCF-7 mER low cells also had occasional punctuate staining, but at a much lower level. Because it is difficult and time-consuming to quantify this staining via immunocytochemistry, we modified our previously developed 96-well plate immunoassay [ 23 ] to measure both membrane and intracellular receptors in breast cancer cells, and to quantitate the relative amounts in these two receptor subpopulations. This assay must be tailored to specific cell types to ensure preservation of the membranes (which have quite different compositions in different cell types) and to optimize for different antibody labeling systems. Although mER-α levels differed between the two cell types named for these differences (high versus low), we discovered that the two subpopulations had the same quantity of total receptor (measured in permeabilized cells). This finding is consistent with intracellular and membrane fractions of ER-α being from the same ER pool, but with a different balance of subcellular distribution. There is disagreement in the literature on the expression of caveolin-1 and -2 in MCF-7 cells. To test whether mER is localized in caveolar membranes, we had first to resolve this uncertainty. Some have reported that in MCF-7 cells caveolin-1 is downregulated and only caveolin-2 is expressed [ 36 , 37 ]. Others reported that caveolin-1 could be upregulated and downregulated in MCF-7 cells in concert with the cells' ability to develop drug resistance [ 38 ]. Some studies used transient transfection to re-express the missing caveolin-1 and test for its function in signaling [ 7 , 39 ]. Apparently, there are several different subpopulations of MCF-7 cells growing in different laboratories [ 40 , 41 ]. Our MCF-7 cells expressed both caveolin-1 and -2, and ER-α was associated with the same gradient fractions in which most of the caveolin proteins were detected. A 67 kDa ER and a variant of lower molecular weight (52 kDa) were detected in both the cell lysate and the caveolar membrane gradient fractions. Until recently the lower molecular weight ER variants were thought to be proteolytic fragments, but evidence that some of those molecules are truncated products of the full-length ER-α was recently presented. It is well documented that in MCF-7 breast cancer cells ER-α mRNA can undergo alternative splicing, generating transcripts with single, double, or multiple exon deletions [ 42 ], and that a 52 kDa protein is translated from the predominant splice variant mRNA that is missing exon 7 [ 43 ]. Others have identified a 46 kDa variant in human endothelial cells [ 44 ] and in MCF-7 cells [ 45 , 46 ]. Some evidence suggests that the 46 kDa receptor is the major functional membrane form of ER [ 44 ]; however, the precise functions of the truncated ER-αs are still under investigation. The traditional detection and quantitation of phosphorylated MAPKs via Western blot analysis is laborious and somewhat arbitrary in the designation of appropriate densitometric backgrounds for comparison, especially in situations where the significant activation response is not large. We developed an enzyme-linked immunosorbent assay to deal with these experimental problems using fixed cells [ 47 ], which allowed convenient testing of large numbers of different conditions. Ours is the first report that different levels of mER in MCF-7 cells influence the different temporal and dose-dependent estrogen-induced phosphorylations of ERKs. The subpopulation of cells with high mER levels exhibited early and more robust activation, peaking at 10 min with a reactivation at 60 min, whereas the subpopulation of cells with low mER levels were only capable of weakly activating ERKs at one early time point (6 min). Furthermore, in the ER-α-negative cell line MDA-MB-231, E 2 could not activate these kinases. The physiologic significance of early ERK1/2 activation was confirmed by showing that short pulse E 2 treatment also stimulated cell proliferation. We and others [ 46 ] have confirmed that mER-α in breast cancer cells is responsible for this effect by showing that E 2 -peroxidase can stimulate proliferation and that this effect can be abolished by prior blocking of ER-α with ligand-binding domain-specific antibody. The inability to activate ERKs with E 2 in an ER-α-negative cell line, and the ability to do so in MCF-7 ER-α-positive cells, was assigned by others to the orphan G-protein-coupled receptor GPR30 [ 48 ]. However, recent studies with antisense GPR30 knockdowns revealed that E 2 -stimulated MCF-7 cells proliferate as well as cells with normal levels of GPR30 [ 49 , 50 ]. ERK activation associated with the ability of these cells to respond to estrogens by proliferating thus does not appear to be a function of GPR30 levels. Different levels of mER also determined the E 2 dose-dependent phosphorylation of ERKs. Cells with low levels of mER responded to a limited range of concentrations (10 -13 to 10 -12 mol/l). However, mER high cells responded to a much wider range of E 2 concentrations (10 -13 to 10 -7 mol/l) with a declining (but still significant) ERK activation at 10–100 nmol/l E 2 . This finding corresponds to our observation that 10 nmol/l and higher E 2 decreases cell proliferation in mER high MCF-7 cells, and is consistent with the idea that E 2 induces cAMP-activated protein kinase A inhibition of MAPK pathways at these higher concentrations (see accompanying paper [ 28 ]). Activation of ERKs has been linked to the proliferative cellular response [ 51 ]. Our accompanying report [ 28 ] addresses other rapid estrogen-elicited signaling responses that, in concert with ERK activation, can affect and perhaps balance cell proliferation responses. The classical antiestrogen ICI182,780 has been well defined as an antagonist of the action of estrogen at the transcriptional level. In our studies this compound was also a potent and rapid ERK activator. With different kinetics, the transcriptionally inactive 17α-estradiol also activated ERKs. It has been demonstrated that pure antiestrogen–ER complex can bind to estrogen response elements, but that the resultant transcriptional unit is inactive [ 52 ]. This binding and inactivation is generally used for pharmacological identification of ER-α participation in gene transcription. However, in a yeast reporter system antiestrogens (ICI182,780 and tamoxifen) induce ER dimerization and transcriptional activity [ 53 ]. Thus, although the published data disagree, it remains possible that antiestrogen–ER complexes can exert rapid effects such as induction of ERK phosphorylation and other signaling effects [ 54 , 55 ]. Other differences between findings may stem from effects assessed at different time points [ 56 - 58 ]. The length of time between estrogen or antiestrogen administration and kinase assessment can clearly influence outcomes, as evidenced by our time- and compound sequencing-dependent changes in ERK activation. It is also known that lengthy exposures to ICI182,780 can dramatically decrease ER-α protein levels [ 47 , 52 ], which could explain the decreased ERK activation observed in some studies. Long-term estrogen-deprived MCF-7 cells express higher levels of ER than do wild-type MCF-7 cells and are hypersensitive to E 2 ; ICI182,780 did not alter the pattern of response to E 2 -stimulated growth in these cells [ 59 , 60 ]. To reconcile this observation with the expected inhibitory effect of ICI182,780, the authors suggested that ICI182,780 blocked only the effect of the residual E 2 coming from plastic tissue culture flasks, without affecting the added E 2 . We can speculate that the observed effect of ICI182,780 was genuine and resulted from potentially increased levels of mER in long-term estrogen deprived MCF-7 cells, because our previous studies suggested that serum deprivation elevates mER-α levels [ 61 ]. In other cell types ICI182,780 behaved as a potent agonist of ERK1/2 activation (e.g. in rat cerebellar neurons expressing plasmid-generated ERs [ 60 ]), or as a partial ER agonist (in the superficial stroma and glandular epithelium of the sheep uterine endometrium [ 62 ]). It is likely that subtle differences in the shape of estrogen- and antiestrogen-liganded ERs, whether nuclear or membrane, present different interaction platforms for a variety of co-modulators, and that each signaling pathway must be considered in the context of a particular cell type's repertoire of partnering proteins. In addition to having positive effects by itself, ICI182,780 was able to alter the E 2 response differently, dependent on the timing of its administration. If the cells were pretreated for only 30 min (the time point when ICI182,780 alone did not change basal ERK activation), then the initial E 2 activation and the late re-activation at 60 min were preserved, whereas the after-peak deactivation was enhanced. If ICI182,780 and E 2 were added simultaneously, then the activation was delayed to 20 min, but weakly persisted up to 1 hour. Similar observations that ICI182,780 can change the time of ERK activation were reported in the case of a human thyroid carcinoma cell line, in which it reduced the 30 min to 1 hour activation, but enhanced later sustained activation at 6 hours [ 63 ]. The same authors showed that, in differentiated thyroid gland carcinoma cells (XTC133), addition of ICI182,780 induces a small decrease in the sustained ERK phosphorylation. Thus, in addition to demonstrating that ICI182,780 is not just an antagonist for this response, these studies and ours may point to cell-specific differences in ERK regulation. Phosphorylations leading to activation and subsequent inactivation of proteins are important regulatory mechanisms for control of cell growth and differentiation [ 64 , 65 ]. Although ERKs are thought to play a key role in cell proliferation, it has been suggested that persistent activation of ERKs might result in cell cycle arrest and differentiation in PC12 cells, or cell growth inhibition and apoptosis in normal rat hepatocytes and several human tumor cell lines [ 65 ]. However, a detailed time course and transient ERK activations, interspersed with phosphatase-mediated inactivations, are rarely addressed. We (in the present study) and others [ 66 ] showed that the serine/threonine phosphatases PP1 and PP2A participate in the dephosphorylation of ERKs. It is interesting that OA (a PP1 and PP2A inhibitor) was more effective in mER high cells, whereas cyclosporin A (a PP2B inhibitor) was almost equally efficient in both cell populations. These data suggest that the level of mER-α expression/activation can be associated with activation of different types of ERK-modulating phosphatases. Other phosphatases, such as tyrosine phosphatase and dual specificity phosphatases (which dephosphorylate tyrosine, threonine, and serine residues), have been implicated in ERK dephosphorylation [ 67 , 68 ] and will be the subject of our future studies. Additional studies of ERK activations, deactivations, and stability will be needed before we can formulate a more global picture of the post-translational modifications that lead to function of this important group of regulators in proliferation and differentiation. However, it is clear that nongenomic estrogen actions and the membrane receptors through which they act participate in this regulation. Conclusion E 2 -induced changes in breast cancer cell number [ 28 ] can be directly related to ERK1/2 activation/deactivation patterns and interacting signaling mechanisms. The differential behavior of cell lines expressing different levels of mER-α suggests a role for this receptor in the temporal coordination of phosphorylation/dephosphorylation events affecting the mitogen-activated kinases ERK1 and ERK2. Abbreviations CV = crystal violet; DCSS = medium with 4 × dextran-coated charcoal-stripped serum; E 2 = 17β-estradiol; EGF = epidermal growth factor; ER = estrogen receptor; ERK = extracellular signal-regulated kinase; FACS = fluorescence-activated cell sorting; OA = okadaic acid; MAPK = mitogen-activated protein kinase; MEK = mitogen-activated protein kinase kinase; mER = membrane estrogen receptor; pNp = paranitrophenol; PP = protein phosphatase; tER = total estrogen receptor. Competing interests The author(s) declare that they have no competing interests.

D:\keerthana\PMC001xxxxxx\PMC1064105.xml

1064108

Alphavirus replicon particles containing the gene for HER2/neu inhibit breast cancer growth and tumorigenesis

Introduction Overexpression of the HER2/ neu gene in breast cancer is associated with an increased incidence of metastatic disease and with a poor prognosis. Although passive immunotherapy with the humanized monoclonal antibody trastuzumab (Herceptin) has shown some effect, a vaccine capable of inducing T-cell and humoral immunity could be more effective. Methods Virus-like replicon particles (VRP) of Venezuelan equine encephalitis virus containing the gene for HER2/ neu (VRP- neu ) were tested by an active immunotherapeutic approach in tumor prevention models and in a metastasis prevention model. Results VRP- neu prevented or significantly inhibited the growth of HER2/ neu -expressing murine breast cancer cells injected either into mammary tissue or intravenously. Vaccination with VRP- neu completely prevented tumor formation in and death of MMTV-c- neu transgenic mice, and resulted in high levels of neu -specific CD8 + T lymphocytes and serum IgG. Conclusion On the basis of these findings, clinical testing of this vaccine in patients with HER2/ neu + breast cancer is warranted.

Introduction The management of breast cancer currently relies on surgery, chemotherapy and radiotherapy. Despite recent advances in clinical management of breast cancer once metastasis has occurred, the probability of a complete cure is greatly reduced. Of the women who have no detectable lymph node metastases at the time of diagnosis, up to one-third later develop metastases [ 1 ]. In patients with metastatic disease that does not respond to radiotherapy or chemotherapy, immunotherapy may offer an additional form of cancer control [ 2 - 4 ]. Clinical trials of trastuzumab, a monoclonal antibody specific for HER2/ neu , have demonstrated the utility of an immunologic approach for breast cancers that overexpress this gene [ 5 - 7 ]. A drawback to 'passive' immunotherapy using monoclonal antibodies is that the effect is short-lived. An alternative approach is active vaccination that could induce neu -specific cytotoxic T cells with the ability to control the growth of the primary tumor and metastases. However, unlike passive immunotherapy whose effectiveness quickly wanes, effector and memory T cells induced by vaccination may remain present and be able to respond to any metastatic cells expressing HER2/ neu that arise after treatment. HER2/ neu is an excellent target for gene vaccines, and several preclinical studies have shown the effectiveness of plasmid vaccines encoding neu in murine models [ 8 - 16 ]. Using a plasmid markedly different from those previously described [ 8 - 16 ], we created an effective gene vaccine against HER2/ neu [ 17 ]. The previously described ELVIS plasmid vaccine construct for HER2/ neu contained the cDNA of a replicon RNA from the Alphavirus Sindbis [ 18 , 19 ]. The replicon RNA contained the replication/transcription genes of the parent virus, but the structural protein genes were replaced by the gene for rat neu . From this plasmid construct, the replicon RNA is synthesized in the nucleus of the host cell and is transported to the cytoplasm for replication and transcription. The neu gene product is produced at high levels in the cytosol. Since the structural protein genes from the parent virus are not encoded by the replicon, progeny infectious Sindbis virions are not generated [ 20 - 22 ]. Alternatives to Alphavirus replicon plasmid vaccines [ 17 , 19 , 23 - 25 ] are Alphavirus-based virus-like replicon particles (VRP). As already mentioned, the replicon RNA does not contain the structural genes from the parent virus. It is therefore a single-cycle, propagation-defective RNA and replicates only within the cell into which it is introduced. To generate VRP from an attenuated strain of the Alphavirus Venezuelan equine encephalitis virus (VEE), the replicon RNA is packaged into particles by co-transfecting the replicon RNA and two separate helper RNAs, which together encode the full complement of VEE structural proteins [ 26 ]. Although the VRP can infect target cells in culture or in vivo , and can express the foreign gene to a very high level, the VRP are defective since they lack critical portions of the VEE genome – they lack the VEE structural protein genes necessary to produce infectious virus particles capable of spreading to other cells [ 21 , 27 ]. Several reports have demonstrated that VRP are extremely effective vaccine vectors [ 28 - 39 ]. The VEE VRP vaccine vectors are particularly attractive because the VEE envelope glycoproteins target the VRP to cells of lymphoid tissue [ 40 ], because they can be administered multiple times [ 39 ], because they induce both cellular and humoral immune responses, and because pre-existing immunity to VEE in humans should not be problematic since the incidence of VEE infection is low. In the current study, we sought to determine whether vaccination with VEE-derived VRP containing the gene for HER2/ neu would inhibit tumor growth in prevention models in which HER2/ neu -expressing tumor cells had been injected either into a mammary fat pad or intravenously. We also sought to determine whether vaccination could prevent spontaneous tumorigenesis in HER2/ neu transgenic mice. VRP- neu vaccination induced antigen-specific CD8 + T-cell and IgG responses that corresponded with the lack of tumor growth in both tumor models. In light of the clinical benefit of trastuzumab, a safe and effective vaccine that can induce cellular and humoral immunity, VRP- neu warrants clinical evaluation. Materials and methods Tumor cell line and reagents The A2L2 cell line that expresses high levels of rat HER2/ neu has been previously described in detail [ 17 ]. The A2L2 cell line has consistently expressed high levels of HER2/ neu for more than 5 years and consistently induces tumors in Balb/c mice when injected into a mammary fat pad or intravenously. The A2L2 cell line was maintained in Eagle's MEM containing 5% FCS, sodium pyruvate, nonessential amino acids, L-glutamine, and vitamins (GIBCO, Carlsbad, CA, USA). The monolayer cultures were subdivided at approximately 75% confluence by treatment for 1–3 min with 0.25% trypsin and 0.02% EDTA at 37°C. Virus-like replicon particles The pSV2- neu plasmid containing the gene for rat HER2/ neu was provided by Dr Mien-Chie Hung (The University of Texas MD Anderson Cancer Center, Houston, TX, USA). VEE VRP were constructed at AlphaVax, Inc. (Research Triangle Park, NC, USA) according to the published procedure [ 26 ]. Control VRP containing the gene for A/PR/8/34 influenza hemagglutinin (HA) were also prepared following the same procedure. VRP preparations were screened in a sensitive in vitro Vero cell cytopathic effect assay for the detection of replication competent virus. Prior to their use in these studies, the preparations were shown to be devoid of detectable replication competent virus. Flow cytometry A2L2 cells were incubated for 1 hour at 37°C with either immune serum or control serum diluted in PBS. FITC-labeled goat anti-mouse IgG diluted 1:1000 in PBS was added to the cells and incubation was continued for 1 hour at 37°C. The cells were washed three times in PBS and were analyzed by flow cytometry using an EPICS Profile Analyzer (Beckman Coulter, Inc., Fullerton, CA, USA). Mice Female Balb/c mice, aged 6–8 weeks, were obtained from the Frederick Cancer Research Facility (Frederick, MD, USA). Mice were allowed to acclimate for at least 1 week before use. Six-week-old MMTV-c- neu transgenic mice were obtained from Charles River Laboratories Inc. (Wilmington, MA, USA). This strain of mice is called the 'Oncomouse' by Charles River Laboratories due to the spontaneous generation of breast cancer. These mice are of the FVB/N strain, and the transgene is the 'activated' rat c- neu oncogene preceded by the MMTV promoter [ 41 ]. Vaccination with VRP Mice were vaccinated subcutaneously in the hind foot pad with VRP suspended in 50 μl normal saline. Repeat vaccinations, when administered, were performed using alternate hind feet. Tail vein blood was removed and tumor challenge performed 2 weeks after the final vaccination. Serum was separated from the blood by centrifugation after overnight storage at 4°C. Mammary fat pad tumor prevention model A2L2 cells were harvested from subconfluent cultures with trypsin and EDTA as described earlier. The cells were washed once in serum-containing culture medium and were washed once in PBS. Mice were anesthetized by inhalation of isoflurane using a special apparatus developed by the veterinarians at MD Anderson Cancer Center. The fur was shaved over the lateral thorax, and a 5-mm-long incision was made to reveal mammary fat pad 2 as previously described [ 42 ]. A 0.1-ml sample containing 2.5 × 10 4 A2L2 cells in normal saline was injected into the fat pad. The incision was closed with a wound clip. Wound clips that had not already fallen off were removed after 7 days. The mice were then observed daily and their tumors measured in perpendicular directions with a pair of calipers. Mice with tumors 10 mm in the greatest dimension were killed according to our approved Institutional Animal Care and Use protocol. At the termination of the experiment, all mice were killed by CO 2 inhalation and all tumors were excised and weighed. Experimental metastasis prevention model A 0.1-ml sample containing 2.5 × 10 4 A2L2 cells in normal saline was injected into the tail vein of each immunized mouse. The mice were killed 30 days later, and the surface lung metastases in each animal were counted. Tetramer analysis of immune spleen cells A K(d) tetramer containing the peptide sequence PYVSRLLGI [ 43 ] was prepared by the National Institutes of Health Tetramer Facility at Emory University (Atlanta, GA, USA). The incorporated peptide was synthesized at the peptide synthesis facility of MD Anderson Cancer Center. We received a K(d) tetramer specific for A/PR/8/34 influenza HA containing the peptide IYSTVASSL from Linda Sherman at the Scripps Research Institute (La Jolla, CA, USA) [ 44 ]. A single-cell suspension of the spleen from a naïve mouse was prepared in R10S medium (RPMI 1640, 10% HI-FCS [Summit Biotechnology, Fort Collins, CO, USA], 1% nonessential amino acids, 100 mM MEM sodium pyruvate, 2 mM L-glutamine, 100 U/ml penicillin, 1 ml streptomycin, and 50 mM 2-mercaptoethanol) by gently swirling the spleen between frosted glass slides. Debris was removed from the cell suspension by filtration through nylon mesh into a 10-ml centrifuge tube. The cell suspension was centrifuged for 10 min at 700 × g and the pellet resuspended in 5 ml ACK solution (0.15 M NH 4 Cl, 1.0 mM KHCO 3 , 0.01 mM NaEDTA, pH adjusted with 1 N HCl to 7.2–7.4) before being gently rocked for 5 min to lyse red blood cells. An additional 5 ml R10S medium was added to the cell suspension, and the cells were washed once with R10S medium. The naïve spleen cells were cultured with either PYVSRLLGI or IYSTVASSL (70 ng/ml) on a rocker for 2 hours at 37°C and were γ-irradiated with 20 Gray. The mice were vaccinated three times with a 2-week interval with 10 6 infectious units (IU) VRP- neu or 10 6 IU VRP-HA, and the immune spleens were harvested 3 weeks after the final vaccination. The immune spleens were prepared to produce a single cell suspension in R10S medium and were co-cultured with the peptide-pulsed stimulator cells at a ratio of one stimulator to eight responders. The cell mixture was cultured for 7 days at 37°C, washed, and was resuspended in FACS buffer (PBS with 1% BSA) at a concentration of 5 × 10 7 cells/ml. Twenty microliters of the cell mixtures were added to 20 μl PE-conjugated Her2/ neu- specific or HA-specific tetramers to make the final dilutions of the tetramers 1:50, 1:100, and 1:200. One microliter of PerCP-conjugated anti-mouseCD3e (PharMingen, San Diego, CA, USA) and FITC-conjugated anti-mouse CD8a (Caltag, Burlingame, CA, USA) was added to the mixture and incubated for 1 hour at 4°C in the dark. The cells were then suspended in 150 μl FACS buffer and transferred to a polystyrene round-bottomed tube. The cells were washed twice with FACS buffer and suspended in 200 μl of 1% paraformaldehyde in PBS. List mode data were acquired with a FACScan (BD Biosciences, San Jose, CA, USA). Dead cells and monocytes were excluded from the analysis by forward scatter and side scatter gating. A total of 10,000–30,000 events were typically acquired, and compensation was optimized using unstained cells, cells stained with only PerCP-conjugated anti-mouse CD3e, cells stained with only FITC-conjugated anti-mouseCD8a and cells stained with only PE-conjugated anti-mouse CD4. The CD3 + cells were gated from the total population of live cells, and the CD8 + cells were gated from the CD3 + cells. From the CD3 + cells and the CD8 double-positive cells, the percentages of PE-tetramer-positive cells were calculated. Isotype controls for the anti-CD3e and anti-CD8a were subtracted from the acquired data. List mode files were analyzed using CELLQUEST software (BD Biosciences, San Jose, CA, USA). Intracellular interferon-γ analysis of immune spleen cells The procedure for this technique was obtained from PharMingen, whose reagents were used whenever possible. Spleens from VRP- neu -immunized and VRP-HA-immunized mice were prepared as already described for tetramer analysis. The spleen cells were stimulated with PYVSRLLGI at a concentration of 70 ng/ml at 37°C in R10S medium for 5 days as described. On the sixth day, the cells were suspended at 2 × 10 6 /ml in R10S medium containing 10 ng/ml phorbol-12-myristate acetate (Sigma-Aldrich, St Louis, MO, USA) and 250 ng/ml ionomycin (Sigma-Aldrich). Brefeldin (1 μl/ml; Sigma-Aldrich) was added at the same time to block cytokine secretion. The cells were washed after 5 hours and resuspended at 10 7 cells/ml in staining buffer (Dulbecco's PBS without Mg 2+ or Ca 2+ , 1% heat-inactivated FCS, w/v 0.09% sodium azide; pH adjusted to 7.4–7.6). The cells were then incubated for 15 min with purified 2.4 G2 antibody 10 μg/ml (PharMingen) to block nonspecific staining by fluorochrome-conjugated antibodies via Fc receptors. The cells were washed twice with staining buffer, and 10 6 cells were stained with 1 μg FITC-labeled anti-mouse CD8a and PerCP-labeled anti-mouse CD3e (PharMingen) in 50 μl staining buffer at 4°C for 30 min. The cells were again washed twice with staining buffer, followed by fixation and permeabilization with Perm/Wash (Cytofix/Cytoperm Kits; PharMingen) for 20 min at 4°C. The cells were washed twice and resuspended in 50 μl of the same solution. PE-conjugated anti-interferon (IFN)-γ monoclonal antibody (0.5 μg/10 6 cells; PharMingen) was added and the suspension was incubated for 30 min at 4°C. The cells were then washed twice with Perm/Wash solution and resuspended in staining buffer. The FACScan was used to analyze the percentage of intracellular IFN-γ-containing cells among the CD3 + and CD8 + cells. Isotype controls for anti-CD3a, anti-CD8e, and anti-IFN-γ (PharMingen) were subtracted from the acquired data. ELISPOT analysis of IFN-γ production by immune spleen cells The procedure for the ELISPOT technique was obtained from PharMingen, whose reagents were used whenever possible. The wells of an ELISPOT plate (CTL Immunospot plate; Cellular Technology Ltd, Cleveland, OH, USA) were coated overnight at 4°C with 100 μl anti-mouse IFN-γ (2 μg/ml; PharMingen) diluted at 1:200 in coating buffer (PBS, pH 7.2). The coated wells were washed with blocking solution (R10S medium), fresh blocking solution was added to each well, and then the plate was incubated for 2 hours at room temperature. The blocking solution was then discarded and 100 μl A2L2 cells (2 × 10 5 -1 × 10 6 ) was added to each well of the ELISPOT plate. Immune spleens were dissected from vaccinated mice and were prepared as already described for tetramer analysis. Increasing numbers of spleen cells (5 × 10 6 - 2 × 10 7 ) in 100 μl R10S medium were added to the wells, and the plate was incubated for 24 hours at 37°C in a 5% CO 2 atmosphere at 99% humidity. Wells containing only spleen cells served as negative controls, and spleen cells from VRP- neu -vaccinated mice cultured overnight with 5 μg/ml concanavalin A (Sigma-Aldrich) served as a positive control. The cell suspension was aspirated, and the wells were washed twice with deionized water and were then soaked with deionized water for 5 min. The wells were washed three times with wash buffer I (PBS containing 0.05% Tween-20). The detection antibody, biotinylated anti-mouse IFN-γ (PharMingen), was diluted to 2 μg/ml in dilution buffer (PBS containing 10% FBS), and 100 μl was added to each well. The plate was incubated at room temperature for 2 hours, and the wells were washed three times with buffer I. Avidin–horseradish peroxidase reagent (PharMingen) was diluted to 1:100 in dilution buffer, and 100 μl was added to each well, which was then incubated at room temperature for 1 hour. The wells were washed four times with wash buffer I and twice with wash buffer II (PBS). A stock solution containing 100 mg 3-amino-9-ethyl-carbazole (Sigma) dissolved in 10 ml N , N -dimethylformamide (Sigma-Aldrich) was prepared. The final substrate solution was made by adding 333 μl 3-amino-9-ethyl-carbazole stock solution to 10 ml of 0.1 M sodium acetate (pH 5.0), followed by filtering through a 0.45-μm filter. Five microliters of 30% H 2 O 2 was added to the substrate solution immediately before use. One hundred microliters of the final substrate solution was added to each well, and the plate was incubated in the dark for 5–60 min at room temperature. The reaction was stopped by washing the wells with deionized water. The plate was air-dried overnight at room temperature in the dark and sent to ZellNet Consulting, Inc. , where the spots were enumerated automatically using an ImmunoSpot Series I analyzer (BD Biosciences). If overlapping spots (confluence) were present in the wells, the number of spots in a nonconfluent area of that well was determined. To estimate the total number of spots in each well with confluence, the following equation was used: total spot number = spot count + 2 × (spot count × % confluence / [100% - % confluence]). Statistical analysis Student's t test was performed using Prism 4.0 Graphpad Software (San Diego, CA, USA). Statistical analysis, power analysis and the sample size per group were evaluated and found to be statistically acceptable by Dr Lyle Broemling (Associated Professor of Biostatistics, The University of Texas MD Anderson Cancer Center). Results Induction of antigen-specific IgG by vaccination with VRP- neu Groups of Balb/c mice ( n = 5 per group) were vaccinated once subcutaneously in one hind leg footpad with either 10 6 IU VRP- neu or 10 6 IU VRP-HA suspended in PBS. The HER2/ neu -specific humoral response of serum pooled from mice in each group was measured 14 days later by flow cytometry using A2L2 cells. Compared with the mice vaccinated with VRP-HA, the mice vaccinated with VRP- neu had a strong IgG response (Fig. 1 ). Pre-immune sera for both groups were nonreactive with A2L2 cells, and immune sera from both vaccinated groups were nonreactive with 66.3 cells, the parental cell line from which A2L2 was derived by transfection with neu (data not shown). Protection from tumor challenge in a mammary fat pad prevention model following vaccination with VRP- neu Groups of mice ( n = 7 per group) were vaccinated subcutaneously with 10 5 IU or 10 6 IU VRP- neu or with 10 6 IU VRP-HA three times at 14-day intervals. Two weeks after the final vaccination, the mice were challenged with 2.5 × 10 4 A2L2 cells injected into a mammary fat pad. Five weeks after tumor challenge, the largest tumor dimension was measured and the mice were killed. If a tumor was present, its mass was determined. All of the mice vaccinated with VRP-HA had a measurable tumor, whereas only one mouse in each group vaccinated with 10 6 IU VRP- neu or 10 5 IU VRP- neu had a measurable tumor (Fig. 2a,2b ). These findings clearly demonstrate that vaccination three times with either 10 5 IU or 10 6 IU VRP- neu protected mice from challenge with A2L2 cells. VRP-HA failed to provide protection for any of the mice, and therefore the protective effect was specific for the vaccine containing the gene for HER2/ neu . Determination of the minimal effective vaccine dose in two tumor prevention models Because vaccination three times with 10 5 IU VRP- neu prevented tumor growth in a mammary fat pad, we next determined the minimum number of VRP- neu particles and the minimum number of vaccinations necessary to significantly inhibit tumor growth. In the mammary fat pad prevention model, vaccination twice with 10 5 IU VRP- neu or vaccination three times with 10 4 IU VRP- neu completely prevented tumor growth in many mice and significantly reduced the tumor mass in the entire group compared with the tumor mass of the mice vaccinated three times with VRP-HA (Fig. 3a ). Identical results were obtained in the experimental lung metastasis prevention model, in which mice were injected with A2L2 cells intravenously in the tail vein after vaccination (Fig. 3b ). These results demonstrate that, in both tumor models, vaccination three times with 10 4 IU VRP- neu or twice with 10 5 IU VRP- neu significantly reduced the tumor mass and lung metastasis. In addition, several mice in each vaccinated group were tumor free in mammary tissue or lungs. Vaccination of MMTV-c- neu transgenic mice MMTV-c- neu transgenic mice contain the activated rat neu gene under the control of the MMTV promoter and spontaneously develop neu + breast tumors within 110–120 days [ 45 ]. Without intervention, all of the mice die of breast cancer. We vaccinated groups of mice ( n = 10 per group) three times at 14-day intervals with 10 6 IU VRP- neu or 10 6 IU VRP-HA and determined the effect on survival. Eight of the 10 mice vaccinated with VRP-HA were killed by 140 days owing to moribundity, and the remaining two mice were killed on day 195 (Fig. 4 ). None of the mice vaccinated with VRP- neu showed any sign of illness at 240 days, and breast tumors were not evident on palpation. The mice in the VRP- neu -vaccinated group were killed at this time, and gross pathologic examination of the breasts following retraction of the skin did not reveal any tumors. These results demonstrate that vaccination with VRP- neu prevented spontaneous formation of tumor in the breasts of neu transgenic mice and that tolerance to the neu transgene was broken by vaccination with VRP- neu . Induction of antigen-specific IgG by vaccination of MMTV-c- neu transgenic mice with VRP- neu Immune serum was drawn from mice described at 133 days or 55 days after the third vaccination with VRP- neu or VRP-HA (Fig. 4 ). Flow cytometric analysis of the immune sera using A2L2 cells demonstrated that vaccination with VRP- neu induced a strong IgG response that was evident at dilutions of 1:25 (Fig. 5a ) and 1:100 (Fig. 5b ). Vaccination with the control VRP-HA failed to induce a neu -specific antibody response (Fig. 5a,5b ). These findings indicate that vaccination with VRP- neu induced humoral immunity to the protein product of the neu -transgene in neu -transgenic mice, thus breaking any existing tolerance to p185. Tetramer analysis of CD8 + T cells following vaccination with VRP- neu or VRP-HA Ikuta and colleagues [ 43 ] identified a K(d)-restricted peptide for mouse HER2/ neu . The identical sequence (PYVSRLLGI) is present in rat HER2/ neu . We therefore ordered a tetramer containing this sequence from the National Institutes of Health Tetramer Facility at Emory University. A K(d) tetramer specific for A/PR/8/34 influenza HA was used as a positive control for VRP-HA-vaccinated mice. Balb/c mice were vaccinated three times at 2-week intervals with VRP- neu or VRP-HA, and the spleens from two mice were collected 3 weeks after the third injection. We found that 2.38% of the pooled spleen cells from the mice that had been vaccinated three times with VRP- neu were stained by both the anti-CD8 antibody and the HER2/ neu tetramer (Table 1 ). This value is in excellent agreement with the percentage of dual-positive cells from mice that had been vaccinated with VRP-HA and stained with the HA tetramer (2.79%). In contrast, the percentage of dual-positive cells from the mice that had been vaccinated with VRP-HA and stained with the HER2/ neu tetramer was only 0.22%, and the percentage from the mice that had been vaccinated with VRP- neu and stained with the HA tetramer was only 0.37%. These results clearly demonstrate that vaccination with VRP- neu produced antigen-specific CD8 + T cells. Intracellular IFN-γ analysis of CD8 + T cells following vaccination with VRP- neu or VRP-HA An alternative assay to tetramer analysis is measurement of the percentage of CD8 + T cells that also contain intracellular IFN-γ after in vitro stimulation with an antigenic peptide. We therefore vaccinated Balb/c mice three times at 2-week intervals with VRP- neu or VRP-HA, and then removed the spleens 3 weeks after the third injection. The in vitro culture procedure was similar to that used for tetramer analysis except that the peptide PYVSRLLGI was cultured directly with the immune spleen cells rather than with naïve spleen cells. We found that 2.80% of the spleen cells from the mice that had been vaccinated three times with VRP- neu stained positive for intracellular IFN-γ (Fig. 6a and Table 2 ) compared with only 0.27% for spleen cells from mice vaccinated with VRP-HA (Fig. 6b and Table 2 ). This percentage (2.80%) is in excellent agreement with that of tetramer-positive cells after vaccination with VRP- neu (2.38%) (Table 1 ). Analysis of immune spleen cells by IFN-γ ELISPOT after vaccination with VRP- neu or VRP-HA We vaccinated Balb/c mice three times at 2-week intervals with VRP- neu or VRP-HA and removed the spleens 3 weeks after the third injection. The number of spleen cells secreting IFN-γ in response to overnight co-culture with A2L2 cells in an ELISPOT assay were then determined. At all three effector-to-stimulator ratios, vaccination with VRP- neu resulted in a statistically significant increase in the number of spots per 10 6 spleen cells (Table 3 ). This finding demonstrates that secretion of IFN-γ in response to co-culture with A2L2 cells was dependent on vaccination with VRP- neu and did not result from vaccination with the control VRP-HA. Discussion Although more than a decade has elapsed since the original descriptions of gene vaccines [ 46 , 47 ], a clinically approved gene vaccine for either an infectious disease or for cancer has yet to be developed. Gene vaccines containing elements of the Alphaviruses VEE, Sindbis virus and Semliki Forest virus offer substantial clinical potential and safety [ 22 , 27 ]. Vaccine vectors incorporating genetic elements of Alphaviruses can be divided into two major categories [ 48 ]: expression plasmids containing viral genes that are essential for replication and transcription of the positive-strand RNA genome of the viruses [ 18 , 24 ], and infectious but replication-incompetent viral particles in which the genes for the viral structural proteins have been replaced by the gene for a protein antigen [ 20 - 22 ]. We have previously described a gene vaccine for HER2/ neu of the first category [ 17 ]. In the present article, we describe results of a gene vaccine for HER2/ neu of the second category. We report here that VRP vaccine vectors derived from an attenuated strain of VEE containing the gene for rat HER2/ neu were highly immunogenic when used to vaccinate both conventional mice and mice transgenic for the rat neu gene. Immune serum from mice vaccinated once with VRP- neu was reactive with A2L2 cells, demonstrating induction of an antigen-specific IgG response, whereas serum from mice that had been vaccinated with VRP-HA was nonreactive (Fig. 1 ). To determine whether vaccination with VRP- neu could protect mice from challenge with a breast cancer cell line engineered to overexpress HER2 /neu , mice that had been vaccinated three times with either VRP- neu or VRP-HA were injected in a mammary fat pad with A2L2 cells. Vaccination with either 10 5 IU or 10 6 IU VRP- neu protected all but one mouse in each group from developing tumors (Fig. 2 ). Because our previous experiment demonstrated that a single vaccination with 10 6 IU VRP- neu induced an IgG response, we next vaccinated mice once, twice or three times with 10 4 IU or 10 5 IU VRP- neu or the control VRP-HA. Vaccination three times with 10 4 IU VRP- neu ( P = 0.0265) or twice with 10 5 IU VRP- neu ( P = 0.0079) was sufficient to prevent tumor growth of A2L2 cells injected into a mammary fat pad (Fig. 3a ). Mice similarly vaccinated were also challenged with A2L2 cells injected intravenously in the tail vein. Vaccination three times with 10 4 IU VRP- neu ( P = 0.0317) or twice with 10 5 IU VRP- neu ( P = 0.0159) prevented tumor growth in the lungs (Fig. 3b ). Some mice in the vaccine groups had no visible lung metastases 5 weeks after the tumor challenge. This finding clearly demonstrates that vaccination with VRP- neu prevented tumor growth in two tumor prevention models. There is an important point to be considered regarding our experimental models. We are vaccinating mice with the gene for rat neu and we are challenging mice with a cell line also overexpressing the gene for rat neu . We must therefore consider the possibility that mice recognized rat p185 (the protein product of the HER2/ neu gene) as a xenoantigen and that vaccination with VRP- neu merely boosted, but did not initiate, an immune response. We addressed this question by vaccinating rat neu transgenic mice, which are immunologically tolerant to rat p185. MMTV-c- neu transgenic mice express the rat neu transgene under the control of the murine mammary tumor virus promoter and spontaneously develop neu + breast cancer. Vaccination of these transgenic mice with VRP- neu very clearly demonstrated ( P < 0.0001) that the mice survived to 240 days old (Fig. 4 ), at which time the experiment was concluded. On postmortem examination, no tumor was detected in the breast of any mouse that had been vaccinated with VRP- neu . All mice in the control group that had been vaccinated with VRP-HA were moribund owing to extensive breast cancer by 200 days. This result is much more dramatic than our previous finding that vaccination of MMTV-c- neu transgenic mice with the Sindbis/DNA plasmid-based ELVIS replicon vector increased survival of the transgenic mice but failed to protect against tumor formation and death [ 17 ]. To determine whether vaccination of the MMTV-c- neu transgenic mice with VRP- neu induced an IgG response, as we found in Balb/c mice (Fig. 1 ), we tested the serum of the vaccinated neu -transgenic mice at 133 days old. Vaccination with VRP- neu induced a strong, antigen-specific IgG response in these transgenic mice (Fig. 5 ). Vaccination with VRP- neu therefore overcame tolerance to p185 (Fig. 5 ). In our present study, we performed three separate in vitro T-cell assays to determine whether vaccination with VRP- neu induces antigen-specific T cells. In the first assay we used tetramers containing an antigenic peptide of HER2/ neu to analyze spleen cells from mice vaccinated with VRP- neu . As a positive control we used a tetramer containing an antigenic peptide of HA. We found that 2.38% of the CD8 + spleen cells resulting from vaccination with VRP- neu were positive for tetramer binding (Table 1 ), a value in excellent agreement with the 2.79% of the spleen cells from VRP-HA-vaccinated mice that were positive for an HA-specific tetramer. We further analyzed the induction of antigen-specific CD8 + T cells resulting from VRP- neu vaccination by performing intracellular IFN-γ analysis. We found that 2.80% of CD8 + cells from VRP- neu -vaccinated mice were positive for intracellular IFN-γ (Table 2 ), a value in excellent agreement with the 2.38% tetramer-binding cells (Table 1 ). In the third assay we measured the number of IFN-γ secreting cells by ELISPOT analysis (Table 3 ). At all three effector:stimulator ratios, the number of IFN-γ secreting cells in the VRP- neu -vaccinated spleen cells was significantly greater than that in the VRP-HA-vaccinated spleen cells. These three independent assays clearly demonstrate that vaccination with VRP- neu induced antigen-specific CD8 + T lymphocytes. The tetramer and intracellular IFN-γ assays further indicate that immune spleen cells were able to recognize an antigenic peptide of p185 presented on the A2L2 cells used for the ELISPOT assay. The in vivo and in vitro experiments described demonstrate that vaccination induced antigen-specific cell-mediated and humoral immunity, but they do not indicate the role that each type of immunity may have played in the resulting antitumor effect. Therefore, although vaccination with VRP- neu produced antigen-specific IgG in both conventional and transgenic mice, whether this antibody played a role in the antitumor effect remains unclear. Pilon and colleagues [ 49 ] reported that antibody was not required for the antitumor effect of a plasmid vaccine against HER2/ neu . Lindencrona and colleagues [ 50 ] similarly induced antitumor immunity against HER2/ neu in B cell-deficient mice. Although T-cell immunity alone may be sufficient in tumor prevention models in mice, a clinical trial showed that trastuzumab clearly benefited patients and increased the antitumor effect of a whole-cell vaccine to HER2/ neu [ 51 ]. Conclusion Our findings demonstrate that vaccination with VRP- neu inhibited or eliminated tumor growth in prevention models in which breast tumor cells had been injected either in the mammary fat pad or intravenously. Vaccination with VRP- neu also prevented tumorigenesis in transgenic mice in which the neu gene was expressed in the breasts under the control of an MMTV promoter. Vaccination with VRP- neu induced antigen-specific CD8 + , and this finding corresponded with the absence or inhibition of tumor growth. Furthermore, vaccination with VRP- neu induced antigen-specific IgG in both conventional and transgenic mice, and tolerance to HER2/ neu in neu -transgenic mice was broken by vaccination with VRP- neu . These findings suggest that VRP- neu constitutes a powerful gene vaccine that induces both cellular and humoral immunity against HER2/ neu . We speculate that such vaccination could be more effective than passive immunotherapy using monoclonal antibodies such as trastuzumab in patients with breast cancer. Abbreviations BSA = bovine serum albumin; FACS = fluorescence-activated cell sorting; FCS = fetal calf serum; FITC = fluorescein isothiocyanate; HA = hemagglutinin; IFN = interferon; IU = infectious units; MEM = modified Eagle's medium; MMTV = mouse mammary tumor virus; PBS = phosphate-buffered saline; VEE = Venezuelan equine encephalitis virus; VRP = virus-like replicon particles. Competing interests MFM is an employee of AlphaVax, Inc. and has options to purchase the company's stock. None of the other authors have any financial interest in AlphaVax, Inc. or in any products that could result from this research. All of the research was conducted independently at the University of Texas MD Anderson Cancer Center and was not influenced in any way by employees or the management of AlphaVax, Inc. Authors' contributions The authors' contributions to this research are reflected in the order shown with the exception of LBL, who supervised all aspects of this research and the preparation of this report.

D:\keerthana\PMC001xxxxxx\PMC1064108.xml

1064109

Immunohistochemical expression of insulin-like growth factor binding protein-3 in invasive breast cancers and ductal carcinoma in situ: implications for clinicopathology and patient outcome

Introduction Insulin-like growth factor binding protein-3 (IGFBP-3) differentially modulates breast epithelial cell growth through insulin-like growth factor (IGF)-dependent and IGF-independent pathways and is a direct (IGF-independent) growth inhibitor as well as a mitogen that potentiates EGF (epidermal growth factor) and interacts with HER-2. Previously, high IGFBP-3 levels in breast cancers have been determined by enzyme-linked immunosorbent assay and immunoradiometric assay methods. In vitro , IGFBP-3's mechanisms of action may involve cell membrane binding and nuclear translocation. To evaluate tumour-specific IGFBP-3 expression and its subcellular localisation, this study examined immunohistochemical IGFBP-3 expression in a series of invasive ductal breast cancers (IDCs) with synchronous ductal carcinomas in situ (DCIS) in relation to clinicopathological variables and patient outcome. Methods Immunohistochemical expression of IGFBP-3 was evaluated with the sheep polyclonal antiserum (developed in house) with staining performed as described previously. Results IGFBP-3 was evaluable in 101 patients with a variable pattern of cytoplasmic expression (positivity of 1+/2+ score) in 85% of invasive and 90% of DCIS components. Strong (2+) IGFBP-3 expression was evident in 32 IDCs and 40 cases of DCIS. A minority of invasive tumours (15%) and DCIS (10%) lacked IGFBP-3 expression. Nuclear IGFBP-3 expression was not detectable in either invasive cancers or DCIS, with a consistent similarity in IGFBP-3 immunoreactivity in IDCs and DCIS. Positive IGFBP-3 expression showed a possible trend in association with increased proliferation ( P = 0.096), oestrogen receptor (ER) negativity ( P = 0.06) and HER-2 overexpression ( P = 0.065) in invasive tumours and a strong association with ER negativity ( P = 0.037) in DCIS. Although IGFBP-3 expression was not an independent prognosticator, IGFBP-3-positive breast cancers may have shorter disease-free and overall survivals, although these did not reach statistical significance. Conclusions Increased breast epithelial IGFBP-3 expression is a feature of tumorigenesis with cytoplasmic immunoreactivity in the absence of significant nuclear localisation in IDCs and DCIS. There are trends between high levels of IGFBP-3 and poor prognostic features, suggesting that IGFBP-3 is a potential mitogen. IGFBP-3 is not an independent prognosticator for overall survival or disease-free survival, to reflect its dual effects on breast cancer growth regulated by complex pathways in vivo that may relate to its interactions with other growth factors.

Introduction Insulin-like growth factors (IGF-I and IGF-II) regulate the cellular growth of normal and malignant breast epithelial cells with a role in malignant transformation [ 1 - 3 ]. IGFs are potent mitogens and are synergistic with oestrogen and epidermal growth factor (EGF) to stimulate cellular proliferation[ B4 ]. The IGFs are modulated by a family of six high-affinity IGF binding proteins, of which IGFBP-3 predominates in serum and is upregulated in breast cancer cell lines, including breast epithelium [ 1 , 5 , 6 ]. Both IGFs (IGF-I and IGF-II) have a preferential stromal expression and together with epithelial IGFBP-3 have a significant paracrine influence on breast epithelial growth [ 1 , 7 ]. IGFBPs have multiple and complex functions that can be either IGF dependent or IGF independent. With respect to IGF-dependent function, IGFBP-3 preferentially binds IGFs to either inhibit or activate IGF mitogenic effects in vitro , through blocking the IGF-receptor interaction, in contrast to a prolongation of IGFs' half-life and their protection from degradation [ 1 - 3 ]. IGFBP-3 is pro-apoptotic in an IGF-dependent manner, as well as an IGF-independent manner, and enhances the p53 DNA damage response in vitro [ 8 , 9 ]. The particular significance of IGFBP-3 in regulating epithelial cell growth has been highlighted because the actions of many growth inhibitors, apoptotic agents and anti-cancer treatments (transforming growth factor-β, retinoids, p53 and anti-oestrogens) are, at least in part, mediated by their ability to stimulate local IGFBP-3 production [ 1 - 3 ]. Through complex and as yet poorly understood mechanisms, IGFBP-3 is a direct (IGF-independent) growth inhibitor as well as a mitogen on breast epithelial cells[ B10 , 11 ]. Further, IGFBP-3 inhibits oestradiol-stimulated cell proliferation in breast cancer cell lines, with the potential to accentuate ceramide and paclitaxel-induced apoptosis directly [ 12 - 14 ]. By contrast, accumulating evidence in vitro suggests the potential mitogenicity of IGFBP-3 through its interactions with EGF receptor (EGFR) and Ras-p44/42 mitogen-activated protein kinase (MAPK) signalling in breast epithelial cells [ 15 - 18 ]. Postulated mechanisms for IGFBP-3's direct intrinsic actions on cell growth and apoptosis have not as yet characterised a definitive cell membrane receptor or the necessity for IGFBP-3 cell surface interaction or nuclear translocation[ B19 ]. Intracellular trafficking of IGFBP-3 with nuclear localisation in T47D breast cancer cells is explicable through a carboxy-terminal nuclear localisation signal and importin-β-mediated nuclear transport[ B15 , 20 , 21 ]. At present, the functional implications of nuclear IGFBP-3 are unknown[ B22 ]. Complex IGFBP-3 modulation of breast cancer growth has prompted several studies to examine levels of IGFBP-3 in breast cancer tissues in relation to clinicopathological characteristics and patient outcome [ 23 - 26 ]. Circulating IGFBP-3 may avert breast cancer development, with the clinical paradox that increasing IGFBP-3 levels in breast tumours may indicate adverse prognostic cancers [ 2 , 23 - 26 ]. This reflects the complexity of IGFBP-3 effects on cell proliferation and its potential role as a mitogen as well as a growth inhibitor. Poor prognostic tumours with increasing IGFBP-3 expression may relate to recent evidence in vitro in which the pro-apoptotic action of IGFBP-3 is reversed by the extracellular matrix protein fibronectin[ B27 , 28 ]. In keeping with these findings, high levels of fibronectin expression are associated with poor prognostic breast cancers [ 6 , 29 ]. IGFBP-3 interacts with integrin-receptor signalling with modulation by fibronectin to increase cell attachment and possible resistance to apoptosis[ B27 ]. The aim of this first immunohistochemical study was to evaluate breast epithelial IGFBP-3 expression in relation to clinicopathological parameters and prognosis in breast cancer. IGFBPs are upregulated in malignant breast epithelial cells with evidence in vivo that IGFBP-5 is overexpressed in the cytoplasm of breast cancers and their lymph node metastases on tissue microassay immunohistochemistry (IHC)[ B6 ]. Accumulating evidence in vitro supports the dual effects of IGFBP-3 on the cellular growth of breast epithelial cells to emphasise the importance of selectively analysing breast epithelial IGFBP-3 expression in comparison with the stroma. By contrast, previous studies have collectively assessed breast epithelial and stromal IGFBP-3 levels by using immunoassay and immunoblot or ligand blot methods [ 23 - 26 ]. Moreover, both the histological location of IGFBP-3 and variations in its immunoreactivity have been compared in this series of invasive ductal breast cancers with concomitant evaluation of IGFBP-3 expression in synchronous ductal carcinoma in situ (DCIS) within the same tumour specimens. Increasing IGFBP-3 levels in breast tumours may indicate adverse prognostic cancers, with contradictory implications on patient outcomes[ B24 , 25 ] to reflect the complexity of IGFBP-3 effects on cell proliferation[ B27 , 28 , 30 ]. We have shown a varying pattern of epithelial IGFBP-3 cytoplasmic expression (1+/2+ score) in 85% of invasive ductal cancers (IDCs) and 90% of DCIS components, without detectable nuclear immunoreactivity. Increasing levels of IGFBP-3 expression showed a trend with increased proliferation, oestrogen receptor (ER) negativity and HER-2 overexpression to suggest its association with poor prognostic tumours. Methods Patients The study included 103 patients aged from 26 to 88 years (median 59 years) with IDC of the breast, in association with concomitant DCIS diagnosed between 1996 and 2000 at the Bristol Royal Infirmary, Bristol, UK (Table 1 ). Patient numbers in clinicopathological subgroups reflect those in whom IGFBP-3 was evaluable. Regional Ethics Committee approval was granted before the start of the study. Axillary lymphadenopathy was evaluable in 88 patients; 37 (42%) were lymph-node-negative and 51 (58%) were lymph-node-positive (N1, mobile ipsilateral, or N2, fixed ipsilateral) patients. No axillary surgery was undertaken in the remaining 13 patients because of age-related co-morbidity. Clinicopathological subgroups were analysed in accordance with the Nottingham Prognostic Index (NPI) and divided into good (GPG), moderate (MPG) and poor (PPG) prognostic groups as described, with a modification that included no assessment of the internal mammary lymph nodes[ B31 ]. Evaluation of the NPI was precluded in 15 patients because of non-evaluable regional lymphadenopathy and tumour size. Similarly, subgroups of DCIS were analysed in accordance with the Van Nuys Pathologic Classification (VNPC) (Table 1 ), which was not assessable in five patients [ 32 ]. The design of the study to include tumour representative samples of synchronous IDC and DCIS precluded the analysis of the Van Nuys Prognostic Index[ B32 ]. Adjuvant treatment groups comprised the following: tamoxifen in 60 patients (27 GPG, 16 MPG, 4 PPG and 13 no NPI), and CMF-containing and anthracycline-containing regimes in 17 and 21 patients, respectively (4 GPG, 19 MPG, 13 PPG and 2 no NPI). Five patients received no adjuvant treatment. The median follow-up duration was 51 months (range 4–120 months). All were primary tumours with the exception of six local tumour recurrences, which were excluded from the analysis of patient outcome (see Table 3 and Fig. 2 ). Tumour samples were collected and freshly fixed in buffered formalin in accordance with a standardised protocol at a single institution. Tumours were classified in accordance with NHSBSP guidelines [ 33 ]. Invasive ductal carcinomas were graded by the modified Bloom's grading system described by Elston and Ellis[ B34 ]. ER immunostaining was performed with a standard three-layered streptavidin-avidin-biotin horseradish peroxidase method with a mouse anti-human ER primary antibody (M0747, 1:100 dilution; DAKO, Ely, Cambridgeshire, UK) and a biotinylated rabbit anti-mouse secondary antibody (E354, 1:350 dilution; DAKO). Expression of ER was assessed with the quick-score (0–8) and classified as positive (4–8; >3) or negative (0–3; ≤ 3) in five high-power fields (HPFs) [ 35 ]. Tumour proliferation was assessed with nuclear Ki67 immunostaining (polyclonal rabbit anti-human Ki67 antigen; A0047, 1:100 dilution; DAKO). A goat anti-rabbit biotin-labelled polypeptide (E432, 1:400 dilution; DAKO, Glostrup, Denmark) was used as a secondary antibody. Tonsillar tissue was used as a positive control and primary antibody was replaced with Tris-buffered saline (TBS) as a negative control. Ki67 staining was evaluated as percentage of positive tumour cells, with low proliferation indicative of <10% of positive-staining cells, compared with high proliferation with ≥ 10% positivity [ 36 ]. HER-2 immunostaining was performed with the mouse monoclonal anti-HER-2 antibody (RTU-CB11; Novocastra/Vector, Newcastle upon Tyne, UK), and the Envision Plus HRP system (K4006; DAKO). HER-2 expression was scored according to the degree and proportion of membrane staining, with a score of 0 or 1+ defined as negative, and 2+ or 3+ as HER-2 positive[ B37 ]. Lymphovascular invasion was assessed as present or not, and together with ER, HER-2 and Ki67 was analysed in the Department of Pathology (by CS and CC). Immunohistochemistry IGFBP-3 immunoreactivity was evaluated with the in-house sheep polyclonal antiserum (Professor JMP Holly, IGF Research Group, University of Bristol, Bristol, UK) at 1:800 dilution[ B9 ]. IGFBP-3 immunostaining of IDC and DCIS was compared with formalin-fixed normal liver tissue as a positive control for IGFBP-3, and a replacement of the primary antibody with TBS as a negative control. Validation and specificity of the sheep polyclonal in-house antiserum has previously been demonstrated with an IGFBP-3 peptide on immunocytochemistry of Hs578T breast cancer cells[ B38 ]. Formalin-fixed paraffin sections of breast cancer tissue and normal liver tissue were mounted on glass slides coated with 3-aminopropyl-triethoxysilane (APES; Sigma, Poole, Dorset, UK) and were baked for 30 min at 56–60°C, before being dewaxed in Clearene (Surgipath Europe, Peterborough, UK). The tissue was rehydrated by sequential immersion in 100% and 50% ethanol to distilled water. Tissue sections were subjected to heat antigen retrieval for 3 min in citrate buffer (pH 6) in a pressure cooker, and after cooling were incubated for 5 min in 0.3% (v/v) hydrogen peroxide. Subsequently, sections were washed in tap water and TBS (pH 7.45). Before incubation with IGFBP-3 primary antibody, sections were exposed to avidin and biotin blocking solutions (Vector Laboratories, Burlingame, CA, USA) for 15 min, respectively. Further blocking was achieved through exposure to normal rabbit serum (diluted with TBS) for 30 min at room temperature (20–22°C). Primary antibody was applied and incubated overnight at 4°C (18 hours). After washing with TBS, biotinylated rabbit anti-goat secondary antibody, together with the Strept-AB Complex/HRP (0377, DAKO, Glostrup, Denmark) was applied for 30 min at room temperature. Staining was revealed by development in the chromogen 3,3-diaminobenzidine tetrahydrochloride (DAB) for 5–10 min before counterstaining with haematoxylin in preparation for mounting. Immunostaining was assessed with a Zeiss Axioskop microscope with a 40× Achrostigmat lens (× 400 overall magnification) and a field diameter of 0.46 mm. In the neoplastic cell population for IDC and DCIS, the degree of staining intensity and the proportion of cells with IGFBP-3 immunoreactivity in the nucleus and cytoplasm were graded semi-quantitatively to produce an intensity distribution score for each localisation, with invasive and pre-invasive components given separate scores. Initial scoring was of 10 HPFs; however, in view of the homogeneous staining, this was reduced to 5 HPFs. Sections were scored independently by two observers and were scored as follows: negative (0), weak/moderately positive (1+) or strongly positive (2+), with DCIS and invasive components scored independently. Scores were assessed as a continuum for the purposes of statistical correlation, unless otherwise stated. Statistical analysis Data were analysed with the SPSS 10.0 for Windows statistics software and summarised with descriptive statistics. The associations between IGFBP-3 and patient characteristics were assessed with the Spearman non-parametric test for continuous variables and the χ 2 test for categorical factors. Analyses of survival data were performed with the log-rank test and the Cox regression model, and survival curves were computed with the Kaplan-Meier method. For IGFBP-3, univariate and multivariate analyses were performed, the latter adjusting for NPI score and treatment received (tamoxifen/chemotherapy/none). Because the NPI is based on nodal involvement, on tumour size and on grade, patients ( n = 11) with non-evaluable lymphadenopathy and tumour size were excluded from the multivariate regression analyses (see Table 3 ). Results IGFBP-3 expression in IDCs and DCIS and their relationships to clinicopathological factors IGFBP-3 immunoreactivity was evaluable in 101 (98%) cases of IDCs and in 102 cases of DCIS, and scored positively (1+/2+) as a homogeneous cytoplasmic expression in 86 (85%) of invasive and 92 (90%) of DCIS components (Fig. 1 ; Tables 1 and 2 ). Strong (2+) IGFBP-3 expression was seen in 32 invasive breast cancers and 40 cases of DCIS. A weak/moderate (1+) expression of IGFBP-3 was evident in the majority of invasive ( n = 54) and DCIS ( n = 52) components. There was no clear evidence of nuclear IGFBP-3 expression on IHC in IDC or DCIS. Consistently low levels of IGFBP-3 were evident throughout the stroma, without strong expression except in vascular endothelial cells. A comparison showed that the levels of IGFBP-3 expression in DCIS were similar to those in invasive disease (Table 2 ). We investigated the relationships between the levels of IGFBP-3 expression and clinicopathological parameters in IDCs and DCIS. IGFBP-3 scores were analysed as a continuum for the purposes of statistical analysis. There were no significant associations (where data was assessed as a continuum) between IGFBP-3 and established prognostic indicators in invasive disease (lymph node involvement, increasing tumour size, increasing tumour histological grade, ER negativity [quick-score 0–8], lymphovascular invasion and NPI). There was a possible trend between increasing IGFBP-3 levels and increasing cellular proliferation (Ki67) in invasive disease (Spearman correlation coefficient [cc] 0.166, P = 0.096) (where assessed as a continuum) that was not observed in DCIS ( P = 0.8) (data not shown), or where Ki67 was categorised as <10% versus ≥ 10% (Table 1 ). A comparison of categorical IGFBP-3 expression with ER-positive (quick-score 4–8) and ER-negative (quick-score 0–3) IDCs showed a trend ( P = 0.06) with ER-negative tumours and HER-2-positive invasive tumours ( P = 0.065) (Table 1 ). In Table 1 , 94% (33 of 35) of ER-negative cancers expressed either 1+ or 2+ IGFBP-3 versus 80% of ER-positive tumours (χ 2 test, P = 0.06). Similarly, 94% (15 of 16) of HER-2-positive tumours expressed IGFBP-3 (1+/2+) versus 84% of HER-2-negative cancers ( P = 0.065). There were no associations between IGFBP-3 expression and pathological variables on logistic regression in DCIS (VNPC, HER-2 expression and increased proliferation/Ki67). A categorical analysis of individual IGFBP-3 scores (negative versus 1+ versus 2+) in DCIS showed that 97% (32 of 33) of IGFBP-3-positive DCIS were ER-negative versus 80% of ER-positive DCIS ( P = 0.037) (Table 1 ). We demonstrated an inverse correlation between local IGFBP-3 expression and patient age, in which IGFBP-3 immunoreactivity analysed on a continuum decreased with age (cc -0.214, P = 0.03) (data not shown); Table 1 shows a significant association with age ( P = 0.04) when IGFBP-3 and age are analysed categorically in the invasive components, with a possible trend in DCIS (cc -0.174, P = 0.08) (data not shown). Relationships of clinicopathological factors to prognosis and the predictive potential of IGFBP-3 expression Overall survival (OS) and disease-free survival (DFS) were determined in 87 and 95 patients, respectively, with a median follow-up of 51 months (range 4–120 months). Disease relapses (local or distant recurrences) occurred in 30 women; of these, deaths were confirmed in 23 patients, with 8 suspected deaths in the absence of a recorded mortality date. Locoregional recurrence occurred at a median duration of 28.5 months (range 3–156 months) from diagnosis. Breast cancer-related mortality occurred at a median of 26 months (range 8–98 months) from presentation. The mean durations of OS and DFS were 93 months and 87 months, respectively. Four-year DFS and OS were 70% and 77%, respectively. The relationship of established clinicopathological features with OS and DFS were analysed with Cox's regression analysis (Table 3 ). Generally poor prognostic factors such as large tumour size, high tumour grade, lymphovascular invasion, lymph node metastases, ER negativity, HER-2 overexpression and NPI were significantly associated with decreased OS and DFS. High tumour proliferation (Ki67 on IHC), although associated with a lower percentage of patients remaining disease-free (DFS) and alive (OS) at 4 years, did not reach statistical significance. Univariate and multivariate analysis with the continuous score variables were used to investigate possible relationships between patient outcome data and levels of expression for IGFBP-3. IGFBP-3 scores (assessed as a continuum) were not predictive for OS or DFS after univariate or multivariate analysis (Table 3 ), adjusted for NPI and adjuvant treatment (tamoxifen/chemotherapy/none) with respect to the multivariate analysis. Where IGFBP-3 was categorised as positive (1+/2+) or negative (0), the Kaplan-Meier survival curves (Fig. 2 ) demonstrate a possible trend towards a more favourable outcome for IGFBP-3-negative tumours, although this failed to reach statistical significance for OS ( P = 0.168) or DFS ( P = 0.269), perhaps related in part to the small numbers of IGFBP-3-negative tumours observed in this study. Discussion IGFBP-3 has direct intrinsic actions, as well as regulating IGFs to influence cellular growth, survival and apoptosis of breast epithelial cells. Breast cancer cells typically express several IGFBPs, with IGFBP-2, IGFBP-3, IGFBP-4 and IGFBP-5 being observed most often [ 5 , 6 , 22 ]. Although previous studies in vivo have suggested a tumour-related upregulation of IGFBP-3 levels, clinical studies have yet to determine IGFBP-3 epithelial protein expression in breast cancers; this is the first IHC study. IHC of tissue microassays confirms a tumour-specific upregulation of IGFBP-5 and IGFBP-2 in primary breast cancers and their lymph node metastases [ 6 ]. The significance of these findings in the context of a decreased mRNA expression for IGFBP-5 and IGFBP-2, respectively, highlights the value and clinical contribution of an immunohistochemical evaluation[ B6 ]. Furthermore, a tumour-specific pathway is implicated, with negligible levels of IGFBP-5 and IGFBP-2 in normal breast epithelial cells[ B6 , 39 , 40 ]. Increasing levels of IGFBP-3 in breast cancer tissues correlate with poor prognostic features, as demonstrated in four studies in vivo through enzyme-linked immunosorbent assay and immunoradiometric assay [ 23 - 26 ]. The admixture of tumour with normal and stromal breast tissue is a feature of both quantitative methods with possible limitations regarding the evaluation of a predominant tumour-specific epithelial protein. No studies have yet clarified IGFBP-3 expression on IHC in invasive breast cancers in comparison with DCIS, although there has been a limited review of colorectal carcinomas and normal colonic mucosa[ B9 ]. This study suggests that most breast cancers express epithelial IGFBP-3 (1+/2+), mostly with a weak/moderate immunoreactivity (1+). Fewer tumours expressed IGFBP-3 strongly (2+) and it was evident more frequently in DCIS. Similar genomic aberrations may occur in DCIS and IDC to highlight possible similarities in protein expression [ 41 , 42 ]. This study, in large part, shows a consistent similarity in the levels of IGFBP-3 expression in DCIS and IDC (cc 0.789; P < 0.001) (Table 2 ). Nuclear localisation of IGFBP-3 has been described in several cell lines in vitro , including breast cancer cells[ 20 , 21 ]. This observation is supported by the direct interaction of IGFBP-3 with the nuclear receptor retinoid X receptor and the ability of IGFBP-3 to act as a nuclear-import carrier for IGF-I[ B20 , 22 , 43 ]. This has raised considerable interest in the potential nuclear actions of IGFBP-3 and its functional implications. There have been very few observations of nuclear localisation of IGFBPs in vivo , with cytoplasmic IGFBP-5 expression in breast tumours and, similarly, cytoplasmic IGFBP-3 in normal colonic crypts[ B6 , 9 ]. Moreover, there is further evidence in vitro suggesting that IGFBP-3 growth modulation might be independent of nuclear translocation[ B19 ]. In the 101 samples of IDC and DCIS analysed in the study, we observed no evidence of nuclear staining for IGFBP-3. IGFBP-3 modulates cellular proliferation with dual actions that either enhance IGFs or inhibit their actions. By contrast, in IGF-unresponsive Hs578T breast cancer cells, IGFBP-3 is predominantly growth-inhibitory and pro-apoptotic[ B11 , 27 ]. Similarly, there is a potential for IGFBP-3 to switch its action on cell survival in Hs578T cells through changes in the extracellular matrix, with a clear reversal of IGFBP-3 accentuation of apoptosis when cells are changed from being grown on either plastic, collagen or laminin to fibronectin[ B27 ]. This suggests that IGFBP-3 might be preferentially activating integrin receptors that bind fibronectin in a pro-survival growth stimulatory pathway[ B44 , 45 ]. Upregulation of fibronectin expression is a feature of breast cancer metastases and poor prognostic tumours, with evidence of IGFBP-3 binding to this mesenchymal extracellular matrix glycoprotein[ B29 , 46 , 47 ]. Tumour-related upregulation of IGFBP-3 levels in aggressive breast cancers is only partly explicable by the described findings in vitro . Evidence is accumulating that IGFBP-3 is a potential mitogen that interacts with EGFR and HER-2 signalling pathways [ 16 - 18 , 28 ]. The potential to switch the IGFBP-3 action on cell growth suggests that IGFBP-3 has a role in malignant progression of breast cancer cells with insensitivity to IGFBP-3 growth inhibition through the expression of oncogenic Ras [ 17 ]. This is further supported by four clinical studies of breast tumours, including the present study demonstrating a spectrum of increased IGFBP-3 levels, with the highest expression indicative of a more malignant phenotype [ 23 - 26 ]. Increasing passages of T47D cells switch their response to IGFBP-3 with increasing tumorigenicity, although initially growth-inhibited by this binding protein [ 18 ]. Dual growth modulation by IGFBP-3 is demonstrated in MCF10A cells, in which IGFBP-3 changes from a growth inhibitor to a mitogen through Ras-induced malignant transformation and activation of Ras-p44/42 MAPK [ 16 , 17 ]. Normal breast epithelial MCF10A cells exposed to increasing doses of IGFBP-3 show a similar biphasic response, with preliminary growth inhibition followed by IGFBP-3 mitogenicity in the context of an IGF-I receptor antagonist, or a serine phosphorylation domain peptide (SPD, a non-IGF binding peptide), to verify these IGF-independent effects[ B11 , 48 ]. LNCaP prostate cancer cells are similarly growth-stimulated by IGFBP-3 independently of IGFs [ 49 ]. In this study we found a possible association with increased proliferation, together with other adverse prognostic features such as ER negativity and HER-2 overexpression. Elevated IGFBP-3 expression might therefore suggest an abundance of IGFs sequestered by the binding protein, with further implications of an IGF-independent mitogenic role. IGFBP-3 mitogenicity may also relate to candidate proteases, such as cathepsin D, prostate-specific antigen and matrix metalloproteinases, with IGFBP-3 proteolysis potentially releasing IGFs to enhance their mitogenicity. The precise mechanism for this regulation remains unknown because studies so far have shown no clear correlations between IGFBP-3 and these proteins in tissue extracts. This study suggests an association between IGFBP-3 expression and ER negativity and confirms previous findings in vivo [ 24 , 25 ]. Clearly, there is a significant synergism between the ER and the IGFs in breast cancer cells [ 1 - 3 ]. Many members of the IGF system are under transcriptional control of the ER, with IGF-I similarly enhancing the transcriptional activity of ER[ B1 , 2 ]. Oestrogens transcriptionally downregulate IGFBPs in breast tissue and increase IGFBP-3 proteases, which may in part explain the inverse association between IGFBP-3 and ER expression [ 24 , 25 ], as well as perhaps reflecting the disruption of common pathways characteristic of poor prognostic tumours [ 2 , 50 , 51 ]. HER-2 overexpression predicts aggressive and poor prognostic breast tumours that are likely to be ER negative and tamoxifen resistant[ B52 ]. Recent evidence in vitro suggests a functional interaction between the IGF-I receptor and HER-2, with the potential for IGFBP-3 to modulate this response[ B53 , 54 ]. Our finding of an association between IGFBP-3 and HER-2 expression is preliminary and, although based on a limited number of HER-2-positive tumours, may suggest a role for HER-2 in IGFBP-3 growth modulation. In vitro , IGFBP-3 promotes EGF in HER-2-overexpressing T47D and Hs578T breast cancer cells [ 18 , 28 , 30 ]. Ageing is associated with profound changes in the growth hormone/IGF regulatory pathways, with diminished circulating IGFBP-3, as well as decreased tissue levels previously observed in vivo and confirmed in this study [ 2 , 25 ]. Although high levels of IGFBP-3 in breast cancers are associated with poor prognostic features of the tumour, few studies have substantiated any significant implications on patient outcome[ B24 , 25 ]. Tissue IGFBP-3 concentrations have been reported to predict a reduced OS, but this was not associated with breast cancer recurrence [ 25 ]. This study suggests that IGFBP-3 expression in breast cancers might be associated with a shorter OS and DFS, although few patients were negative for IGFBP-3 expression. Other studies, including the present one, show a consistent negative association between IGFBP-3 expression and favourable prognostic markers underlining the potential importance of this pathway in tumorigenesis [ 23 - 26 ]. IGFBP-3 regulates breast cancer epithelial growth through IGF-dependent and IGF-independent pathways that involve both growth inhibition and enhanced apoptosis with the potential to switch to growth stimulatory pathways interacting with EGFR, HER-2 and fibronectin. Defining these mechanisms merits further studies in vitro and in vivo . Conclusion IGFBP-3 is important in tumorigenesis because of its effects on cellular proliferation, survival and apoptosis. We have demonstrated a tumour-associated upregulation of cytoplasmic IGFBP-3 epithelial expression in invasive and non-invasive breast cancers, with similar patterns of immunoreactivity. Despite evidence in vitro and functional implications, no nuclear IGFBP-3 expression was detectable on IHC. Invasive breast cancers expressing IGFBP-3 showed an association with poor prognostic features including increased proliferation, ER negativity and HER-2 overexpression, with possible implications for patient outcome. IGFBP-3 is a growth modulator with the potential to switch from a growth inhibitor to a mitogen that interacts with the EGFR family. Abbreviations cc = Spearman correlation coefficient; DCIS = ductal carcinoma in situ ; DFS = disease-free survival; ER = oestrogen receptor; GPG = good prognostic group; HPF = high-power field; IDC = invasive ductal cancer; IGF = insulin-like growth factor; IGFBP-3 = insulin-like growth factor binding protein-3; IHC = immunohistochemistry; MAPK = mitogen-activated protein kinase; MPG = moderate prognostic group; NPI = Nottingham Prognostic Index; OS = overall survival; PPG = poor prognostic group; TBS = Tris-buffered saline; VNPC = Van Nuys Pathologic Classification. Competing interests The author(s) declare that they have no competing interests. Authors' contributions SBV carried out the immunohistochemistry for IGFBP-3 as well as the clinical and statistical analyses. CMP established conditions for testing the IGFBP-3 antibody and together with JMPH has contributed to the development of the in-house antibody. CS evaluated all scoring of IGFBP-3 expression and performed all histopathology review. CJC evaluated the IGFBP-3 scoring and re-reviewed all histopathology. JMPH developed the IGFBP-3 antibody and conditions for its application, as well as conceiving the study. ZEW conceived the study, and participated in the study design, statistical analyses and wrote the manuscript. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064109.xml

1064111

Hypersensitive K303R oestrogen receptor-α variant not found in invasive carcinomas

Introduction Genetic abnormalities or mutations in premalignant breast lesions may have a role in progression toward malignancy or influence the behaviour of subsequent disease. The A908G (Lys303→Arg) change in the gene encoding oestrogen receptor-α (ER-α) creates a hypersensitivity to oestradiol and would have significant consequences if present in breast carcinoma, especially those treated with endocrine therapy. We have therefore examined a panel of endocrine-treated invasive carcinomas for the presence of this mutation. Methods Sequencing of control DNA was shown to detect mutation present in as little as 15% of the starting material. Enrichment for the mutation by using Mbo II restriction digestion allowed the detection of mutant present at 1% or less. We applied these techniques to genomic DNA and cDNA from 136 invasive breast carcinomas. Results No evidence of the A908G mutation was found with either technique. The incidence of this mutation in our panel of tumours is therefore significantly less than previously reported. Conclusion The fact that the mutation was not found leads us to believe that this mutation is absent from most cells in invasive carcinomas and furthermore that the major expression product of the ER-α gene in cancers does not contain the K303R mutation. It is therefore unlikely to influence the effectiveness of endocrine treatment.

Introduction Most breast cancers overexpress oestrogen receptor-α (ER-α), and this molecule is the major target of the anti-estrogens, such as tamoxifen, used to treat the disease. Not all patients respond to anti-oestrogen treatment and many who do respond later relapse. The reasons for treatment failure are still unclear. The suggestion that mutation of ER-α might have a role in the formation of breast cancer and subsequent response to treatment was raised by the detection of a somatic A908G (Lys303→Arg; K303R) mutation in the gene encoding ER-α. This mutation was reported in a significant proportion of breast hyperplasia [ 1 ] and also in the majority of invasive cancers and all metastases tested [ 2 , 3 ]. The K303R ER-α variant apparently exhibits a hypersensitivity to oestradiol [ 1 ], a characteristic that might allow breast cancers to respond to much lower levels of oestrogenic stimulation with a subsequent impact on malignant progression and the effectiveness of anti-oestrogen treatment. This would be of particular importance in post-menopausal women and in women receiving anti-oestrogen therapy. We have therefore studied a cohort of post-menopausal, endocrine-treated breast cancers for the presence of this mutation. Furthermore, because expression of the hypersensitive mutant would be required for activity, we also examined mRNA. The detection of mutations in breast cancers is not a trivial task, owing in part to the heterogeneity of tumour tissue and also to the reporting of false positive and false negative results. Although microdissection allows the purity of selected cellular populations to be enhanced, the sensitivity and specificity of standard DNA sequencing approaches still limit our ability to assess mutation status accurately. We now show that by performing a second round of sequencing after enrichment for the mutation, we were able to increase the sensitivity and specificity of our assay and apply it to invasive carcinomas without the need for microdissection. Methods Determination of sensitivity for mutation detection To confirm the sensitivity of this technique for detection of the mutant allele in the presence of wild-type DNA, experiments were performed with the use of mixtures of cloned wild-type and mutant polymerase chain reaction (PCR) products, generated with primers described by Fuqua and colleagues [ 1 ] (ERmut1, 5'-CAA GCG CCA GAG AGA TGA TG-3'; ERmut2, 5'-ACA AGG CAC TGA CCA TCT GG-3'). Plasmid clones of PCR products with either an A or a G at position 908 of ER-α were mixed in various ratios (100% to 0% mutant) and diluted to the equivalent of 1000 copies per PCR reaction before being amplified and sequenced. A two-stage PCR was performed on 2 μl of mixed DNA in the 20 μl first-round reaction (20 cycles) and 4 μl of first-round product in the 40 μl second round (50 cycles). For each PCR an initial 13 min 95°C denaturation step was followed by cycles of 95°C for 30 s, 68°C for 60 s and a single 3 min final extension at 72°C. Other PCR conditions were also the same in both rounds: 0.2 mM dNTPs, each primer at 0.5 μM, 0.5 units of HotstarTaq DNA polymerase (Qiagen) and 1× PCR Buffer (containing 1.5 mM MgCl 2 ; Qiagen). PCR products were sequenced (Fig. 1 ) by using primers ERmut1 and ERmut2 and subjected to Mbo II digestion, reamplification and further sequencing as described below. Digestion with Mbo II and reamplification by PCR The Mbo II restriction enzyme has a GAAGA recognition site; this sequence is present at the K303R mutation with the second A being mutated to G. Digestion with Mbo II can therefore be used as an assay for the presence of mutation, with non-digested DNA indicating mutation within the recognition site [ 4 ]. PCR product (5–10 μl) was digested in a total volume of 20 μl including 1× Buffer2 (New England Biolabs) and 5 units of Mbo II restriction enzyme (New England Biolabs) by incubation at 37°C for 90 min. PCR products were separated by electrophoresis on gels containing 3% Seakem Agarose (Flowgen) and TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 7.6). Molecular mass markers (φX174/ Hae III; Abgene) were included on each gel and DNA was revealed by the inclusion of 0.5 μg/ml ethidium bromide, scanning with a Molecular Dynamics FluorimagerSI and analysis with ImageQuant version 4.1 software (Molecular Dynamics). After digestion the non-mutant 158-base-pair (bp) ERmut1–ERmut2 PCR product gives rise to 118-bp and 40-bp bands, detectable by agarose gel electrophoresis (Fig. 2 ), whereas 158-bp PCR products containing mutant remain undigested. Similarly ERADNA1–ERADNA2 genomic DNA and ERARNA1–ERARNA2 cDNA PCR products give different patterns of digestion products depending on the presence of a mutation affecting the Mbo II recognition site. Rather than relying on the detection of non-digested PCR products by agarose gel electrophoresis, we further increased the sensitivity and specificity of this assay by sequencing amplified non-digested DNA after digestion. PCR reamplification of non-digested DNA (20 cycles) was performed as above for primers ERmut1 and ERmut2 with 2 μl of a 1:10 dilution of the restriction digest. Reamplified PCR products were sequenced with ERmut1 primer. In all cases with any evidence of a mutant G at the relevant position, alternative PCR reactions were examined and reverse sequencing with primer ERmut2 was performed, to rule out PCR or sequencing anomalies. ER-α PCR For ESR1 genomic DNA (RefSeq NM_000125) from breast cancer cases, primers were designed that were specific for genomic DNA (because they included intronic sequence) and that better allowed sequence determination in both directions (ERADNA1, 5'-AAG TGG CCT GAA GTT TGT TA-3'; ERADNA2, 5'-CTT ACC TGG CAC CCT CTT-3'). PCR reactions contained 1 ng of genomic DNA, 0.2 mM dNTPs, each primer at 1 μM, 0.5 units of HotstarTaq DNA polymerase (Qiagen), 1× PCR Buffer (containing 1.5 mM MgCl 2 ; Qiagen) and additional MgCl 2 to a final concentration of 2.5 mM. An initial 13 min 95°C denaturation step was followed by 40 cycles of 94°C for 40 s, 60°C for 40 s, 72°C for 60 s and a single 10 min final extension at 72°C. The 609-bp genomic PCR product is digested by Mbo II to give a 296-bp product, a 6-bp product and products of 130 and 177 bp; the last two of these fail to digest if a mutation is present and instead result in a 307-bp product that can be reamplified by ERmut1 and ERmut2. For RNA analysis from breast cancer cases, reverse transcription (RT) was performed in duplicate on 0.5 μg of RNA in accordance with the manufacturer's instructions (Invitrogen), after a DNAaseI digestion step (Invitrogen). RT reactions incorporated Superscript II Reverse Transcriptase (Invitrogen), 0.5 μg of oligo (dT) 17 and 0.5 μl of Prime Recombinant Ribonuclease Inhibitor (Eppendorf). Parallel reactions were performed in which the RT enzyme was omitted; these acted as controls for genomic DNA contamination. RT–PCR reactions were performed with cDNA-specific primers designed to cross intron–exon boundaries (ERARNA1, 5'-AAG TGG GAA TGA TGA AAG GT-3'; ERARNA2, 5'-CAA GAG CAA GTT AGG AGC AA-3') and that better allowed sequence determination in both directions. These primers are located in exons 4 and 6 of ER-α and generate a 491-bp RT–PCR product. The same primers occasionally amplify additional, minor, shorter products arising from splice variants in which exon 5 is absent. PCR reactions contained 2 μl of a 1:20 dilution of the RT reaction, 0.2 mM dNTPs, each primer at 1 μM, 0.5 units of HotstarTaq DNA polymerase (Qiagen) and 1× PCR Buffer (containing 1.5 mM MgCl 2 ; Qiagen). PCR cycling conditions were the same as for genomic DNA from breast cancers. The 491-bp genomic PCR product is digested by Mbo II to give a 135-bp product, a 49-bp product and products of 130 bp and 177 bp; the last two of these fail to digest if a mutation is present and instead result in a 307-bp product that can be reamplified by ERmut1 and ERmut2. DNA sequencing PCR products were treated with ExoSAP (Amersham Biosciences) before sequencing. Fluorescent DNA sequencing (Fig. 1 ) was performed with DYEnamic ET Dye Terminator Cycle Sequencing Kit for MegaBACE (Amersham Biosciences) and analysed on a MegaBACE 1000 (Amersham Biosciences). For RT–PCR products, the primers used for PCR were also used for sequencing. When sequencing the genomic DNA from breast cancers the reverse primer ERADNA2 and an additional forward sequencing primer (ERADNA3, 5'-TAC GAA AAG ACC GAA GAG-3') were used. Primer ERADNA3 is located in exon 5 of ER-α and overcomes difficulties in sequencing through an A and T rich tract in intron 4 of ER-α; the same primer can also be used to sequence RT–PCR products and overcomes any difficulties caused by exon-5-deleted PCR products. DNA sequencing of genomic and cDNA PCR products was performed in both directions and the position of the putative mutation was analysed individually for each PCR product. To call a mutation we adopted the standard practice that it should be observed in at least two independent PCR reactions (avoiding possible PCR misincorporation due to Taq polymerase infidelity) and in both sequencing directions (avoiding sequencing anomalies due to misincorporation of dideoxy terminators). Patients and samples Post-menopausal patients undergoing treatment for invasive breast cancer during the period between 1993 and 1999 were identified within the database of the Cancer Tissue Bank Research Centre (CTBRC) at the Royal Liverpool University Hospital. The diagnoses of invasive cancers were made in accordance with the pathology guidelines of the NHS Breast Screening Program [ 5 ]. DNA and/or RNA from 136 cases were selected for analysis and provided by CTBRC along with further details as given in Table 1 . All tissue donated to CTBRC is fully consented for research purposes and permission was granted by all required local research ethics committees. Results Our ability to detect variable amounts of mutant 908G ER-α in the presence of wild-type 908A ER-α was confirmed by sequencing PCR products from mixtures of these two variants (Fig. 1 ). Direct sequencing was routinely sensitive enough to detect mutant ER-α when present in as little as 15% of the starting DNA and was frequently able to detect the mutation at even lower levels (for example 10%). This was true of sequencing in either direction, although for the original primers described by Fuqua and colleagues [ 1 ] the reverse sequence was more difficult to read because the mutation was closer to the reverse PCR primer (ERmut2). The new primers designed for use on cDNA overcame this limitation, while ensuring that only cDNA was amplified. PCR primers for genomic DNA were similarly specific for genomic DNA and, when used in conjunction with a sequencing primer (ERADNA3), again gave sequence in both directions. Digestion with Mbo II restriction enzyme allowed the detection of as little as 1% mutant DNA by agarose gel electrophoresis in control reactions (Fig. 2 ). However, no evidence of undigested PCR product was visible for any breast tumour assayed in this way. When sequencing control DNA reamplified with ERmut1 and ERmut2 after enrichment for mutant DNA by Mbo II digestion (Fig. 1 ), the mutant G base was clearly detected even when only present in 1% of the original DNA. The wild-type A was also detected, either because of inefficient digestion or because of heterodimer formation in the PCR reaction. Notably in control reactions containing only wild-type ER-α, any post-digest reamplified DNA was clearly wild type. Therefore, despite a failure to detect non-digested PCR product on gels stained with ethidium bromide, reamplification and sequencing confirms that restriction digestion is seldom 100% efficient and that by PCR we are able to reamplify products originating from 1% or less of the starting DNA. It is therefore possible to apply this reamplification and sequencing technique to all samples, because the PCR used routinely amplifies non-digested DNA enriched for mutant PCR product. The enhanced sensitivity after enrichment using digestion with Mbo II and reamplification is not without drawbacks, in that we detected a greater number of PCR and sequencing anomalies with this technique. For non-enriched sequence analysis, evidence of a very minor G in genomic DNA from two breast cancers was subsequently shown to be due to sequencing anomalies because they were not present either in repeat PCR products from the same cases or in sequence generated in the reverse direction. After enrichment, similar errors were noted in eight genomic DNA PCR products and six cDNA PCR products. Ten of these anomalies were apparently due to Taq polymerase infidelity (that is, they were identified in reverse sequencing but not repeatable in replicate PCR) and four were sequencing anomalies (that is, they were not identifiable in reverse sequencing). Notably, errors due to Taq polymerase infidelity were seen only with the more sensitive assay based on enrichment with Mbo II, and no such anomalies of either type were seen in any non-mutant controls at the relevant base position. We were unable to confirm that the proposed ER-α A→G mutation leading to a K303R amino acid substitution was present in any case of invasive cancer examined. In all, we examined 136 cases of invasive cancer (Table 1 ) with our Mbo II-enriched sequencing assay, having also sequenced 130 of these before enrichment. With the exception of the 14 cases noted previously, in every sequenced PCR product the A at the base position of the variant was clearly an A with no evidence of any G substitution. For 60 of these cases we examined both genomic DNA and cDNA, for 71 cases genomic DNA only was examined (for example, for ER-α-negative cancers) and for 5 cases only cDNA was examined. All 100 cases of invasive cancer for which a pathologist assessed the proportion of tumour cells contained at least 50% tumour, and 63% contained at least 90% tumour cells. Given the proven sensitivity of our techniques we would therefore have expected to detect the hypersensitive ER-α mutation even if present in only a minority of cancer cells or contaminating normal cells. Discussion The abundant expression of ER-α in most breast cancers is fundamental to both our understanding of this disease and its treatment. The observations that ER-α is overexpressed in a proportion of premalignant lesions [ 6 , 7 ] and is possibly related to an increased risk of progression [ 8 ] further raise the importance of oestrogenic activity in the establishment and behaviour of breast carcinoma. It is therefore important to understand whether other mechanisms for increased ER-α activity are also present in breast tumours or their putative precursors. One such mechanism is the hypersensitivity apparently inferred by a K303R mutant ER-α reported to be present in a significant proportion of breast hyperplasia [ 1 ]. We have been unable to detect the reported A908G mutation of ER-α in genomic DNA from any case of invasive carcinoma in our study, despite applying a sensitive and robust assay capable of detecting mutant if present in as little as 1% of the starting DNA. The absence of this hypersensitive ER-α variant suggests that this mutation is not present in the majority of cells in such lesions, or indeed even in a significant minority. Furthermore, the absence of the mutation in RT–PCR products confirms that non-mutant ER-α is the major expression product in the breast cancers. Zhang and colleagues also failed to detect the mutation in a variety of breast lesions from Japanese women [ 4 ], and while this manuscript was under review two further papers have reported a lack of evidence for the K303R mutation [ 9 , 10 ]. We have previously been unable to detect the mutation by sequencing a series of 56 microdissected ductal carcinomas in situ from an unrelated cohort, confirming that this mutation is also apparently absent from these premalignant lesions. Case selection can influence detection rates, and the cohort of patients tested here was selected by virtue of being post-menopausal and treated with endocrine therapies but not chemotherapy, so as to allow any effect on endocrine treatment to be more easily defined. Although it is possible that the mutation occurs in other groups of breast cancers, our cohort is broadly representative of subgroups defined by ER-α status, nodal status, grade, tumour size and histology. Other studies reporting an inability to detect the mutation [ 4 , 9 , 10 ] have applied no obvious selection criteria and it therefore seems unlikely that any selection bias is to blame for the lack of detectable mutation. The low level of risk of subsequent cancer attached to hyperplasia implies that many of these lesions fail to contribute to disease progression and it is perhaps therefore unsurprising that a mutation so far reported in only about one-third of hyperplasias by a single laboratory is found to be absent from more advanced lesions in the present study. Preliminary reports of the K303R mutation in a higher proportion of breast cancers [ 2 , 3 ] from the same group as reported the original observations in hyperplasia are perplexing. Although confirmatory data have not so far been published, these results continue to be reported at scientific meetings and referenced in reviews [ 11 ]. Many previous studies of mutation in breast cancers have failed to report this mutation, and more recent reports [ 4 , 9 , 10 ] have also produced evidence that if it exists it does so below the level of detection of the techniques used. It is therefore important to consider the sensitivity and specificity of the different assays used to detect the K303R mutant. The fact that Zhang and colleagues analysed 7 breast hyperplasias [ 4 ] and Tebbit and colleagues analysed 25 hyperplasias [ 9 ] and found no mutation rules out the use of hyperplasia as 'positive' controls and challenges the original observations. Here the sequencing of PCR products with fluorescent dideoxy terminators is shown to be sensitive enough to detect a 908G mutant when present at 15% or less of the starting DNA, in keeping with the results of Tebbit and colleagues [ 9 ]. This is probably suitable for use with microdissected lesions as used by Tebbit and colleagues for 36 invasive cancers, or lesions with a relatively high tumour content as used here (Table 1 ), but might not be sensitive enough for use with non-microdissected cancers used elsewhere [ 4 , 10 ] or if the mutation is present in only a small minority of cancer cells. We therefore also used gel-based PCR with restriction-fragment-length polymorphism and were able to detect as little as 1% mutant by agarose gel electrophoresis of PCR products from control mixing experiments. Although the same technique was used by Zhang and colleagues for 215 cancers [ 4 ], they did not report their detection sensitivity or confirm their specificity. Given that this technique potentially detects very low levels of mutation at any position in the Mbo II recognition site, it is important to confirm sensitivity and specificity by a further round of DNA sequencing. A very significant enhancement of sequencing sensitivity (to 1% mutant or less) was seen with this Mbo II digestion and reamplification technique. We can therefore state that this mutation is not present in the vast majority of tumour cells of any cancer tested, even without the benefit of enrichment for tumour cells by microdissection. Although assay sensitivity is important, specificity is equally crucial. We never reported any mutant in either assay used for our non-mutant control DNA, indicating that our specificity is high. However, in our experience, false positive mutations by DNA sequencing, as seen here in 16 of the 136 cancers tested, are to be expected when PCR reactions are performed on very low levels of DNA as found in many microdissected samples, or after digestion with restriction enzyme. It is therefore important to perform rigorous confirmatory assays. To this end, we always sought to confirm mutations by sequencing in both directions to avoid anomalies that often occur as a result of the misincorporation of dideoxy terminators in unidirectional sequencing. We also sequenced multiple PCR products to avoid anomalies due to Taq polymerase infidelity. Originally, the K303R mutation was reported in 34% of microdissected hyperplasias tested by unidirectional, non-fluorescent sequencing [ 1 ]. This assay is prone to false positives and it is unclear whether replicate PCR or other confirmatory assays were performed. The fact that the mutation was always found in addition to the wild type is also to be expected with artefacts due to dideoxy terminator misincorporation. The possibility that observations of the K303R mutation in hyperplasia result from a lack of specificity must therefore be considered and some doubt cast over other positive results in cancers. Conclusion We have developed and validated a highly sensitive and specific assay for the detection of the A908G (K303R) mutation of ER-α. Having tested 136 different breast tumours, we find no evidence to support the hypothesis that the A908G mutation of ER-α is present or expressed in breast carcinoma cells from post-menopausal, endocrine-treated patients. It is therefore unlikely to have an impact on the effectiveness of such treatments. Abbreviations ER-α = oestrogen receptor-α; PCR = polymerase chain reaction; RT = reverse transcription. Competing interests The author(s) declare that they have no competing interests. Authors' contributions All authors listed contributed to the production of this manuscript. PAO and HI provided clinical review of cases studied; PAO prepared the cDNA used; HI provided genomic DNA; mutation detection was performed by MPAD, and DRS provided both technical insight and critical review. The manuscript was produced by MPAD. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064111.xml

1064114

Atm heterozygous deficiency enhances development of mammary carcinomas in p53 heterozygous knockout mice

Introduction Ataxia-telangiectasia is an autosomal-recessive disease that affects neuro-immunological functions, associated with increased susceptibility to malignancy, chromosomal instability and hypersensitivity to ionizing radiation. Although ataxia-telangiectasia mutated ( ATM ) heterozygous deficiency has been proposed to increase susceptibility to breast cancer, some studies have not found excess risk. In experimental animals, increased susceptibility to breast cancer is not observed in the Atm heterozygous deficient mice ( Atm +/- ) carrying a knockout null allele. In order to determine the effect of Atm heterozygous deficiency on mammary tumourigenesis, we generated a series of Atm +/- mice on the p53 +/- background with a certain predisposition to spontaneous development of mammary carcinomas, and we examined the development of the tumours after X-irradiation. Methods BALB/cHeA- p53 +/- mice were crossed with MSM/Ms- Atm +/- mice, and females of the F 1 progeny ([BALB/cHeA × MSM/Ms]F 1 ) with four genotypes were used in the experiments. The mice were exposed to X-rays (5 Gy; 0.5 Gy/min) at age 5 weeks. Results We tested the effect of haploinsufficiency of the Atm gene on mammary tumourigenesis after X-irradiation in the p53 +/- mice of the BALB/cHeA × MSM/Ms background. The singly heterozygous p53 +/- mice subjected to X-irradiation developed mammary carcinomas at around 25 weeks of age, and the final incidence of mammary carcinomas at 39 weeks was 31% (19 out of 61). The introduction of the heterozygous Atm knockout alleles into the background of the p53 +/- genotype significantly increased the incidence of mammary carcinoma to 58% (32 out of 55) and increased the average number of mammary carcinomas per mouse. However, introduction of Atm alleles did not change the latency of development of mammary carcinoma. Conclusion Our results indicate a strong enhancement in mammary carcinogenesis by Atm heterozygous deficiency in p53 +/- mice. Thus, doubly heterozygous mice represent a useful model system with which to analyze the interaction of heterozygous genotypes for p53 , Atm and other genes, and their effects on mammary carcinogenesis.

Introduction Ataxia-telangiectasia is an autosomal-recessive disease that affects neuro-immunologic functions, and is associated with increased susceptibility to malignancy, chromosomal instability and hypersensitivity to ionizing radiation [ 1 , 2 ]. ATM (ataxia-telangiectasia mutated) heterozygous deficiency has been proposed to increase susceptibility to breast cancer [ 3 - 7 ]. However, those early studies were limited by the lack of reliable assays with which to identify carriers [ 8 ]. In fact, in a later study a lack of association of heterozygous ATM mutations with early onset of breast cancer was found [ 9 ]. More recent epidemiological studies suggested that missense mutation in the ATM gene, rather than a protein-truncating mutation, which accounts for the majority of mutations in patients with ataxia-telangiectasia, confers increased risk for breast cancer [ 10 ]. Thus, cancer risk in ATM heterozygotes varies depending on the mutation type (i.e. some missense-type mutations are associated with early onset of breast carcinoma whereas truncation-type mutations are not) [ 11 , 12 ]. Recently, epidemiological studies on excess risk for breast cancer in ATM heterozygosity were reported [ 13 ]. In experimental animals, no tumours were observed in Atm heterozygous mice carrying a knockout null allele of Atm [ 14 ]. In contrast, Atm knock-in heterozygous mice harbouring an in-frame deletion corresponding to the human mutation exhibit increased susceptibility to a wide variety of tumours [ 14 ]. Thus, data in humans and mice suggest that the type of Atm mutation determines susceptibility to cancer in heterozygous individuals. Heterozygosity for a null knockout allele of Atm in mice and protein-truncating alleles of ATM in humans was thought not to increase susceptibility to mammary cancer. On the other hand, haploinsufficiency at the Atm gene has a phenotype of increased sensitivity to ionizing radiation in mice [ 15 ]. In contrast to the Atm gene, the p53 null allele exhibits haploinsufficiency for the development of tumours in mice, mainly lympho-haematopoietic malignancies [ 16 , 17 ]. The p53 heterozygotes of BALB/c genetic background develop mammary tumours [ 18 - 20 ]. Mice doubly null for the p53 and Atm genes were reported to exhibit a dramatic acceleration in tumour formation relative to singly null mice, indicating that the genes cooperate in a significant manner to prevent tumourigenesis [ 21 ]. However, the authors noted no mammary carcinoma in any of the four genotypes studied ( p53 +/+ Atm -/- , p53 +/- Atm -/- , p53 -/- Atm +/- , and p53 -/- Atm -/- ). Thus, the significance of haploinsufficiency of the Atm null allele in mammary carcinogenesis is obscure at present. In order to determine the effect of Atm heterozygous deficiency on mammary tumourigenesis, we generated a series of Atm +/- mice on the background of p53 +/- mice with a certain predisposition to spontaneous development of mammary carcinomas, and we examined the development of tumours after X-irradiation. Our results indicate a strong enhancement of mammary carcinogenesis in the Atm heterozygous deficient mice under the p53 heterozygous deficiency. Materials and Methods Mice The p53 targeted allele generated by Donehower and coworkers [ 22 ] was introduced into the BALB/cHeA mouse at The Netherlands Cancer Institute (Amsterdam). The p53 heterozygous deficient mice ( p53 +/- ) were repeatedly backcrossed to BALB/cHeA mice more than 30 times, and maintained at the animal facility of Osaka Prefecture University. The Atm targeted mouse (129/SvEv- Atm tm1Awb /+ mouse) was originally generated in the Jackson Laboratory [ 23 ]. The Atm heterozygous deficient mice ( Atm +/- ) were repeatedly backcrossed more than 10 times to MSM/Ms mice. The BALB/cHeA- p53 +/- mice were crossed with MSM/Ms- Atm +/- mice, and females of the F 1 progeny ([BALB/cHeA × MSM/Ms]F 1 ) with four genotypes (i.e. p53 +/- Atm +/- , p53 +/- Atm +/+ , p53 +/+ Atm +/- and p53 +/+ Atm +/+ ) were used in the experiments. The conditions for breeding were described previously [ 24 ]. X-irradiation Mice were exposed at 5 weeks of age to X-rays (5 Gy; 260 kV, 12.0 mA, 0.3 mm Cu + 0.5 mm Al filter; 0.5 Gy/min) from an X-ray generator (Radioflex 350; Rigaku Industrial Co., Takatsuki, Japan). All animal experiments were carried out in accordance with the standards relating to the care and management of experimental animals (Japan) and Osaka Prefecture University's guidelines for animal care and use. Histopathological examination Moribund mice were killed by cervical dislocation for autopsy. In cases of thymic lymphoma, the enlarged thymuses were examined as previously described [ 25 ]. In tumour-bearing mice the tumours were fixed in 10% buffered formalin, processed histologically, and stained with haematoxylin and eosin. The processed tumour specimens were evaluated by medical and veterinary pathologists using the Annapolis guidelines established by Cardiff and coworkers [ 26 ]. DNA isolation and genotyping Normal and mammary carcinoma tissues were removed. Isolation of DNA, PCR amplification, electrophoresis of PCR products and assessment of allelic losses were performed according to a procedure described previously [ 24 ]. Genotypes for the wild-type and targeted alleles of p53 and Atm genes were determined by analyzing the PCR products for these alleles. Amplification for the p53 alleles was done as described elsewhere [ 27 ]. The wild-type and the targeted alleles of the p53 gene were amplified by PCR using primers p53 -4F (5'-CGACCTCCGTTCTCTCTCCTCTCTT-3') and p53 -6R (5'-AGACGCACAAACCAAAACAAAATTACA-3'), and primers p53 -NF (5'-GCCTTCTATCGCCTTCTTGACGAGT-3') and p53 -6R, respectively. Similarly, amplification of the wild-type and the targeted allele of the Atm gene were performed by using primers IMR0640F (5'-GCTGCCATACTTGATCCATG-3') and IMR0641R (5'-TCCGAATTTGCAGGAGTTG-3'), and primers IMR0640F and AtmNeo410R (5'-CGGTGGATGTGGAATGTGTG-3'), respectively. Statistical analysis Statistical significance was evaluated for the incidence of mammary carcinoma and number of carcinomas per mouse by χ 2 analysis and Mann–Whitney U-test, respectively. Comparison of latency in mammary carcinoma development was examined by unpaired Student's t-test. Results and discussion Histopathological features of tumours developed in F 1 mice doubly heterozygous for p53 and Atm null alleles Mammary carcinomas occurred in BALB/c mice of the p53 +/- genotype, and tumours of similar macroscopic morphologies were also observed in the (BALB/c × MSM/Ms)F 1 – p53 +/- mice. The histological features of these tumours in the mammary glands are shown in Fig. 1a,1b,1c,1d,1e,1f , which indicates that they are adenocarcinomas. The histopathology of the tumours in nonirradiated mice (Fig. 1a,1b ) exhibited more hyperplastic lesions than did those in irradiated mice (Fig. 1c,1d,1e,1f ), but basically there were no marked differences between the two groups. There was also no remarkable difference in histopathological features between Atm +/- (Fig. 1c,1d ) and Atm +/+ mice (Fig. 1e,1f ) in the irradiated groups. These tumours formed glands lined by highly pleomorphic cells exhibiting frequent mitosis, and were classified as glandular adenocarcinomas, high grade according to the Annapolis Pathology Classification [ 26 ]. Lymphomas (Fig. 1g ), mainly thymic lymphomas, were efficiently induced by exposure to X-rays, regardless of p53 and Atm genotype, although only a few lymphomas were observed in nonirradiated mice. Ovarian carcinomas, osteosarcomas (Fig. 1h ) and hepatomas developed spontaneously. Squamous cell carcinomas, basal cell carcinomas, histiocytic sarcomas and granulocytic leukaemias were also observed in irradiated mice. Spontaneous tumour development in mice with four genotypes for p53 and Atm Twenty-eight p53 +/- Atm +/- , 22 p53 +/- Atm +/+ , 11 p53 +/+ Atm +/- and 25 p53 +/+ Atm +/+ mice were examined for spontaneous development of tumours until age 26 months (113 weeks). Fourteen out of 28 p53 +/- Atm +/- mice (50%) and seven out of 22 p53 +/- Atm +/+ mice (32%) developed mammary carcinomas during the period of observation (Table 1 , Fig. 2 ). The incidence of mammary carcinomas in p53 +/- Atm +/- mice appeared to be higher than that in p53 +/- Atm +/+ mice, but the incidences did not differ significantly between Atm +/- and Atm +/+ genotypes ( P = 0.32, by Fisher's exact probability test). Among the tumours that developed in the p53 +/- mice, mammary carcinoma were the most common. These mammary carcinomas were mainly observed at 41–77 weeks after birth (Fig. 2 ). No significant difference in latency in the nonirradiated groups was observed between doubly heterozygous mice and p53 singly heterozygous mice ( P = 0.47, by unpaired Student's t-test). None of 11 p53 +/+ Atm +/- mice and 25 p53 +/+ Atm +/+ mice developed mammary carcinomas. Several lymphomas and a few other tumours developed in the four genotypes (Table 1 ). Thus, mammary carcinoma development depended strongly on p53 heterozygous deficiency in (BALB/c × MSM/Ms)F 1 mice, and p53 +/- mice of both Atm +/+ and Atm +/- genotypes developed mammary carcinoma. Enhancement of mammary carcinogenesis in Atm heterozygous deficient mice by X-irradiation To test the effect of haploinsufficiency of the Atm gene on mammary carcinogenesis after X-irradiation in p53 +/- mice, 55 p53 +/- Atm +/- , 61 p53 +/- Atm +/+ , 47 p53 +/+ Atm +/- and 53 p53 +/+ Atm +/+ mice (for a total of 216 mice) were exposed to X-rays (5 Gy) at age 5 weeks. Only one out of 53 p53 +/+ Atm +/+ mice and none of 47 p53 +/+ Atm +/- mice developed mammary carcinoma, indicating that almost all p53 +/+ mice fail to develop mammary carcinomas, despite X-irradiation and irrespective of Atm gene status. In contrast, 32 out of 55 p53 +/- Atm +/- mice (58%) and 19 out of 61 p53 +/- Atm +/+ mice (31%) developed mammary carcinomas (Table 2 , Fig. 2 ). The proportion of mice developing mammary carcinomas in the p53 +/- Atm +/- group was significantly greater than that in the p53 +/- Atm +/+ group ( P = 0.0034, by χ 2 test). A total of 52 mammary carcinomas developed in 55 p53 +/- Atm +/- mice (average number of mammary carcinomas/mouse = 0.95), whereas 28 mammary carcinoma developed in 61 p53 +/- Atm +/+ mice (average number of mammary carcinomas/mouse = 0.46; Table 2 ). The average number of mammary carcinomas per mouse in the p53 +/- Atm +/- group was significantly greater than that in p53 +/- Atm +/+ mice ( P = 0.0052, by Mann–Whitney U-test). Thus, Atm heterozygous deficiency enhanced development of mammary carcinoma in irradiated p53 heterozygous knockout mice. Spring and coworkers [ 14 ] observed no tumours in Atm knockout ( Atm +/- ) heterozygous mice. Mice bearing a knockout allele of Atm and humans carrying a mutant allele of truncated type in the ATM gene have been shown not to have obviously elevated susceptibility to mammary carcinogenesis. Our findings show that heterozygosity for a null knockout allele of Atm enhances mammary carcinogenesis under p53 +/- status, although the Atm mutation is not a dominant-negative type. Heterozygous deficiency of p53 might make clear the effect on mammary carcinogenesis of haploinsufficiency in the Atm gene. These mammary carcinomas were observed significantly earlier (at 18–38 weeks after irradiation; i.e. 23–43 weeks of age) than in the nonirradiated group (age 41–75 weeks; Fig. 2 ). In particular, mammary carcinomas frequently developed 23–28 weeks after irradiation. The mean (± standard deviation) latency periods were 32.6 ± 4.8 and 29.8 ± 3.6 weeks in p53 +/- Atm +/- and p53 +/- Atm +/+ mice, respectively. Thus, X-irradiation at 5 Gy at age 5 weeks considerably shortened the latency period of mammary carcinoma development in these two groups with different genotypes. As shown in Tables 1 and 2 , the incidences of mammary carcinoma in p53 +/- Atm +/- mice were 58% (32 out of 55) and 50% (14 out of 28) in irradiated and nonirradiated groups, respectively; in p53 +/- Atm +/+ mice the incidence in the irradiated group was 31% (19 out of 61) and that in the nonirradiated group was 32% (7 out of 22). The incidence of mammary carcinoma for each genotype was similar between irradiated and non-irradiated groups. Thus, irradiation may not elevate the incidence of the tumours. Altogether, irradiation markedly hastened mammary carcinoma development in the p53 +/- mice, in which mammary carcinomas developed spontaneously. Furthermore, irradiation also induced lymphomas, mainly thymic lymphomas, in all four genotypes of mice. The incidence of the lymphomas did not differ significantly among the four groups, with different genotypes for p53 and Atm genes. A high incidence of thymic lymphoma was observed in previous studies performed using p53 heterozygous deficient F 1 mice [ 17 , 28 ]. In the present study an extremely high incidence of tumours, most of which were mammary carcinomas and thymic lymphomas, was observed in irradiated p53 +/- Atm +/- mice. Status of wild-type alleles of p53 and Atm in mammary carcinoma The wild-type alleles of the p53 and Atm genes were examined in mammary carcinoma tissue from heterozygous mice. The wild-type p53 allele was lost in 25 (96%) out of 26 mammary carcinomas from irradiated p53 +/- mice and in all of 15 mammary carcinomas from nonirradiated p53 +/- mice, regardless of Atm gene status. Only one mammary carcinoma in irradiated p53 +/+ Atm +/+ mice (Table 2 ) was found to retain the p53 wild-type allele. On the other hand, wild-type Atm allele was preserved in all of 17 mammary carcinomas from irradiated Atm +/- mice and in all of 10 mammary carcinoma s from irradiated Atm +/+ mice, regardless of p53 gene status. Wild-type Atm allele was also retained in nine out of 10 mammary carcinomas from nonirradiated Atm +/- mice and in all of five mammary carcinomas from nonirradiated Atm +/+ mice. These results suggest that the homozygous loss of the p53 allele was a necessary condition for the development of mammary carcinomas, whereas the Atm null allele exhibited haploinsufficiency. Haploinsufficiency for tumour development was reported for the Nbn knockout mice, the mouse homologue of NBS1 . Heterozygosity for the Nbn knockout allele rendered the mice susceptible to tumour development, yet the Nbn wild-type allele was fully retained in all 12 tumours examined [ 29 ]. p27 heterozygotes are also predisposed to tumours in multiple tissues when challenged with irradiation or a chemical carcinogen, and in the developed tumours the remaining wild-type allele is neither mutated nor silenced, indicating that p27 is haploinsufficient for tumour suppression [ 30 ]. In that study, heterozygous Atm knockout enhanced mammary carcinoma development in p53 -heterozygous deficient mice, but the effect of Atm deficiency was not as profound. A previous study showed that heterozygosity for the Atm knockout allele did not enhance tumour development but that dominant-negative type missense mutations in the Atm gene did [ 14 ]. In humans, heterozygosity of the truncation-type mutations, which represent the majority of Atm mutations that occur in humans, had no effect on carcinogenesis, suggesting that enhancement in tumourigenesis depends strongly on mutation type. However, in the present study we demonstrated that haploinsufficiency does occur for the Atm null allele in combination with heterozygous deficiency in the p53 gene. The p53 heterozygous deficient BALB/c mice, which developed mammary carcinomas early and efficiently, may represent a useful model for the study of effects of genes other than Atm on mammary carcinogenesis. F 1 mice between different subspecies may also provide an experimental system for precise genome-wide allelotype analysis of genes that cooperate with p53 . Conclusion Tumourigenesis is strongly enhanced in mice with homogeneous deficiency in the p53 or Atm gene. In the present study we tested the effect of haploinsufficiency of the Atm gene on mammary tumourigenesis after X-irradiation in p53 +/- mice of the BALB/cHeA × MSM/Ms background. Singly heterozygous p53 +/- mice X-irradiated (5 Gy) at age 5 weeks developed mammary carcinomas at around 25 weeks of age, and the final incidence of mammary carcinoma at 39 weeks was 31% (19 out of 61). Introduction of the heterozygous Atm alleles into the background of the p53 +/- genotype significantly increased the incidence of mammary carcinomas to 58% (32 out of 55) and increased the average number of mammary carcinomas per mouse. However, it apparently did not change the latency of mammary carcinoma development. In nonirradiated mice, introduction of the Atm +/- allele into p53 +/- mice also tended to increase spontaneous incidence of mammary carcinoma. In contrast, almost none of the p53 +/+ mice developed mammary carcinoma, regardless of the Atm gene status and whether mice were subjected to irradiation. In almost all of the spontaneous and radiation-induced mammary carcinomas, the wild-type p53 allele was found to be lost whereas the wild-type Atm allele was retained, suggesting haploinsufficiency of the latter gene in mammary carcinoma development. Thus, doubly heterozygous mice represent a useful model system with which to analyze the interaction of heterozygous genotypes for p53 , Atm and other genes and their effect on mammary carcinogenesis. Abbreviations PCR = polymerase chain reaction. Competing interests The author(s) declare that they have no competing interests. Authors' contributions SU carried out design of the study, observed mammary carcinoma development and other diseases, and drafted a manuscript. KF prepared tissue specimens and conducted histopathological examinations (veterinarian). SI carried out DNA isolation and genotyping of mice. NM carried out X-irradiation of mice and statistical analysis. MT performed DNA isolation and genotyping of mice. DH carried out production of heterozygous deficient mice. CS participated in designing the study and discussion of data on mammary carcinoma development. SH contributed to histopathological examination. SI conducted histopathological examinations (medical doctor). ON participated in discussion of data and contributed to preparation of the final manuscript. MO carried out design of the study and X-irradiation, and wrote the final manuscript. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064114.xml

1064115

Letrozole sensitizes breast cancer cells to ionizing radiation

Introduction Radiotherapy (RT) is considered a standard treatment option after surgery for breast cancer. Letrozole, an aromatase inhibitor, is being evaluated in the adjuvant setting. We determined the effects of the combination of RT and letrozole in the aromatase-expressing breast tumour cell line MCF-7CA, stably transfected with the CYP19 gene. Methods Irradiations were performed using a cobalt-60 source with doses ranging from 0 to 4 Gy. Cells were incubated with androstenedione in the presence or absence of letrozole. Effects of treatment were evaluated using clonogenic assays, tetrazolium salt colorimetric (MTT) assays, and cell number determinations. Cell-cycle analyses were conducted using flow cytometry. Results The survival fraction at 2 Gy was 0.66 for RT alone and was 0.44 for RT plus letrozole ( P = 0.02). Growth of MCF-7CA cells as measured by the cell number 6 days after radiotherapy (2 and 4 Gy) was decreased by 76% in those cells treated additionally with letrozole (0.7 μM) compared with those receiving radiotherapy alone ( P = 0.009). Growth inhibition, assessed either by cell number ( P = 0.009) or by the MTT assay ( P = 0.02), was increased after 12 days of the combination treatment. Compared with radiation alone, the combination of radiation and letrozole produced a significant decrease in radiation-induced G 2 phase arrest and a decrease of cells in the S phase, with cell redistribution in the G 1 phase. Conclusions These radiobiological results may form the basis for concurrent use of letrozole and radiation as postsurgical adjuvant therapy for breast cancer.

Introduction The aromatase inhibitor letrozole has been shown to be superior to tamoxifen (TAM) in the first-line treatment of metastatic breast cancer [ 1 , 2 ]. A randomized clinical trial comparing these two drugs in the adjuvant setting is currently ongoing, and should soon provide insight into the relative efficacy of letrozole and TAM. Another adjuvant trial, the MA.17 study, evaluated patients from the National Cancer Institute of Canada Clinical Trials Group who were disease-free after initially being treated with 5 years of TAM and then being randomized to receive either 5 years of placebo or 5 years of letrozole treatment. As compared with placebo, letrozole therapy after the completion of TAM treatment significantly improved disease-free survival [ 3 ]. Postoperative radiotherapy (RT) decreases the risk of locoregional recurrence. RT is also associated with improved survival in high-risk premenopausal and postmenopausal women with breast cancer given either adjuvant chemotherapy [ 4 ] or TAM [ 5 ], respectively. Whether letrozole sensitizes breast cancer cells to RT has not been determined. The present study has examined this question in vitro , using a human breast cancer cell line stably transfected with the human aromatase gene. We present herein the capacity of letrozole to sensitize these cells to RT using clonogenic, MTT, and cell-count assays. Cell-cycle analyses showed a G 1 phase arrest and a decrease of the S phase with letrozole as compared with control cells. Moreover, compared with RT alone, combined RT and letrozole produced a significant decrease in radiation-induced G 2 phase arrest and in the number of cells in the S phase, with cell redistribution in the G 1 phase. Materials and methods Cell line, culture, and letrozole MCF-7 human breast cancer cells stably transfected with the human aromatase/CYP19 gene (MCF-7CA) were kindly provided by Dr S Chen (City of Hope, Duarte, CA, USA [ 6 ]). These cells were maintained in DMEM/NUT MIX F-12 containing 10% heat-inactivated FCS (Gibco Laboratories, Cergy Pontoise, France), 500 μg/ml geneticin (G418), 300 μg/ml glutamine, 0.25 μg/ml fungizone, 100 μg/ml streptomycin, and 100 units/ml penicillin G. The two cell lines, MCF-7 wild type and MCF-7CA, were adherent and grew as monolayers at 37°C in a humidified 5% CO 2 incubator. The cells were harvested with 0.5 g/l trypsin (Gibco Laboratories) and 0.2 g/l EDTA (Gibco Laboratories) for 3 min. Cultures were checked for the absence of mycoplasma every month. Steroids were removed from the FCS by two treatments with dextran-coated charcoal (DCC) [ 7 ]. Cells were cultured as monolayers in 60-mm Petri dishes with phenol-red-free medium and DCC FCS. Letrozole used in the study was synthesized in the laboratories of Novartis Pharma AG (Basel, Switzerland). In all experiments, letrozole was added to cultures that were nonconfluent at concentrations of 7 nM–0.7 μM in 10 μl ethanol 6 days before RT. Aromatase tritiated water assay Aromatase activity was determined using the 3 H 2 O release method reported by Zhou and colleagues [ 6 ], based on the procedure described by Thompson and Siiteri [ 8 ]. Cells were maintained for 5 days in 10% FCS phenol-red-free medium and then seeded at 700,000 cells per well in six-well dishes with fresh DCC-treated medium. Three days later, culture plates were washed twice with PBS. One millitre of serum-free medium containing 40 nM 1β- 3 H(N)-androst-4-ene-3,17-dione (NET-926; NEN PerkinElmer Life Sciences, Inc., Courtaboeuf, France) as substrate (specific activity, 25.9 Ci/mmol) with or without 100 nM letrozole was then added to each well. After 6 hours of incubation at 37°C, the reaction mixture was removed and extracted with two volumes of chloroform to terminate the reaction and to extract unused substrate and steroids. After 5 min centrifugation at 100 × g , the aqueous phase was then treated with an equal volume of 5% charcoal suspension to eliminate residual steroids. After 5 min centrifugation at 15,000 × g , radioactivity was assessed by liquid scintillation counting. The protein concentration was determined by the Bradford method [ 9 ] after cell extraction with Reporter Lysis 5X Buffer according to the manufacturer's instructions (PROMEGA Corp., Lyon, France). Aromatase activity was then calculated from the disintegrations per minute as picomoles of oestrogen produced by milligrams of protein per hour Radiation modalities Cells were plated in 10 ml DCC and phenol-red-free medium (to ensure homogeneous energy deposition within each dish using 60-mm Petri dishes) and irradiated with a cobalt-60 source (γ irradiation, ELITE 100; Theratronics, Ottawa, ON, Canada) in the Radiation Department. The radiation was delivered as a single dose ranging from 0 to 4 Gy in an 11 cm × 11 cm field size at a dose rate of 0.5 Gy/min. A 3-cm polystyrene block was used under the Petri dishes during each irradiation to allow homogeneous backscattering γ-rays. The source-half-depth distance was initially calculated to obtain a constant dose rate of 0.5 Gy/min, and was adapted monthly to the cobalt-60 source radioactivity decay. Control cells were removed from the incubator and were placed for the same period of time under the cobalt-60 source but without RT. In the combined treatment modality studies, letrozole was added 6 days prior to RT. Cytotoxicity assays Cultures were trypsinized and washed, and cells were plated in quintuplicate at a density of 100 cells per 60-mm Petri dish after dilution. Letrozole was added at concentrations ranging from 7 nM to 0.7 μM 24 hours after the cells were plated to allow for cell attachment. Cells were incubated at 37°C in a humidified chamber containing 5% CO 2 for 12 days. The colonies were then fixed with a 1:3 (v/v) acetic acid methanol solution and were stained with 10% Giemsa (Sigma Chemical Co., St Louis, MO, USA); colonies of greater than 50 cells were scored. The plating efficiency was calculated with and without letrozole. The cytotoxicity of RT against asynchronous, exponentially growing cells was also determined from colony formation assays. Before irradiation, the cell density was determined using appropriate dilutions (100, 100, 200, 300, and 400 cells for 0, 1, 2, 3, and 4 Gy, respectively), and cells were plated onto five replicate 60-mm Petri dishes. Cells were irradiated as already described 24 hours after plating to allow for cell attachment before the administration of RT. The letrozole-containing medium was given at a concentration of 0.7 μM 6 days before RT. Cultures were irradiated with the drug present in the medium, and were immediately returned to the incubator after irradiation. Colonies were counted after 12 days. Experimentally derived data points are the mean of five experiments. The dose–response curves were fitted to a four-parameter model, where the response R varies with the dose D according to the equation: R = a /[1 + ( D / b ) c ] + R ∞, where a is the difference between the maximum and minimum response, b is the concentration of drug needed to obtain 50% of the maximal effect, c is a slope factor, and R ∞ is the maximal effect. The multitarget model survival curves were fitted to the data using a least-squares regression to the linear-quadratic model: S = S 0 exp(-α D 1 - β D 1 2 ), where D 1 is the radiation dose, S is the surviving fraction, and S 0 is a normalizing parameter. MTT and cell number assays The antiproliferative effect of letrozole with or without RT on the MCF-7CA cell line was evaluated using the MTT assay as described by Mosmann [ 10 ]. Briefly, exponentially growing cells were seeded into 96-well plates and were incubated in medium containing letrozole from 7 nM to 0.7 μM for 48 hours at 37°C. Duplicate plates containing six replicate wells/assay condition were seeded in 0.1 ml medium at densities of 5000, 15,000, and 35,000 cells per well, and were then treated with letrozole alone, with RT (2 Gy) ± letrozole (0.7 μM), and with RT (4 Gy) ± letrozole (0.7 μM), respectively. Twelve days after letrozole ± RT (2 or 4 Gy) treatment, the viability of the cells was analysed using the tetrazolium salt (Sigma) MTT colorimetric assay. Briefly, 50 μl of a 0.5% MTT solution was added to each well, followed by incubation for 4 hours at 37°C to allow MTT metabolization. The crystals formed were dissolved by adding 100 μl/well isopropylic alcohol, 1 N HCl. The absorbance at 540 nm was measured on a Microtiter ® Plate Reader MRX (Dynatech Laboratories, Chantilly, VA, USA). The results were expressed with respect to control values (cells without any treatment). The antiproliferative effect of letrozole ± RT on the MCF-7CA cell line was also evaluated using a cell-count assay. MCF-7CA cells were allowed to grow for 24 hours, at which time letrozole was added at a concentration of 0.7 μM. RT (2 and 4 Gy) was delivered to those cells receiving the combination treatment 6 days after the addition of letrozole to the medium. Cells were counted every 6 days (days 6, 12, and 18) over an 18-day period after removing the cells from the plates with 0.5 g/l trypsin. Flow cytometry Cells were plated in 60-mm Petri dishes at a density of 2 × 10 6 cells/dish. Treatment consisted of letrozole (0.7 μM) alone, RT (2 and 4 Gy) alone, or letrozole (0.7 μM) plus RT (2 and 4 Gy). Cells were collected 15 days after plating, and were processed for cell-cycle analysis. Cells were harvested by trypsinization, were washed with PBS, and then 1 × 10 6 cells/dish of treated cells were fixed in 70% ethanol for 2 min. After removal of ethanol by centrifugation, cells were then stained with a solution containing 40 μg/ml propidium iodide (Sigma) and 0.1 mg/ml RNase A (Roche, Indianapolis, IN, USA). Stained nuclei were analysed for DNA-PI fluorescence using a Becton-Dickinson FACScan flow cytometer (Mountain View, CA, USA). Resulting DNA distributions were analysed using the CellQUEST software (Becton Dickinson) for the proportion of cells in the sub-G 0 phase, the G 1 phase, the S phase, and the G 2 –M phase of the cell cycle. In a second series of experiments, cells were treated with letrozole (0.7 μM) alone and were then cultured for 21 days. Cells were stained at different time points up to 21 days and were analysed for DNA content on a FACScan as already described. Statistical analysis The nonparametric Wilcoxon signed-rank test was used to compare the surviving fraction for each dose between the two groups (RT alone and RT plus letrozole). All experiments were performed three times and the results are expressed as the mean ± standard error. All statistical tests were two-sided with an alpha level of 0.05. Data were analysed using the STATA 7.0 software (Stata Corporation, College Station, TX, USA). Results Letrozole inhibits MCF-7CA aromatase activity The aromatase activity was induced by the androgenic substrate Δ4 androstenedione at a 40 nM concentration. The effects of letrozole on aromatase activity in cultured breast cancer cells are shown in Fig. 1 . In the MCF-7CA transfected cell line, letrozole (100 nM) markedly inhibited aromatase activity. Letrozole inhibits MCF-7CA proliferation The cytotoxic effects of several concentrations of letrozole (7 nM–0.7 μM) on asynchronous, exponentially growing MCF-7CA cells were determined by cell-count assay. Letrozole at a concentration of 7 nM did not inhibit the growth of MCF-7CA cells during 18 days of incubation, whereas 0.7 μM resulted in about 50 ± 10% inhibition after 6, 12, and 18 days of incubation (Fig. 2 ). No effect of letrozole was seen on MCF-7 wild-type cells. Letrozole enhances radiosensitivity Cell survival following RT in aerated medium fits a linear quadratic model, as described in Materials and methods. Figure 3 depicts normalized results from clonogenic experiments. The survival fraction at 2 Gy was 0.66 ± 0.05, and the D 0 value (dose of radiation producing a 37% survival rate) was 3.2 Gy when irradiation was used alone. Letrozole added 6 days before RT led to a greater decrease of the surviving fractions as compared with those obtained when letrozole was added 3 days before RT, when letrozole was added 3 days after RT or when RT was delivered alone. Treatment at 6 days before RT led to a survival fraction at 2 Gy and a D 0 value of 0.46 ± 0.05 and 2.2 Gy, respectively. The survival fraction at 2 Gy was thus nearly 30% (±4%) lower for the combination treatment, with a significant test result ( P = 0.02). When the data were analysed according to the linear quadratic model, the α and β components were, respectively, nearly 0/Gy and 0.098 ± 0.0038/Gy 2 without letrozole, and were nearly 0/Gy and 0.154 ± 0.014/Gy 2 for the combination treatment. These data indicate that treatment with letrozole results in a steeper decline in cell survival due both to a higher initial slope of the dose–response curve and to a major decrease of the quadratic parameter. These results thus show possible additive effects for the combined treatment. Growth of MCF-7CA cells measured by MTT assay was inhibited to a 40 ± 6% greater extent with letrozole plus 2 Gy RT, and to a 76 ± 3% greater extent with letrozole plus 4 Gy RT, compared with radiation alone ( P = 0.02) (Fig. 4 ). Growth of MCF-7CA cells measured by cell-count assay was determined for an 18-day period after initial treatment. It was inhibited to a 76 ± 2% greater extent after 12 days, and to an 85 ± 4% greater extent after 18 days, for letrozole treatment plus 2 or 4 Gy RT compared with RT alone ( P = 0.009) (Fig. 5 ). Letrozole induces G 1 cell-cycle arrest The effect of letrozole treatment on cell-cycle phase distribution in the MCF-7CA cell line was evaluated using flow cytometry (Fig. 6 ). Treatment with 0.7 μM letrozole for 6 days induced accumulation of cells in the G 1 phase (77.5 ± 1.5%), with a significant decrease in the percentage of cells in the S phase (9.0 ± 1%) relative to controls (20.4 ± 0.7%). No cells with subdiploid DNA content were observed, demonstrating that letrozole does not induce apoptosis in these cell lines. After 6 days of treatment, cells were further cultured for 21 days in the presence of letrozole. A G 1 cell-cycle arrest (nearly 70%) was observed at days 12 and 21. One day after RT alone, we observed a cell-cycle arrest in the G 2 phase (32 ± 0.3%) with a decrease in the percentage of cells in the G 0 /G 1 phase and the S phase as compared with the control (59 ± 0.8% versus 63 ± 1.2%, and 9 ± 0.4% versus 20.4 ± 0.7%, respectively). When letrozole was added 6 days before RT, the radiation-induced G 2 /M phase arrest decreased compared with RT alone (21 ± 0.6% versus 32 ± 0.3%, respectively), with a letrozole-induced G 0 /G 1 phase blockade and a very low proportion of cells in the S phase. Discussion TAM is the most widely used endocrine agent for the adjuvant therapy of early breast cancer in postmenopausal women [ 11 ]. TAM use nevertheless remains limited, most notably by its recognized pharmacological properties and side effects. For instance, TAM has been associated with an increased risk of both thromboembolic events and endometrial changes, including endometrial cancer [ 12 , 13 ]. It is possible, therefore, that other endocrine drugs may provide at least equivalent, if not superior, efficacy together with improved tolerability in patients with early disease. In this context, the third-generation aromatase inhibitors are now under investigation, and the initial results may substantially change our current clinical practice patterns [ 3 , 14 , 15 ]. The efficacy of breast-conserving surgery with axillary dissection and breast radiation has been known for a long time [ 16 ]. Overgaard and colleagues more recently demonstrated that postoperative radiotherapy decreases the risk of locoregional recurrence, and that it is associated with improved survival in high-risk patients with breast cancer given either adjuvant chemotherapy (in premenopausal women) [ 4 ] or tamoxifen (in postmenopausal women) [ 5 ]. In the present study, we wanted to evaluate in vitro the effects of the combination of ionizing radiation and letrozole, one of the third-generation aromatase inhibitors. To our knowledge, our results demonstrate for the first time that letrozole could act as a potent radiosensitizer. A significant effect was observed in all three assays used in our experiments (clonogenic, MTT, and cell count). The effects of associating other hormonotherapies, such as TAM, with radiotherapy have been rarely described and remain poorly defined. TAM appears to exert its cytostatic activity at least partly through competitive inhibition of oestrogen binding at the oestrogen receptor, resulting in segregation of cells into the G 0 /G 1 phase of the cell cycle [ 17 ]. Because relatively less radiosensitivity was observed in the early G 1 phase [ 18 ], a hypothetical concern was raised in several studies that the use of TAM during RT might result in the radioprotection of tumour clonogens of both hormonally responsive and unresponsive breast carcinoma cells at dose levels typically used in clinical practice [ 19 ]. In these studies, however, cell cultures were grown in a medium containing phenol red and foetal bovine serum, two sources of exogenous estrogenic compounds [ 20 - 23 ]. This fact complicates the interpretation of the resultant radiation survival curves [ 24 ]. In contrast to these reports, no significant differences were observed in radiosensitivity for oestradiol-stimulated or 4-hydroxytamoxifen-inhibited cultures plated into growth-stimulating conditions immediately after irradiation or following an additional 24 hours in oestrogen-free conditions [ 25 ]. Clearly, under defined hormonal conditions [ 26 ], no protective effect of the active TAM metabolite 4-hydroxytamoxifen was observed [ 25 ]. Our data suggest that letrozole inhibits cell proliferation by invoking a transition delay or by blocking in the G 1 phase of the cell cycle. This delay may require more than one complete passage of some cells through the cell cycle, since cell numbers more than doubled before the plateau growth phase was reached. In addition, two to three generation times were required for maximal accumulation of cells in the G 1 phase. We therefore added letrozole 6 days before RT. Under defined hormonal conditions, we found more than an additive effect [ 27 ] using the combination of letrozole and RT at each dose tested. This result was reinforced by the poor effect of letrozole used alone at high concentration (7 μM) and tested in all combination treatments. Isobologram analyses are planned to confirm the existence of either an additive effect or a supra-additive effect for the association of these two treatment modalities in a future study, as proposed by Steel [ 28 ]. A small percentage of cells was insensitive to letrozole, and remained in the proliferative fraction (9% in the S phase). The mechanism of unresponsiveness of these cells to letrozole will require further study. As shown with anti-oestrogens [ 17 ], these data are consistent with the hypothesis that aromatase inhibitors such as letrozole induce regression of breast cancer in vivo by simply blocking progression of cells through the cell cycle rather than by a direct cytotoxic effect. With cell replication inhibited, tumours might then regress because of ongoing cell shedding or death, or because of interaction with normal host defences. In the RT–letrozole combination treatment, we observed a 50% decrease of MCF-7CA cells arrested in the G 2 phase as compared with RT alone, a proportional redistribution in the G 1 phase, and an interrupted synthesis phase for an 18-day period. This cell-cycle redistribution phenomenon may also explain the decrease in the surviving fraction in the combination treatment presented in the present study (Fig. 3 ). In addition, associating irradiation and letrozole may modify the cytosolic oestrogen and progesterone receptor content in breast cancer cells, which might explain the observed changes in these cells' radiation sensitivity during hormonal treatment, as published with TAM [ 29 ]. These results may have important clinical implications for the treatment of breast cancer. Since the adjuvant use of aromatase inhibitors is still an emerging treatment option [ 3 , 14 , 15 ], the most effective time at which to commence the use of these drugs after surgery is not yet known. Also, if letrozole is simply placing hormone-responsive breast cancer cells into a 'resting' phase of the cell cycle (G 0 –G 1 phase), then treatment of patients after surgery for primary breast cancer with adjuvant letrozole may have to be continued at least long enough for host defences or normal cell attrition to eradicate residual tumour cells. Premature discontinuation of therapy might lead to tumour regrowth. Finally, where letrozole is used concurrently with RT, we recommend starting letrozole at least 3 weeks before the start of RT in an attempt to obtain an optimal effect, consistent with the biological rationale described earlier. Furthermore, and in so far as the preclinical data with breast cancer cells can be extrapolated to the clinical situation, a radiosensitizing effect may be observed in the healthy breast. Nevertheless, Miller and colleagues showed a rising aromatase activity level in adipose tissue obtained from patients with breast cancer and within the quadrant involved with the tumour [ 30 - 32 ]. Conclusions These results are the first preclinical findings demonstrating the radiosensitization effect of letrozole, and thus may provide a basis for the use of combined letrozole and radiation for the adjuvant therapy in breast cancer patients. Abbreviations DCC = dextran-coated charcoal; DMEM = Dulbecco's modified Eagle's medium; FCS = foetal calf serum; MTT = 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; PBS = phosphate-buffered saline; RT = radiotherapy; TAM = tamoxifen. Competing interests The author(s) declare that they have no competing interests. Authors' contributions DA initiated the project and developed all first experiments. CL confirmed all previous results and performed all FACS analyses. SC and PP performed the aromatase activity assay. MO and SG made the statistical analyses. PM constructed survival curves of the clonogenic assay. DA, DBE, GR, and AP managed scientific experiments and contributed to the elaboration of the discussion.

D:\keerthana\PMC001xxxxxx\PMC1064115.xml

1064126

The androgen receptor CAG repeat polymorphism and modification of breast cancer risk in BRCA1 and BRCA2 mutation carriers

Introduction The androgen receptor ( AR ) gene exon 1 CAG repeat polymorphism encodes a string of 9–32 glutamines. Women with germline BRCA1 mutations who carry at least one AR allele with 28 or more repeats have been reported to have an earlier age at onset of breast cancer. Methods A total of 604 living female Australian and British BRCA1 and/or BRCA2 mutation carriers from 376 families were genotyped for the AR CAG repeat polymorphism. The association between AR genotype and disease risk was assessed using Cox regression. AR genotype was analyzed as a dichotomous covariate using cut-points previously reported to be associated with increased risk among BRCA1 mutation carriers, and as a continuous variable considering smaller allele, larger allele and average allele size. Results There was no evidence that the AR CAG repeat polymorphism modified disease risk in the 376 BRCA1 or 219 BRCA2 mutation carriers screened successfully. The rate ratio associated with possession of at least one allele with 28 or more CAG repeats was 0.74 (95% confidence interval 0.42–1.29; P = 0.3) for BRCA1 carriers, and 1.12 (95% confidence interval 0.55–2.25; P = 0.8) for BRCA2 carriers. Conclusion The AR exon 1 CAG repeat polymorphism does not appear to have an effect on breast cancer risk in BRCA1 or BRCA2 mutation carriers.

Introduction A CAG length polymorphism within exon 1 of the androgen receptor ( AR ) gene encodes a string of 9–32 glutamines. Even within this normal range, CAG repeat number is inversely associated with AR-mediated transcriptional activation in vitro [ 1 , 2 ]. Involvement of the AR in breast tumourigenesis is suggested by the existence of inactivating germline mutations in the hormone-binding domain in male breast cancer patients [ 3 , 4 ], and by splice variants that disrupt the transactivation domain in female breast tumours and tumour cell lines [ 5 ]. There is evidence that suggests an association between longer AR CAG repeat length – representative of less active AR – and breast cancer risk at the population level (for review, see Lillie and coworkers [ 6 ]). Using slightly variable definitions of shorter and longer allele size across studies, one study reported a significant twofold increased risk for breast cancer, another three studies reported a slightly increased risk for breast cancer (1.2- to 1.4-fold), and another reported a 1.7-fold increased risk limited to individuals with a first-degree family history of breast cancer [ 6 ]. Several studies have been undertaken to assess the role of the CAG repeat polymorphism as a modifier of breast cancer risk in BRCA1 and BRCA2 mutation carriers. The hypothesis-generating study [ 7 ] examined AR CAG length in 304 female BRCA1 mutation carriers (54% with breast cancer), and assessed breast cancer risk associated with CAG length as a continuous variable, and at a number of different cut-points. That study reported a 1.8-fold relative risk (95% confidence interval [CI] 1.1–3.1; P = 0.03) among the small subgroup of women with at least one AR allele of 28 or more CAG repeats in length, with relative risks of 2.6 (95% CI 1.5–4.7; P < 0.001) and 4.5 (95% CI 1.3–15.2; P = 0.02) for the cut-points ≥ 29 CAG repeats and ≥ 30 CAG repeats, respectively. Subsequent studies have reported conflicting results. A relative risk of 1.1 (95% CI 0.5–2.6) was observed for women with 28 or more CAG repeats from a pooled analysis of 188 BRCA1 and BRCA2 mutation carriers [ 8 ], whereas a study of 227 BRCA1 and BRCA2 mutation carriers reported shorter mean CAG repeat number in women with breast cancer diagnosed before age 42 years as compared with those diagnosed after this age [ 9 ]. It has been hypothesized that AR alleles with decreased transactivation might act directly to result in decreased breast cell proliferation, or possibly indirectly via an endocrine or paracrine mechanism whereby altered levels of circulating or stromal hormones might affect mammary epithelial growth [ 7 ]. Subsequent biochemical studies reported a protein–protein interaction between the AR protein and particular regions of both the BRCA2 and BRCA1 proteins. BRCA2 has been reported to enhance androgen-dependent AR activity [ 10 ]. Although that study also reported that an expressed truncated BRCA2 protein encoded by the BRCA2 L1042X mutation failed to enhance AR transactivation [ 10 ], it did not examine the effect of BRCA2 mutation position in general, or which domains of BRCA2 are required for the BRCA2–AR interaction. BRCA1 has also been reported to enhance androgen-dependent AR transactivation [ 11 , 12 ]. Mammalian two-hybrid assays have shown that BRCA1 amino acids 231–1314 and 1560–1863 are responsible for this direct interaction, with BRCA1 region 758–1064 observed to bind specifically to the AR amino-terminal domain containing the glutamine repeat [ 11 ]. Androgen response assays indicated that, although BRCA1 mutations across the gene all reduce AR activity enhancement as compared with the wild-type, the effect was more marked for mutations up to amino acid 1365 [ 11 ]. This was particularly true for a mutation that caused truncation at amino acid 772 within the BRCA1–AR amino-terminus interaction domain, which had 20% activity relative to the wild-type BRCA1. The effect of the AR CAG repeat length on BRCA1–AR interactions has not yet been investigated. However, molecular studies provide some support for a biochemical interaction between AR CAG repeat length and BRCA1 mutation status, in that the decreased AR transactivation observed in vitro with increasing glutamine length was only observed in the absence of coexpressed BRCA1 [ 12 ]. These data suggest that an association of AR CAG repeat length with increased breast cancer risk may be found only in BRCA1 or BRCA2 mutation carriers (and not individual germline without BRCA1 or BRCA2 mutations), and that AR -dependent modification of cancer risk in BRCA1 and BRCA2 mutation carriers may differ according to which gene is mutated, and the mutation position relative to AR-binding site. To evaluate further the evidence for an association between AR CAG repeat length and breast cancer risk in BRCA1 and BRCA2 mutation carriers, we genotyped the polymorphism in a large series of female mutation carriers. Methods Subjects The distribution of samples according to source, gene and cancer status is shown in Table 1 . A total of 604 living female Australian and British carriers of pathogenic BRCA1 or BRCA2 mutations were identified in 376 families from the following sources: the Epidemiological study of BRCA1 and BRCA2 Mutation Carriers (EMBRACE; ), the Kathleen Cuningham Consortium for Research into Familial Breast Cancer (kConFaB; ), the Australian Jewish Breast Cancer Study (AJBCS) [ 13 ], and the Australian Breast Cancer Family Study (ABCFS) [ 14 , 15 ]. EMBRACE recruits participants from among women and men referred for genetic testing at clinical genetics centres in the UK and Eire. kConFaB recruits participants from multiple-case breast and ovarian cancer families referred for genetic testing at family cancer clinics in Australia and New Zealand. AJBCS recruits Ashkenazi Jewish women reporting a personal or family history of breast or ovarian cancer in a first- or second-degree relative, and living in Melbourne or Sydney, Australia. Finally, ABCFS is a population-based case–control-family study that includes women with a first primary breast cancer recruited through the Victorian and New South Wales cancer registries, and their affected and unaffected relatives. Apart from index cases recruited through cancer registries for the ABCFS, the cancer status of participants was based on self-report. For samples recruited through EMBRACE, a pathogenic mutation was defined as an established disease-causing mutation under the classification scheme used by Breast Cancer Information Core . For samples recruited through kConFaB, AJBCS and ABCFS, mutations were classified as pathogenic according to the criteria established by kConfab . Specifically, the criteria specify the following as being pathogenic: all truncating mutations, unless there is clear evidence that the mutation is a single nucleotide polymorphism (e.g. terminal BRCA2 variant); and any variant that is well characterized in family studies of multiple generations, and not found in control individuals, that results in a nonconservative amino acid substitution, and occurs in a residue conserved across species and in a functional domain. All mutations included in the study that were shared across sites were classified as pathogenic according to both routes of definition. Within Australia, ethical approvals were obtained from the ethics committees of the Peter MacCallum Cancer Institute, The Prince of Wales Hospital, The University of Melbourne, The Cancer Council New South Wales, The Cancer Council Victoria and the Queensland Institute of Medical Research. Ethical approval for the EMBRACE study was obtained from the Eastern Multicentre Research Ethics Committee and the relevant local ethics committees. Written informed consent was obtained from each participant. Molecular methods The AR exon 1 CAG repeat length was measured by fluorescent polymerase chain reaction PAGE methodology, using the ABI Prism 373 Genescan (Applied Biosystems, Foster City, CA, USA) and Genotyper systems (Applied Biosystems). Details of this method were previously reported [ 16 ]. Genotyping was successful for 375 out of 382 (98%) BRCA1 carriers, 218 out of 221 (99%) BRCA2 carriers, and the single carrier of both a BRCA1 and a BRCA2 mutation. Statistical methods Individuals with a first diagnosis of primary invasive breast cancer were considered to be affected, whereas individuals with no reported breast or ovarian cancer were censored at age at interview. Individuals with a first diagnosis of primary ovarian cancer were censored as unaffected at age at onset of ovarian cancer, and selected analyses were also performed in which individuals with a first primary diagnosis of ovarian cancer were excluded. All individuals were censored at age of prophylactic mastectomy. Individuals reporting prophylactic surgery included a single BRCA2 carrier who was subsequently diagnosed with multiple breast cancers 4 and 5 years after surgery, and an additional 12 unaffected individuals (7 BRCA1 and 5 BRCA2 carriers) with surgery 1–11 years before interview (average 3 years). Prophylactic oophorectomy of affected and unaffected individuals was controlled for by adjustment as a time-dependent covariate, as described below. Linear regression was used to assess the association of AR CAG repeat length (smaller allele size, larger allele size and average allele size) with potential confounders within the subset of 364 BRCA1 carriers and 209 BRCA2 carriers for whom information was available. The potential confounders included year of birth (categorized into subgroups 1910–1949, 1950–1959 and 1960–1979), age at menarche (categorized as ≤ 11, 11.5–12, 12.5–13, 13.5–14 or ≥ 14.5 years), oral contraceptive pill use (ever/never) and parity (categorized as 0 or ≥ 1 live births before censored age). Questionnaire information was available from participants on age at first and last live birth, but not age at each live birth. Hence, it was not possible to assess association with parity as an absolute number of live births before censored age, but rather only as a never/ever variable. Associations were assessed separately for affected and unaffected women. The primary analyses of association between AR genotype and disease risk were performed using Cox regression with time to breast cancer onset as the end-point. AR CAG repeat length was defined as follows: a binary variable, defined by cut-points investigated in the hypothesis-generating study conducted by Rebbeck and coworkers [ 7 ] (namely one or more allele of ≥ 28 CAG repeats, ≥ 29 CAG repeats, or ≥ 30 CAG repeats); or a continuous variable, using the length of the smaller of the two alleles ( AR small CAG), the larger of the two alleles ( AR large CAG), and the average length of a participant's two alleles ( AR average CAG). Rate ratios (RRs) and 95% CIs were estimated with adjustment for source group (as indicated in Table 1 ) and ethnicity (non-Jewish Caucasian, Jewish, other). Analyses were complicated by the fact that more than one mutation carrier could come from the same family and could not therefore be considered independent. Standard Cox regression provides unbiased RR estimates but their standard errors and CIs are incorrect. This was rectified by computing the confidence limits for the RRs using Huber's sandwich estimator of the covariance matrix [ 17 ]. This allows for variation between carriers from the same family without modelling their dependence explicitly. Further analyses adjusted for oophorectomy and parity as time-dependent covariates, and for age at menarche, oral contraceptive pill use and year of birth. Oophorectomy before censored age at interview or diagnosis of breast cancer was reported by 39 BRCA1 carriers (10 with primary breast cancer) and 21 BRCA2 carriers (9 with primary breast cancer) with genotype information available. RRs were estimated separately for BRCA1 and BRCA2 carriers, including in both analyses the single individual with a mutation in both genes. Models adjusting for only group and ethnicity included all individuals with genotype information, namely 376 BRCA1 carriers (with 200 events) and 219 BRCA2 carriers (with 122 events). The sample size for BRCA1 carriers with a putative risk allele was 28 (14 events) for the ≥ 28 CAG cut-point, 26 (13 events) for the ≥ 29 CAG cut-point, and 11 (4 events) for the ≥ 30 CAG cut-point. Similarly, for BRCA2 carriers it was 17 (10 events) for the ≥ 28 CAG cut-point, 14 (7 events) for the ≥ 29 CAG cutpoint, and 11 (5 events) for the ≥ 30 CAG cut-point. Full models adjusting for year of birth and additional hormonal variables included the 364 BRCA1 and 219 BRCA2 carriers with full information on potential confounders, comprising 193 and 116 events, respectively. In addition, analyses were carried out separately for subgroups of BRCA1 and BRCA2 carriers defined by mutation position in relation to proposed AR-binding domains, and/or in vitro data regarding mutation effect on AR transactivation [ 10 , 11 ]. For BRCA1 , subgroups were defined by the mutation position either relative to amino acid 1365 (< or ≥ nucleotide 4213), because mutations 5' of amino acid 1365 have been shown to have a markedly decreased effect on AR transactivation [ 11 ]. This created subgroups including 314 and 62 individuals. In addition, BRCA1 subgroups were defined by mutation relative to amino acid 1065 (< or ≥ nucleotide 3311), because this defines the 3' end of the BRCA1 fragment shown in vitro to bind the AR amino-terminal domain containing the CAG-encoded polyglutamine tract [ 11 ], creating subgroups of 210 and 166 individuals. BRCA2 subgroups were defined by mutation relative to amino acid 1042 (< or ≥ nucleotide 3352), because it has been shown that the BRCA2 L1042X mutation does not enhance AR transactivation [ 10 ]. Subgroup sample sizes were 42 and 177. For both BRCA1 and BRCA2 subgroup analyses, the 3' and 5' subgroups were termed domain 1 and domain 2, respectively. Protein truncating and splice mutations were stratified into domain 1 or domain 2 according to their nucleotide/amino acid position, whereas all missense mutations were included in domain 2 because these nontruncating mutations may act in a dominant-negative manner. Although the primary analysis provides a valid test of the association between a genotype and disease risk, it may not provide a consistent estimate of the RR because the disease status of the individuals may have affected the likelihood of ascertainment (for the non-population-based studies). Oversampling of affected individuals is likely, as is presentation of affected carriers at later mean age than unaffected carriers. To correct for this potential bias, we also conducted secondary analyses using the weighted Cox regression approach as described by Antoniou and coworkers (unpublished data), in which individuals are weighted such that the observed breast cancer incidence rates in the study sample are consistent with established breast cancer risk estimates for BRCA1 and BRCA2 mutation carriers. Antoniou and coworkers (unpublished data) have shown that this approach gives estimates that are close to unbiased, but with some loss of power as compared with the standard unweighted approach. Weights were computed separately for BRCA1 and BRCA2 mutation carriers using the breast cancer incidence rate estimates reported in the meta-analysis conducted by Antoniou et al . [ 18 ]. A global set of weights was computed because the number of mutation carriers by study was too small to compute reliable study-specific weights. Moreover, Antoniou et al . [ 18 ] found no significant differences in the BRCA1 and BRCA2 cancer risks by country or study centre. As for unweighted analyses, confidence limits for the risk ratio were calculated using a robust variance approach to allow for the dependence among individuals. We evaluated the power of detecting the effects reported by Rebbeck and coworkers [ 7 ] in our samples of BRCA1 and BRCA2 mutation carriers using simulations. For this purpose, we assumed the age distribution of the affected and unaffected carriers in our sample (Table 1 ) and simulated among them risk factors with risk ratios 1.8 and 2.6 and the frequencies for the ≥ 28 and ≥ 29 CAG cut-points observed in our sample. The data were then analyzed using unweighted Cox regression. We conducted 1000 simulations per model. More details about the simulations are available from the authors of the present report. The power of detecting risk ratios of 1.8 and 2.6 was estimated to be 51% and 92%, respectively, for the sample of BRCA1 mutation carriers and 28% and 78% for the sample of BRCA2 mutation carriers. R version 1.9.0 (R Foundation for Statistical Computing, Vienna, Austria) was used to perform the unweighted Cox regression, and STATA version 7 (Stata Corporation, College Station, TX, USA) was used for the weighted analyses. Results The AR CAG length ranged from 8 to 36 repeats. There was little power to assess potential confounding or risk associated with the cut-point ≥ 30 CAG repeats, because only 11 BRCA1 and 11 BRCA2 carriers had at least one allele of this size. There was no evidence for an association between AR CAG repeat length and year of birth, age at menarche, or parity. There was marginal evidence for a preponderance of smaller alleles among affected BRCA2 carriers who reported using the oral contraceptive pill ( P = 0.03), but this association was not seen in unaffected individuals or in BRCA1 affected or unaffected carriers ( P > 0.2). The estimated rate ratios associated with AR CAG repeat length are given in Tables 2 and 3 . Risk estimates using the weighted Cox regression approach were very similar to the unweighted estimates, and for simplicity the unweighted estimates are shown. No associations were observed for alleles ≥ 28 CAG repeats or ≥ 29 CAG repeats – cut-points previously reported to be associated with risk. Estimated RRs were close to and not significantly different from 1 (all P > 0.3) and were in most instances less than 1. The number of individuals with ≥ 30 CAG repeats was too small to provide reliable risk estimates, but point estimates (0.49 [ P = 0.1] for BRCA1 , 0.69 [ P = 0.5] for BRCA2 ) provided no evidence for increased risk associated with these large alleles. When AR CAG repeat length was considered as a continuous variable, there was no association either with average repeat length or with the length of the shorter or longer allele ( P ≥ 0.2). There was little difference between the estimates adjusted only for source group and ethnicity, and those adjusted also for year of birth, and hormonal variables oophorectomy, parity, age at menarche and contraceptive pill use. Risk estimates were not markedly different when women with a first primary diagnosis of ovarian cancer were excluded, with a RR (95% CI) for the ≥ 28 CAG cut-point of 0.85 (0.49–1.47) for BRCA1 mutation carriers ( P = 0.6), and 1.12 (0.0.55–2.27) for BRCA2 mutation carriers ( P = 0.8). Results using the weighted Cox regression approach indicated that RR estimates were not materially affected by possible ascertainment biases. For example, for the ≥ 28 CAG cut-point, the RR (95% CI) adjusted for group and ethnicity was 0.74 (0.33–1.66) for BRCA1 carriers ( P = 0.5) and 0.94 (0.32–2.77) for BRCA2 carriers ( P = 0.9). For average CAG length, the RR (95% CI) adjusted for group and ethnicity was 1.01 (0.94–1.10) for BRCA1 carriers ( P = 0.8) and 0.96 (0.82–1.11) for BRCA2 carriers ( P = 0.6). These results are consistent with the findings of Antoniou and coworkers (unpublished data), in which both weighted and unweighted Cox regression analyses give similar estimates when the true RR is 1.0. There was also no compelling evidence for an effect of mutation position on risk associated with AR CAG repeat length. BRCA1 mutations were firstly divided by position relative to amino acid 1365 (nucleotide 4213), because mutations 5' of this have been reported to exhibit markedly decreased AR transactivation ability [ 11 ]. The RR (95% CI) for the ≥ 28 CAG cut-point analyses were 0.93 (0.53–1.64) for domain 1 ( P = 0.8) and 0.35 (0.12–1.02) for domain 2 ( P = 0.06), with marginal evidence for an interaction ( P = 0.1). BRCA1 mutations were also divided by position relative to amino acid 1065 (nucleotide 3311), because the 3' end of the BRCA1 fragment has been shown in vitro to bind the AR amino-terminal domain containing the CAG-encoded polyglutamine tract [ 11 ]. The RR (95% CI) for the ≥ 28 CAG cut-point analyses were 0.98 (0.54–1.78) for domain 1 ( P = 0.9) and 0.45 (0.16–1.23) for domain 2 ( P = 0.1), with no evidence for an interaction ( P = 0.3) There was no convincing rationale for stratification of the BRCA2 mutation carriers by mutation position, in that there is no published information available with regard to the domains of BRCA2 required for BRCA2–AR interaction or to the effect of BRCA2 mutation position on the BRCA2–AR interaction. However, because a single report has shown that the BRCA2 L1042X mutation does not enhance AR transactivation [ 10 ], mutations 5' to this mutation site might be expected to have similar drastic consequences, and to convey increased risk among carriers with large AR CAG alleles. Stratification by mutation position relative to amino acid 1042 (nucleotide 4213) divided the BRCA2 carrier sample into two subgroups of 42 and 177 individuals. None of the 42 individuals with mutations in domain 1 had alleles with repeat length ≥ 28 CAG, precluding estimates of risk for this subgroup. However, there was no evidence for increased risk among BRCA2 carriers with mutations in the domain 5' to amino acid L1024, with a RR (95% CI) for the ≥ 28 CAG cut-point of 1.11 (0.54–2.30) for the 5' domain ( P = 0.8). Discussion Our study found no evidence to support the previously reported association of AR allele length with increased breast cancer risk in BRCA1 carriers [ 7 ]. The hypothesis-generating study estimated a 1.8-fold risk (95% CI 1.1–3.1) in BRCA1 carriers with at least one AR allele length ≥ 28 CAG repeats, and increasing risks of 2.7-fold and 4.5-fold for the ≥ 29 CAG repeat and ≥ 30 CAG repeat cut-points, respectively [ 7 ]. The RR estimated by our study for ≥ 28 repeats was 0.74, and the upper 95% confidence limit (1.28) excludes the effect size reported by Rebbeck and coworkers [ 7 ]. Moreover, there was no evidence for increased RRs at the ≥ 29 or ≥ 30 cut-points. Given that we had approximately 80% or more power to detect risk estimates previously reported for the ≥ 29 cut-point [ 7 ], these results suggest that if there is any increased risk for large number of CAG repeats among BRCA1 mutation carriers, then it is of much lower magnitude than was first reported. Rebbeck and coworkers [ 7 ] found no effect of AR CAG repeat length when considered as a continuous variable, and our study likewise found no evidence for an association with shorter, larger, or average CAG repeat length. Other published studies have found no evidence to support an association between AR CAG repeat length and breast cancer risk in BRCA2 mutation carriers [ 8 , 9 ]. Our study also found no evidence in support of an association. Although our sample of 220 BRCA2 carriers was not sufficiently large to provide precise estimates of risk, the upper 95% confidence limits imply that a 2.5-fold risk associated with ≥ 28 or ≥ 29 repeat lengths is unlikely. Conclusion Our analyses provide no support for an association between AR CAG repeat length and breast cancer risk in either BRCA1 or BRCA2 mutation carriers. An increased risk associated with ≥ 28 or ≥ 29 repeat lengths is not compatible with our data. Weak associations cannot be excluded, but analyses involving much larger numbers of carriers would be required to evaluate this possibility. Abbreviations AR = androgen receptor; CI = confidence interval; RR = rate ratio. Competing interests The author(s) declare that they have no competing interests. Authors' contributions ABS was responsible for supervising genotyping design and laboratory work, for data cleaning and analysis, and leading the manuscript preparation. ACA, DLD, NP, DMP and BN assisted in the analytical design and analysis of this study. XC and LK were responsible for assay design and optimization, laboratory work and genotype data cleaning. SP and MRC were responsible for the coordination of the EMBRACE study. PLS assisted with the data management and analysis of this study. GSD, CA and MCS assisted with management and provision of data and DNA from the ABCFS and AJBCS. JLH and GGG initiated the ABCFS, and have been instrumental in the ongoing execution of this study. GC-T initiated and obtained funding for this study, initiated collaborative efforts and was involved in the supervision of laboratory work. DFE initiated the EMBRACE study, and assisted in the development of the analytical design of this study and the manuscript preparation. All authors reviewed the manuscript and offered comments.

D:\keerthana\PMC001xxxxxx\PMC1064126.xml

1064127

Haplotype analysis of common variants in the BRCA1 gene and risk of sporadic breast cancer

Introduction Truncation mutations in the BRCA1 gene cause a substantial increase in risk of breast cancer. However, these mutations are rare in the general population and account for little of the overall incidence of sporadic breast cancer. Method We used whole-gene resequencing data to select haplotype tagging single nucleotide polymorphisms, and examined the association between common haplotypes of BRCA1 and breast cancer in a nested case-control study in the Nurses' Health Study (1323 cases and 1910 controls). Results One haplotype was associated with a slight increase in risk (odds ratio 1.18, 95% confidence interval 1.02–1.37). A significant interaction ( P = 0.05) was seen between this haplotype, positive family history of breast cancer, and breast cancer risk. Although not statistically significant, similar interactions were observed with age at diagnosis and with menopausal status at diagnosis; risk tended to be higher among younger, pre-menopausal women. Conclusions We have described a haplotype in the BRCA1 gene that was associated with an approximately 20% increase in risk of sporadic breast cancer in the general population. However, the functional variant(s) responsible for the association are unclear.

Introduction Truncation mutations in the BRCA1 gene are high-penetrance, low-prevalence factors in risk of breast cancer. BRCA1 is hypothesized to be a locus under recombinational inhibition, and very few haplotypes have been described. In fact, only one haplotype block and two major haplotypes have been shown to exist in Caucasians. Because of the size of the gene (more than 80 kilobases), polymorphism discovery screenings have focused on exons. Although many non-synonymous polymorphisms are known in the gene, the degree of linkage disequilibrium (LD) across the entire region limits genetic variability. This limited variability has led to inconclusive results in the risk of sporadic breast cancer associated with variants in the BRCA1 gene [ 1 , 2 ]. The high degree of LD at BRCA1 led Huttley and colleagues [ 3 ] to investigate the possibility of recent selective pressure being exerted on this gene. They found that whereas the ratio of non-synonymous to synonymous nucleotide substitutions is the same between the chimpanzee and humans, this ratio is different from other primates, and greater than 1. They also note that these differences occur in the region of BRCA1 that interacts with Rad51, suggesting that it is the role of BRCA1 in maintaining genome integrity that has driven this selection. Paradoxically, BRCA1 has a large number of Alu repeat sequences. These are repetitive elements that are thought to be involved in recombination and evolution of the genome [ 4 , 5 ]. Given that knocking out brca1 in mice is embryonic lethal [ 6 ], it can be hypothesized that the apparent suppression of recombination at BRCA1 in the human is due to the non-viability of recombinants. Recently, resequencing information over the entire region of the gene, including most introns, has become publicly available [ 7 ]. We used these data to select haplotype tagging single nucleotide polymorphisms (htSNPs), to test the association of these haplotypes with breast cancer risk in a nested case-control study within the Nurses' Health Study. Method Resequencing information from the Environmental Genome Project of the National Institute of Environmental Health Sciences (NIEHS) at the University of Washington was used to generate haplotypes for the selection of htSNPs [ 7 ]. There were 90 individuals with 301 SNPs in the whole data set. SNPs were excluded from analysis if they were out of Hardy–Weinberg equilibrium ( P < 0.05), had a minor allele frequency of less than 5%, or had more than 25% missing data. Haplotypes were reconstructed with PHASE [ 8 ], and htSNPs were determined with BEST [ 9 ]. Four htSNPs were selected, at positions 33,420 (rs799917, P871L), 38,085 (rs8176166), 44,059 (rs3737559), and 64,646 (rs8176267, base pairs reported as on GenBank sequence AY273801). These htSNPs were genotyped in cases and controls using the TaqMan system (Applied Biosystems, Foster City, CA). Primer and probe sequences are available from the authors on request. Our study consisted of 1323 breast cancer cases and 1910 controls, nested within the prospective Nurses' Health Study. The Nurses' Health study was initiated in 1976, when 121,700 United States registered nurses between the ages of 30 and 55 years returned an initial questionnaire reporting medical histories and baseline health-related exposures. Updated information has been obtained by questionnaire every 2 years. Incident breast cancers were identified by self-report and confirmed by medical record review. Between 1989 and 1990, blood samples were collected from 32,826 women. Follow-up has been about 98% in all subsequent questionnaire cycles for this subcohort. Eligible cases in this study consisted of women with incident breast cancer from the subcohort who gave a blood specimen. Cases with a diagnosis any time after blood collection up to 1 June 2000 with no previously diagnosed cancer except for nonmelanoma skin cancer were included. Controls were randomly selected participants who gave a blood sample and were free of diagnosed cancer (except nonmelanoma skin cancer), and were matched to cases on the basis of age, menopausal status, recent post-menopausal hormone use, and time, day, and month of blood collection. Table 1 shows basic characteristics of cases and controls. Haplotype frequencies were estimated with the EM algorithm, as implemented in SAS PROC Haplotype (SAS Institute, Cary, NC). Omnibus tests of haplotype association and haplotype-specific odds ratios (ORs) were calculated by haplotype replacement regression [ 10 ], assuming an additive model using the probability of carrying each pair of haplotypes provided by PROC Haplotype. The most common haplotype was used as the reference, and rare haplotypes (combined frequency less than 0.5%) were dropped from analysis. Unconditional logistic regression analyses were used to determine relative risk, controlling for age, family history of breast cancer, history of benign breast disease, post-menopausal hormone use, parity, age at first birth, and age of menopause. We assumed an additive model, where haplotype-specific parameters represent the per-haplotype increase in log odds of disease. Departures from a multiplicative gene × environment interaction model were tested by means of likelihood ratio tests. A fifth SNP (Q356R, rs1799950), previously described as being associated with a reduced risk of breast cancer [ 1 ], was also examined with a TaqMan assay. This SNP was not present at more than 5% in the resequencing data and therefore was not included in our haplotype analysis. All P values reported are two sided. Sequence alignments were performed with base pairs 64,601–64,700 on GenBank sequence AY273801. Blast alignments were performed over the web at using the blastn program against the alu_repeats database. No filtering was used, and expected values were set at 10 -20 to limit the number of hits. All other default values were used. AluSp repeats on contig NT_010755 were selected from the AluGene database . These sequences were aligned with ClustalW at , using all default values. Results and Discussion The polymorphism at codon 356 in the BRCA1 gene had previously been described as being inversely associated with breast cancer risk (Gln356→ Arg, OR 0.88, 95% confidence interval [CI] 0.63–1.23; Arg356→ Arg, OR 0.00, 95% CI 0.00–0.56) [ 1 ]. We were unable to reproduce these results in our data set. Dunning and colleagues did not observe any homozygotes of the Arg allele at this codon among cases ( n = 765). In contrast, we observed homozygotes among both cases and controls, and did not detect any association (Table 2 ). We had about 80% power to detect a relative risk of 0.73 assuming a log additive model. This polymorphism was not detected above the 5% threshold for inclusion as a htSNP in the NIEHS database, and was not included in our haplotype analyses. We did explore its inclusion in the haplotype analyses, and it did not materially alter the risk estimates for other haplotypes. Five haplotypes of more than 5% frequency were described from the 39 polymorphisms meeting the selection criteria. BRCA1 exists as one haplotype block, with significant LD along the entire gene. Only four SNPs were needed to tag these haplotypes. To test the hypothesis that a difference in haplotype frequencies is seen between cases and controls, a global test was performed ( P = 0.08). This test is not formally significant; this should be kept in mind while interpreting results based on haplotype analysis. Table 3 shows the results of the regression trend test of haplotypes. Haplotype 2 (C A G G) was associated with a small, though significant, increase in risk (OR 1.18, 95% CI 1.02–1.37; Table 3 ). When considering the diplotype of haplotype 2, a significant increase in risk was observed among the homozygous carriers (OR 1.62, 95% CI 1.05–2.48; Table 4 ). A nearly significant interaction was seen between haplotype 2 and family history of breast cancer ( P = 0.05). A large increase in risk (OR 10.83, 95% CI 2.39–49.2) was observed in women homozygous for haplotype 2 and having a positive family history of breast cancer (Table 5 ). Similar, although not statistically significant, interactions were seen for age of diagnosis (less than 50 or more than 50, interaction P = 0.36) and menopausal status at diagnosis (pre-menopausal or post-menopausal, interaction P = 0.19, data not shown). Additional studies focusing on breast cancer incidence in younger, pre-menopausal women would be of interest, to improve the definition of risk associated with this haplotype. Little is known about the actual effects on the expression or function of these polymorphisms in BRCA1 . Because of the low complexity of the gene at the haplotype level, we can describe haplotype 2 by using just one SNP, at base pair 64,646. This is in the intron between exons 19 and 20, in the middle of an Alu repeat sequence. This is a rather long intron, spanning 6 kilobases (63,044–69,242). The Alu repeat surrounding base pair 64,646 is a member of the AluSp family. Aligning this sequence against the Alu database at NCBI shows that the consensus nucleotide for this family at this position is G, which is the risk allele. Alignment with other AluSp repeats on the same contig as BRCA1 shows that those most similar to this region also have a G at this position. This implies that the G allele might recombine more readily than the A allele with other Alu repeats in this region. It could therefore be hypothesized that this SNP is influential in Alu-mediated non-homologous recombination and other rearrangements of the BRCA1 gene. These sorts of aberrations are responsible for roughly 10% of BRCA1 disease-causing mutations, and could be involved in somatic alteration of the structure of the BRCA1 gene [ 11 ]. However, it is important to note that this SNP was selected not because of any prior knowledge of potential function but because it tags a common haplotype. This risk haplotype is a subset of the wild-type haplotype, and no coding or potential splice-site SNPs in the NIEHS Environmental Genome Project database are in LD with the SNP defining this haplotype. It should be noted that the sequencing data reported by the NIEHS Environmental Genome Project are limited to about 1 kilobase of sequence 5' to the start of transcription, and although the entire 3' untranslated region has been sequenced, only about 800 base pairs beyond the poly(A) site are included. Additionally, 13,403 of 82,899 base pairs (16%) of the genomic region of BRCA1 was not sequenced. However, all the unsequenced regions are intronic. This leaves the possibility that a potentially functional SNP in the unsequenced regions of the BRCA1 gene resides on haplotype 2. Osorio and colleagues [ 12 ] examined the occurrence of mutations in BRCA1 among the index cases of familial breast and ovarian cancers. They found that mutations occur more readily on the rarer of the two common haplotypes of BRCA1 (their haplotype II). These haplotypes are the third, fourth and fifth listed in Table 3 , not the haplotype for which we observed an increase in risk, so the relevance of their observation for our findings is unclear. Although we cannot rule out the possibility that these results are spurious or due to population stratification, the Nurses' Health Study consists almost entirely of Caucasian women; population stratification should therefore be minimal. Two additional hypotheses that need further examination are that there are functional polymorphisms in the BRCA1 gene that are not in the coding sequence, and/or that variants in BRCA1 are in LD with functional variants in neighboring genes. The LD block around BRCA1 is quite extensive [ 13 ], and includes a BRCA1 pseudogene, as well as the genes NBR1 and NBR2 . Potentially functional variation in these genes also needs to be described. Conclusions We have described a haplotype associated with the BRCA1 gene that is associated with an approximately 20% increase in risk of sporadic breast cancer in the general population. However, the functional variant(s) responsible for the association are unclear. Abbreviations CI = confidence interval; htSNPs = haplotype tagging single nucleotide polymorphisms; NIEHS = National Institute of Environmental Health Sciences; LD = linkage disequilibrium; OR = odds ratio. Competing interests The author(s) declare that they have no competing interests. Authors' contributions DGC carried out analyses and wrote the manuscript, SEH provided funding for the Nurses' Health Study blood cohort and participated in manuscript editing, PK provided statistical support and participated in manuscript editing, DJH provided funding for genotyping and participated in manuscript editing. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064127.xml

1064128

Breast implants following mastectomy in women with early-stage breast cancer: prevalence and impact on survival

Background Few studies have examined the effect of breast implants after mastectomy on long-term survival in breast cancer patients, despite growing public health concern over potential long-term adverse health effects. Methods We analyzed data from the Surveillance, Epidemiology and End Results Breast Implant Surveillance Study conducted in San Francisco–Oakland, in Seattle–Puget Sound, and in Iowa. This population-based, retrospective cohort included women younger than 65 years when diagnosed with early or unstaged first primary breast cancer between 1983 and 1989, treated with mastectomy. The women were followed for a median of 12.4 years ( n = 4968). Breast implant usage was validated by medical record review. Cox proportional hazards models were used to estimate hazard rate ratios for survival time until death due to breast cancer or other causes for women with and without breast implants, adjusted for relevant patient and tumor characteristics. Results Twenty percent of cases received postmastectomy breast implants, with silicone gel-filled implants comprising the most common type. Patients with implants were younger and more likely to have in situ disease than patients not receiving implants. Risks of breast cancer mortality (hazard ratio, 0.54; 95% confidence interval, 0.43–0.67) and nonbreast cancer mortality (hazard ratio, 0.59; 95% confidence interval, 0.41–0.85) were lower in patients with implants than in those patients without implants, following adjustment for age and year of diagnosis, race/ethnicity, stage, tumor grade, histology, and radiation therapy. Implant type did not appear to influence long-term survival. Conclusions In a large, population-representative sample, breast implants following mastectomy do not appear to confer any survival disadvantage following early-stage breast cancer in women younger than 65 years old.

Introduction Over the past 30 years, an estimated 1.5–2 million women have received breast implants in the United States [ 1 ]. Starting in the 1980s, widespread public health concern arose regarding their potential adverse health effects [ 2 ]. Numerous epidemiologic investigations have focused on systemic complications, particularly cancer and connective tissue disease, but have found no significantly increased short-term risk for these diseases [ 3 , 4 ]. Approximately 20% of breast implants are used for reconstruction in breast cancer patients following mastectomy [ 1 ]. In this population, however, the use of breast implants, while increasing, has not been well documented [ 5 ]. Furthermore, little research has addressed long-term survival, although the few studies conducted suggest that use of implants for breast reconstruction does not impact patient survival [ 6 - 9 ]. However, these studies were limited by comprising clinic-based samples not necessarily representative of all patients, by small sample sizes, by lack of information on type of implant, and by short durations of follow-up. In 1993, the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute implemented the Breast Implant Surveillance Study to document and validate postmastectomy breast implant usage in a population-based series of young, early-stage breast cancer cases. Since the SEER program also monitors patient vital status for life, survival up to 17 years exists in the study cohort. The purpose of the current analysis was to describe the use of breast implants and to examine the impact of breast implants on survival after breast cancer in this cohort. Methods Breast Implant Surveillance Study The Breast Implant Surveillance Study was conducted during the period 1993–1994 in the United States. Patients were identified through the population-based SEER cancer registries in San Francisco–Oakland, CA, in Seattle–Puget Sound, WA, and in the state of Iowa. Eligible patients included the 5862 females diagnosed younger than age 65 with early-stage or unstaged first primary breast cancer in 1983, 1985, 1987, or 1989 and treated with mastectomy during their first course of therapy. Participation involved completing a standardized questionnaire inquiring about implant status (right breast only, left breast only, both breasts, no implant, or unknown) and implant type (silicone gel, saline, double lumen [consisting of silicone gel and saline], other, and unknown type), and providing signed consent for release of medical records for validation of implant usage and implant type. Patients in the Seattle–Puget Sound region and the state of Iowa were mailed a self-administered questionnaire, and nonrespondents were contacted for a telephone interview. Women from the San Francisco–Oakland region were administered questionnaires by telephone. Next-of-kin were asked about the deceased patient's implant status and for consent for medical record review. For women reporting breast implants, a medical record review of breast implant usage (including the date and type of implant received, and removal and replacement status) was conducted by trained abstractors for women who reported having a breast implant. Patient and tumor characteristics at diagnosis, including age, race/ethnicity, year, stage, histology, grade, radiation treatment, vital status, and cause of death, were obtained from the SEER database. Although socioeconomic status (SES) is not routinely collected by SEER, we were able to assign census block group level measures of SES to a subset of subjects (1989 San Francisco Bay Area patients). Using data from the 1990 US Census, we examined the impact of living in census block groups characterized by low education, by poverty, and by occupation (median income, < 20% below poverty versus ≥ 20% below poverty; education, no high school diploma versus high school graduate; and occupation, blue collar versus nonblue collar [ 10 ]). Patients who reported a breast implant were classified as having a particular type of implant if they had a unilateral implant, or if they had bilateral implants of the same type. Nineteen women with bilateral implants and discordant information about implant type were excluded from analyses stratified by the type of implant. An additional 133 women reporting implants but lacking implant information were excluded from all analyses requiring detailed implant information. The vital status (obtained annually through patient contact, death records, motor vehicle departments, voters' registration records, and Social Security files) was determined from the December 1999 SEER Public Use Tape. The outcome variables were death due to breast cancer and death due to nonbreast cancer causes, as routinely ascertained by SEER and as defined by the International Classification of Disease, Ninth Revision, codes 174.0–174.9 for deaths occurring between 1983 through 1998, and by the International Classification of Disease, 10th Revision, code C509 for deaths occurring in 1999. Survival time was calculated from the date of diagnosis to the earliest date: death, last known to be alive, or 31 December 1999 (study cutoff date). Statistical analysis For descriptive analyses, women with breast implants were compared with those without implants on characteristics at diagnosis (SEER region of residence, age, year, race/ethnicity, marital status, stage, grade, and histology), using chi-square tests and Fisher's exact test to assess differences. Two-sided P < 0.05 was considered statistically significant. Of the 5862 patients eligible for the study, 4968 (84.7% of those eligible) patients participated in the study. After restricting the sample to women with primary invasive breast cancer and known survival time, a final sample size of 4385 patients were used for all survival analyses. Survival estimates were computed using the Kaplan–Meier method, and differences in survival were compared using the two-sided log-rank test. To adjust for patient and tumor characteristics, the risks of nonbreast cancer mortality and death due to breast cancer were modeled using Cox proportional hazards regression after censoring deaths from other or unknown causes. The assumption of proportionality was tested and met for all covariates used in the Cox analysis. All analyses were performed using SAS Version 8.02 (SAS Institute, Inc., Cary, NC, USA). Results Study population Eighty-five percent of study-eligible women completed the interview. Responders were slightly older compared with nonresponders. They also were more often non-Hispanic white, were diagnosed with in situ cancers, were less likely to have more than one primary tumor and were more likely to have lived until the end of the study period (Table 1 ). Among the responders, 20% received a breast implant (Table 1 ). Their mean age at diagnosis was 47 years, and they were younger, on average, than women without implants. Somewhat higher proportions of women with implants were of non-Hispanic white race/ethnicity, resided in the San Francisco–Oakland region, were diagnosed in the late 1980s, and had tumors of lobular histology than women without implants. The percentage of women with in situ breast cancer was twice as high in women with breast implants as in nonimplanted women. In addition, women with implants were less likely to receive radiation therapy or to be diagnosed with more than one primary tumor. Implant characteristics and usage Implant information was obtained for 866 women; these women were slightly older and more likely to be living than women with unknown implant information. Among the 1143 breast implants received (Table 2 ), the most common types were silicone gel and double lumen. The majority of women (67%) received a unilateral implant a median of 9.6 months after breast cancer diagnosis. Approximately one-third of the women had an implant removed; the majority of these women chose to have the implant replaced. Saline-filled implants were removed for 48% of women, although saline-filled tissue expanders (temporary implants) may be incorrectly included in this category. Fifty-four percent of women with 'other' implants had them removed, and 49% had them replaced. Survival At the end of the follow-up period, 231 (5.3%) patients did not have complete follow-up. Twenty-eight percent of all patients died, with nearly two-thirds of these deaths due to breast cancer (Table 3 ). Women with implants had a similar distribution of causes of death to those without implants (Table 3 ), except for a significantly larger proportion of deaths due to suicide – although this was based on a small number of deaths (0.4% versus 0.03% of all patients, respectively; P = 0.02). Among the 4385 patients with invasive breast cancer and known survival time, the 817 (19%) receiving a breast implant experienced better survival than women without implants, after adjustment for age at diagnosis (Fig. 1 ). The multivariate Cox proportional hazards model showed that implant status was a significant factor associated with improved survival for deaths due to breast cancer and for nonbreast cancer mortality (Table 4 ). Risk of breast cancer death in women with implants was approximately one-half of that for women without implants, after adjustment for multiple clinical and sociodemographic factors. Age at diagnosis, stage, grade, histology, and radiation therapy were significant predictors of breast cancer death in this cohort, as they were for women without implants. With the exception of age, results were similar when modeled for nonbreast cancer mortality, although hazard rate ratios for women with implants were slightly higher overall. Among women with implants, the type of breast implant did not significantly impact survival, although women receiving saline implants had marginally lower risks than women receiving silicone gel implants. For the subsample of 384 San Francisco Bay Area study subjects for whom we were able to assign census-level SES indicators, risk of death among women with breast implants compared with that among women without implants continued to be reduced (hazard ratio, 0.50; 95% confidence interval, 0.20–1.25) after adjustment for census block-group level SES variables. Discussion In this large population-based study of breast cancer patients treated with mastectomy, risks of breast cancer death and nonbreast cancer mortality were lower in women with implants than in women without implants, after adjustment for potential confounders. Postmastectomy breast implants were used by one-fifth of patients who were slightly younger at diagnosis and were more likely to be of white race/ethnicity and to have in situ disease than women without implants. The silicone gel-filled implant was the most common type of implant received. Although breast reconstruction has been shown to provide psychosocial benefits to breast cancer survivors [ 11 - 13 ], concerns have been raised that breast implants may increase the risk of local complications and systemic diseases, including certain cancers and autoimmune diseases [ 2 , 3 , 14 , 15 ]. Breast implants have been suggested to interfere with mammography, thereby facilitating delayed detection of breast tumors, and, consequently, decreased survival [ 16 ]. Despite recent Institute of Medicine recommendations to continue monitoring women with breast implants and to evaluate the potential long-term health effects [ 1 ], few research studies have addressed long-term health outcomes in this group. Moreover, these studies were often conducted on small, nonrepresentative samples without detailed information on implant type and history of use. Georgiade and colleagues found that the survival time for 101 women undergoing breast reconstruction with breast implants was nonsignificantly better than that for 377 women without reconstruction, after adjustment for tumor grade, histology, lymph node involvement, and age at diagnosis, and after a median of 3 years of follow-up [ 6 ]. With a median of 13 years of follow-up, Petit and colleagues found that the risk of breast cancer death was marginally lower in 146 women who underwent breast reconstruction with silicone gel-filled implants than in a matched group without implants (relative risk, 0.6; 95% confidence interval, 0.3–1.1) [ 7 ]. Vandeweyer and colleagues compared 49 women who received saline-filled breast implants following mastectomy with a matched group of women who did not. They found no difference in the number of breast cancer deaths between the two groups [ 8 ]. In a matched analysis of 176 women with a mean of almost 6 years of follow-up, Park and colleagues found that women with breast implants after mastectomy had approximately a 70% reduced risk of death compared with women without implants (relative risk, 0.33; 95% confidence interval, 0.11–0.92) [ 9 ]. Our finding of better survival in women with breast implants is consistent with most of this research [ 6 - 8 ]. However, our study has the substantial advantages of being population-based, being large, having a long follow-up (median, 12.4 years), and including information on implant type and the implant removal and replacement. It is thus well suited to address public health concerns regarding the long-term survival and use of breast implants in women with early-stage, mastectomy-treated breast cancer. Such concerns have recently been re-evaluated in conjunction with the Food and Drug Administration hearings regarding the safety of silicone gel breast implants and their availability for the general market [ 17 , 18 ]. One explanation for our finding of reduced mortality in patients with breast implants may relate to self-selection rather than to a causal role of implants. Although most women who receive mastectomy are eligible to receive breast implants as part of breast reconstruction, surgeons may not recommend this surgery to women with health conditions such as obesity or a recent history of smoking that may contribute to postoperative complications, and may thus impact on survival [ 19 , 20 ]. In our data, the possibility of self-selection based on smoking is supported by the higher proportions of deaths from respiratory cancers and chronic obstructive pulmonary diseases in women without breast implants (Table 3 ). Further investigation is warranted for lifestyle factors (e.g. smoking, diet) and for comorbidities that may account for the survival advantage seen in women with breast implants. In the present study, women with breast implants had a significant excess proportion of deaths due to suicide compared with women without implants. This finding, albeit based on small numbers, is consistent with observations from studies conducted in cosmetic breast implant patients [ 21 , 22 ] and suggests psychiatric consultation should also be considered for breast cancer patients seeking reconstructive surgery with breast implants. In any case, future studies with larger sample sizes are needed to confirm this finding in the breast reconstruction population. An important bias of common concern in retrospective cohort studies is loss to follow-up. A total 231 (5.3%) of the 4385 patients included in the survival analysis did not have complete follow-up at the end of the study period. However, because of the relatively small percentage of patients lost to follow-up, we know that bias due to loss to follow-up has little impact on our survival findings since we found no substantial change in hazards ratios when we assumed the worst-case scenario that all patients lost to follow-up had all died or assumed that all patients lost to follow-up all lived until the end of the study period. Furthermore, although our response rate was relatively high, differences between nonresponders and responders on several patient and tumor characteristics could have biased our findings. Although we were able to adjust for reported patient and tumor characteristics in our multivariate analyses, 356 women (nearly 40% of study-eligible patients) who did not participate in the study were deceased. In the unlikely event that all 356 deceased women had received breast implants, it is possible that the exclusion of these cases from our analysis could bias our results towards and beyond the null, and thereby overestimate the protective effects. Although our survival analyses were adjusted for various demographic and clinical characteristics, our finding of better survival in women with breast implants could reflect uncontrolled confounding by social class, medical care, and psychological factors related to implant usage and survival. Among breast cancer patients treated with mastectomy, those choosing to have breast reconstruction have been shown to differ from women without breast reconstruction on SES, which may be an important factor affecting survival [ 23 , 24 ]. In a convenience sample of more than 200,000 breast cancer patients undergoing mastectomy between 1985 and 1995, Morrow and colleagues found that patients with a family income of $40,000 or more were twice as likely as patients with a family income of less than $40,000 to receive postmastectomy breast reconstruction [ 25 ]. Higher income may be a predictor of better survival after breast cancer, as women with higher incomes may have better access to cancer care and treatment. In the present study, SES did not alter the effect of implants on survival in the subset of women for whom SES measures were available. Differences in these area-level measures of SES are thus not likely to contribute substantially to the survival differences between breast cancer patients with and without implants in this study. Additional unmeasured confounders related to the increased medical care of women with breast implants could explain the protective association of breast implants with cancer survival. Because women with breast implants may be more closely followed in their medical care, they may have recurrences diagnosed and treated earlier; thus they may experience better survival than women without implants. Although our study lacked information on breast cancer recurrence, we were able to examine the impact of subsequently diagnosed primary breast tumors. We observed that the proportion of women with two or more primary breast tumors was lower in women with breast implants than in women without (15% and 21%, respectively). To address the possibility that a higher incidence of subsequently diagnosed primary breast tumors impacted survival in women without breast implants, we limited survival analyses to women with only one primary tumor ( n = 3535) and found a consistently reduced risk of breast cancer death associated with breast implant usage (hazard ratio, 0.54; 95% confidence interval, 0.42–0.68), after adjusting for similar prognostic factors. Our findings also are consistent with results from studies showing a reduced risk of death in augmentation mammoplasty patients with at least 10 years of follow-up compared with the general population [ 26 - 29 ]. Furthermore, psychological factors underlying a woman's decision to obtain breast implants [ 30 , 31 ], including body image concerns and self-esteem, may play a role in lifestyle behaviors relevant to survival, although the extent to which they directly impact survival is unclear. Several biological mechanisms have been proposed to explain how breast implants may influence survival outcomes [ 16 , 26 , 32 , 33 ]. Breast implants may stimulate a local immune response in which cancer cells are more likely to be destroyed [ 34 ]. Breast implants may compress breast tissue, reducing the flow of blood and thereby slowing the rate of cell or tumor growth. Breast implants may decrease the temperature of the breast by separating the breast tissue from the body, thereby decreasing the metabolic rate and slowing the growth rate of residual breast cancer cells [ 35 ]. These mechanisms may provide important clues in cancer prevention and warrant further investigation. Conclusions With a median of more than 12 years of follow-up on patients, our population-based study shows that the risk of breast cancer mortality in patients with breast implants following mastectomy is about one-half of that for patients without implants, after adjustment for prognostic characteristics. Although patients in our study received implants in the 1980s and early 1990s, breast implants continue to be an integral part of breast reconstruction in breast cancer patients and have not changed dramatically in design. Thus, despite an overall decrease in implant use among breast cancer patients, breast implants remain an important and commonly used option for women considering reconstruction. Certainly, further research is needed to explain the survival differential in women with breast implants and those without, by examining potentially explanatory factors such as SES, comorbidity, smoking, or other lifestyle factors. However, based on this large, representative sample of breast cancer patients with extensive follow-up, we found that breast implants following mastectomy do not appear to confer any survival disadvantage following early-stage breast cancer in women younger than 65 years old. Abbreviations SEER = Surveillance, Epidemiology, and End Results; SES = socioeconomic status. Competing interests The author(s) declare that they have no competing interests. Authors' contributions SLG, DWW, JLS, and CFL implemented the study and acquired the data. GML, SLG, and CDO participated in the design and conceptualization of the study. GML performed the statistical analysis and drafted the manuscript. GML, SLG, CDO, and THMK participated in the analysis and interpretation of the data. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064128.xml

1064129

Knockdown of c-Myc expression by RNAi inhibits MCF-7 breast tumor cells growth in vitro and in vivo

Introduction Breast cancer is the leading cause of cancer death in women worldwide. Elevated expression of c-Myc is a frequent genetic abnormality seen in this malignancy. For a better understanding of its role in maintaining the malignant phenotype, we used RNA interference (RNAi) directed against c-Myc in our study. RNAi provides a new, reliable method to investigate gene function and has the potential for gene therapy. The aim of the study was to examine the anti-tumor effects elicited by a decrease in the protein level of c-Myc by RNAi and its possible mechanism of effects in MCF-7 cells. Method A plasmid-based polymerase III promoter system was used to deliver and express short interfering RNA (siRNA) targeting c- myc to reduce its expression in MCF-7 cells. Western blot analysis was used to measure the protein level of c-Myc. We assessed the effects of c-Myc silencing on tumor growth by a growth curve, by soft agar assay and by nude mice experiments in vivo . Standard fluorescence-activated cell sorter analysis and TdT-mediated dUTP nick end labelling assay were used to determine apoptosis of the cells. Results Our data showed that plasmids expressing siRNA against c- myc markedly and durably reduced its expression in MCF-7 cells by up to 80%, decreased the growth rate of MCF-7 cells, inhibited colony formation in soft agar and significantly reduced tumor growth in nude mice. We also found that depletion of c-Myc in this manner promoted apoptosis of MCF-7 cells upon serum withdrawal. Conclusion c-Myc has a pivotal function in the development of breast cancer. Our data show that decreasing the c-Myc protein level in MCF-7 cells by RNAi could significantly inhibit tumor growth both in vitro and in vivo , and imply the therapeutic potential of RNAi on the treatment of breast cancer by targeting overexpression oncogenes such as c- myc , and c- myc might be a potential therapeutic target for human breast cancer.

Introduction Breast cancer is the leading cause of cancer death in women worldwide. Despite advances in detection and chemotherapy, many women with breast cancer continue to die of this malignancy [ 1 ]. Therefore, an understanding of the molecular mechanisms involved in breast cancer formation and progression should be helpful in developing more effective treatments for breast cancer. c-Myc is believed to participate in most aspects of cellular function, including replication, growth, metabolism, differentiation, and apoptosis [ 2 ]. Previous studies indicate that c-Myc activates a variety of known genes as part of a heterodimeric complex with Max [ 2 ]. A frequent genetic abnormality seen in breast cancer is the elevated expression of c-Myc [ 3 , 4 ]. The importance of c-Myc expression in breast cancer is demonstrated both by studies of transgenic mice and by clinical research [ 3 , 5 ]. Abnormal expression of c- myc transgenes in the mouse mammary gland is associated with an increased incidence of breast carcinomas [ 5 ]. Moreover, clinical studies have indicated that c-Myc is important in the development and progression of breast cancer, in that overexpression of c-Myc was found in most breast cancer patients and was correlated with poor prognosis in those patients [ 3 ]. The role of c-Myc in breast cancer has been extensively examined in many studies for the past decade [ 6 ]; however, specifically reducing its level by genetic means in established breast cancer cell lines is still helpful for a better understanding of its role in maintaining the malignant phenotype. Thus, in this study, we investigated whether specifically decreasing the protein level of c-Myc in a breast cancer cell line in which this protein was overexpressed might result in the inhibition of cell growth in vitro and in vivo . For this purpose, RNA interference (RNAi) directed against c- myc was used. RNAi is the sequence-specific gene silencing induced by double-stranded RNA (dsRNA). This phenomenon is conserved in a variety of organisms: Caenorhabditis elegans , Drosophila , plants, and mammals. RNAi is mediated by short interfering RNAs (siRNAs) that are produced from long dsRNAs of exogenous or endogenous origin by an endonuclease of the ribonuclease-III type, called Dicer. The resulting siRNAs are about 21–23 nucleotides (nt) long and are then incorporated into a nuclease complex, the RNA-inducing silencing complex, which then targets and cleaves mRNA containing a sequence identical to that of the siRNA [ 7 ]. Rapid progress has been made in the use of RNAi [ 8 ]. More recently, a technical breakthrough came from the demonstration that dsRNA of 19–29 nt expressed endogenously with RNA polymerase III promoter induced target gene silencing in mammalian cells [ 9 ]. The expression of siRNA from DNA templates offers several advantages over chemically synthesized siRNA delivery. Hairpin siRNAs transcribed from a vector are thought to suppress the expression of targeted genes more efficiently, less expensively and more easily than synthesized siRNA [ 10 ]. Here we used a plasmid-based polymerase III promoter system to deliver and express siRNA targeting c- myc to determine whether this technique could be used for the specific inhibition of oncogene overexpression and whether this inhibition resulted in antitumor effects. We showed in our study that specific downregulation of c-Myc by RNAi was sufficient to inhibit the growth of MCF-7 cells in vitro and in vivo , and that c- myc might serve as a therapeutic target for human breast cancer. Method Plasmid construction To generate c-Myc knockdown vector, one annealed set of oligonucleotides encoding short hairpin transcripts corresponding to nt 1906–1926 of c- myc mRNA (GenBank accession no. NM-002467) [ 11 ] were cloned into p Silencer 1.0_U6 (Ambion; hereafter abbreviated to p Silencer ). In brief, the short-hairpin-RNA-encoding complementary single-stranded oligonucleotides, which hybridized to give overhangs compatible with Apa I and Eco RI, were designed with a computer program available on the Internet . Oligonucleotides encoding short hairpin RNAs were then ligated into p Silencer . Bacterial colonies were pooled and used for plasmid preparation. The positive clones were confirmed by sequencing. The resulting plasmid was designated as p Silencer –c-Myc. Cell line and cell culture The breast cancer adenocarcinoma cell line MCF-7 was obtained from the American Type Culture Collection (Manassas, VA). The cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Gibco BRL), 50 units/ml penicillin, and 50 μg/ml streptomycin. The MCF-7 cells were maintained in a humidified 37°C incubator with 5% CO 2 , fed every 3 days with complete medium, and subcultured when confluence was reached. Transfection of cells A total of 2 × 10 5 cells were seeded into each well of a six-well tissue culture plate (Costar). The next day (when the cells were 70–80% confluent), the culture medium was aspirated and the cell monolayer was washed with prewarmed sterile phosphate-buffered saline (PBS). Cells were transfected with the appropriate plasmids by using LipofectAMINE reagent (Invitrogen) in accordance with the manufacturer's protocol. The cells were harvested at different time points. Western blot analysis or other experiments were performed. Western blot analysis Cells were harvested at different time points and lysed in mammalian cell lysis buffer, then western blot analysis was performed with the use of conventional protocols as described previously [ 12 ]. In brief, the protein concentration was determined with a bicinchoninic acid kit with bovine serum albumin as a standard (Pierce). Equal amounts of total protein were then separated on 12% polyacrylamide gels by using standard sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) techniques, then transferred to nitrocellulose membranes (PROTRAN). The antibodies and dilutions used included anti-c-Myc (9E10; 1:1000 dilution; Santa Cruz) and anti-β-actin (AC-15; 1:5000 dilution; Sigma), and after being washed extensively the membranes were incubated with anti-mouse IgG–horseradish peroxidase conjugate antibody (Zhongshan Company) for 1 hour at room temperature and developed with a Luminol chemiluminescence detection kit (Santa Cruz). Membranes probed for c-Myc were reprobed for β-actin to normalize for loading and/or quantification errors and to allow comparisons of target protein expression to be made. Protein expression was quantified with a Gel EDAS analysis system (Cold Spring USA Corporation) and Gel-Pro Analyzer 3.1 software (Media Cybernetics). Cell growth assay At 2 days after transfection, MCF-7 cells transfected with indicated plasmids were harvested and replated at a density of 50 cells/mm 2 in triplicate. The total cell number was quantified every 2 days with a hematocytometer and an Olympus inverted microscope. Cell viability was assessed by using trypan blue. Soft agar colony assay At 2 days after transfection, MCF-7 cells (300 cells per well) transfected with indicated plasmids were mixed with tissue culture medium containing 0.7% agar to result in a final agar concentration of 0.35%. Then 1 ml samples of this cell suspension were immediately plated in six-well plates coated with 0.6% agar in tissue culture medium (2 ml per well) and cultured at 37°C with 5% CO 2 . After 2 weeks the top layer of the culture was stained with 0.2% p -iodonitrotetrazolium violet (Sigma). The culture was analyzed in triplicate, and colonies larger than 100 μm in diameter were counted. Tumor growth in nude mice Equal numbers (10 6 or 2 × 10 6 ) of MCF-7 cells transfected with p Silencer –c-Myc or p Silencer were harvested by trypsinization 2 days after transfection, washed twice with 1 × PBS, and resuspended in 0.2 ml of saline. Two groups of five 4–6-week-old female nude mice were then given bilateral subcutaneous injections with control cells or cells transfected with plasmids against c-Myc. The mice were kept in pathogen-free environments and checked every 2 days. The date at which a palpable tumor first arose and the weight of the tumor were recorded. Cell cycle analysis Standard fluorescence-activated cell sorting analysis was used to determine apoptosis of the cells. In brief, MCF-7 cells were transfected with p Silencer –c-Myc or p Silencer ; 24 hours later, cells were deprived of serum for 36 hours. Then cells were harvested, washed once in PBS and stained with propidium iodide (BD Biosciences). The apoptotic cells were assessed by flow cytometric detection of sub-G1 DNA content. TdT-mediated dUTP nick end labelling assay Apoptotic cells were confirmed with the in situ cell death detection kit, Alkaline Phosphatase (Roche Applied Science), in accordance with the manufacturer's instructions. In brief, MCF-7 cells were grown on coverslips. The next day, cells were transfected with p Silencer –c-Myc or p Silencer . At 24 hours after transfection, cells were deprived of serum for 36 hours. Coverslips with adherent cells were fixed in 4% paraformaldehyde for 1 hour at room temperature and permeabilized with 0.1% Triton X-100 for 2 min on ice. DNA fragments were labeled with the TdT-mediated dUTP nick end labelling (TUNEL) reaction mixture for 60 min at 37°C in a humidified atmosphere in the dark. The coverslips were then incubated with Converter alkaline phosphatase for 30 min at 37°C in a humidified chamber, rinsed in PBS, and incubated with nitro blue tetrazolium/5-bromo-4-chloroindol-3-yl phosphate (Roche Applied Science) for 10 min. Cells were mounted cell side downward on a microscope slide, and the apoptotic cells (dark blue staining) were counted under a microscope. Three fields were randomly counted for each sample. Statistical analysis SPSS for Windows (SPSS Inc.) was used to analyze the data and plot curves. A two-tailed unpaired t -test was used to compare the statistical significance of the differences in data from the two groups. Results Suppression of c-Myc overexpression in MCF-7 cells by RNAi We designed and synthesized one pair of oligonucleotides encoding short hairpin transcripts directed against a portion of the c- myc mRNA. This pair of oligonucleotides was then ligated into p Silencer . The plasmid of p Silencer –c-Myc contains a U6 promoter that directs the synthesis of oligonucleotides in an inverted repeat with 9 nt for its loop, with six T bases added at the end to serve as a termination signal for RNA polymerase III. The RNA is expected to fold back to form a hairpin loop structure after being transcribed; the hairpin dsRNA can then be further cleaved by Dicer to generate a 21-nucleotide siRNA, the active form for the RNAi effect, which will form dsRNA–endonuclease complexes and will bind and destroy c- myc mRNA inside cells [ 8 ] (Fig. 1 ). p Silencer –c-Myc was transfected into MCF-7 cells and its effects on c-Myc protein levels were determined by comparison with p Silencer -transfected cells by western blot at the time points indicated. We found that the c-Myc expression levels were suppressed by up to 80% in MCF-7 cells at day 5 after transfection. Actually the levels of c-Myc were decreased as early as 24 hours after transfection and remained at low levels until 12 days after transfection (Fig. 2 ). The inhibitory effect was shown to be specific because transfection with p Silencer did not alter c-Myc levels. In addition, RNAi did not cause a nonspecific downregulation of gene expression, as determined by the β-actin control (Fig. 2 ). Similar results with c-Myc RNAi were also seen in other cell lines including HCT116 and HepG2 (data not shown). These data indicated that vector-based RNAi could effectively suppress c-Myc overexpression and resulted in prolonged decreases in specific cellular gene expression without marked effects on other cellular proteins. Decreased levels of c-Myc significantly alter the growth rate of MCF-7 cells Previous studies have shown that c-Myc is important in cellular proliferation and cell growth [ 6 ]. Thus, increased levels of c-Myc might have a function in the growth advantage seen in breast tumors. In our study, MCF-7 cells were transfected with p Silencer –c-Myc or p Silencer . The number of MCF-7 cells was then counted every 2 days after transfection. Our data showed that RNAi directed against c- myc significantly decreased the growth rate of MCF-7 cells, with a 50–60% decrease at different time points repeatedly in three separate experiments (Fig. 3 ). Decreases in c-Myc protein inhibit colony formation We then tested whether RNAi-mediated reductions in c-Myc levels could influence the ability of MCF-7 cells to form colonies in soft agar. MCF-7 cells were transfected with p Silencer –c-Myc or p Silencer . At 48 hours after transfection, the cells were placed into medium with soft agar, and colonies were counted after 2 weeks. RNAi directed against c- myc resulted in a significant decrease (about 65%) in colony formation in MCF-7 cells (Fig. 4 ). The smaller number of colonies in the p Silencer –c-Myc group than in the control group was statistically significant ( P < 0.001). These results showed that the reduction in c-Myc protein level decreased the ability of breast cancer cells to form colonies in soft agar. RNAi directed against c-Myc reduces tumor growth in nude mice To address the potential effects of RNAi in vivo on inhibiting the growth of breast cancer cells, equal numbers (10 6 or 2 × 10 6 ) of MCF-7 cells transfected with p Silencer –c-Myc or p Silencer were injected into female nude mice (five animals for each treatment). At 5 or 8 weeks after injection of these cells, the mice were killed and the weights of the tumors were recorded. As seen in Table 1 , when 10 6 cells were injected into nude mice, tumors were seen 8 weeks later in three of five at the left lateral where MCF-7 cells transfected with p Silencer were injected, whereas none were seen at the right lateral where MCF-7 cells transfected with p Silencer –c-Myc were injected. In addition, 5 weeks after 2 × 10 6 cells were injected into nude mice, tumors were seen in four of five at the left lateral where MCF-7 cells transfected with p Silencer were injected, whereas none were seen at the right lateral where MCF-7 cells transfected with p Silencer –c-Myc were injected (Table 1 ). Thus, c-Myc RNAi significantly suppressed tumor growth in nude mice in comparison with control, indicating that targeting c- myc by RNAi could exert a strong antitumor effect in vivo on MCF-7 cells. Induction of apoptosis in MCF-7 cells by RNAi depletion of c-Myc upon serum deprivation The above data demonstrated that knockdown of c-Myc in MCF-7 cells could significantly inhibit the growth of tumor cells both in vitro and in vivo . To determine whether depletion of c-Myc could promote the death of tumor cells, flow cytometry and TUNEL assays were performed. At 24 hours after transfection with p Silencer –c-Myc or p Silencer , MCF-7 cells were deprived of serum for 36 hours. These cells were then analyzed by flow cytometry or TUNEL assay. Significant sub-G1 (apoptotic) populations were observed in the flow cytometry assay. We found that 31.1% of MCF-7 cells transfected with p Silencer –c-Myc underwent apoptosis after serum starvation, compared with 5.8% in the control group (Fig. 5a ). We also confirmed the apoptosis of MCF-7 cells by TUNEL assay (Fig. 5b ). About 40% cells were TUNEL-positive in the p Silencer –c-Myc group, compared with 6% in the control group ( P < 0.01). These data suggested that depletion of c-Myc by RNAi in MCF-7 cells made the cells more sensitive to apoptosis after serum deprivation. Discussion Cancer cells often show alteration in the signal-transduction pathways, leading to proliferation in response to external signals. Oncogene overexpression is a common phenomenon in the development and progression of many human cancers. Oncogenes therefore provide a potential target for cancer gene therapy [ 13 ]. The important oncogene c- myc is expressed in a high proportion of most human cancers, including breast, prostate, gastrointestinal cancer, lymphoma, melanoma, and myeloid leukemia [ 14 ]. In its physiological role, c-Myc is broadly expressed during embryogenesis and in tissue compartments of the adult that possess high proliferative capacity. Altered expression of c-Myc seems to define a common event associated with the pathogenesis of most human cancers [ 6 ]. Previous studies demonstrated that the continued presence of c-Myc was required for cancer development and not just for initiation, and inactivation of c-Myc resulted in the sustained regression of tumors [ 15 - 17 ]. Similar results were also observed in breast cancer. D'Cruz and colleagues demonstrated that overexpression of c-Myc by an inducible system in the mammary epithelium of transgenic mice resulted in the formation of invasive mammary adenocarcinomas, many of which regressed fully after c-Myc deinduction [ 18 ]. Therefore, specific downregulation of c-Myc might be a potential therapeutic strategy against human cancers, including breast cancer. In fact, the antagonists of c-Myc, including full-length antisense mRNA [ 19 ], oligonucleotides against c- myc mRNA [ 20 ] or a dominant-negative mutant [ 21 ], were previously reported to inhibit proliferation of cancer cell lines in vitro . However, it was only successful in some situations; these technologies have been difficult to apply universally [ 22 ]. Recently the advent of RNAi-directed 'knock-down' has sparked a revolution in somatic cell genetics, allowing the inexpensive, rapid analysis of gene function in mammals, and might be exploited for gene therapy [ 7 , 8 ]. Some studies directly compared RNAi with antisense RNA and found that RNAi seemed to be quantitatively more efficient and durable in cell culture and in nude mice [ 23 ]. By means of the RNAi method, in the present study, cellular growth assays, both in vitro and in vivo , were used to determine the functional consequences of RNAi-mediated decreases in of c-Myc in established breast cancer cells. Our results demonstrated that RNAi can effectively downregulate oncogene overexpression with great specificity. We showed that the plasmids endogenously expressing siRNA could successfully deplete up to 80% of c-Myc expression in MCF-7 cells at day 5 after transfection. Furthermore, the tumor inhibition effects persisted for at least 12 days after transfection in dishes and for 2 months in nude mice as shown by experiments in vitro and in vivo , even though the protein level of c-Myc in silenced clones expressing siRNA was back to almost the same level as in the control cells by day 12 after transfection. Our data were consistent with the results of Jain and colleagues [ 17 ]. They showed that within 24 hours of c- myc inactivation, the osteogenic sarcoma cells flattened and showed less cell division. And even after reactivation of c- myc expression in these cultured cells, total cell numbers continued to be lower. Less than 1% of the tumor cells regained their neoplastic growth properties [ 17 ]. All of these data indicated that brief inactivation of c-Myc could induce a sustained loss of neoplastic phenotypes. Moreover, other groups using chemically synthesized siRNAs to knock down their favored oncogenes also found that a transient decrease in oncogene expression could inhibit the growth of tumor cells in vitro and/or in vivo [ 8 ]. Nevertheless, the underlying mechanism of this phenomenon in MCF-7 cells should be further investigated. Although some studies previously revealed that the effects of inactivation of c-Myc in some cell lines were modest [ 6 ], other groups using different approaches to reduce the protein level of c-Myc found that a decrease in c-Myc expression could inhibit the growth of these tumor cells, including breast tumor cells [ 19 - 21 ]. Nevertheless there were still conflicting results on whether c-Myc expression was necessary to maintain tumorigenesis in different animal models from different laboratories [ 15 - 18 , 24 , 25 ]. For example, some studies showed that the role of oncogenic c-Myc in tumor maintenance was essential and that all effects of c-Myc in vivo were reversible, in that without continuous c-Myc activation there would even be regression of established tumors back to phenotypically normal in transgenic mouse models [ 15 , 16 , 24 ]. Similar phenomena were observed by other groups focusing on other oncogenes, such as bcr / abl [ 26 ] and H - ras [ 27 ]. However, there were also conflicting reports, mainly showing that brief inactivation of c-Myc could induce sustained loss of neoplastic phenotypes in certain animal models [ 17 , 18 , 25 ]. It was notable that without secondary oncogenic mutation, spontaneously or selectively, such as Kras2 oncogene mutation, in nearly all breast tumors induced by conditionally expressing the human c-Myc in the mammary epithelium of a transgenic mouse model, deinduction of c-Myc protein could lead to full regression of tumors [ 18 ]. Similar data were also obtained by Karlsson and colleagues that the inactivation of c-Myc alone was found to be sufficient to cause sustained tumor regression in c-Myc-induced hematopoietic tumors; in contrast, tumor cells that acquired novel chromosomal translocations relapsed independently of Myc to maintain their neoplastic phenotype [ 25 ]. It was therefore not surprising for us to show here that a transient reduction of c-Myc protein level by RNAi could significantly inhibit the growth rate of MCF-7 cells and its ability to form colonies in soft agar. Additionally, the remarkable effect in nude mice supported the effectiveness of this treatment. Our data also suggested that knockdown of c-Myc by RNAi in MCF-7 cells could increase the sensitivity of these cells to apoptotic stimuli, such as serum starvation. This was most probably one of the reasons for the anti-tumor effects. There have previously been conflicting reports about the role of c-Myc in apoptosis [ 3 ]. Constant overexpression of c-Myc might induce apoptosis [ 2 , 6 ], and a decrease in c-Myc levels by techniques brought about by, for example, an antisense approach might also cause apoptosis of certain tumor cells [ 28 - 30 ] or might increase the sensitivity of the cells to apoptotic stimuli [ 31 ]. These conflicting observations suggested that c-Myc was capable of both inducing and suppressing apoptosis in different types of tumor cell, under different conditions, and in different systems. In MCF-7 cells, suppression of c-Myc expression in response to aromatase inhibitors or topoisomerase α inhibitors could induce these cells to apoptosis [ 32 , 33 ]. However, the pathways that c-Myc controls and/or that are involved in the observed apoptosis remain obscure. D'Agnano and colleagues suggested that in melanoma cells the downregulation of c-Myc by an antisense approach could activate apoptosis by increasing the levels of p27 Kip1 [ 28 ]. Overexpression of the cyclin-dependent kinase inhibitor p27 Kip1 was able to promote apoptosis in several mammalian tumor cell lines [ 34 ]. However, understanding the precise pathway by which a decrease in c-Myc in MCF-7 cells by RNAi was able to induce apoptosis upon serum deprivation needs further study. So far, RNAi has been used to inhibit virus-induced diseases (for example HIV [ 35 ] and influenza [ 36 ]), oncogenic K-ras , H-ras -induced tumorigenesis [ 37 , 38 ], activation of oncogenes resulting from chromosomal translocations (for example bcr / abl in chromic myeloid leukemia [ 39 ]), cancers caused by viral infections [ 40 ], and so on. Recently, retroviral-based approaches to deliver siRNA into tissue-cultured mammalian cells have been proved to be powerful [ 37 , 38 ], and doxycycline-regulated inducible knockdown of gene expression by RNAi has been shown to be particularly useful for the analysis of genes that are essential for cellular survival [ 41 , 42 ]. These studies have marked a new era in the genetic manipulation of human cancer development by allowing oncogenes to be downregulated by RNAi. Conclusion In summary, RNAi has been used in this study to demonstrate that decreases in c-Myc levels can inhibit tumor growth in assays both in vivo and in vitro . In addition, these data indicate that RNAi provides a useful method with which to study the role of genes that control the growth of cancer cells; and, given its specificity and the lower doses needed to inhibit gene expression than those required for antisense oligonucleotides, RNAi might have potential therapeutic utility in a variety of disease states, including cancers. Future studies could further test whether c-Myc overexpression can be efficiently depleted by siRNA expressed from a DNA-based expression vector combined with a tumor-specific promoter, such that RNAi can specifically target oncogenes in cancer cells without affecting normal cells. Abbreviations dsRNA = double-stranded RNA; nt = nucleotides; PBS = phosphate-buffered saline; RNAi = RNA interference; SDS–PAGE = sodium dodecyl sulfate–polyacrylamide gel electrophoresis; siRNA = short interfering RNA. Competing interests The author(s) declare that they have no competing interests. Authors' contributions YW designed the experiments, constructed the plasmids, performed transfections, and wrote the manuscript. SL constructed the growth curve and performed the nude mice experiments. GZ performed the western blots and the soft agar assay. CZ, HZ, and XZ performed the flow cytometry and TUNEL experiments. LQ and JB conducted the cell culturing. NX is the corresponding author. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064129.xml

1064130

SULT1A1 genotype, active and passive smoking, and breast cancer risk by age 50 years in a German case–control study

Introduction Sulfotransferase 1A1 (encoded by SULT1A1 ) is involved in the metabolism of procarcinogens such as heterocyclic amines and polycyclic aromatic hydrocarbons, both of which are present in tobacco smoke. We recently reported a differential effect of N -acetyltransferase ( NAT ) 2 genotype on the association between active and passive smoking and breast cancer. Additional investigation of a common SULT1A1 genetic polymorphism associated with reduced enzyme activity and stability might therefore provide deeper insight into the modification of breast cancer susceptibility. Methods We conducted a population-based case–control study in Germany. A total of 419 patients who had developed breast cancer by age 50 years and 884 age-matched control individuals, for whom risk factor information and detailed smoking history were available, were included in the analysis. Genotyping was performed using a fluorescence-based melting curve analysis method. Multivariate logistic regression analysis was used to estimate breast cancer risk associated with the SULT1A1 Arg 213 His polymorphism alone and in combination with NAT2 genotype in relation to smoking. Results The overall risk for breast cancer in women who were carriers of at least one SULT1A1*2 allele was not significantly different from that for women with the SULT1A1*1 / *1 genotype (adjusted odds ratio 0.83, 95% confidence interval 0.66–1.06). Risk for breast cancer with respect to several smoking variables did not differ substantially between carriers of the *2 allele and noncarriers. However, among NAT2 fast acetylators, the odds ratio associated with passive smoking only (3.23, 95% confidence interval 1.05–9.92) was elevated in homozygous carriers of the SULT1A1*1 allele but not in carriers of the SULT1A1*2 allele (odds ratio 1.28, 95% confidence interval 0.50–3.31). Conclusion We found no evidence that the SULT1A1 genotype in itself modifies breast cancer risk associated with smoking in women up to age 50 years. In combination with NAT2 fast acetylator status, however, the SULT1A1*1 / *1 genotype might increase breast cancer risk in women exposed to tobacco smoke.

Introduction Epidemiologic evidence linking cigarette smoking to increased risk for development of breast cancer is mounting (for review [ 1 , 2 ]). In addition, findings from both epidemiology and molecular biology indicate that there is differential susceptibility within the population to development of malignant neoplasms following exposure to certain xenobiotics because of polymorphisms in genes that encode metabolizing enzymes. Previously, we reported a differential effect of N -acetyltransferase ( NAT ) 2 genotype on the association between active and passive smoking and breast cancer risk [ 3 ]. The identification of passive smoking as a breast cancer risk factor, particularly for fast acetylators, implied that heterocyclic aromatic amines (HCAs) are among the responsible carcinogens. HCAs are particularly abundant in side-stream tobacco smoke [ 4 ] and are activated by O -acetylation catalyzed by NATs [ 5 ]. Because sulfotransferase (SULT)1A1 (encoded by SULT1A1 ) is also involved in the metabolism of pro-carcinogens from tobacco smoke, the additional investigation of a common polymorphism in the SULT1A1 gene might provide deeper insight into the modification of susceptibility to breast cancer. The SULT1A1 enzyme is generally associated with detoxification of xenobiotic compounds and has been implicated in oestrogen metabolism. However, Glatt and coworkers [ 6 , 7 ] showed that several substances can be activated by the conjugation reaction with SULT1A1, among which are pro-carcinogens such as polycyclic aromatic hydrocarbons and HCAs, both of which are present in tobacco smoke [ 8 , 9 ]. In contrast to earlier assumptions [ 10 ], there is increasing evidence that the SULT1A1 enzyme apparently does not play an important role in oestrogen metabolism in vivo . Results from in vitro studies showed that only the SULT1E1 enzyme is capable of the sulfonation of oestradiol, oestrone and catecholestrogens at physiologically relevant concentrations [ 11 , 12 ]. For instance, Adjei and Weinshilboum [ 11 ] showed that K m values for the sulfonation of oestradiol with SULT1A1 were in the micromolar range, which is clearly above physiological concentrations, whereas the K m value for SULT1E1 was considerably lower (0.029 ± 0.01 μmol/l). Large interindividual variations in the biochemical and metabolic properties of the SULT1A1 enzyme have been observed that can partly be explained by a G to A polymorphism at nucleotide 638 (Arg 213 His), referred to as the *2 allele. The *2 allele has been associated with lower activity and lower thermal stability of the SULT1A1 enzyme [ 13 , 14 ], and thus reduced bioactivation of mutagens [ 6 ]. Thus far, three case–control studies that investigated the association between SULT1A1 genotype and breast cancer risk [ 15 - 17 ] have been reported. Results were not consistent and the effect of smoking was not considered in any of those case–control studies. Saintot and coworkers [ 18 ] recently reported a positive interaction between smoking and the variant allele for SULT1A1 with respect to breast cancer risk in a case-only study. We conducted the present study to elucidate the potential role of SULT1A1 genotype alone and in combination with NAT2 genotype as a modifier of susceptibility to breast cancer associated with exposure to tobacco smoke among predominantly premenopausal women. Methods Study population The present study is based on a case–control study that is described in greater detail elsewhere [ 19 , 20 ]. In brief, between January 1992 and December 1995 a population-based case–control study on breast cancer was conducted in two regions (Rhein-Neckar-Odenwald and Freiburg regions) in southern Germany. Women with a diagnosis of in situ or invasive breast cancer were identified by surveying all of the hospitals that serve the two study regions. Women were eligible for inclusion in the study if they spoke German, if they lived in the study region and if the neoplasm was diagnosed before their 51st birthday. During the period of study 1020 women were identified, of whom 1005 were alive at the time of identification. Of the living, eligible patients, 706 (70.2%) completed a self-administered questionnaire. For every patient, two controls were selected randomly from lists of female residents obtained by the population registries of the study regions and matched according to exact age and residence. Of the 2257 eligible control individuals who were contacted by letter, 1381 (61.2%) participated in the study. After giving written informed consent, all participants completed a self-administered questionnaire and were asked to provide a blood sample. The study is in compliance with the Declaration of Helsinki and was reviewed and approved by the ethics committee of the University of Heidelberg. The study participants were re-contacted in August 1999 and were invited to participate in a computer-assisted telephone interview to assess comprehensively their history of active and passive smoking [ 20 ]. Of the original study population, 66.3% of cases and 78.9% of controls took part in this additional investigation. In short, women were asked when they began smoking, the type of product, the amount and frequency of tobacco usage, the intensity of inhalation, and the date of cessation or changes in their smoking habits. Exposure to passive smoking was assessed in childhood, in the adult household and at work. For passive smoking in adult life, women who had lived with a smoking partner were asked the onset, end, or changes to smoking exposure, daily amount and type of product smoked, and number of hours and days of passive exposure. For childhood exposure as well as exposure at work and that due to other household members, questions pertained to number of smokers living in the household, onset of exposure, and the number of hours and days of smoke exposure that the participant experienced in the presence of each smoking person. All information was truncated at the reference date, which was the date of diagnosis for patients and the date of recruitment for control individuals. Menopausal status was defined as the reported state at half a year before the reference date. The status of women with previous hysterectomy not accompanied by bilateral oophorectomy was not ascertainable and therefore classified as unknown. Because the study participants were all aged 50 years or younger, these women were included in the analysis restricted to the subgroup with premenopausal status. Blood samples were available for 95% of cases and 82% of controls in the original study population. This analysis was restricted to women who had either both (97.8%) or at least one parent of German nationality (2.2%) in order to achieve ethnic homogeneity of the study population. In total, 419 patients with breast cancer and 884 control individuals, for whom full genotype information and detailed history of tobacco smoke exposure were available, were included in the analysis. Genotyping DNA was extracted from EDTA blood samples using a standard method based on salt precipitation. SULT1A1 -specific primers and hybridization probes were used to detect G638A in exon 7. The primers for DNA amplification were previously described by Coughtrie and coworkers [ 21 ]. As sensor and anchor probes, we used LCRed640-CAgggAgCgCCCCACAA-p and gAACCATgAAgTCCACggTCTCCTCT-x, respectively. PCR and melting curve analyses were performed in 10 μl volumes in glass capillaries (Roche Diagnostics, Mannheim, Germany) using the following: 1× PCR buffer, 2.5 mmol/l MgCl 2 , 200 μmol/l dNTPs, 0.1% bovine serum albumin, 0.5 U Taq polymerase, 0.15 μmol/l of each probe (TIB MOLBIOL, Berlin), 1 μmol/l of the sense primer (CF) and 0.1 μmol/l of the reverse primer (CR; asymmetric PCR). Approximately 10 ng gDNA was used as a template. The cycling conditions were as follows: initial denaturation at 95°C for 2 min followed by 45 cycles of denaturation at 95°C for 0 s, annealing at 63°C for 5 s and elongation at 72°C for 10 s, with a ramping rate of 20°C/s. Melting curve analyses were performed with an initial denaturation at 95°C for 10 s, 20 s at 40°C, followed by slow heating of the samples to 80°C with a ramping rate of 0.1°C/s and continuous fluorescence detection. The melting curves were converted to melting peaks by plotting the negative derivatives of fluorescence against temperature (–dF/dT). Melting peaks were mostly unambiguous, but certain samples exhibited abnormal peaks due to two rare silent genetic polymorphisms, one covered by the sensor (G645A [Leu215]) and another covered by the anchor probe (G654A [Glu218]). Forty-eight such samples were additionally digested with Hha I [ 15 ], allowing unambiguous genotyping at position 638. A further 160 samples selected randomly for quality control exhibited no discrepancies between genotyping results obtained with both methods. Two additional rare genetic variants of the SULT1A1 gene (i.e. the *3 [Met 223 Val] and *4 [Arg 37 Gly] alleles), which have been observed in Caucasian populations with allele frequencies of 0.01 and 0.003, respectively, were not accounted for in the present study [ 22 ]. Detection of polymorphic sites in the NAT2 gene was also carried out by capillary-based PCR followed by melting curve analysis. The method was described in detail previously [ 3 ]. Statistical analysis The association between active/passive smoking and breast cancer by SULT1A1 genotype was assessed by multivariate conditional logistic regression analysis. We computed maximum likelihood estimates for the odds ratios (ORs) and their 95% confidence intervals (CIs) using the PHREG procedure of the statistical software package SAS release 8.2 (SAS Institute, Cary, NC, USA). 'Ever active smoking' was defined as having smoked more than 100 cigarettes in one's life. Among ever active smokers, women were termed current smokers if they had smoked regularly within the year preceding the interview; otherwise, they were classified as former smokers. If women were on average exposed to passive smoke for more than 1 hour/day for at least 1 year, then they were defined as ever passive smokers. The average exposure was obtained by multiplying the average hours/day by the duration in years for each exposure phase and dividing the sum over all phases separately for childhood and adulthood by the total years of passive exposure. Missing data on hours/day for 7.7% of cases and 5.7% of controls were replaced with the mean hours/day of exposed controls for the particular source of exposure. A detailed description of the quantification of lifetime exposure to passive smoke can be found elsewhere [ 20 ]. In the multivariate model, we included several relevant variables that influence breast cancer risk, such as first-degree family history of breast cancer, total duration of breastfeeding, body mass index, average daily alcohol intake, education level, number of full-term pregnancies and menopausal status. Variables that did not change the estimates substantially, such as study region or age at menarche, were not adjusted for in the analyses presented here. Statistical interaction between genotype and smoking variables was tested by using multiplicative interaction terms and evaluated using the likelihood ratio test. We performed the multivariate analyses with stratification in 5-year age groups to ensure sufficient numbers of subjects in the subgroups for genotypes and smoking characteristics. Results The women included in the present study, for whom a comprehensive history of active and passive smoking was available, closely resemble the original study population with respect to the distributions of several sociodemographic characteristics and putative risk factors, such as age, family history of breast cancer, body mass index, education level, parity, menopausal status, alcohol consumption, smoking and breastfeeding (data not shown). Selected characteristics of the present study population are summarized in Table 1 . The mean (± standard deviation) age for breast cancer patients was 42.9 ± 5.5 years and that for control individuals was 42.7 ± 5.6 years. The frequency of the SULT1A1*2 allele was 0.33 among cases and 0.35 among controls (0.32 and 0.34 in the original population). Of cases and control individuals, 52.7% and 57.7%, respectively, were carriers of at least one SULT1A1*2 allele. The distribution of SULT1A1 genotypes was in Hardy–Weinberg equilibrium ( P = 0.92 for control individuals, P = 0.09 for cases). The overall risk for breast cancer among carriers of the SULT1A1*2 allele was not significantly different from that in women with the SULT1A1*1/*1 genotype (adjusted OR 0.83, 95% CI 0.66–1.06). The distributions of potential risk factors, such as first-degree family history of breast cancer, body mass index, alcohol consumption, menopausal status, parity and breastfeeding, were similar in carriers and noncarriers of the SULT1A1*2 allele. There was also no major effect of SULT1A1 genotype in combination with NAT2 acetylator status on breast cancer risk (data not shown). We assessed the effect of SULT1A1 genotype on the association between smoking and breast cancer risk, initially comparing ever active smokers with nonsmokers (i.e. passive-only smokers were included in the reference group). The ORs for variables such as smoking status (current or former active smoker), duration and pack-years of smoking did not differ by SULT1A1 genotype (data not shown). We then considered a separate category of only passively exposed women, with a reference group comprising women with neither active nor passive cigarette smoke exposure (Table 2 ). Associations of breast cancer risk with smoking variables were apparent, but the risk estimates were similar for carriers and for noncarriers of the SULT1A1*2 allele. In the analysis of passive smoking among never active smokers, we observed a tendency toward higher ORs in women with the SULT1A1*1 / *1 genotype compared with carriers of the SULT1A1*2 allele (Table 2 ). The test for interaction between SULT1A1 genotype and passive smoking was not statistically significant ( P = 0.6). We investigated the combined effect of SULT1A1 and NAT2 genotype with respect to smoking, and observed elevated ORs associated with passive smoking only (OR 3.23, 95% CI 1.05–9.92) in NAT2 fast acetylators with the SULT1A1*1 / *1 genotype but not in NAT2 fast acetylators carrying the SULT1A1*2 allele (Table 3 ). There was also a difference in OR for 11 or more pack-years of active smoking by SULT1A1 genotype, but the risk estimates were not significant. The test for interaction between SULT1A1 genotype and active/passive smoking among NAT2 fast acetylators did not reach statistical significance ( P = 0.4). Among NAT2 slow acetylators, risk estimates for active and passive smoking did not differ by SULT1A1 genotype. The results were generally similar when the analysis was restricted to the subgroup of women with premenopausal status, although the confidence intervals were wider. With regard to the differential effect of SULT1A1 and NAT2 genotype, the OR for passive smoking was 2.24 (95% CI 0.68–7.35) in NAT2 fast acetylators with the SULT1A1*1 / *1 genotype and 1.03 (95% CI 0.39–2.71) in fast acetylators carrying the SULT1A1*2 allele. Discussion Our data do not suggest a strong influence of SULT1A1 genotype alone or in combination with NAT2 on the risk for breast cancer. There is no clear evidence that the SULT1A1 Arg 213 His single nucleotide polymorphism investigated in this study in itself is an important effect modifier of breast cancer risk associated with active/passive smoking among women up to age 50 years. Differences in risk estimates for carriers and noncarriers of the SULT1A1*2 allele associated with smoking were apparent among NAT2 fast acetylators but not among slow acetylators. The observed estimates indicated that fast acetylators with the SULT1A1*1 / *1 genotype were at higher risk for breast cancer than were carriers of the SULT1A1*2 allele when exposed to tobacco smoke, with a particularly prominent increase in risk for passive smokers versus never active/passive smokers. We cannot rule out the possibility that the observed risk elevation for SULT1A1*1 / *1 among fast acetylators was due to chance, because confidence intervals were wide for the combined analysis of genotypes. However, it seems biologically plausible that the combination of NAT2 and SULT1A1 'fast' genotypes is unfavourable. Both enzymes have been shown to be capable of bioactivating several pro-carcinogens. NAT2 is thought to play a major role in the activation of N -hydroxy derivatives of HCAs by O -acetylation [ 5 ]. The sulfonation of a variety of xenobiotics or their metabolites, such as polycyclic aromatic hydrocarbons, HCAs and aromatic amines, can lead to short-lived conjugates that may react with DNA and other cellular nucleophiles [ 23 ]. Studies that investigated the genotype–phenotype correlation for SULT1A1 clearly indicate that the SULT1A1*2 allele is associated with decreased catalytic activity of the respective allozyme as compared with the sulfonation activity of the wild-type SULT1A1 enzyme [ 6 , 13 , 14 ]. Consequently, in individuals with this genotype combination, reactive metabolites from acetylation and sulfonation might accumulate and lead to greater DNA damage and increase tumourigenesis. For instance, DNA adducts of 2-amino-3-methylimidazo [4,5- f ]quinoline (IQ) and 2-amino-1-methyl-6-phenylimidazo [4,5- b ]pyridine (PhIP) have been detected in human breast milk [ 24 ]. These HCAs, which are also present in tobacco smoke, have been classified as probably and possibly carcinogenic to humans, respectively [ 25 ]. In mutagenicity assays after heterologous expression of NAT2 and SULT1A1 in Salmonella typhimurium , N -hydroxy-IQ was found to be efficiently activated by NAT2 whereas N -hydroxy-PhIP was specifically activated by the SULT1A1 enzyme [ 26 ]. Likewise, mutagenicity of 2-amino-3-methyl-9 H -pyrido [2,3- b ]indole, another abundant HCA, was strongly enhanced in a Salmonella typhimurium strain expressing SULT1A1 [ 27 ]. Consistent with a role of greater bioactivation of HCAs associated with SULT1A1*1 rather than SULT1A1*2 in carcinogenesis are previous reports of greater risk for breast cancer and prostate cancer associated with intake of well done red meat [ 14 , 15 ]. In accord with this notion are the findings of three studies that investigated the formation of DNA adducts of heterocyclic and aromatic amines [ 28 - 30 ]. All of them found a tendency toward a higher capacity for adduct formation for the SULT1A1*1 enzyme as compared with the *2 allozyme, which is in agreement with a more efficient activation of several pro-mutagens by SULT1A1*1 than by the *2 allelic variant reported by Glatt and coworkers [ 6 ]. Two previous studies [ 15 , 17 ] reported an increased risk for breast cancer associated with the SULT1A1*2 allele per se . Risk estimates were statistically significant only in the study conducted by Zheng and coworkers [ 15 ], a nested case–control study in postmenopausal women, which found an 80% elevated risk for homozygous carriers of the variant allele. We found no association with the SULT1A1*2 variant although our study had 93% power to detect an OR of equivalent magnitude at a significance level of α = 0.05. In a recent case-only study, an interaction between the SULT1A1 polymorphism and tobacco smoke exposure with an OR for interaction of 2.55 (95% CI 1.21–5.36) for current smokers carrying the *2 allele was found [ 18 ]. Results from our case–control study did not provide an indication for such a strong interaction between SULT1A1 genotype and smoking. Accordingly, we failed to detect a significant interaction in a case-only analysis of our data, although the power of our study is similar and the precondition of independence between genotype and exposure in the general population was fulfilled. The ORs (95% CIs) for interaction were 1.04 (0.50–2.14) for passive smoking only, 1.34 (0.63–2.86) for former smoking, and 1.15 (0.55–2.39) for current smoking. We feel confident that our study population is representative of the general German population. The observed allele frequencies for SULT1A1 are in accordance with previous studies conducted in Caucasian populations [ 31 ] and the SULT1A1 genotype distribution did not deviate from Hardy–Weinberg equilibrium. Re-contacting the study participants for the telephone interview might have introduced selection bias. However, the participants in the present study closely resemble the original study population with regard to the distributions of relevant characteristics. Also, we do not believe that recall bias is a major concern because smoking was not known to be associated with breast cancer at the time of the interviews, and the correlation of reported active smoking between the original study and the present study is high [ 20 ]. Moreover, previous studies showed that the validity for self-reported active smoking, as well as passive smoke exposure, is high and nondifferential in cases and controls [ 32 , 33 ]. Although in vitro data suggest that SULT1A1 may not play an important role at physiologically relevant oestrogen concentrations [ 11 , 12 ], the question regarding whether SULT1A1 genotype actually has an effect on oestrogen metabolism in vivo deserves further study. Concerning the expression of SULT1A1 and SULT1E1 enzymes, for instance, there is some controversy in the literature. Falany and coworkers [ 34 , 35 ] observed SULT1E1 expression in normal breast epithelial cells, whereas Williams and coworkers [ 36 ] reported that only SULT1A1 was expressed at detectable levels. Because we cannot definitely rule out a potential role of SULT1A1 in the metabolism of oestrogens, we analyzed our data also with respect to use of oral contraceptives and various reproductive factors that may alter exposure to oestrogens. The results did not provide any indication for a modification of breast cancer risk related to oestrogens by SULT1A1 genotype (data not shown) and corroborate recent evidence indicating that sulfonation of oestrogens catalyzed by SULT1A1 is less relevant in normal breast tissue in physiological conditions [ 11 , 12 ]. The inconsistent findings of previous studies, which also considered the possible involvement of the SULT1A1 gene in other cancer sites (summarized by Glatt and Meinl [ 23 ]), and the broad substrate specificity of the SULT1A1 enzyme indicate the complexity of the issue. Elucidation of the potential effects of SULT1A1 genotype on a hormone-related cancer, such as breast cancer, is rendered more complicated by the fact that smoking may alter oestrogen levels in the body [ 37 - 40 ]. Moreover, and as suggested by our findings, it is possible that SULT1A1 genotype only exerts a detectable effect in combination with other genes, not to mention several polymorphic genes that are involved in oestrogen metabolism. Further determinants such as varying levels of enzyme expression or enzyme induction, which cannot easily be assessed in epidemiological studies, might also be of importance. Nevertheless, we cannot exclude that, because of limitations in statistical power, we were unable to detect a potential weak or moderate association or interaction between SULT1A1 genotype, smoking and breast cancer, independent of NAT2 genotype. Conclusion In summary, the results of our study do not suggest that there is a strong association between the SULT1A1 Arg 213 His genetic polymorphism and risk for breast cancer in women who had developed breast cancer by age 50 years. We did not find any evidence for a significant interaction of SULT1A1 with smoking. The SULT1A1*1 / *1 genotype in combination with NAT2 fast acetylator status, however, appeared to increase breast cancer risk in women exposed to tobacco smoke. Hence, further biochemical investigations and large molecular epidemiologic studies are required to evaluate the effects of multiple genes and exposures on susceptibility to breast cancer. Abbreviations CI = confidence interval; HCA = heterocyclic aromatic amine; IQ = 2-amino-3-methylimidazo [4,5– f ]quinoline; NAT = N -acetyltransferase; OR = odds ratio; PCR = polymerase chain reaction; PhIP = 2-amino-1-methyl-6-phenylimidazo [4,5– b ]pyridine; SULT = sulfotransferase. Competing interests The author(s) declare that they have no competing interests. Authors' contributions CL performed the statistical analysis and drafted the manuscript. AR was responsible for the genotyping assays, and contributed to study design and manuscript preparation. SK conducted the re-contacting of study participants in 1999 and participated in the statistical analyses. JCC conceived the study and supervised the project. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064130.xml

1064131

High expression of focal adhesion kinase (p125FAK) in node-negative breast cancer is related to overexpression of HER-2/neu and activated Akt kinase but does not predict outcome

Introduction Focal adhesion kinase (FAK) regulates multiple cellular processes including growth, differentiation, adhesion, motility and apoptosis. In breast carcinoma, FAK overexpression has been linked to cancer progression but the prognostic relevance remains unknown. In particular, with regard to lymph node-negative breast cancer it is important to identify high-risk patients who would benefit from further adjuvant therapy. Methods We analyzed 162 node-negative breast cancer cases to determine the prognostic relevance of FAK expression, and we investigated the relationship of FAK with major associated signaling pathways (HER2, Src, Akt and extracellular regulated kinases) by immunohistochemistry and western blot analysis. Results Elevated FAK expression did not predict patient outcome, in contrast to tumor grading ( P = 0.005), Akt activation ( P = 0.0383) and estrogen receptor status ( P = 0.0033). Significant positive correlations were observed between elevated FAK expression and HER2 overexpression ( P = 0.001), as well as phospho-Src Tyr-215 ( P = 0.021) and phospho-Akt ( P < 0.001), but not with phospho-ERK1/2 ( P = 0.108). Western blot analysis showed a significant correlation of FAK Tyr-861 activation and HER2 overexpression ( P = 0.01). Conclusions Immunohistochemical detection of FAK expression is of no prognostic significance in node-negative breast cancer but provides evidence that HER2 is involved in tumor malignancy and metastatic ability of breast cancer through a novel signaling pathway participating FAK and Src.

Introduction Breast cancer is a major cause of death among women. Adjuvant systemic therapy may considerably improve survival rates, but it is associated with severe toxic side effects. Especially in patients with node-negative breast cancer, the pros and cons of adjuvant systemic therapy are always critically weighted out. The identification of novel markers for node-negative high-risk patients who would benefit from adjuvant therapy is therefore of major importance. The aim of the present study was to assess the prognostic relevance of focal adhesion kinase (FAK) expression in node-negative breast cancer. In addition, the present report was designed to investigate the consequence of FAK expression on two major downstream targets (namely, Akt and Erk1/2) and to elucidate its connection with the HER2 signaling pathway. The invasion and metastasis of cancer is a complex process including changes in cell adhesion and motility that allow tumor cells to invade and migrate through the extracellular matrix. Some of these alterations may occur at focal adhesions, which are cell/extracellular matrix contact points containing membrane-associated, cytoskeletal and intracellular signaling molecules [ 1 ]. The survival of normal epithelial cells critically depends on cell–cell and cell–matrix contact. Without these contacts epithelial cells die through the controlled process of apoptosis, termed anoikis [ 2 ]. FAK is a tyrosine kinase considered a central molecule in integrin-mediated signaling, and it is involved in cellular motility and protection against apoptosis [ 3 - 7 ]. The importance of FAK in epithelial cell biology is underlined by the findings that epithelial cells lines expressing constitutively active FAK survive in suspension and that cells derived from FAK -/- mouse embryos exhibit reduced migration [ 8 , 9 ]. High levels of FAK have been found in a variety of tumors, including head and neck carcinomas, ovarian carcinomas, thyroid carcinomas and colon carcinomas [ 10 - 12 ]. Only two studies have so far described elevated FAK expression in human breast cancer and preinvasive lesions in contrast to benign breast lesions in small cohorts [ 13 , 14 ]. The prognostic value of FAK expression in breast cancer remains unclear. Regarding other cancer types, three studies demonstrated varying results in colon cancer, squamous cancer and hepatocellular carcinoma [ 15 - 17 ]. Recent studies have identified HER2, a prognostic marker for aggressiveness in breast cancer, to influence the migratory behavior of breast cancer cells in vitro through a novel signaling pathway involving phosphorylation of FAK at tyrosine 861 by its upstream kinase c-Src [ 18 , 19 ]. These findings suggest that the Her-2/neu signaling pathway may influence migration and metastasis of breast cancer cells through activating the Src/FAK signaling pathway. There are several downstream targets of FAK such as the Ras-ERK signaling cascade, which is activated by extracellular, frequently mitogenic ligands and results in increased cellular proliferation and malignancy in vitro and in vivo [ 20 , 21 ]. FAK also impacts on the phosphotidylinositol-3-kinase (PI3K)/Akt-signaling pathway that plays a central role in tumorigenesis [ 22 , 23 ]. In node-negative breast cancer, we have previously shown that activated Akt is a prognostic parameter [ 24 ]. The aim of the present study was to assess the prognostic relevance of FAK expression in node-negative breast cancer. In addition, the present report was designed to investigate the correlation of FAK expression with two major downstream signaling pathways, Akt and Erk1/2, and to elucidate its connection with the HER2 signaling pathway. Materials and methods Patients This study comprised 162 female breast cancer patients (mean age, 59 years) from the Department of Gynecology, University of Essen-Duisburg, Germany, who underwent surgery and further adjuvant therapy between 1989 and 1996. Complete clinical records and follow-up information were available in all cases. The negative lymph node status was confirmed by axillary dissection. All surgical material was fixed in 4% formalin and routinely processed. The tumors were classified according to the pTNM system (sixth edition) and were graded according to Elston and Ellis [ 25 ]. Twenty-seven of the 162 patients died during follow-up, the cause of death being unknown in two cases. Altogether 19 patients died of breast cancer, and six patients were excluded from survival analysis having died from either benign or other cancer diseases. Statistical analysis was based on a mean follow-up period of 7.48 years. Table 1 summarizes the clinicopathological parameters of this study. Immunohistochemistry The primary antibodies used in this study are presented in Table 2 . Immunohistochemistry was performed on paraffin sections 5 μm thick. The alkaline phosphatase anti-alkaline phosphatase method was used for antibody demonstration. Antigen retrieval was carried out with 0.01 M citrate buffer at pH 6.1 for 20 min (both phospho-Akt antibodies), for 40 min (phospho-ERK1/2) and for 20 min (FAK, phospho-Src Tyr-215 and phospho-Src Tyr-416) in a hot water bath (95°C). Antibodies were incubated overnight in a humidified chamber at 4°C. Positive controls were included in each staining series. No significant staining was observed in the negative controls using mouse immunoglobin replacing the primary antibody. The DAKO HerceptTest™ (DakoCytomation, Glostrup, Denmark) and a HER2/ErbB2 polyclonal antibody at a 1:50 dilution (rabbit polyclonal antibody; Cell Signaling Technology, Beverly, MA, USA) were used for the detection of HER2 protein expression. The estrogen receptor (ER) status of the tumors was determined using a monoclonal anti-human antibody as previously described [ 24 ]. Evaluation of immunostaining FAK immunostaining The constant positive staining of smooth muscle cells of tumor vessels served as the positive internal control as previously described [ 26 ]. If more than 20% of carcinoma cells in a given specimen were stained more intensely than smooth muscle cells, the sample was classified as strong FAK overexpression (3+). In the case of equal FAK immunostaining compared with that of vascular smooth muscle cells, the sample was classified as intermediate expression (2+). When FAK immunostaining was weaker than that of the internal control, the tumor was classified as low FAK expression (1+). Tumors lacking FAK immunostaining were classified as negative (0). For statistical analysis, negative (0), intermediate (1+ and 2+) and strong (3+) groups were created. Phospho-Akt and phospho-ERK1/2 immunostaining The number of positive immunoreactive tumor nuclei was counted within 300 cells at the invasive tumor front, and is expressed as a percentage. Cytoplasmatic phospho-Akt staining was noticed but was not part of the scoring system. Tumor cells with easily detectable specific phospho-ERK1/2 immunostaining, independent of the amount of stained cells, were scored as strongly positive (2+). Tumors exhibiting a detectable but faint immunostaining were scored as weak (1+), whereas tumors with a minimal, hardly detectable or missing staining pattern were classified as phospho-ERK1/2-negative (0). Phospho-Src Tyr-215 and phospho-Src Tyr-416 immunostaining Most tumor samples showed a membranous staining with varying intensity. Four samples revealed a strong cytoplasmatic staining. Nuclear staining was absent. Tumor samples with a readily detectable membranous staining or with a strong cytoplasmatic staining were classified as positive. Samples lacking or with a weak membranous staining or with a weak cytoplasmatic staining were classified as negative. HER2 protein and ER status The Her-2/neu status was determined according to the evaluation system of the HerceptTest™. For statistical analysis, negative (DAKO score 0 and 1+) and positive (DAKO score 2+ and 3+) groups were created. A tumor was regarded as ER-negative if none or less than 10% of the tumor cells showed weak or missing nuclear immunostaining. Western blot analysis Total cell protein of seven representative frozen human breast cancer samples was extracted in lysis buffer and the protein concentration was measured to load equal protein amounts. Each lane consisted of 50 μg protein, and was size fractioned by electrophoresis on 10% polyacrylamide-SDS gels and electrotransferred to a polyvinylidene fluoride membrane using a tank-blotting system. After staining the membranes with reversible Ponceau red solution to confirm equal protein loading, the membranes were blocked with 5% non-fat dry milk in Tris-buffered saline for 1 hour at room temperature. The membranes were then incubated with the anti-HER2/ErbB2 antibody (polyclonal rabbit HER2/erbB2 antibody, 1:1000; Cell Signaling Technology, Inc.), anti-phospho-FAK antibody (Tyr-861, polyclonal goat antibody, 1:50; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and anti-phospho-Src antibody (Tyr 215, polyclonal rabbit antibody, 1:2000; Sigma-Aldrich, Inc., St Louis, MO, USA) for 16 hours at room temperature. After incubating the membrane for 1 hour with goat anti-rabbit/donkey anti-goat antibody conjugated to horseradish peroxidase, the antigen–antibody complex was visualized by chemiluminescence (Amersham Biosciences, Piscataway, NJ, USA). Statistical analysis FAK, phospho-Erk1/2, Her-2/neu, phosph-Src Tyr-215/416 and ER immunostainings were assessed by a set of pathologists (KJS, FG, JWO, FO) in a blind-trial fashion without knowledge of the clinical outcome. In cases of disagreement, the slides were evaluated by two pathologists and a final decision was made. Two of the authors assessed the percentage of phospho-AKT-positive tumor cells and the mean percentage of both counts was used for statistical analysis. Staining scores of both phospho-Akt antibodies (Santa Cruz Biotechnology Inc. and Cell Signaling Technology Inc.) revealed a high, significant interobserver concordance (Pearson's r = 0.671, P < 0.001). Results obtained from phospho-Akt immunostaining performed with the antibody from Santa Cruz showed less background staining, and were therefore used for statistical analysis. All data were converted to a PC and statistically analyzed using SPSS Version 10 for Windows (SPSS Inc., Chicago, IL, USA). Relationships between ordinal parameters were investigated using two-tailed chi-squared analysis (or Fisher's exact test where patient numbers were small). The relationship between FAK expression and the percentage of phospho-Akt-positive tumor cells was determined using the Kruskal–Wallis test. Overall survival curves were estimated using the Kaplan–Meier method, and any differences in the survival curves were compared by the log-rank test. Western blot results were digitized with Image Master software from Amersham Biosciences. Results Immunohistochemical analysis of FAK Staining results were obtained from all 162 cases. In normal breast epithelium, a weak to missing FAK expression was detected at the cell membrane and in the cytoplasm. Strong FAK overexpression was detected in 30 patients (18.5%), intermediate staining in 119 patients (73.5%) and negative staining in 13 patients (8%). Higher FAK expression was significantly associated with a poorer differentiation grade. No statistically significant differences were found among cases with different tumor sizes or different tumor types (Table 1 ). The staining intensity of an adjacent in situ component was equal to or stronger than that of the invasive tumor, with only one sample exhibiting a weaker FAK staining in the in situ component than in the corresponding invasive carcinoma (Fig. 1 ). Correlation between FAK expression and survival The impact of FAK expression and clinicopathological features on patient survival was assessed using univariate Kaplan–Meier survival analysis. Except for histological grading ( P = 0.005), ER status ( P = 0.0033) and activation of Akt ( P = 0.0383), none of the other investigated parameters – including FAK expression (irrespective of ER expression) or HER-2/neu status – proved to be of prognostic significance in the node-negative breast cancers examined. Correlation of FAK protein expression with HER-2/neu status, ER status and phospho-Src Tyr-215 and phospho-Src Tyr-416 overexpression A total of 159 tumors were analyzed for HER-2/neu protein expression: 59 tumors (37.1%) were classified as negative (0), 56 tumors (35.2%) as negative (1+), 21 tumors (13.2%) as weakly positive (2+) and 23 tumors (14.5%) as strongly positive (3+). Tumors with a score of 2+ and 3+ were defined as Her-2/neu-positive. A total 51.7% of tumors classified as strongly FAK-positive exhibited Her-2/neu protein overexpression, but intermediate classified tumors totaled only 24.5% of Her-2/neu overexpressing tumors. All of the FAK-negative tumors were Her-2/neu-negative. The ER status was obtained from 152 tumors. Ninety tumors (59.2%) showed a significant ER expression whereas 62 tumors (40.8%) were classified as negative. Statistical analysis revealed a significant relationship of elevated FAK expression ( P = 0.001) with Her-2/neu overexpression (2+ and 3+; Table 3 ) but not with the ER status. Phospho-Src Tyr-215 staining was obtained from 137 patients and phospho-Src Tyr-416 staining was obtained from 140 samples. Normal breast tissue exhibited no phospho-Src Tyr-215 immunostaining. Membranous phospho-Src Tyr-416 immunostaining was noticed in sporadical epithelial cells. Myoepithelial cells lacked both phospho-Src Tyr-215 and phospho-Src Tyr-416 immunostaining. Tumor tissue of the remaining samples was not available due to lack of paraffin material. Statistical analysis revealed a significant direct correlation of FAK overexpression with phospho-Src Tyr-215 staining results ( P = 0.021) but not with phospho-Src Tyr-416 staining scores (Table 3 ). On the other hand, positive phospho-Src Tyr-215 staining was significantly associated with HER2 overexpression ( P = 0.011) whereas phospho-Src Tyr-416 was not. Representative immunohistochemical stainings are shown in Figs 2 and 3 . Correlation between FAK expression and Akt/ERK activation Normal breast tissue revealed mostly a cytoplasmatic Akt-staining pattern, with a low mean percentage of positive nuclear stained cells in normal breast tissue of 28%. Prominent nuclear and partly cytoplasmatic phospho-Akt immunoreactivity was noticed in tumor cells. Staining heterogeneity was occasionally apparent at the invasive tumor front. The percentage of phospho-AKT-positive nuclei in the tumor samples ranged from 0% to 85%. Tumors with more than 45% of positive nuclei were classified as phospho-Akt-positive. Erk1/2 immunohistochemistry of tumor samples revealed strong nuclear immunostaining and partly cytoplasmatic immunostaining, while non-neoplastic tissue only occasionally revealed a weak staining. Similar to phospho-Akt, heterogeneous immunostaining for phospho-ERK1/2 was detected at the invasive tumor front. Phospho-Akt-positive classified tumors exhibited stronger FAK immunostaining intensities more frequently ( P < 0.001) than phospho-Akt-negative classified tumors. No correlation was observed between different FAK immunostaining scores and phospho-ERK1/2 immunostaining scores ( P = 0.108) (Table 3 ). Statistical analysis revealed a significant direct correlation of different FAK immunostaining intensities with the percentage of phospho-Akt-positive stained tumor cells ( P < 0.001). Correlation between HER2 protein overexpression and phosphorylation of FAK at Tyr-861 in western blot analysis We next analyzed the relationship of HER2 overexpression with the levels of phospho-Src Tyr-215, phospho-FAK Tyr-861 and total FAK using seven samples of fresh-frozen breast cancer tissue. Of the seven samples analyzed, four tumors with high HER2 levels exhibited increased levels of phosphorylation of FAK on tyrosine 861 compared with tumors without HER2 overexpression ( P = 0.01; Fig. 4 ). An increase of total FAK and phospho-Src Tyr-215 was noticed within the HER2 overexpressing tumors but reached not statistical significance. These results provide evidence that the HER/FAK Tyr-861 pathway is activated in human breast cancer. Discussion Several studies have demonstrated upregulation of FAK expression in malignant tumors, including breast cancer, but the prognostic value of FAK expression in human cancer remains unclear. From the three studies that have previously analyzed this, one study denies the prognostic value of FAK in adenocarcinomas of the colon [ 16 ], whereas the other two studies have provided evidence for the prognostic value of immunohistochemically detected FAK expression in esophageal squamous carcinoma and in hepatocellular carcinoma, respectively [ 15 , 17 ]. The present study is the first to clarify the prognostic relevance of FAK expression in a cohort of 162 invasive node-negative breast cancer cases with long-term follow-up. Our findings demonstrate that FAK overexpression does not predict patient outcome in this setting; although, similar to esophageal squamous cancers, levels of FAK expression were strongly correlated with poorer tumor differentiation [ 17 ]. FAK is a nonreceptor protein kinase involved in integrin-mediated signaling that has a profound impact on cell proliferation, survival and migration. One downstream target is the PI3K/Akt signaling pathway. In the present study we found a significant association of elevated FAK expression with Akt phosphorylation in support of the notion that Akt may be a downstream target of FAK-mediated signaling [ 27 ]. However, while Akt activation is associated with a worse prognosis in our study cohort, FAK expression levels are not. This may be due to the central role of Akt in tumorigenesis, where a number of stimuli and pathways besides FAK contribute to the common end point of increased Akt activity. Akt, also known as protein kinase B, is a serine/threonine protein kinase that has been shown to regulate cell survival signals in response to a diversity of signals generated by growth factors, cytokines and oncogenic ras. Alternative activation mechanisms for Akt besides growth factor-mediated PI3K stimulation are mutational inactivations of PTEN or activation via integrin-linked kinase [ 28 - 30 ]. The difference in prognostic relevance between Akt and FAK may therefore not be surprising. Several studies have demonstrated upregulation of FAK in human cancer including breast cancer and have suggested that FAK overexpression is an early event in tumorigenesis [ 13 , 14 ]. Our data support this hypothesis as all (but one) of the in situ carcinomas adjacent to the invasive cancer exhibited FAK overexpression. Interestingly, in every cancer specimen analyzed the adjacent preinvasive component exhibited an equal or even higher FAK immunostaining than the corresponding invasive carcinoma. In recent studies, Vadlamudi and colleagues utilized human breast cancer cell lines in vitro to establish a novel signaling pathway involving HER2, phospho-Src Tyr-215 and phospho-FAK Tyr-861 leading to increased cellular motility [ 18 , 19 ]. The authors showed that heregulin-induced HER2 activation resulted in phosphorylation of FAK at tyrosine 861, while six breast tumor samples exhibited increased level of phospho-Src Tyr-215 in HER2 neu-overexpressing breast cancers. From these in vitro data it was tempting to hypothesize that HER2 influences metastasis of breast cancer in vivo via a similar pathway. To test for the existence of this novel pathway in a large cohort of human breast cancer tissue we evaluated total FAK protein and phospho-Src Tyr-215 expression as well as phospho-Src Tyr-416 expression by immunohistochemistry. Additionally, we investigated the levels of phospho-FAK Tyr-861, phospho-Src Tyr-215 and Her-2/neu in a set of seven of freshly frozen breast cancer tissues using western blot analysis. Our findings support the hypothesis that HER2-overexpressing tumors exhibit higher levels of phospho-FAK Tyr-861 but also of total FAK levels. This is also supported by the immunohistochemical data where Her-2/neu protein expression levels coincide with FAK protein expression. Interestingly, increased levels of phospho-Src Tyr-215 but not of phospho-Src Tyr-416 were significantly associated with HER2 overexpression, supporting the role of an activatory Src Tyr-215 phosphorylation as an intermediate between HER2 and FAK signaling. Although Src comprises several phosphorylation sites, phosphorylation of tyrosine 215 and not tyrosine 416, the 'classical' activatory phosphorylation site, appears to be essential for FAK activation via HER2 [ 19 ]. Our data confirm the specificity of this pathway as src phosphorylation at tyrosine 416 was not associated neither with HER2 overexpression. The present study was designed to investigate this novel signaling pathway in vivo via an immunohistochemical approach allowing the determination of HER2 and FAK protein expression. A recent study identified frequent polysomic patterns for chromosome 1, chromosome 8 and chromosome 17 that are indicative for increased tumor malignancy in breast cancer [ 31 ]. As the focal adhesion kinase and the Her-2/neu gene are located on chromosome 8 and chromosome 17, respectively, one might conclude that such a polysomic pattern causes a combined HER2/FAK protein overexpression. However, the identification of potential underlying causative genetic alterations should be a topic of further investigation. Our results propose that among the many biological effects of Her-2/neu overexpression the activation of the Akt survival pathway via the FAK Tyr-861–Src Tyr-215 pathway might contribute to the more aggressive behavior of Her-2/neu-overexpressing breast cancers. In summary, FAK overexpression is not a prognostic marker in this series of 162 node-negative breast cancers. This might be due to the composition of our cohort, which mainly contains early-stage cancers without lymph node metastasis. Recent studies demonstrated a higher level of FAK expression in colorectal liver metastasis than in the primary tumor, and provided evidence for an upregulation of FAK protein in hepatocellular cancers with portal invasion [ 15 ]. Nevertheless, FAK protein overexpression coincides with Her-2/neu overexpression and may mediate HER2 signaling via Src, resulting in PI3K/Akt activation. These data might contribute to the understanding of how Her-2/neu overexpression in human breast cancer affects tumor malignancy and metastasis (Fig. 5 ). Owing to the relatively small number of breast cancer samples exhibiting FAK protein overexpression in this cohort, however, these promising results need to be confirmed in larger cohorts. Abbreviations ER = estrogen receptor; ERK = extracellular regulated kinase; FAK = focal adhesion kinase; PI3K = phosphotidylinositol-3-kinase. Competing interests There has been no prior publication of the content of this study. This paper has not been submitted to any other journal. No financial relationship exists that might lead to conflict concerning the content of this study. The paper has been read and approved by all of the authors. The authors declare that they have no competing interests. Authors' contributions KJS carried out the primary immunohistochemical staining evaluation, performed statistical analysis and wrote the article. FG, FO and JW carried out the secondary immunohistochemical staining evaluation to rule out interindividual observer discrepancies. RC and RK provided the clinical data. BL contributed substantially to the conception and design of the study and to the drafting of the manuscript. KWS performed critical revision of the manuscript for important intellectual content. HAB provided administrative, technical and material support and supervision, and analyzed western blot data. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064131.xml

1064132

Downregulation of the anaphase-promoting complex (APC)7 in invasive ductal carcinomas of the breast and its clinicopathologic relationships

Introduction The anaphase-promoting complex (APC) is a multiprotein complex with E3 ubiquitin ligase activity, which is required for the ubiquitination of securin and cyclin-B. Moreover, the mitotic spindle checkpoint is activated if APC activation is prevented. In addition, several APC-targeting molecules such as securin, polo-like kinase, aurora kinase, and SnoN have been reported to be oncogenes. Therefore, dysregulation of APC may be associated with tumorigenesis. However, the clinical significance and the involvement of APC in tumorigenesis have not been investigated. Methods The expression of APC7 was immunohistochemically investigated in 108 invasive ductal carcinomas of the breast and its relationship with clinicopathologic parameters was examined. The expression of APC7 was defined as positive when the summed scores of staining intensities (0 to 3+) and stained proportions (0 to 3+) exceeded 3+. Results Positive APC7 expression was less frequent than its negative expression when histologic ( P = 0.009) or nuclear grade ( P = 0.009), or mitotic number ( P = 0.0016) was elevated. The frequency of APC7 negative expression was higher in high Ki-67 or aneuploid groups than in low Ki-67 or diploid groups. Conclusion These data show that loss of APC7 expression is more common in breast carcinoma cases with poor prognostic parameters or malignant characteristics. They therefore suggest that dysregulation of APC activity, possibly through downregulation of APC7, may be associated with tumorigenesis in breast cancer.

Introduction The anaphase-promoting complex (APC) is an E3 ubiquitin ligase that controls mitotic progression [ 1 , 2 ]. APC is a polymeric protein complex composed of at least 11 subunits, which contains tetratricopeptide repeat proteins (APC3, 5, 6, 7, and 8), a cullin homolog (APC2), and a ring-H2 finger domain (APC11). APC requires two WD40 repeat-containing coactivators, Cdc20 and Cdh1, to recruit and select various substrates at different stages of the cell cycle, and it was recently suggested that APC3 and APC7 interact with these APC activators [ 3 ]. APC promotes metaphase/anaphase transition by ubiquitizing and degrading securin, an inhibitor of separase that participates in the degradation of the chromatic cohesion complex. APC also ubiquitinates cyclin-B and accelerates its degradation during the late mitotic to the G 1 phase, which results in mitotic exit. In addition, APC is known to target various cell cycle regulatory molecules, including spindle-associated protein, DNA replication inhibitors, and mitotic kinases. Several molecules targeted by APC have been reported to promote transformation. Pituitary tumor-transforming gene (PTTG), a vertebrate analog of securin, has been reported to be an oncogene [ 4 ], and cancerous tissues from patients with leukemia, lymphoma, or testicular, ovarian, breast, or pituitary cancer were found to over-express PTTG [ 5 - 7 ]. It was further reported that the constitutive expression of polo-like kinase (PLK), a serine/threonine kinase that is involved in spindle formation, centrosome cycles, and chromosome segregation [ 8 ], may induce tumor formation [ 9 ]. Several reports have suggested a role for PLK in the progression and/or malignancy of human cancers, such as glioma, and endometrial carcinoma, breast, ovarian, and esophageal carcinoma [ 10 - 13 ]. Aurora kinase, another serine/threonine kinase that is involved in chromosome segregation and centrosome maturation [ 14 ], has also been reported to be amplified in bladder, gastric, breast, and colorectal cancers [ 15 - 18 ] and to have the ability to transform NIH3T3 cells [ 19 ]. Recently, SnoN, a negative regulator of Smad that is involved in the transforming growth factor-β signaling pathway, was shown to be a target molecule for the APC [ 20 , 21 ] and to have transforming potential [ 22 ]. It was also found that SnoN is amplified in stomach, thyroid, and lung carcinoma and lymphoma [ 23 ]. APC-regulating molecules have also been reported to be involved in transformation. RASSF-1A and Mad2, which inhibit APC activity, were reported to be tumor suppressors [ 24 , 25 ]. Chromosome instability is believed to contribute to malignant transformation because the majority of malignant human cancers exhibit chromosomal gain or loss [ 26 ] and because mitotic defects including chromosome aberrations are frequently found in malignant cancers [ 27 - 29 ]. Because of the roles played by APC in mitotic cell cycle progression, the timely activation of APC is thought to be important for maintaining accurate chromosome separation. In addition, a report indicating that the mitotic spindle checkpoint was reached by preventing APC activation [ 30 ] suggests that the dysregulation of APC may give rise to abnormal chromosome segregation, resulting in aneuploidy. The recent finding that APC5 deficiency in Drosophila is accompanied by a mitotic defect, which included aneuploidy, suggests a role for APC in the maintenance of chromosome stability [ 31 ]. It can therefore be hypothesized that the abnormal regulation of APC may be involved in malignant transformation through chromosome instability. However, it is not known whether the abnormal regulation of APC, possibly through genomic mutation or the modulation of APC components, is related to tumorigenesis. Furthermore, whether dysregulation of APC is related to clinical parameters in various human cancers is yet to be determined. Thus, we investigated immunohistochemically the levels of APC7 in various cancer tissues and found weak APC7 expression in high-grade ductal carcinomas of breast. Therefore, we were encouraged to investigate the expression of APC7 in 108 breast carcinomas and to examine the relationship between the expression of APC7 and clinicopathologic parameters. Methods Production of polyclonal antibodies against APC7 Polyclonal antibodies against mouse APC7 were raised in a NZW rabbit by immunization with recombinant APC7 protein. Briefly, recombinant mouse APC7 proteins were produced in Escherichia coli using a pET32 expression vector system (Novagen, Madison, WI, USA). The resulting 6× histidine-tagged APC7 proteins were purified by Ni-NTA affinity chromatography (Qiagen, Hilden, Germany). A NZW rabbit was then immunized with the purified APC7 protein and boosted twice. Blood was collected from the auricular artery, and serum was prepared by clotting and differential centrifugal separation (10,000 g for 10 min). APC7-specific antibodies were further purified by binding serum to APC7-coupled nitrocellulose and eluting with 100 mmol/l glycine–HCl buffer (pH 2.5). Immunoblotting and immunoprecipitation Protein extracts were prepared by solubilizing cells in RIPA buffer (150 mmol/l NaCl, 1% NP40, 0.5% deoxycholate, 0.1% sodium dodecylsulfate, 50 mmol/l Tris.Cl, pH 7.5, protease inhibitors) and differential centrifugation (10,000 g for 10 min). Of the protein fractions obtained, 30 μg was resolved by 12% SDS-PAGE, and then the separated proteins were electrotransferred onto Immobilon membranes (Millipore, Bedford, MA, USA). After preblocking these membranes with 5% skimmed milk, they were treated with anti-mouse APC7 or human APC7 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) antibodies as primary antibody and horseradish peroxidase-conjugated anti-rabbit antibodies as secondary antibody. Immunoreactive bands were developed using an electrogenerated chemiluminescence system (Amersham Pharmacia Biotech, Uppsala, Sweden). Immunoprecipitation was carried out with anti-human APC3 antibodies (Transduction Laboratory, San Diego, CA, USA) or anti-mouse APC7 antibodies. Subconfluent cells were collected and then lysed by incubation on ice for 15 min in EBC buffer (0.5% Nonidet P-40, 40 mmol/l Tris.HCl, pH 8.0, 120 mmol/l NaCl, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 10 μg/ml PMSF). Cell lysates (1 mg) collected by differential centrifugation (12,000 g for 15 min) were mixed with 1 μg anti-human APC3 antibodies or anti-APC7 antibodies, and these mixtures were then further incubated for 1 hour at 4°C. Immune complexes were collected by incubating with 30 μl of 50% protein A-sepharose slurry for 1 hour and centrifugation (10,000 g for 15 s). After washing three times with ice-cold EBC buffer, the pellets were suspended in 2× loading buffer (125 mmol/l Tris.HCl, pH 6.8, 4% SDS, 20% glycerol, 14.4 mmol/l 2-mercaptoethanol) and boiled for 5 min. Immunoblotting was carried out with anti-human APC3, anti-human APC6 (Santa Cruz Biotechnology), or anti-mouse APC7 antibodies. Tissue samples Paraffin wax embedded blocks containing breast tumor tissues resected from 108 patients diagnosed as having invasive ductal carcinoma of breast at Wonju Christian Hospital (Wonju, Korea) between January 1996 and May 2001 were used in this study. Patient ages ranged from 24 to 81 years (mean 47.5 years). All procedures were performed in accordance with our hospital's ethical guidelines, and approval for the study was granted by the university hospital's ethics committee. All patients provided informed consent. Pathologic examination Hematoxylin–eosin stained slides were reviewed, and histologic grade was determined in terms of tubule formation, nuclear pleomorphism and mitosis, using the criteria described by Bloom and Richardson [ 32 ]. Tumor size, lymphatic metastasis, and clinical stage were determined according to the American Joint Committee on Cancer criteria [ 33 ]. Immunohistochemistry and evaluation Specimens were fixed in 10% buffered formaldehyde and embedded in paraffin using routine methods. Sections 5 μm thick were placed on silane-coated glass slides, dried at 50°C for 2 hours, deparaffinized in xylene, rehydrated in graded ethanol, and then washed in distilled water. To retrieve antigenicity, the sections were dipped in citrate buffer (10 mmol/l, pH 6.0) in a tender cooker (Nordic Ware, Minneapolis, MN, USA) and then warmed for 15 min in a microwave oven. Endogenous peroxidase activity was blocked by pretreating with 0.3% hydrogen peroxide for 10 min. After washing with 50 mmol/l Tris buffer (pH 7.5), primary antibodies, namely anti-mouse APC7, human APC7, human Ki-67 (DAKO, Copenhagen, Denmark), or estrogen receptor (ER) antibodies (Novocastra, Newcastle, UK), were applied overnight at a dilution of 1:50 or 1:100. The sections were then further incubated for 20 min in a 1:50 dilution of biotinylated goat anti-rabbit or rabbit anti-mouse antibody (DAKO) as secondary antibody. Color was developed by incubating with streptavidin peroxidase (DAKO) for 20 min and staining with 3-amino-9-ethylcarazole. Counter-staining was carried out with hematoxylin before mounting. To obtain relevant staining equivalence of APC7 in different carcinoma tissues, an unstained tissue sample and a strongly stained tissue sample were used as negative and positive control, respectively. Whenever a staining procedure was performed, negative and positive control tissues were simultaneously stained with new battery of tissues and then the control tissues were used as a staining reference. All slides were examined by three pathologists and scores were determined by consensus. The immunohistochemical intensity of APC7 was awarded an 'intensity' score of 0 to 3+, with 0 represented an unstained nucleus and 3+ the strongest staining intensity. The 'proportion' score represented the estimated percentage of stained cells as a fraction of all tumor cells in the microscopic field (i.e. 0 = 0%, 1+ = 0–25%, 2+ = 25–50%, and 3+ = >50%). Because true negative expression with a staining intensity of 0 was very rare (six cases/108 tissue samples), and therefore the statistical significance between APC7 expression and clinicopathologic parameters could not obtained, we included a weak APC7 expression group (34 cases/108 tissue samples) with staining intensity of 1 and proportion score of 1 in the negative group. Therefore, a summed intensity and proportion score of ≥ 3+ was defined as positive APC7 expression whereas a score of ≤ 2+ was defined as negative. The Ki-67 labeling index was defined as the percentage of positively stained cells in five to seven high power fields (×400). At least 1000 cells per field were counted. Nuclear ER staining was also examined at ×400 and compared with a strong positive control. ER staining intensity was designated weak, moderate, or strong. Positive reactivity was defined when the proportion of cells exhibiting moderate to strong staining exceeded 10%. DNA analysis Two 50-μm sections were cut from each paraffin block, deparaffinized in xylene, rehydrated in a descending ethanol series, and then washed in phosphate-buffered saline. The sections were then placed in 10 mmol/l citrate solution (pH 6.0) and incubated for 2 hours at 80°C. After cooling, 1 mg/ml pepsin in 0.1 N HCl was added and the sections were digested for 30 min. The resulting suspension (2 × 10 6 nuclei) was filtered through 50 μ-mesh and further suspended in 500 μl of 1% bovine serum albumin solution. DNAs were stained using a Cycle TEST PLUS DNA Reagent Kit™ (Becton Dickinson, Ontario, Canada). Stained cells were analyzed using a FACscan (Becton Dickinson, San Jose, CA, USA) and the fraction of aneuploid cells was calculated using Cell Fit software (Becton Dickinson). Statistical analysis Statistical analysis was performed using the SPSS ver. 10.0 program (SPSS Inc., Chicago, IL, USA). The association between APC7 expression and clinicopathologic parameters was analyzed using χ 2 tests. P ≤ 0.05 was considered statistically significant. Results Characterization of polyclonal antibodies against APC7 In this study we isolated a novel gene (GenBank Accession Number: AF076607) and identified it as the mouse APC7 gene (GenBank Accession Number: BC006635). It was found to have 97.7% homology with its human counterpart (GenBank Accession Number: AF191340). Polyclonal antibodies were raised by immunizing a NZW rabbit with recombinant mouse APC7 proteins (amino acids 88–565), and the APC7-specific antibodies so obtained were then purified by affinity binding to APC7-coupled nitrocellulose. Immunoblotting analysis of MCF-7 human breast carcinoma extracts showed that these purified antibodies and human APC7 antibodies (sc-20987; Santa Cruz) recognized a distinct 63-kDa band, and that this immune reactivity was APC7-specific (Fig. 1A , panels a and b). The antibodies recognized same sized antigens from mouse and human cells (Fig. 1A , panel c). Moreover, APC7 antibodies precipitated both APC3 and APC6 components in mouse and human derived cells, whereas human CDC27 (APC3) antibodies precipitated APC6 and APC7 components (Fig. 1B ), thus demonstrating that the antigen recognized by our purified antibody is the APC7 component of the APC. Immunohistochemical studies on the paraffin-embedded sections of normal breast tissues showed that the purified APC7 antibodies recognized antigens located in the nucleus (Fig. 1C , panel a), which is in accordance with the finding that most APC antigens are localized in nucleus during the interphase [ 33 ]. Immune reactivity, as shown by immunohistochemistry, was also APC7 specific (Fig. 1C , panel b). Expression of APC7 in various human tissues To search for differentially expressed APC7 in normal and cancerous tissues, we performed immunohistochemical analyses with the purified mouse APC7 antibodies using tissue array slides containing 50 normal or 50 tumor tissue cores. We compared the APC7 expressions of the cores by assessing the averaged staining intensities (0 to 3+). Staining of ≥ 2+ was defined as positive expression and of ≤ 1+ as negative expression. Table 1 lists the APC7 expression of 17 normal and 22 tumor tissues with multiple cores. Positive staining was observed in rapid growing normal epithelial tissues. In contrast, slow growing but more differentiated tissues such as skeletal muscle, adipocytes, spinal cord, brain, and basal stromal tissues near epithelial cells exhibited no or weak immune reactivity to APC7. In addition, slowly growing tumors such as chondrosarcomas, lipomas, low-grade urothelial carcinomas, and renal cell carcinomas tended to show weak reactivity to APC7, whereas most tumor tissues with high proliferation rate were positive. Interestingly, some ductal carcinomas of the breast with an undifferentiated high histologic grade exhibited weak reactivity to APC7. Relationship between APC7 expression and clinicopathologic parameters in breast carcinomas To determine whether the loss of APC7 expression is related to tumorigenesis in breast cancer, we scrutinized the expression level of APC7 immunohistochemically in 108 invasive ductal carcinomas of the breast and then searched for correlations with clinocopathologic parameters. Figure 2A shows the representative features of APC7 staining scores of 0 to 3+, and Fig. 2B shows the immunoblotting results for three representative tissues with different intensity scores. These data show that APC7 staining intensity was proportional to band intensity by immunoblotting, demonstrating that immunohistochemical staining intensities represent APC7 expression. Typical negative and positive APC7 expressions in breast carcinoma are shown in Fig. 2C panels a and b. The ratios of APC7-positive to APC-negative expression and their relationships with various clinicopathologic parameters are summarized in Table 2 . Of breast carcinomas, 63% exhibited positive APC7 expression and 37% were negative. APC7 expression did not correlate with tumor size ( P = 0.8180) or metastasis ( P = 0.9703). No statistical significance was found between clinical stage (0.2798) and APC7 expression, or between ER expression (0.1031) and APC7 expression. Nevertheless, the frequency of positive APC7 expression tended to be lower in clinical stage III (58.3%) than in stage I (78.9%) tumors, and in patients who were ER negative (52.9%) than in those who were ER positive (70.6%). In contrast, negative APC7 expression was highest in stage III tumors and in those who were ER negative. On the other hand, the negative expression of APC7 was positively correlated with higher histologic grade ( P = 0.001), nuclear grade ( P = 0.0090), mitotic number ( P = 0.0016), Ki-67 index ( P = 0.0078), and aneuploidy ( P = 0.0095). Moreover, APC7 expression was more frequent in those with a low histologic grade (84.8%) than in those with a high grade (35.0%). Because histologic grade is determined by nuclear pleomorphism, mitotic number, and tubule formation, those with a high nuclear grade and a high mitotic number also exhibited a similar negative correlation with APC7 expression. The frequency of positive APC7 expression was lower in the high Ki-67 group than in the low Ki-67 group. About 82% (77/94) of tissue samples were classified as aneuploid. Nearly half of the aneuploid group exhibited a low level of APC7 expression (42.9%), whereas most of the diploid group showed positive APC7 expression (94.1%), indicating that breast carcinomas with normal ploidy express higher levels of APC7. Immunoblotting analysis of APC expression in breast carcinomas To determine whether the expression of the APC7 component is exclusively modulated in breast carcinoma, we investigated the expression levels of other APC components. We first performed immunohistochemic analysis using anti-human APC3 antibodies or anti-human APC6 antibodies. However, we could not obtain meaningful data because nuclei in all breast carcinoma tissues were strongly stained by these antibodies, probably because of nonspecific cross-reactivity. Next, we compared the expression levels of APC3 and APC6 components by immunoblotting. However, immunoblotting analysis with anti-human APC6 antibodies also failed to exhibit a distinct band in tumor tissues because of the weak immune reactivity and the nonspecific reactivity of the APC6 antibody. Thus, we were able to obtain expression data on APC3 in various breast carcinoma tissues, in conjunction with APC7 expression. Figure 3 shows immunoblotting results for APC3 and APC7 in 24 representative breast carcinoma tissues. The expression levels of APC3 and APC7 in these tissues was variable, which might have been due to variable APC expression in these tissues. Some tissues (lanes 1, 2, 3, 8, 9, 13, 14, and 15) exhibited relatively high levels of expression of both APC3 and APC7, whereas other tissues (lanes 6, 7, 16, and 22) showed no expression of APC3 and APC7. These data suggest that the expression levels of APC3 and APC7 are simultaneously regulated in some breast carcinoma tissues. On the other hand, several carcinomas (lanes 4, 11, 12, 23, and 24) showed relatively high APC3 expression but low APC7, suggesting that selective downregulation of APC7 is unique to some breast carcinomas. Discussion This work was undertaken to determine whether the expressional modulation of APC7 is related to tumorigenesis in human cancers. We first used immunohistochemistry to investigate the expression of APC7 in tissue array slides mounted with various cancer tissues, and we observed strong immune reactivity to APC7 in the most rapidly growing tumor tissues. However, some breast cancer tissues with a high histologic grade exhibited weak immune reactivity to APC7. Therefore, we scrutinized APC7 expression in 108 invasive ductal carcinomas of the breast and compared these findings with clinicopathologic parameters. Although positive immune reactivity to APC7 was observed in more than 60% of breast carcinomas, negative APC7 expression was frequently observed in breast carcinomas with more aggressive characteristics (i.e. a higher histologic grade, a higher nuclear grade, a higher proliferation rate, or aneuploidy). These findings suggest a possible association between the expression of APC7 and breast cancer tumorigenesis. Most components of APC have been reported to be expressed in growing tissues at fairly constant levels [ 34 ]. On the other hand, Gieffers and coworkers [ 35 ] reported that components of APC (i.e. APC2, APC3, and APC7) are expressed in postmitotic adult brain tissue. However, it is not known how the expressions of APC components are modulated according to growth or cell differentiation. We observed twofold APC7 modulation in mouse NIH3T3 cells according to cell cycle (data not shown). In the present study, immunohistochemical studies using normal and cancer tissue arrays showed that APC7 is highly expressed in most proliferating cells. Strong immunoreactivity to APC7 was restricted to normal epithelial tissues and proliferating cancer tissues, whereas low APC7 immunoreactivity was observed in slow growing and differentiated tissues, such as adipocytes, hepatocytes, muscle cells, brain, and spinal cord, and in slowly growing tumor tissues such as lipoma, pleomorphic adenoma of the salivary gland, adenoid cystic carcinoma, chondrosarcoma, low-grade urothelial carcinoma, and renal cell carcinoma. Interestingly, we found a negative correlation between APC7 expression and some high-grade breast carcinoma tissues, and especially in those with aneuploidy. This negative correlation seems to be unique to some malignant breast carcinomas because we did not observe significant loss of APC7 expression in other aggressive carcinomas. In fact, we further investigated APC7 expression in two representative carcinomas, namely lung and renal carcinomas (i.e. rapid and slow growing carcinomas, respectively; data not shown), and obtained the same result as that obtained using the tissue array. All rapidly growing carcinoma tissues examined showed positive APC7 expression, whereas over 90% of slow growing renal carcinomas showed negative APC7 expression. Reports that the level of APC1 expression in breast cancer tissue is lower than that in other cancer tissues, and that expressed sequence tags of APC7 are reduced in breast tissues [ 36 ] support our observation that the loss of APC7 expression seems to be restricted in some high-grade breast carcinomas. Immunoblotting analysis showed that the expression of APC7 was variable in different breast carcinomas, which appears to be due to differences in expression of APC7 between individual breast carcinomas. However, the possibility of epithelial cell contamination during tissue extraction cannot be excluded. In several breast carcinomas, the loss of APC7 expression was accompanied by the loss of another APC component, such as APC3, but other tissues exhibited selective APC7 downregulation. These observations demonstrate that a unique loss of APC7 expression occurs in some breast carcinomas, and suggest that regulation of APC components in breast carcinomas is heterogeneous. Our data agree with reports that the expressional patterns of APC components are not simultaneously modulated [ 37 ] and that APC components can be individually modulated by environmental stimuli [ 38 ]. It has been reported that chromosome instability through abnormal mitotic progression plays a critical role in tumor malignancy [ 38 , 39 ]. Therefore, the dysregulation of APC activation, which probably perturbs mitotic progression, may affect malignant transformation or tumor progression. Moreover, the finding that APC is required for the G 2 and mitotic checkpoints suggests that malignant transformation can be caused by chromosome instability through the dysregulation of APC activation [ 40 ]. Recently, Wang and coworkers [ 41 ] reported a genetic alteration in APC6 and APC8 in human colon cancer cells, and suggested their involvement in colon carcinogensis. Aneuploidy is frequently observed in breast cancer tissues [ 42 ], and APC target molecules such as PTTG, PLK, and aurora kinase are often upregulated in the same tissues [ 7 , 12 , 17 ], supporting the notion that dysregulation of APC may play a role in the tumorigenesis of breast cancer. Our data concerning the negative correlation between APC7 expression and a high histologic grade with aneuploidy supports a possible linkage between the downregulation of APC7 and malignant transformation in breast cancer. As several papers have reported lethality induced by the loss of APC components [ 43 , 44 ], our observation raises the question as to how the cell cycle can progress in the absence of APC7. Although defects in the spindle checkpoint could elicit cell death, cancer cells in conjunction with p53 mutation could override the mitotic checkpoint and the cell lethality elicited by abnormal mitosis [ 41 ]. In this situation, it is believed that an abnormality in APC regulation could induce unscheduled mitotic progression [ 31 ]. One interesting observation was that the cell cycle of mutant APC8-harboring cells progressed, even with a disturbed pattern [ 45 ]. A recent report that the expression of cyclin-B 1 in breast cancer cells was sustained in the G 1 phase suggests that the cell cycle can progress in the presence of an abnormal mitotic cell cycle machinery [ 46 ]. Therefore, we hypothesize that cells with a functional APC defect can drive cell cycle progression in a situation in which cyclin-B has accumulated. However, the mechanism that allows breast cancer cells to progress through mitosis with functionally defective APC remains unknown. We did not determine whether APC7 expression is correlated with disease-free survival because most of the patients had recently been diagnosed. However, our data demonstrate that the downregulation of APC7 is more common in those who exhibit markers of poor prognosis. A number of parameters, namely lymph node metastases, tumor size, histologic grade, ER and progesterone receptor expression, lymphovascular invasion, proliferation rate, DNA content, and expression of oncogenes, have been reported to influence the prognosis of women with breast cancer [ 47 ]. On the other hand, in the present study poor prognostic indications such as high histologic grade, high proliferation rate, and aneuploidy were found to be related to downregulation of APC7, which suggests that breast cancer patients exhibiting weak APC7 expression would have poor survival. Yoshimoto and coworkers [ 48 ] reported that histologic grade and clinical stage, including lymph node metastasis, are important prognostic markers in breast cancer patients, which supports the notion that the downregulation of APC7 can be a viable prognostic marker. An earlier report that the Ki-67 index is related to histologic grade also supports the prognostic significance of the negative correlation between APC7 and proliferation [ 49 ]. Conclusion We found that downregulation of APC7 in breast carcinoma is more common in those with high histologic grade and high proliferation rate, and in those showing aneuploidy. Therefore, downregulation of APC7 may be a marker of a poor prognosis and may contribute to breast cancer tumorigenesis via chromosome instability and/or accelerated oncogenic signaling. Abbreviations APC = the anaphase-promoting complex; ER = estrogen receptor; PLK = polo-like kinase; PTTG = pituitary tumor transforming gene. Competing interests The author(s) declare that they have no competing interests. Authors' contributions KP carried out immunohistological studies and statistical analysis. SC produced mouse APC7 proteins and prepared APC7 antibodies. She also performed immunoblotting studies. ME participated in pathologic and immunohistologic studies. YK designed and coordinated this research and prepared the manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064132.xml

1064134

Relative microvessel area of the primary tumour, and not lymph node status, predicts the presence of bone marrow micrometastases detected by reverse transcriptase polymerase chain reaction in patients with clinically non-metastatic breast cancer

About 50% of patients with breast cancer have no involvement of axillary lymph nodes at diagnosis and can be considered cured after primary locoregional treatment. However, about 20–30% will experience distant relapse. The group of patients at risk is not well characterised: recurrence is probably due to the establishment of micrometastases before treatment. Given the early steps of metastasis in which tumour cells interact with endothelial cells of blood vessels, and, given the independent prognostic value in breast cancer of both the quantification of tumour vascularisation and the detection of micrometastases in the bone marrow, the aim of this study was to determine the relationship between vascularisation, measured by Chalkley morphometry, and the bone marrow content of cytokeratin-19 (CK-19) mRNA, quantified by real-time reverse transcriptase polymerase chain reaction, in a series of 68 patients with localised untreated breast cancer. The blood concentration of factors involved in angiogenesis (interleukin-6 and vascular endothelial growth factor) and of factors involved in coagulation (D-dimer, fibrinogen, platelets) was also measured. When bone marrow CK-19 relative gene expression (RGE) was categorised according to the cut-off value of 0.77 (95th centile of control patients), 53% of the patients had an elevated CK-19 RGE. Patients with bone marrow micrometastases, on the basis of an elevated CK-19 RGE, had a mean Chalkley count of 7.5 ± 1.7 (median 7, standard error [SE] 0.30) compared with a mean Chalkley count of 6.5 ± 1.7 in other patients (median 6, SE 0.3) (Mann–Whitney U -test; P = 0.04). Multiple regression analysis revealed that Chalkley count, not lymph node status, independently predicted CK-19 RGE status ( P = 0.04; odds ratio 1.38; 95% confidence interval 1.009–1.882). Blood parameters reflecting angiogenesis and coagulation were positively correlated with Chalkley count and/or CK-19 RGE. Our data are in support of an association between elevated relative microvessel area of the primary tumour and the presence of bone marrow micrometastases in breast cancer patients with operable disease, and corroborate the paracrine and endocrine role of interleukin-6 and the involvement of coagulation in breast cancer growth and metastasis.

Introduction The development of distant metastases is the primary cause of death in breast cancer patients. The involvement of the axillary lymph nodes, tumour size, histopathological grade and hormone receptor status determine prognosis and treatment options at initial diagnosis [ 1 ]. Nevertheless, these parameters do not accurately predict which patients will relapse after primary treatment, and they give limited information about the effectiveness of adjuvant treatment. About 50% of patients have no involvement of the axillary lymph nodes at diagnosis and can therefore be considered cured after primary locoregional treatment. However, about 20–30% will experience distant relapse within 5–10 years, suggesting outgrowth of disseminated tumour cells present at diagnosis and undetectable by the current diagnostics [ 2 ]. This prompted the refinement of methods able to detect subclinical tumour deposits in various body compartments. Tumour cells residing in bone marrow are considered to mirror the efficacy of the metastatic process throughout the body. Several prospectively designed clinical trials have confirmed the independent prognostic significance of the lodging of tumour cells in the bone marrow [ 3 ], suggesting that this minimal disease is indeed the progenitor of manifest metastasis. It is not clear whether the tumour cells that are part of subclinical metastases have arisen early during progression of the primary tumour or whether they are late and rare metastatic variants as a result of the cumulative acquisition of malignant phenotypic traits such as self-sufficiency in growth signals, insensitivity to anti-growth signals, evasion of apoptosis, limitless replicative potential, genomic instability, tissue invasion and sustained angiogenesis [ 4 - 6 ]. Bone marrow micrometastasis can be detected by immunocytochemical analyses with antibodies directed at epithelial markers. Polymerase chain reaction (PCR)-based techniques that amplify epithelial mRNA are more sensitive but need the introduction of cut-off values for positivity to correct for the inevitable loss of specificity. The concept of dependence on vascularisation of growth, invasion and metastasis of malignant tumours has been challenged by the description of angiogenesis-independent mechanisms [ 7 - 10 ]. Nevertheless, the growth of most primary tumours needs angiogenesis. There is accumulating evidence that angiogenesis is intrinsically linked with the process of haemostasis [ 11 ]. Both angiogenesis and haemostasis are tightly regulated in physiological circumstances, for example during wound healing, but are deregulated when involved in tumour growth, invasion and metastasis. We have previously demonstrated the prognostic importance of the angiogenic cytokine interleukin (IL)-6, and the fibrin degradation product D-dimer, in patients with metastatic breast cancer [ 12 , 13 ]. A reproducible method of quantifying vascularisation, by assessing the relative microvessel area, is Chalkley point overlap morphometry [ 14 ]. Significant associations between the Chalkley count and axillary lymph node metastasis, increasing tumour size, high grade and histological subtype have been reported in patients with breast cancer. Moreover, an independent prognostic value has been demonstrated in patients with breast cancer by a meta-analysis of 87 published studies [ 15 ]. Chalkley counting is done in selected areas of high microvessel density, so-called 'hot spots'. The hypothetical rationale for counting in these highly vascular areas is that they predict the presence of more angiogenic subclones of the tumour [ 16 ]. These subclones might therefore have a higher metastatic efficiency. The primary aim of this study was to assess the association of tumour vascularity, quantified by a well-standardised morphometrical method, and the presence of bone marrow micrometastases, detected by a sensitive and quantitative real-time reverse transcriptase PCR (RT–PCR) technique, in patients with operable breast cancer before treatment, to evaluate the relative microvessel area as a surrogate marker of bone marrow micrometastasis, as suggested by Fox and colleagues [ 17 ]. Materials and methods Patients Blood and bone marrow samples were collected preoperatively in 68 consecutive patients with localised, untreated, operable breast cancer. Eleven patients with a haematological malignancy who underwent bone marrow sampling for diagnostic purposes were entered as control patients for cytokeratin-19 (CK-19) relative gene expression (RGE). Tissue A representative, full cross-section of the tumour sample surrounded by adjacent normal breast tissue was taken from all tumours. The tissue was fixed in buffered formalin and was paraffin-embedded. Sections 5 μm thick were cut and mounted on poly-L-lysine-coated slides. Blood coagulation tests Plasma collection and measurement of fibrinogen, platelet counts and D-dimer levels were performed as described previously [ 12 ]. Chalkley morphometry Tumour vascularisation was evaluated by the Chalkley method, a morphometric point overlap counting system using a microscope eyepiece graticule. It has been shown to be both a rapid and a reproducible method that gives independent prognostic information and has been suggested as a standard method in an international consensus report on the quantification of angiogenesis [ 14 ]. In brief, a representative paraffin section including the tumour border was immunostained with a monoclonal anti-CD34 antibody (clone QBEnd/10; Biogenex, San Ramon, CA, USA) diluted 1:5 with overnight incubation at 4°C and without antigen retrieval procedure. The areas of highest vascular density ('hot spots') were identified at low magnification (× 10 ocular and × 10 objective). On a higher magnification (× 10 ocular and × 20 objective), a 25-point Chalkley eyepiece graticule (Chalkley grid area 0.22 mm 2 ) was applied to each hot spot and orientated to permit the maximum number of points to hit on or in a microvessel. The Chalkley count in the hot spot with highest vascular density (the largest number of hits) was used for further analyses. Enzyme-linked immunosorbent assay (ELISA) of angiogenic cytokines Serum samples were collected in serum separator tubes (type Vacutainer; Becton-Dickinson) and centrifuged at 2000 g for 5 min. Measurements of serum vascular endothelial growth factor (VEGF)-A 165 and IL-6 were performed as described previously [ 18 ]. Plasma (Becton-Dickinson Vacutainer tubes type 9NC 0.129 M, 4.5 ml) VEGF-A 165 was measured with an ELISA kit (R&D Systems, Minneapolis, MN, USA) on trisodium citrate-anticoagulated collected blood. For the ELISA assay of VEGF-A 165 , no cross-reactivity with other VEGF-A isoforms is documented. No cross-reactivity and no interference was found with IL-6-related (for example leukaemia inhibitory factor and oncostatin M) and growth factors not related to IL-6 (for example human platelet-derived growth factor and IL-7) (Quantikine human IL-6; R&D Systems, Minneapolis, MN, USA). Samples were assayed in duplicate. Within-assay variability had been tested before and was limited [ 18 ]. Bone marrow sampling Bone marrow (9 ml) was aspirated from the posterior iliac crest under local anaesthesia into syringes containing heparin as anticoagulant (Becton-Dickinson Vacutainer tubes type NH 170 IU 10 ml). Mononuclear cells were isolated by density-gradient centrifugation through Ficoll-Paque (Amersham Pharmacia Biotech) and washed twice with phosphate-buffered saline. After final centrifugation, the cell pellets were resuspended in a guanidine-containing buffer and stored at -70°C. Cell line The MDA-MB 361 cell line (American Type Culture Collection [ATCC] catalogue code HTB-27) was cultured in Leibovitz's L-15 medium with a free gas exchange with atmospheric air, and supplemented with 20% fetal bovine serum and antibiotics. The medium was replaced every 3–4 days. Cells were harvested in accordance with the ATCC guidelines. RNA isolation and cDNA synthesis Total RNA was extracted from the mononuclear cell fraction with the RNeasy kit (Qiagen). The amount of RNA was measured spectrophotometrically. Only samples with an A 260 / A 280 ratio of more than 1.8, indicating high purity, were used. The RNA integrity was tested with the Agilent Bioanalyzer (Agilent Technologies). Only samples with a lack of degradation on the electrophoretogram and with a 28S/18S ratio of at least 1.9 were analysed. For first cDNA strand generation, 2 μg of total RNA was reverse-transcribed with the High-Capacity cDNA Archive kit (Applied Biosystems) in a total volume of 100 μl. Primers and probe design Primers and probe for CK-19 were designed with the aid of Primer Express software (Applied Biosystems). To avoid amplification of contaminating genomic DNA, primers and probe were located on different exons. The forward primer of CK-19 (5'-CCCGCGACTACAGCCACTA-3') is situated on exon 1, the probe (5'-ACCATTGAGAACTCCAGGATTGTCCTGCA-3', labelled with 6-carboxyfluorescein [FAM] at the 5' end and 6-carboxy-tetramethylrhodamine [TAMRA] at the 3' end) on exon 2 and the reverse primer (5'-CTCATGCGCAGAGCCTGTT-3') on exon 3. RT-PCR with this primer set resulted in a 163-base-pair fragment. The nucleotide sequences of the primers and probe were checked for their specificity in the NCI BLAST ® database. The amplification product was sequenced and was comparable with the predicted nucleotide sequence. PCR amplification All PCR reactions were performed on the ABI Prism 7700 Sequence Detection System (Applied Biosystems) with the fluorescent Taqman method. Mispriming of processed pseudogenes was excluded. The CK-19 mRNA quantities were analysed in triplicate, normalised against β-actin as a control gene and were expressed in relation to a calibrator sample. The calibrator was produced from the blood of a healthy volunteer spiked with five MDA-MB 361 cells per 10 6 mononuclear cells. The calibrator was given a RGE of 100. As described by Livak and colleagues [ 19 ], results are expressed as RGE with the ΔΔ C T method. Ethical considerations The Ethical Committee of both institutions approved this study. Written informed consent was obtained from all patients. Statistical analysis Statistical analysis was performed with Graphpad Prism (version 2.0; Graphpad Software, Inc.). For IL-6 and VEGF-A ELISA assays, half of the detection limit value of the patient samples was used for statistical analysis in case the measured values did not reach the detection limit of the assay. Detection limits were 0.7 and 9 pg/ml, respectively. Comparisons of continuous variables were performed with the Mann–Whitney U -test. Relationship between categorical variables was validated with a χ 2 test. The correlation of continuous variables was analysed with a Spearman rank test. A multiple regression analysis was performed with a stepwise-backward likelihood procedure. P < 0.05 was necessary for statistical significance. Results Patients Age, T status, N status, oestrogen and progesterone receptor-status, tumour differentiation, menopausal status and tumour histology are given in Table 1 . CK-19 RGE Quantification of the mRNA transcripts of CK-19 in the bone marrow aspirates could be performed in all control samples and in 62 bone marrow samples of patients with breast cancer. In six patients, the volume of bone marrow or the quality of RNA was insufficient. A median RGE of 0.57 (range 0.22–0.78) was found for the control samples. Patients with primary breast cancer had a mean RGE of 3.85 ± 20.95 (median 0.79, standard error [SE] 2.66). Values above the 95th centile of the control values (0.77) were considered to indicate the presence of bone marrow micrometastasis. With this cut-off, 33 of 62 patients (53%) had bone marrow micrometastasis. In 49% of patients there was a concordance between lymph node status and bone marrow status. Fifty-five percent of patients with bone marrow micrometastasis had negative lymph nodes. Forty-six percent of patients without bone marrow micrometastasis had positive lymph nodes. Fifty-five percent of patients without lymph node metastasis had bone marrow micrometastasis. Forty-six percent of patients with lymph node metastasis had no bone marrow micrometastases. Chalkley count Quantification of relative microvessel area by Chalkley morphometry was performed in all ( n = 68) breast cancer samples. A mean value of 7.1 ± 1.9 (median 7, SE 0.23) was found. The median value of 7 was taken as a cut-off for categorisation into 'high' (7 or more) and 'low' Chalkley count. There was a positive association of Chalkley count with T-stage, but not with lymph node status (Table 2 ). CK-19 RGE and Chalkley count Considering both Chalkley counts and CK-19 RGE as continuous variables, a significant correlation was found ( r = 0.26, P = 0.04; Fig. 1a ). Sixty-two percent (23 of 37 patients) of breast tumours with a high Chalkley count were related to a bone marrow sample with positive CK-19 RGE. This was true for only 40% (10 of 25 patients) of breast tumours with a low Chalkley count (χ 2 test P = 0.04). Patients with a CK-19 RGE of 0.77 or more had a tumour with a mean Chalkley count of 7.5 ± 1.7 (median 7.0, SE 0.3) compared with 6.5 ± 1.7 (median 6.0, SE 0.3) in patients with a CK-19 RGE of less than 0.77 (Mann–Whitney U -test P = 0.04; Fig. 1b ). A stepwise-backward likelihood multiple regression analysis for the prediction of the presence of bone marrow micrometastases was performed, including Chalkley count, serum IL-6, D-dimers, fibrinogen, platelet count, serum VEGF-A, plasma VEGF-A, T-stage, N-stage, oestrogen receptor status, progesterone receptor status, tumour type and differentiation status. Only Chalkley count, as a continuous variable, independently predicted CK-19 RGE status ( P = 0.04; odds ratio 1.38; 95% confidence interval 1.009–1.882). Coagulation parameters and angiogenesis parameters in blood Values and associations are given in Tables 3 and 4 . In general, parameters reflecting coagulation and angiogenesis in blood had positive significant associations. Chalkley count, coagulation parameters and angiogenesis parameters in blood Serum IL-6 levels were correlated with Chalkley count ( r = 0.2, P = 0.08; Fig. 2a ). The mean IL-6 concentration in the serum of patients with a high Chalkley count was 2.57 ± 3.75 pg/ml (median 1.10, SE 0.57), and 1.52 ± 4.21 pg/ml (median 0.35, SE 0.84; Mann–Whitney U -test P = 0.018; Fig. 2b ) in patients with a low Chalkley count. Platelet count in patients with a high Chalkley count was (282.8 ± 64.6) × 10 3 /μl (median 272, SE 9.8), and (240 ± 73.9) × 10 3 /μl (median 243, SE 14.8; Mann–Whitney U -test P = 0.02; Fig. 3 ) in patients with a low Chalkley count. CK-19 RGE, coagulation parameters and angiogenesis parameters in blood D-dimer levels were correlated ( r = 0.22, P = 0.08; Fig. 4a ) with CK-19 RGE. Forty-one patients had elevated (more than 250 ng/ml) D-dimer levels. A mean D-dimer concentration of 477 ± 362 ng/ml (median 319, SE 67) was found in patients with CK-19 RGE values of less than 0.77, whereas the mean D-dimer level was 674 ± 634 ng/ml in patients with a CK-19 RGE value of 0.77 or more (median 458.5, SE 115.8; Mann–Whitney U -test P = 0.09; Fig. 4b ). No associations between CK-19 RGE and other variables were found (data not shown). Discussion The relation between the relative microvessel area, quantified by Chalkley counting of immunostained blood vessels in sections of primary breast cancer, and the pre-operative quantification of micrometastases in bone marrow by Taqman real-time RT–PCR was investigated in a group of patients with primary breast cancer before the start of treatment. Every increase in Chalkley count independently predicted a higher likelihood of bone marrow epithelial cells (odds ratio of 1.4) in patients with non-metastatic breast cancer. Patients with RGE of CK-19 in the bone marrow above the 95th centile of a control population had a primary tumour with a Chalkley count of 7.5, compared with 6.5 in patients without PCR-detected bone marrow micrometastases ( P = 0.04). Our work confirms the study by Fox and colleagues [ 17 ]. In 214 patients with primary breast cancer, tumour cells were detected through the examination of epithelial membrane antigen expression and the analysis of cell morphology. In immunostained breast cancer sections, a semi-quantitative vascular grade, expressed as being low or high, was determined in so-called vascular hot spots. Absolute concordance by kappa statistics of this vascular grade and Chalkley count was reported in a small subset of 22 patients. High vascular grade and the presence of vascular invasion independently predicted the presence of bone marrow micrometastases (respective odds ratios 2.7 and 2.7). In our study, Chalkley counts were determined in all 68 patients. However, the main difference lies in the higher sensitivity of Taqman quantitative real-time RT–PCR than that of immunocytochemistry on bone marrow cytospins used by Fox and colleagues [ 17 ]. This study corroborates the well-documented prognostic value of the Chalkley count in breast cancer [ 20 ]. Hansen and colleagues [ 20 ] categorised Chalkley counts by using cut-off points of 5 and 7. Patients with Chalkley counts of 7 or more had a hazard ratio for death of 1.46 (95% confidence interval 1.14–1.87) compared with the group of patients with Chalkley counts between 5 and 7. Bone marrow micrometastases are predicted by a high Chalkley count ([ 17 ] and this study) and have been linked to adverse outcome in patients with breast cancer: currently, about 2500 patients with breast cancer have been analysed in five prospectively designed clinical trials to verify associations between the presence of immunostained tumour cells in bone marrow and prognosis [ 3 ]. Braun and colleagues [ 21 ] have included 552 patients with stage I, II and III breast cancer in a prospective study: the presence of tumour cells detected with the antibody A45-B/B3 independently predicted distant metastasis but not locoregional recurrence. The association of a high Chalkley count with the presence of tumour cells in the bone marrow explains at least partly the prognostic value of the former parameter. The presence or absence of lymph node metastases did not predict the bone marrow CK-19 RGE status by multiple regression analysis in this study. Concordance between bone marrow status and lymph node status was present in only 49% of all patients. This is in accord with results of cDNA array analyses of primary breast tumours of bone marrow-positive and bone marrow-negative patients [ 22 ]. Distinct profiles were reported, indicating that bone marrow metastasis is a selective process with a specific molecular signature of the primary breast tumour. In addition, in a prognostic study of patients with operable breast cancer, the bone marrow status provided independent information in addition to tumour size and lymph node status in a multivariate analysis [ 23 ]. The presence of bone marrow micrometastases was related to earlier recurrence. Although we did not compare the CK-19 RGE with the number of immunostained tumour cells in bone marrow cytospins in this study, the Taqman quantitative real-time RT–PCR methodology has been validated by us in a group of patients with metastatic breast cancer [ 24 ]. RT–PCR quantification of mRNA transcripts of epithelial cells has a high sensitivity but a rather poor specificity for the detection of tumour cells in bone marrow or blood. The main reason for this is the expression of some epithelial markers, such as CK-19, by haematopoietic and blood cells. A stringent cut-off for CK-19 RGE positivity has therefore been set in this study. Methodological adaptations to avoid mispriming due to pseudogenes and to avoid amplification of contaminating DNA have also been made. Expression of the CK-19 RGE as an absolute number of tumour cells was not performed, because of unpredictable variations of the expression levels of CK-19 mRNA in tumour cells from individual patients. What are the putative tumour-biological mechanisms to explain the positive association between the extent of tumour vascularisation and the amount of disseminated epithelial cells, and thus probably to a large extent 'tumour' cells, in the bone marrow of breast cancer patients? A first explanation might be that intravasation of breast cancer cells, one of the initial steps of metastasis, is facilitated in tumours with extensive vascularisation [ 25 - 27 ]. McCulloch and colleagues [ 28 ] indeed demonstrated a positive association between, on the one hand, the number of tumour cells in the blood of the veins draining the breast gland during surgery and, on the other, the vascularity of the primary tumour [ 28 , 29 ]. Although mechanical stress due to diagnostic procedures, such as mammography, or due to surgery itself can augment tumour cell shedding, this study links high vascularity to increased intravasation of breast cancer cells in a peri-operative setting. A comparable positive relation between breast tumour vascularity and CK-19 mRNA in the blood was found in patients 2–3 weeks after surgery [ 30 ]. An increased endothelial surface area might provide easier access to the bloodstream. A complementary effect of high vascularity is the more pronounced paracrine interaction between endothelial cells and tumour cells. Activated endothelial cells can indeed produce tumour cell growth factors, for example IL-6 in the vertical growth phase of melanoma [ 31 ], or matrix remodelling enzymes that facilitate tumour cell motility and invasion, as demonstrated in skin carcinogenesis models [ 32 ]. The survival and growth of tumour cells can also be influenced by endocrine interactions with a primary tumour present. In cervical, ovarian and colorectal cancer IL-6 is shed in the blood by the tumour, creating a 2.5-fold arteriovenous gradient [ 33 ]. In this study, patients with a Chalkley count of 7 or more had concentrations of circulating IL-6 that were about 2.5-fold higher than patients with a Chalkley count of less than 7 ( P = 0.018). This is in accord with the angiogenic potential of IL-6. In metastatic breast cancer, high levels of circulating IL-6 predict a shortened survival time [ 13 ]. Unlimited haemostasis is intimately linked to tumour growth and angiogenesis, and tumours can therefore be regarded as non-healing wounds [ 34 ]. Immunohistochemical analyses with antibodies directed at fibrin demonstrated abundant stromal fibrin depositions in inflammatory breast cancer, a highly aggressive and highly angiogenic type of breast cancer, and less or no fibrin in non-inflammatory breast cancer [ 35 ]. A relation of high Chalkley counts and the presence of fibrin in the tumour was also reported in cutaneous breast cancer deposits [ 36 ]. In this study, levels of D-dimers in plasma, as a reflection of active fibrin remodelling, were correlated with CK-19 RGE values ( r = 0.23; P = 0.08). Levels of D-dimers in patients with CK-19 RGE values of 0.77 or more were 1.4-fold those in patients with CK-19 RGE values of less than 0.77 ( P = 0.09). Significantly higher levels of plasma D-dimers have been found in veins draining colorectal adenocarcinomas than in arteries [ 33 ], suggesting that plasma D-dimer levels are a measure of matrix remodelling in the tumour. Elevated levels of circulating D-dimers have been correlated with enhanced progression kinetics and reduced overall survival in metastatic breast cancer [ 12 ]. Taken together, these results strongly suggest interactions between angiogenesis and haemostasis to facilitate metastasis in breast cancer. Conclusion In conclusion, our data show that tumour vascularity, and not lymph node status, independently predicts the presence of PCR-detected bone marrow micrometastases. This is consistent with direct haematogenous spread from the primary breast cancer without orderly progression from local to nodal to distant disease. Abbreviations CK = cytokeratin; ELISA = enzyme-linked immunosorbent assay; IL = interleukin; RGE = relative gene expression; RT–PCR = reverse transcriptase polymerase chain reaction; SE = standard error; VEGF = vascular endothelial growth factor. Competing interests The author(s) declare that they have no competing interests. Authors' contributions IB performed the RT–PCR reactions, the blood coagulation tests and the ELISA analysis of the angiogenic cytokines. She made substantial contributions to analysis and interpretation of data, and drafted the manuscript. RS performed the Chalkley morphometry and made substantial contributions to analysis and interpretation of data, and drafted the manuscript. HE performed the cell line culture, the primers and probe design and optimised the RT–PCR reaction. PvD, EVM and SS made substantial contributions to conception and design. JW performed the statistical analysis. PV conceived of the study and has been involved in drafting the article and revising it critically for important intellectual content. LD conceived of the study and has given final approval of the version to be published. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064134.xml

1064135

Common ERBB2 polymorphisms and risk of breast cancer in a white British population: a case–control study

Introduction About two-thirds of the excess familial risk associated with breast cancer is still unaccounted for and may be explained by multiple weakly predisposing alleles. A gene thought to be involved in low-level predisposition to the disease is ERBB2 (HER2) . This gene is involved in cell division, differentiation, and apoptosis and is frequently amplified in breast tumours. Its amplification correlates with poor prognosis. Moreover, the coding polymorphism I655V has previously been associated with an increased risk of breast cancer. Methods We aimed to determine if common polymorphisms (frequency ≥ 5%) in ERBB2 were associated with breast cancer risk in a white British population. Five single-nucleotide polymorphisms (SNPs) were selected for study: SNP 1 near the promoter, SNP 2 in intron 1, SNP 3 in intron 4, SNP 4 in exon 17 (I655V), and SNP 5 in exon 27 (A1170P). We tested their association with breast cancer in a large case–control study ( n = 2192 cases and 2257 controls). Results There were no differences in genotype frequencies between cases and controls for any of the SNPs examined. To investigate the possibility that a common polymorphism not included in our study might be involved in breast cancer predisposition, we also constructed multilocus haplotypes. Our set of SNPs generated all existing ( n = 6) common haplotypes and no differences were seen in haplotype frequencies between cases and controls ( P = 0.44). Conclusions In our population, common ERBB2 polymorphisms are not involved in predisposition to breast cancer.

Introduction Breast cancer is the most common cause of cancer in women in the United Kingdom and is, after lung cancer, the most common cause of cancer death (Office for National Statistics). Positive family history is a well-established risk factor for the disease: the risk to first-degree relatives of a breast cancer case is about twice the population risk [ 1 ]. Most of the excess familial risk associated with breast cancer is likely to be genetic in origin [ 2 , 3 ]. However, only about a third of this risk is accounted for by known genes, the most important being BRCA1 and BRCA2 , while the remainder might be explained by a combination of weakly predisposing alleles [ 2 - 4 ]. A gene thought to be involved in low-level susceptibility to breast cancer is ERBB2 ( HER2 ). This gene is located on chromosome 17q12–q21, spans 38 kilobases, and comprises 27 coding exons. It is a member of the ERBB family, a family of protein tyrosine kinases involved in cell division, migration, adhesion, differentiation, and apoptosis and consisting of EGFR (ERBB1), ERBB2, ERBB3, and ERBB4 [ 5 ]. ERBB2 amplification or overexpression is seen in about 25% of breast cancers and has been associated with metastatic phenotype, endocrine therapy unresponsiveness, and poor prognosis [ 6 ]. ERBB2 is polymorphic in the transmembrane region of the protein at codon 655 (ATC/isoleucine to GTC/valine [I655V]). The amino acid change could result in increased protein tyrosine kinase activity [ 7 ]. Several association studies of I655V and breast cancer risk have yielded conflicting results. In a study on 700 Han Chinese women, Xie and colleagues first reported a significantly increased risk for carriers of the rare allele (odds ratio [OR] = 1.4) [ 8 ]. Only one of seven subsequent studies showed an overall effect of I655V on breast cancer risk [ 9 - 15 ]. However, of the negative studies, all but one had limited power to detect a risk of this magnitude [ 13 ]. Three groups did report associations in specific subgroup analyses in the absence of overall effect: Wang-Gohrke and Chang-Claude showed an association in women with a positive family history of breast cancer and McKean-Cowdin and colleagues showed an association with localized breast cancer, whereas Millikan and colleagues showed an association in women with a positive family history who were aged 45 years or younger as well as an increased risk of carcinoma in situ [ 12 , 13 , 15 ]. I655V has usually been selected for study because of the possible functional consequences of the amino acid change in the transmembrane region of the protein. Many more single-nucleotide polymorphisms (SNPs) in ERBB2 are known but only one negative study has reported on any of these [ 10 ]. A selected set of sequence polymorphisms can serve as genetic markers to detect association between a particular region and the disease, whether or not the markers themselves have a functional effect [ 16 ]. It is therefore not necessary to test each polymorphism individually. Because most SNPs are correlated with nearby polymorphisms, genotypes at unsassayed, risk-related SNPs will be correlated with one or more assayed SNPs [ 17 ]. If the set of selected markers provides enough information about the remainder of the common polymorphisms in that gene, any susceptibility allele within or close to the gene should be uncovered through the evaluation of the underlying haplotypes [ 18 ]. To clarify the role of ERBB2 in the predisposition to breast cancer, we tested the association of five common polymorphisms (including I655V) with the disease in a large case–control study of white British women. We aimed to identify sufficient SNPs to tag all the common haplotypes across the gene. Materials and methods Patients and controls Cases were drawn from the Anglian Breast Cancer Study, an ongoing population-based study with cases ascertained through the East Anglian Cancer Registry [ 4 ]. All women diagnosed with invasive breast cancer under the age of 55 years between 1 January 1991 and 30 June 1996 and who were alive at the start of the study (prevalent cases) as well as women under the age of 70 who were diagnosed from 1996 onwards (incident cases) were eligible for inclusion. We used prevalent and incident cases in order to maximize sample size; approximately 65% of eligible patients have enrolled in the study. Women taking part in the study were asked to provide a 20-ml blood sample for DNA analysis and to complete a comprehensive epidemiological questionnaire. We carried out genotyping on a subset consisting of the first 2192 (1438 incident and 754 prevalent) enrolled cases. Controls (2257) were randomly drawn from the Norfolk component of the European Prospective Investigation of Cancer (EPIC) [ 19 ]. The ethnic background of both cases and controls is similar, with over 98% being white Anglo-Saxon. Ethical approval was obtained from the Anglia and Oxford Multicentre Research Committee and informed consent was obtained from each patient. SNP identification and selection SNPs with validated frequency data were identified in January 2004 through the dbSNP database . If these data were from a non-Caucasian population, we confirmed the presence of the polymorphism in our population by performing denaturing high-performance liquid chromatography on a set of 48 genomic DNA samples from UK breast cancer patients. We selected all nonsynonymous coding SNPs ( n = 2), SNPs located in the promoter region ( n = 1), and two randomly chosen intronic SNPs [ 20 ]. A total of five SNPs were thus selected for study (Table 1 ). In order to have good power to detect small relative risks, we restricted our attention to SNPs with a frequency of 5% or more. Genotyping Genotyping was carried out using Taqman ® (Applied Biosystems, Warrington, UK) according to the manufacturer's instructions. Primers and probes were either supplied directly by Applied Biosystems in case of Assays-by-Design™ (SNP 1 and SNP 2) and Assays-on-Demand™ (SNP 3) or designed using Primer Express Oligo Design Software v2.0 (Applied Biosystems) (SNP 4 and SNP 5). Sequences are available on request. Reactions were carried out at 54°C (SNP 4) or 60°C (SNP 1, SNP 2, SNP 3, and SNP 5). All assays were carried out in 384-well plates. Each plate contained 384 samples including 2 negative controls with no DNA and 12 samples duplicated on a separate quality-control plate. Plates were read on the ABI Prism 7900 using the Sequence Detection Software (Applied Biosystems). Failed genotypes were not repeated. Statistical methods The characteristics of cases and controls were explored with SPSS © v12.0.1 (SPSS Inc, Chicago, IL, USA). For each SNP, deviation of genotype frequencies in controls from the Hardy–Weinberg equilibrium was assessed by χ 2 test with one degree of freedom (df). Genotype frequencies in cases and controls or within cases stratified by disease stage (stage I vs stages II–IV) or age group (≤ 45 vs >45) were compared by χ 2 test for heterogeneity (2df). Genotype-specific risks were estimated as ORs using standard cross-product ratio. Confidence intervals were calculated using the variance of the log (OR), which was estimated by the standard Taylor expansion. Power was determined using standard statistical methods [ 21 ]. We have over 90% power at the 1% significance level to detect a dominant allele with a frequency of 0.05, which confers a relative risk of 1.5, or a dominant allele with a frequency of 0.2 that confers a relative risk of 1.3. Power to detect recessive alleles at the 1% significance level is more limited: 59% for an allele with a frequency of 0.2 that confers a relative risk of 1.5 or 77% for an allele with a frequency of 0.3 that confers a relative risk of 1.4. The LDA program [ 22 ] was used to calculate pairwise linkage disequilibrium (LD) for each SNP pair in the whole case–control set. The haplo.score program [ 23 ] was used to test for association between haplotypes and breast cancer risk. Haplo.score uses a likelihood that depends on estimated haplotype frequencies to test the statistical association between haplotypes and phenotype. It is based on score statistics, which provide both global tests and haplotype-specific tests [ 23 ]. Results The median age was 48 years (range 25–54) for prevalent cases, 52 years (26–55) for incident cases, and 56 years (25–81) for controls. Incident and prevalent cases were similar regarding breast cancer stage ( P = 0.12) and histological grade ( P = 0.41). Table 2 shows the genotype frequencies in cases and controls as well as genotype-specific risks for the five SNPs assayed. The genotype frequencies were similar in the prevalent and incident cases for all polymorphisms (data not shown). None of the genotype distributions for the controls differed significantly from those expected under Hardy–Weinberg equilibrium. There was no evidence that any of the SNPs is associated with breast cancer; genotype-specific ORs were all close to unity with narrow confidence intervals. We also compared genotype frequencies within cases stratified by disease stage and age group for SNP 4 (I655V). No differences were seen ( P [stage] = 0.61, P [age group] = 0.33). LD was strong (D' > 0.7) across pairs involving SNPs 1, 2, 3, and 5, whereas SNP 4 was in weak LD (D' < 0.3) with all other polymorphisms except SNP 1 (D' [SNP 1-SNP 4] = 0.98) (Fig. 1 ). SNPs 3 and 5 were in nearly perfect LD ( r 2 = 0.92). Of 32 possible haplotypes, only 6 were observed with a frequency greater than 5% (Table 3 ). For the whole case–control set, common haplotypes constituted 98% of all the observed haplotypes. Two haplotypes (haplotypes 3 and 5) contained the SNP 4 (I655V) minor allele. The global test was not significant ( P = 0.44), nor were there any differences between cases and controls for individual haplotypes. Similarly, no differences in haplotype frequencies were seen within cases stratified by disease stage ( P = 0.37) or age group ( P = 0.48). Discussion Our study is the largest case–control study reported on ERBB2 genetic variation. To our knowledge, this is also the first study on ERBB2 reporting results for more than two polymorphisms and looking for involvement of haplotypes in breast cancer predisposition. We performed a study of five common SNPs and found no evidence for association with breast cancer risk. Four of the polymorphisms may be functional: SNP 1 near the promoter region and SNP 2 in intron 1 could be involved in regulatory processes whereas SNP 4 and SNP 5 are nonsynonymous coding SNPs that could affect tyrosine kinase activity or protein structure [ 7 ]. Two association studies have previously reported a positive association between SNP 4 (I655V) and breast cancer risk [ 8 , 14 ]. Both genotyped about 700 individuals and showed a similarly increased risk for carriers of the Val allele (OR = 1.4). We were not able to replicate these findings. We have over 90% power to detect a risk of this magnitude at the 10 -4 level of significance. This suggests that previous positive findings may have been due to type I statistical errors. Neither could we replicate findings associating I655V with low-stage breast cancer or with breast cancer in younger women [ 12 , 13 ]. Positive results from stratified analyses should be treated with caution; very large sample sizes are required to obtain reliable results, the number of possible analyses that can be undertaken is large, and there is a strong possibility that one or more tests will be statistically significant simply by chance [ 24 ]. We could not carry out analyses within cases stratified by family history, because we only had incomplete family history data [ 15 ]. To investigate the possibility that a common polymorphism not included in our study might be involved in breast cancer predisposition, we constructed multilocus haplotypes and observed similar frequencies in cases and controls. We found six common haplotypes. Recently, the NIEHS Environmental Genome Project at the University of Washington released resequencing data based on 90 individuals (the PDR90 population; individual genotypes are available on line: ) and identified nine common SNPs (frequency ≥ 5%) in ERBB2 . All the common haplotypes (frequency ≥ 5%) were tagged by our set of five SNPs, even though, as expected given the multiethnicity of PDR90, differences in frequencies were seen between the two populations (data not shown). Crawford and colleagues resequenced 100 candidate genes involved in inflammation, lipid metabolism, and blood pressure regulation and showed that in a population of European descent the average number of common haplotypes per gene was 4.5, with a maximum number of 8 observed in only two genes [ 25 ]. We are therefore confident that we have detected all common ERBB2 haplotypes present in our population. We limited our study to common polymorphisms. A larger study set would be needed to identify a rarer polymorphism involved in disease predisposition. For example, dominant alleles with a frequency of 2% would require more than 4000 cases and 4000 controls to detect a relative risk of 1.5 significant at the 1% level with 90% power. We cannot exclude the possibility that a common SNP might have a differential effect in another ethnic group via gene–gene or gene–environment interactions, or that a predisposing SNP might be present exclusively in another population [ 26 ]. In summary, we conducted a large case–control study of ERBB2 and breast cancer. We genotyped five common SNPs, including the much-studied I655V polymorphism, and saw no association with the disease. Our set of SNPs generated all common haplotypes, and no differences in haplotype frequencies were seen between cases and controls. Conclusion In our population, common ERBB2 polymorphisms are not involved in predisposition to breast cancer. Abbreviations df = degree of freedom; LD = linkage disequilibrium.; OR = odds ratio; SNP = single-nucleotide polymorphism. Competing interests The author(s) declare that they have no competing interests. Authors' contributions PRB performed the experiments, carried out the analyses, and wrote the manuscript under the supervision of FL, PDP, and BAJP. CL and DMC managed the genotyping process. MS and NED supervised DNA samples collection. DFE was the statistical advisor and AMD was the laboratory manager. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064135.xml

1064136

Claudin-1, -3 and -4 proteins and mRNA expression in benign and malignant breast lesions: a research study

Introduction We compared levels of protein and mRNA expression of three members of the claudin (CLDN) family in malignant breast tumours and benign lesions. Methods Altogether, 56 sections from 52 surgically resected breast specimens were analyzed for CLDN1, CLDN3 and CLDN4 expression by immunohistochemistry. mRNA was also analyzed using real-time PCR in 17 of the 52 cases. Results CLDNs were rarely observed exclusively at tight junction structures. CLDN1 was present in the membrane of normal duct cells and in some of the cell membranes from ductal carcinoma in situ , and was frequently observed in eight out of nine areas of apocrine metaplasia, whereas invasive tumours were negative for CLDN1 or it was present in a scattered distribution among such tumour cells (in 36/39 malignant tumours). CLDN3 was present in 49 of the 56 sections and CLDN4 was present in all 56 tissue sections. However, CLDN4 was highly positive in normal epithelial cells and was decreased or absent in 17 out of 21 ductal carcinoma grade 1, in special types of breast carcinoma (mucinous, papillary, tubular) and in areas of apocrine metaplasia. CLDN1 mRNA was downregulated by 12-fold in the sample (tumour) group as compared with the control group using GAPDH as the reference gene. CLDN3 and CLDN4 mRNA exhibited no difference in expression between invasive tumours and surrounding tissue. Conclusions The significant loss of CLDN1 protein in breast cancer cells suggests that CLDN1 may play a role in invasion and metastasis. The loss of CLDN4 expression in areas of apocrine metaplasia and in the majority of grade 1 invasive carcinomas also suggests a particular role for this protein in mammary glandular cell differentiation and carcinogenesis.

Introduction In epithelial cells, tight junctions (TJs) are the most apical intercellular junctions, and function as selective barriers to macromolecules and prevent diffusion of lipids and proteins between apical and basolateral membrane domains. TJs appear as a network of continuous and anastomosing filaments on the protoplasmic face of the plasma membrane [ 1 ]. The molecular architecture of TJ strands has been described in recent years. The proteins involved in the formation of TJs have been divided into two categories: integral membrane proteins, such as claudins (CLDNs), occludins and JAM (junctional adhesion molecule) [ 2 ]; and peripheral membrane proteins, such as zonulae occludens (ZO)-1, ZO-2 and ZO-3, cingulin, symplekin, pilt, MAGI-1 (MAGUK [membrane associated guanylate kinase] inverted protein 1) and AF-6 (afadin) [ 3 ]. CLDNs, a family comprising 24 members, are the main family of proteins that make up the TJs [ 2 , 4 ]. The tightness of individual TJ strands in situ is determined by the various combinations of CLDN family members. CLDNs have a membrane topology similar to that of occludin, even though they are smaller proteins of about 22 kDa. Their protein structure, which is uniform among family members, contains four transmembrane domains with amino- and carboxyl-termini in the cytoplasm and two extracellular loops [ 5 ]. Until now only few data have been reported regarding the role played by CLDNs in breast lesions, and findings on their function remain controversial [ 6 , 7 ]. There is 38% homology between human CLDN1 and CLDN2 at the amino acid sequence level [ 8 , 9 ]. Expression of the four-exon CLDN1 gene, formerly named senescence associated epithelial membrane protein, was identified in testis [ 10 ], colorectal carcinoma [ 11 ] and lung [ 12 ]. The direct involvement of CLDN1 in the barrier function of TJs has been demonstrated in CLDN1-over-expressing MDCK cells [ 13 ]. In a study conducted in breast carcinomas by Kramer and coworkers in 2000 [ 7 ] it was shown that there are no genetic alterations in the promoter or coding sequences in the CLDN1 gene that could account for the loss of CLDN1 protein expression in tumour cells. Of the relatively few publications concerning CLDN expression in the breast, one study [ 14 ] found that CLDN7 protein and mRNA are downregulated in primary breast cancers as compared with normal breast epithelium. Taken together, these observations indicate that there is a need to study the possible role played by CLDNs in carcinogenesis and tumour differentiation. CLDN3 and CLDN4 can function as receptors for Clostridium perfringens enterotoxin [ 15 ]. The one-exon genes of CLDN3 and CLDN4 are closely linked (CLDN3 7q11 and CLDN4 7q11.23) and share extended homology of approximately 80% at the DNA level. SAGE (serial analysis of gene expression) demonstrated that, of differentially upregulated mRNAs, CLDN3 and CLDN4 were among the six most upregulated in primary ovarian carcinoma cells [ 16 ]. Van Itallie and coworkers [ 17 ] demonstrated the close correlation between the level of CLDN4 expression and change in both conductance and ionic selectivity. Thus far there is no evidence on how CLDN3 and CLDN4 are expressed in breast carcinomas. The present study was designed to determine whether human CLDN1, CLDN3 and CLDN4 are expressed in different types of breast lesions, what differences there are between different CLDNs in breast, and how the protein expression correlates with CLDN mRNA expression. Methods Materials For immunohistochemistry, paraffin-embedded sections were used. For immunofluorescence, paraffin-embedded sections and frozen resected breast tissue samples were analyzed. For real-time PCR RNA isolated from frozen surgical breast tissues were analyzed. Materials used for immunohistochemistry Altogether, 56 paraffin-embedded sections from 52 surgically resected breast specimens were analyzed for expression of CLDN1, CLDN3 and CLDN4 using immunohistochemistry. Tissue samples were collected with the patient's written consent and conforming with national law (23/2002.V.9 EüM) regarding human studies. The age of the 52 female patients ranged between 31 and 80 years. Carcinomas were graded according to the Elston-Bloom and Richardson system [ 18 ]. The histopathological diagnoses of the selected breast lesions were as follows: fibrocystic changes/breast lesions ( n = 12), ductal carcinoma in situ (DCIS; n = 5), invasive ductal carcinoma (IDC) ± DCIS ( n = 27), invasive lobular carcinoma (ILC; n = 5) and special types of breast carcinoma (papillary, mucinous, tubular; n = 7). Materials used for immunofluorescene Immunofluorescence detection of CLDN1, CLDN3 and CLDN4 was performed on frozen sections (IDC and surrounding breast) as well as on tissue microarray of IDCs obtained from paraffin-embedded blocks. Materials used for CLDN1, CLDN3, CLDN4 and GAPDH mRNA expression with real-time PCR In 17 out of the 52 cases analyzed by immunohistochemistry, the level of mRNA expression of CLDN1, CLDN3 and CLDN4 were also analyzed using real-time PCR. The histopathological diagnoses of these 17 breast lesions are as follows: fibrocystic changes/breast lesions ( n = 3), IDC ± DCIS ( n = 12), ILC ( n = 1) and intraductal papilloma ( n = 1). In the following text only the real-time PCR results for the 13 invasive carcinomas (12 IDC ± DCIS and 1 ILC) are discussed in detail. Because relative quantification of the real-time PCR results was used for data analyses in all 17 breast samples, the normal component of the breast tissue was simultaneously investigated for mRNA expression of CLDN1, CLDN3, CLDN4 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA expression. GAPDH was used as the reference gene, and so data are normalized to GAPDH. Breast tissue samples, obtained in accordance with current local and ethical recommendations, were snap frozen in liquid nitrogen within 30 min of surgical removal and stored at -80°C. Immunohistochemistry for CLDN1, CLDN3 and CLDN4 detection Immunohistochemical detection of CLDNs was performed in formalin-fixed, paraffin-embedded sections. Sections (5 μm thick) were cut from each paraffin block, deparaffinized and rehydrated. Immunohistochemical staining for CLDN1, CLDN3 and CLDN4 was performed using the streptavidin-peroxidase procedure. For CLDN1 and CLDN3 proteins, antigen retrieval was performed with two different antigen retrieval solutions in parallel sections: 20 min in Vector (Burlingame, CA, USA) antigen retrieval solution in a microwave oven; and 40 min in DAKO (Carpinteria, CA, USA) antigen retrieval solution in a microwave oven. For CLDN4, antigen retrieval was performed for 20 min in Vector antigen retrieval solution in a microwave oven. Endogenous peroxidase was blocked with 1% H 2 O 2 for 10 min at room temperature. Sections were incubated with primary antibodies (Zymed, San Francisco, CA, USA) to CLDN1, CLDN3 and CLDN4 (concentration 1:100) overnight. The antibodies used were polyclonal rabbit anti-CLDN-1 (Zymed; reactivity: human, rat and dog), polyclonal rabbit anti-CLDN-3 (Zymed; reactivity: human, mouse and dog) and monoclonal mouse anti-CLDN-4 (Zymed; reactivity: human). Antigen-bound primary antibody was detected using standard avidin–biotin immunoperoxidase complex (DAKO LSAB2 Kit). The chromogen substrate was aminoethylcarbazol. In each case, for counter-staining Mayer's haemalaun was used. For each CLDN a negative control with omission of the primary antibody was included. Colon carcinoma was used as positive control. In each case, two independent observers (A-MT and JK) recorded the distribution of staining, intensity and localization. Staining of CLDNs was expressed relative to adjacent normal mammary epithelium. Immunofluorescence microscopy for CLDN1, CLDN3 and CLDN4 detection Sections (10 μm thick) were cut from frozen tissue and from breast tissue array paraffin blocks. The tissue arrays were constructed in our laboratory. The selected breast tumour samples were obtained from paraffin blocks from our files. In brief, the study involved breast tumours from 30 T1N0 and 30 T1N1 cases. Histopathological data included tumour stage according to the UICC TNM system. With the guidance of an experienced pathologist (JK), representative core tissue biopsies (2 mm in diameter) were taken from archival paraffin-embedded primary breast carcinomas and seeded in new paraffin blocks using an instrument provided by Histopathology Company (Pécs, Hungary). Two biopsies were taken from each paraffin block. For paraffin-embedded sections antigen retrieval was performed as described above (under Immunohistochemistry for CLDN1, CLDN3 and CLDN4 detection). Sections were incubated with goat antiserum for 1 hour at room temperature and then incubated with the primary antibodies (Zymed) to CLDN1, CLDN3 and CLDN4 (1:50) overnight. CLDN1 and CLDN3 antigen-bound primary antibody was detected using anti-rabbit IgG conjugated to Alexafluor 488 (Molecular Probes, Eugene, OR, USA) and CLDN4 stained sections was detected using anti-mouse IgG conjugated to Alexafluor 488 (Molecular Probes) for 30 min at room temperature. Cell nuclei were counter-stained with propidium iodide (Molecular Probes) for 10 min at room temperature. For each CLDN a negative control with omission of the primary antibody was included. Coverslips were mounted with fluorescent mounting medium (DAKO). Stained specimens were analyzed by laser scanning confocal microscopy (MRC 1024; Bio-Rad, Hercules, CA, USA). Western blotting The monospecificity of CLDN1, CLDN3 and CLDN4 antibodies was tested by our group in whole tumour cell lysate by Western blot analyses, as described previously [ 14 ]. In brief, from snap frozen tissues of IDC of the breast and normal tissue surrounding the tumour, total protein was extracted using lysis buffer (100 mmol/l NaCl, 1% NP-40, 2 mmol/l EDTA, 50 mmol/l TRIS, 20 μg/ml aprotinin, 20 μg/ml leupeptin, 1 mmol/l PMSF, pH 7.5). Subsequently, the samples were mixed with 2× Laemmli sample buffer and boiled for 10 min, followed by centrifugation at 13,000 rpm for 15 min. Similar amounts of protein were loaded in each lane and run on 10% SDS-PAGE under reducing conditions. Then, gel electrophoresis proteins were transferred to nitrocellulose membranes. For immunodetection, blots were incubated with CLDN1 polyclonal antibody (1:400), CLDN3 polyclonal antibody (1:800) and CLDN4 monoclonal antibody (1:200) overnight. The biotinylated secondary antibody, goat anti-mouse IgG (diluted 1:2000, E433; DAKO), was applied for 1 hour at room temperature. Antibody detection was conducted using an enhanced chemiluminescent reaction system. Real-time PCR for CLDN1, CLDN3, CLDN4 and GAPDH mRNA detection RNA isolation Breast tissue (50–100 mg) was homogenized and RNA was isolated with Trizol (Invitrogen, Carlsbad, CA, USA) [ 19 ], in accordance with the manufacturer's instructions, using a Polytron homogenizer (Kinematica AG, Littau/Lucerne, Switzerland). Briefly, the RNA was precipitated with 0.5 ml isopropyl alcohol in the aqueous phase. The RNA pellet was washed once in 70% ethanol, dried and resuspended in 50 μl of RNAse-free water and kept at -80°C until use. Total RNA integrity was verified by electrophoresis using ethidium bromide staining as well as by measuring optical density (OD). OD 260 nm/OD 280 nm absorption ratio was accepted between 1.70 and 2.8. Reverse transcription of RNA Total RNA (700 ng) was reverse transcribed for 40 min at 48°C in 30 μl with 2.5 unit M-MuLV reverse transcriptase (Applied Biosystems, Foster City, CA, USA) in the presence of RNAse inhibitor (Applied Biosystems) using Random Hexamers (Applied Biosystems). Real-time PCR Conditions for real-time PCRs were optimized in a gradient cycler (Mastercycler Gradient, Eppendorf, Hamburg, Germany). Optimized conditions were transferred to the Bio-Rad real time PCR detection system as follows. Typical real-time PCR reaction was conducted with 2 μl cDNA template in a total volume of 25 μl, using the Bio-Rad iCycler Real Time PCR detection system (BioRad 1708740). For CLDNs and GAPDH the SYBR Green based real-time PCR method was used. Each PCR was conducted in 25 μl volume of 1 × PCR puffer (BioRad 1708882) with 300 nmol/l of each primer (Table 1 ) for 2 min at 95°C for initial denaturing, then 40 cycles of 95°C for 20 s, 63°C for 30 s, and 72°C for 30 s, followed by melting analyses from 55 to 95°C. Melting curve analyses resulted in single product specific melting temperatures for the analyzed CLDN1, CLDN3, CLDN4 and GAPDH genes. No primer–dimer formations were observed during the 40 real-time PCR amplification cycles. PCR reaction for each sample was done in duplicate in 96-well plates. In addition, the resulting real-time PCR product (10 μl) was loaded onto 2% agarose gel prepared in Tris-borate EDTA puffer and stained with ethidium bromide (Bio-Rad). Each PCR was optimized to ensure that no bands corresponding to genomic DNA amplification or primer–dimer pairs were present. The relative quantification method (described in detail by Pfaffl and coworkers [ 20 ] from the Technical University of Munich) was used for data analysis of real-time PCR. The program included a statistical evaluation. Data were analyzed for significant differences by analysis of variance using approximate tests [ 20 ]. The relative expression is based on the expression ratio of a target gene versus a reference gene (GAPDH). Electron microscopy In two cases both the normal breast tissue and the tumour sample were investigated by transmission electron microscopy in order to analyze the presence of TJs in tumours compared with nontumourous tissue, as described previously [ 21 ]. Results Immunohistochemical localization of CLDNs indicated that CLDNs were expressed along the entire length of the lateral plasma membranes between epithelial cells, including apical areas containing TJ structures. Using electron microscopy, TJ structures were identified in normal breast as well as in breast tumour tissue. Cell types in breast tissue are heterogeneous, and it is therefore noteworthy that CLDNs were expressed only in epithelial cells. Surrounding fibroblasts and adipocytes served as negative controls because these cells were negative for CLDN antibodies. Immunohistochemistry of CLDN1 The two antigen retrieval methods used yielded similar results. The monospecificity of the CLDN antibodies was tested by Western blot analyses, and a single band was found at approximately 20 kDa for CLDN1, CLDN3 and CLDN4 (Fig. 1 ). Immunohistochemical localization of CLDN1 indicated that CLDN1 was present in the membranes of normal duct cells in the majority of cases (Fig. 2a ). CLDN1 positivity was also observed in some cell membranes in pure DCIS cases or in the DCIS component of invasive carcinomas. On the other hand, major differences in CLDN1 expression were observed among tumours. Two major differences were noticed. First, in 36 of the 39 carcinomas examined, invasive tumour cells were either negative for CLDN1 or it was present in a scattered distribution among such tumour cells (Fig. 2b ). These findings were confirmed by confocal microscopy images (Fig. 2c ). There was no difference immunohistochemically in CLDN1 expression between different types of invasive carcinomas or between ductal carcinomas of different grade. The second major difference from normal tissue was that CLDN1 was frequently present (eight out of nine samples) in areas of apocrine metaplasia (Fig. 2d ). In these cases positivity was observed only in cells of apocrine metaplasia, and the neighbouring tumour tissue was negative for CLDN1. Immunohistochemistry of CLDN3 and CLDN4 CLDN3 positivity was observed in the majority (49/56) of cases examined (12 fibrocystic breasts, 5 DCIS, 23 IDC, 4 ILC, 2 mucinous, 2 tubular and 1 papillary breast carcinomas) and in all cases analyzed by confocal microscopy. There were no remarkable differences in CLDN3 expression in different types of breast lesions or between normal and tumour components (Fig. 3a,b ). CLDN3 expression is evaluable by standard immunohistochemistry and by comparing two antigen retrieval methods; in the majority of cases CLDN3, similar to CLDN1, was observed along the entire length of the epithelial cell membrane, but in a few cases, especially in normal tissues and in tissue with fibrocystic changes, CLDN3 appeared to be localized close to the apical end of the cells or at the basolateral membrane of the epithelial cells. These findings were seen on both confocal and light microscopy (Fig. 3c,d ). CLDN4 positivity was present in all 56 tissue sections in epithelial cell membranes as well as in the frozen section and breast tumour array analyzed by confocal microscopy. Intense positivity was observed in normal epithelial cells (Fig. 4a ) and in tumours of grade 2 and grade 3 (Fig. 4b ), and was greatly reduced in the majority of tumours of grade 1. CLDN4 was absent or its expression was greatly reduced in tumour cells of 17 out of 21 grade 1 tumours (Fig. 4c ), but it was present in the normal epithelial components within the same blocks (14 IDC NST [no special type], 5 special types, 2 ILC). Positivity was seen in 8 out of 11 grade 2 tumours and in 5 out of 7 grade 3 invasive carcinomas. In the normal tissue surrounding grade 2 and 3 tumours, the CLDNs were consistently present. CLDN4 was absent or its expression was decreased in the seven special-type breast carcinomas (mucinous, tubular, papillary) but it was present in the normal tissue surrounding these tumours. In contrast to CLDN1, CLDN4 was consistently absent in regions of apocrine metaplasia (Fig. 4d ). Real time-PCR results Real-time PCR using CLDN1, CLDN3, CLDN4 and GAPDH specific primers was performed in the 17 samples detailed above. mRNA was expressed in all of the analyzed cases. Relative quantification using GAPDH as the reference gene was applied [ 20 ]. mRNA expression in 13 invasive tumours out of the 17 samples was analyzed as the sample group, and the surrounding breast tissue was used as the control group. During the real-time PCR reaction, CLDN1 mRNA was detected in 12 out of 13 analyzed tumours and in 12 out of 13 corresponding surrounding nontumourous breast tissue samples. In normal, surrounding breast tissue, the mean value of crossing points (cycle threshold) was 34,02 whereas in the tumour tissue samples it was 36,22 (indicating low levels of CLDN1 mRNA). CLDN1 was downregulated by 12-fold in the sample (tumour) group in comparison with the control group using GAPDH as the reference gene. CLDN3 mRNA was present in all tumours and surrounding tissues. CLDN3 exhibited no expressional difference between invasive tumours and nontumourous tissue samples. CLDN4 mRNA was also found to be present in these 13 invasive tumours and surrounding mammary tissues using real time-PCR. According to relative quantification, CLDN4 mRNA expression did not appear to be changed in the invasive tumours in comparison with the surrounding normal tissue. It is important to emphasize that there were only two grade 1 invasive tumour samples among the 13 investigated invasive tumours. The program used by us and described by Pfaffl and coworkers [ 20 ] (Relative Expression Software Tool [REST] for group-wise comparison and statistical analysis of relative expression results in real-time PCR) includes a statistical analysis (analysis of variance). Using this program, we found no statistically significant differences in the distribution of data between the sample and control groups in terms of CLDN expression (the P values for CLDN1, CLDN3 and CLDN 4 were 0.166, 0.928 and 0.996, respectively). However, only 13 tumours were analyzed for RNA expression using real-time PCR because fresh tumour tissue and corresponding normal tissue surrounding the tumour were available in only 13 cases. In these cases the results of immunohistochemistry and real-time PCR appeared to correlate; specifically, CLDN1 appeared to be downregulated in tumours as compared with normal breast, whereas CLDN3 was expressed at similar levels in tumours and normal breast. In the case of CLDN4 expression the differences observed by immunohistochemistry between tumours of different grade were not observed by real-time PCR. CLDN4 mRNA expression did not appear to be changed in the invasive tumours in comparison with the surrounding normal tissue. Real-time PCR remains to be performed in a greater number of cases, including tumours of different grade. Apart from melting analysis, the specificity of real-time PCR results was confirmed on 2% agarose gel electrophoresis. The PCR product was seen at the appropriate size and nonspecific bands were not visualized. The specificity of real-time PCR of CLDN4 was also analyzed by sequencing (data not shown). Discussion In the present study we examined the expression in breast lesions of three members of the CLDN family, namely CLDN1, CLDN3 and CLDN4. The majority of studies published to date have described roles for CLDNs in forming TJs [ 22 ], conferring ionic selectivity [ 17 ] and functioning as a barrier [ 23 ], but only few data have been published on the expression of CLDNs in the breast [ 14 ]. CLDN1 and CLDN2 were the first described CLDN components of TJs, and so the majority of data are related to these proteins. It was demonstrated that CLDN1 and CLDN2 function as major structural components of TJ strands, and that occludin is an accessory protein within this structure [ 8 , 24 ]. Well developed TJs found in the lung contain CLDN3, CLDN4 and CLDN5, which are abundant, whereas levels of expression of CLDN1 and CLDN2 are reduced [ 12 ]. These findings suggest that different members of the CLDN family are involved in the formation of TJ strands in different tissues. The extent to which CLDN expression is tissue dependent remains a subject of research interest, as does the physiological relevance of the existence of multiple CLDN species. In CLDN1-, CLDN2- and CLDN3-transfected mouse L-cells, Kubota and coworkers [ 25 ] showed that CLDNs may act as calcium-independent adhesion molecules, although the adhesion strength is low. It was hypothesized by Rangel and coworkers [ 26 ] that the precise ratio of different CLDNs determined the permeability of TJs. In ovarian tumours, the same group suggested that CLDN3 and CLDN4 are required for signalling through survival or proliferative pathways. In the present study, immunohistochemistry revealed expression of CLDNs in TJs along the lateral plasma membranes of epithelial cells and along the basal plasma membrane of the epithelium, suggesting that CLDNs are not exclusively localized to TJs. Localization of these three CLDNs together in breast tissue sections has not previously been reported, but based on the observations presented here CLDN1, CLDN3 and CLDN4 are not exclusively localized to areas of TJs. Gregory and coworkers [ 27 ] described similar observations for CLDN1 in the rat epididymis. We demonstrated that immunohistochemically detectable CLDN1 protein is absent or its expression is markedly decreased in the majority of different types of invasive breast carcinomas as compared with normal ducts. Of 39 malignant tumours examined by immunohistochemistry, 36 were negative for CLDN1 or it was present in a scattered distribution among the tumour cells. CLDN1 mRNA expression investigated by real-time PCR confirmed this finding. The mRNA level in tumour is considerably lower than that in normal breast tissue. Theoretically, several regulatory pathways could influence CLDN1 protein synthesis and expression. How these regulatory pathways are involved in cellular transformation and oncogenic progression in the breast remains to be elucidated. Previous studies reported that CLDN1 mRNA expression is absent or downregulated in the majority of breast cancer cell lines [ 28 ]. Kramer and coworkers [ 7 ] found no evidence for alterations in the CLDN1 gene being responsible for the loss of expression of CLDN1 protein. The same group found a similar loss of CLDN1 expression in five primary breast tumours. This finding is in contrast to the broad spectrum of CLDN1 expression found in other human tissues, such as liver and pancreas (our unpublished data). This may suggest that CLDN1 might be involved in the development of breast cancer or other epithelial tumours [ 7 ]. Whether TJs are simply lost in breast neoplasia and whether downregulation of CLDN1 is among the causative factors in this process are unknown. Alterations in the number, appearance and permeability of TJs have been described in various tumour types [ 21 , 29 ]. Abnormalities in TJ permeability and number greatly influence the transepithelial flux of growth factors [ 30 ] and hence the development of epithelial tumours [ 31 ]. In two cases, both normal breast tissue and tumour sample were investigated by transmission electron microscopy. Membrane densities corresponding to TJ structures were identified in both normal and neoplastic epithelial cells. The eventual abnormalities of TJs in breast tumours must be analyzed in a large series of breast tumours. Although the similar CLDN3 and CLDN4 TJ proteins are highly expressed in breast neoplasia, we hypothesize that mechanisms other than alteration in TJs are involved in the loss of CLDN1 expression in breast tumours. The role played by CLDN1 positivity in apocrine cells observed in this study requires further investigation. In the present study, CLDN3 appeared to be expressed in a similar manner in primary breast cancers as in normal epithelium. Immunohistochemistry and real-time PCR yielded similar findings without notable upregulation or downregulation in different groups of tumours or compared with normal epithelium. In the majority of cases, CLDN3 was found to be localized along the entire membrane of epithelial cells, but in some cases it was only detected close to the apical end or at the basolateral membrane of the epithelial cells. Although this observation was confirmed by both light and confocal microscopy, in the breast this staining pattern has not previously been reported. Interestingly, Rangel and coworkers [ 26 ] did not find CLDN3 exclusively at the membrane in any of the samples studied; the majority of samples of primary ovarian tissue exhibited a combination of cytoplasmic and membrane staining. In gut, Rahner and coworkers [ 32 ] found CLDN3 to be strongly expressed in the surface epithelial cells of the stomach fundus, and have found it to be distributed predominantly along the basolateral membrane and not at the TJ. That group concluded that CLDNs have different patterns of expression in different gastrointestinal tissues – patterns that should account for differences in paracellular permeability. We found that CLDN4 protein expression was downregulated in grade 1 IDCs. Further sample collection and analysis are required to confirm this result at the mRNA level, because only two grade 1 IDCs were available for PCR examination in our series. However, mRNA and protein were expressed in grade 2 and 3 IDCs to degrees similar to expression in surrounding breast tissue. In a study conducted in ovarian tissue, Hough and coworkers [ 16 ] found that expression of CLDN3 and CLDN4 protein was undetectable in normal ovarian cells and highly expressed on the membranes of ovarian carcinoma cells. Contrary to the findings in ovarian tumours reported by Rangel and coworkers [ 26 ] that CLDN3 and CLDN4 exhibited cytoplasmic staining and that over-expression of these proteins were associated with malignancy, in our study in breast we did not observe clear positivity of CLDNs in cytoplasm. A close relation between CLDN4 expression and changes in both conductance and ionic selectivity has been described. Further important observations were reported by van Itallie and coworkers [ 17 ]; they found that the number of junction strands was increased by over-expression of CLDN4, and that the levels of several other strand proteins were unaffected. Loss of CLDN4 from the cell surface coincides with dramatic changes in the morphology of TJ fibrils. Removal of CLDN4 disrupts fibril organization and increases junction permeability [ 15 ]. A recent study [ 33 ] identified CLDN4 as a potent inhibitor of the invasiveness and metastatic phenotype of pancreatic carcinoma cells by playing role in the transforming growth factor-β and Ras/Raf signal regulated kinase pathway. No studies to date have reported decreased expression of CLDN4 in grade 1 breast carcinomas, in special types of breast tumour, or in apocrine metaplasia compared with normal epithelium, as were observed in our study. The mRNA expression of different CLDNs in a large number of special-type breast carcinomas and tumours of different grades requires investigation so that the decrease in CLDN4 expression in special-type breast carcinomas and IDCs of grade 1 can be unequivocally confirmed. Conclusion Our observations suggest that, in breast tissue, CLDN3 expression is similar in tumours and surrounding normal tissue, as demonstrated by immunohistochemistry and real-time PCR. In contrast, absent or decreased expression of CLDN1 in invasive breast carcinomas was demonstrated at both mRNA and protein levels. These data suggest that CLDN1 is involved in invasion and metastasis, and in mammary carcinogenesis. Furthermore, loss of or decrease in CLDN4 expression in grade 1 IDCs and the absence of CLDN4 in special-type breast carcinomas and in areas of apocrine metaplasia in benign breast tissue, as compared with normal epithelium, suggest a role for CLDN4 in cellular differentiation. The implications of this unique feature require further investigation. The important issue is whether loss of CLDN1 and CLDN4 plays a significant role in cell–cell adhesion and tumour differentiation. Abbreviations CLDN = claudin; DCIS = ductal carcinoma in situ ; GAPDH = glyceraldehyde-3-phosphate dehydrogenase; IDC = invasive ductal carcinoma; ILC = invasive lobular carcinoma; OD = optical density; PCR = polymerase chain reaction; TJ = tight junction; ZO = zonulae occludens. Competing interests The author(s) declare that they have no competing interests. Authors' contributions A-MT and JK selected the cases, performed the pathological analysis (mainly JK), evaluated the immunohistochemical slides, performed the RT-PCR reactions (mainly A-MT) and wrote the manuscript. SP participated in the confocal microscopic investigation of the immunofluorescence slides. ÁS carried out the immunohistochemical reactions. CP participated in the evaluation of the RT-PCR reactions, the statistical analysis of the data and the writing of the manuscript. PKN participated in designing the CLDN primers. LS was the designer of the CLDN primers. AK participated in the evaluation of the RT-PCR reactions and the statistical analysis of the data and in the writing of the manuscript. KB participated in the evaluation of the immunohistochemical reactions and in the writing of the manuscript. ZS participated in the evaluation of the immunohistochemical reactions and in the writing of the manuscript in its final stage.

D:\keerthana\PMC001xxxxxx\PMC1064136.xml

1064138

Claudin 7 expression and localization in the normal murine mammary gland and murine mammary tumors

Introduction Claudins, membrane-associated tetraspanin proteins, are normally associated with the tight junctions of epithelial cells where they confer a variety of permeability properties to the transepithelial barrier. One member of this family, claudin 7, has been shown to be expressed in the human mammary epithelium and some breast tumors. To set the stage for functional experiments on this molecule, we examined the developmental expression and localization of claudin 7 in the murine mammary epithelium and in a selection of murine mammary tumors. Method We used real-time polymerase chain reaction, in situ mRNA localization, and immunohistochemistry (IHC) to examine the expression and localization of claudin 7. Frozen sections were examined by digital confocal microscopy for colocalization with the tight-junction protein ZO1. Results Claudin 7 was expressed constitutively in the mammary epithelium at all developmental stages, and the ratio of its mRNA to that of keratin 19 was nearly constant through development. By IHC, claudin 7 was located in the basolateral part of the cell where it seemed to be localized to discrete vesicles. Scant colocalization with the tight-junction scaffolding protein ZO1 was observed. Similar results were obtained from IHC of the airway epithelium and some renal tubules; however, claudin 7 did partly colocalize with ZO1 in EPH4 cells, a normal murine mammary cell line, and in the epididymis. The molecule was localized in the cytoplasm of MMTV- neu and the transplantable murine tumor cell lines TM4, TM10, and TM40A, in which its ratio to cytokeratin was higher than in the normal mammary epithelium. Conclusion Claudin 7 is expressed constitutively in the mammary epithelium at approximately equal levels throughout development as well as in the murine tumors examined. Although it is capable of localizing to tight junctions, in the epithelia of mammary gland, airway, and kidney it is mostly or entirely confined to punctate cytoplasmic structures, often near the basolateral surfaces of the cells and possibly associated with basolateral membranes. These observations suggest that claudin 7 might be involved in vesicle trafficking to the basolateral membrane, possibly stabilizing cytoplasmic vesicles or participating in cell–matrix interactions.

Introduction The claudins comprise a large family of tetraspanin membrane proteins thought to be the major barrier-forming proteins of tight junctions, the cell–cell contacts at the apical border of epithelial cells that control the paracellular movement of solutes. These proteins are highly conserved, with four transmembrane domains and two hydrophobic extracellular loops; the latter are thought to mediate cell–cell adhesion [ 1 ] and to confer specific paracellular permeability properties on cell monolayers [ 2 , 3 ]. Claudin 7 shares the general structural characteristics of the family, differing primarily in its amino-terminal cytoplasmic tail [ 4 ]. The molecule has been shown to be associated with epithelial cells in the human breast [ 5 ], and its loss is associated with some breast and head and neck malignancies [ 5 , 6 ]. It has been shown to be expressed in parts of the renal tubule [ 7 ] and the airway epithelium [ 8 ], where it is localized to the basolateral aspects of the cells. Here we show that claudin 7 is constitutively present in the epithelium of the murine mammary gland, again localized, not to tight junctions but to punctate structures at or near the basolateral surfaces of the cells. It was present at all cell borders of several murine mammary tumors. Nonetheless, the protein can localize to tight junctions, as shown by its partial colocalization with ZO1 in cultured mammary epithelial cells and epididymis, suggesting a possible dual function depending on tissue type. Method Animals and tissue preparation CD-1 mice, purchased from Charles River Breeding Laboratory (Wilmington, DE), were maintained in the USDA-approved Animal Resource Center of the University of Colorado Health Sciences Center. All procedures were approved by the Institutional Animal Care and Use Committee. The fourth mammary glands of virgin female mice at 3, 6, and 12 weeks of age, of female mice during early gestation (5–7 days), mid-gestation (12 days) and late gestation (18 days), at days 2 and 10 of lactation, and at days 21 and 29 of involution were collected after killing with a lethal dose of pentobarbital. Liver, lung, and kidneys were obtained from virgin female mice and epididymis from male mice. The day on which vaginal plugs were observed was counted as day one of pregnancy. Mice overexpressing the Her2 / neu oncogene [ 9 ] were obtained from Jackson Laboratories, and mammary tumors were dissected and frozen when they reached about 0.5 cm in diameter. The transplantation tumor models TM4, TM10L, and TM40 are maintained in the Medina laboratory as described [ 10 ]. Tissues were flash-frozen in liquid nitrogen in Tri-Zol reagent (Gibco BRL, Life Technologies) for total RNA purification. For protein extractions, the tissue was washed twice in 1 ml of phosphate-buffered saline (PBS) and then placed in 3 volumes of 1% Nonidet P40 buffer (25 mM Hepes-NaOH, pH 7.4; 150 mM NaCl, 4 mM EDTA, 1% Nonidet P40) containing protease inhibitors (10 μg/μl leupeptin, 10 μg/μl aprotinin, 10 μM pepstatin A, 1 mM phenylmethylsulphonyl fluoride, in dimethyl sulphoxide). For hybridizations and immunohistochemistry of frozen tissue in situ , tissues were placed in Tissue-Tek OCT 4583 Compound (Sakura, Torrance, CA) and flash-frozen in isopentane and liquid nitrogen. All samples were stored at – 70°C until ready for use. Characterization of an antibody against claudin 7 A rabbit antibody was custom-made against the carboxy-terminal peptide of murine claudin 7, APRSYPKSNSSKEYV, by Zymed Laboratories (San Francisco, CA). This affinity-purified antibody bound only a 23 kDa peptide by western blotting, did not cross-react with claudin 1, its closest congener, in claudin-1-transfected Cos cells or at the apical border of the mammary epithelium where claudin 1 is present at junctional complexes, and stained only epithelia where its mRNA has been demonstrated. Staining was blocked by preincubation of the antibody with the peptide. Cos cells transfected with full-length claudin 7 were stained by this antibody but not by an antibody against claudin 1. Protocol for real-time polymerase chain reaction For real-time reverse transcriptase polymerase chain reaction, RNA was purified with the Qiagen system: 1 μg of DNase-treated RNA was used for each reaction. Triplicate tissue samples were assayed in duplicate with the ABI Prism 7700 sequence detection system, with lung as the positive control. Primer sequences were as follows: murine claudin 7, forward 5' -CGAAGAAGGCCCGAATAGCT-3' (338-337), reverse 5' -GCTACCAAGGCAGCAAGACC-3' (407–388), probe5' -GCCACAATGAAAACAATGCCTCCAGTCA-3' (359–386); murine cytokeratin 19, forward 5' -TTTAAGACCATCGAGGAC-3', reverse 5' -TCATACTGACTTCTCATCTCAC-3'. Immunohistochemistry and digital confocal microscopy Rat anti-mouse ZO1 antibody was obtained from Chemicon International Inc. (Temecula, CA). Tissue sections were cut at 10 μm thickness from frozen blocks with a Damon/IEC division minotome set at – 18 to -20°C. Sections were collected onto Cell-Tak coated coverslips and were further vapor-fixed with paraformaldehyde for 15 min. Sections were never allowed to dry. PBS was added carefully so as to not disrupt the sections. Tissue was permeabilized with 1% Triton X-100 for 15 min, rinsed well with PBS and blocked with sterifiltered 10% normal donkey serum for 20 min. All antibody solutions were microfuged for 20 min before use. The claudin antibody was diluted 1:1000. Primary incubations were for 1 hour at 21–22°C, followed by extensive washes in PBS, generally six times for 5 min each. Secondary antibodies were diluted in accordance with the manufacturer's instructions, in PBS alone. The host species of all secondary antibodies was donkey and all secondary antibodies were cross-adsorbed against mouse serum proteins. Antibodies used were conjugated to fluorescein, CY3 or CY5. Secondary antibodies were combined with 0.6 μg/ml 4,6-diamidino-2-phenylindole, and incubated on the tissue for 1 hour. Coverslips were rinsed briefly and permitted to soak overnight in PBS. Images were collected with SlideBook software (Intelligent Imaging Innovations, Inc., Denver, CO) on a Nikon Diaphot TMD microscope equipped for fluorescence with a xenon lamp and filter wheels (Sutter Instruments, Novato, CA), fluorescent filters (Chroma, Brattleboro, VT), cooled charge-coupled device camera (Cooke, Tonawanda, NY) and stepper motor (Intelligent Imaging Innovations, Inc., Denver, CO). Multi-fluor images were merged, deconvolved, and renormalized with SlideBook software. Results Developmental expression of claudin 7 mRNA in the mammary gland Figure 1a shows claudin 7 gene expression as a function of developmental stage in the mouse mammary gland from the virgin animal (non-pregnant non-lactating) through pregnancy (days P5, P12, and P18), lactation (days L2, L10, and L19), and involution (days L22 and L29). Gene expression increased more than 1000-fold between the virgin and early lactating gland, leveling out through lactation and decreasing at late involution (day L29) with the loss of epithelial cells. To determine whether expression of claudin 7 was a function of developmental stage or epithelial cell number, we also examined the expression of keratin 19, found only in luminal epithelial cells [ 11 ]. As can be seen, claudin 7 expression parallels that of keratin 19. Further, the ratio of claudin 7 to keratin 19 (Fig. 1a , dotted line) is relatively constant through pregnancy and early lactation, increasing only during later lactation when the expression of keratin 19 mRNA decreases significantly. At late involution, day L29, claudin 7 expression was lower than that of keratin 19, for reasons that are not clear but probably have to do with the types and characteristics of epithelial cells present during late glandular remodeling. Similar results were obtained from microarray analysis (data not shown). In situ hybridization with 35 S-labeled RNA probes to claudin 7 showed that the mRNA was localized to the epithelium in virgin, pregnant, and lactating animals (Fig. 1b ). We conclude that, in the normal mammary gland, claudin 7 is an epithelial cell marker expressed at approximately constant levels through development. The very large changes in expression levels during pregnancy most probably reflect an increase in cell number as the mammary epithelium expands from a system of sparse ducts in the virgin gland, to a dense mass of lactating cells in a gland in which the adipose tissue is largely obliterated. Immunolocalization of claudin 7 in the mammary gland We made and characterized an affinity-purified antibody against the cytoplasmic tail of claudin 7 (see Methods) that proved quite satisfactory for both western blots (Fig. 2a ) and immunohistochemistry on both frozen and paraffin sections. Given the well-known association of claudins with tight junctions, we were surprised to find that claudin stained the basolateral region of mammary cells in all stages of development (Fig. 2b,c ). A ductal structure surrounded by adipose stroma is shown in the virgin; alveoli from the pregnant and lactating glands are also shown (Fig. 2b ). Staining was blocked when the antibody was preabsorbed with the claudin 7 peptide against which it was made (Fig. 2b , lower right). As with the in situ analysis, stain was observed only in luminal epithelial cells. Although there seems to be some colocalization of claudin 7 with stain for the tight-junction scaffolding protein ZO1 in the lower-power images of Fig. 2b , at higher magnification (Fig. 2c ) claudin 7 was entirely localized to the basal and lateral cytoplasmic regions, where it often appeared punctate in nature, particularly when it was not closely apposed to the cell border. Stain was excluded from nuclei (blue) and largely from tight junctions (green), as shown by the minimal overlap between the green and red stain in Fig. 2c . It seems likely that there is basolateral membrane staining in this tissue, but the basal and lateral membranes of the mammary epithelial cells are deeply infolded and membrane localization cannot be assessed at the magnifications possible with the light microscope. When immunohistochemistry is used in the characterization of claudins, there is always a concern that the antibody cross-reacts with another claudin species. We have found claudins 1, 3, and 8 in the murine mammary epithelium (MC Neville, unpublished data); however, claudin 7 stain shows entirely different patterns from these claudins. As shown in Fig. 2a , claudin 1 is localized at the position of the junctional complexes; claudin 3 is present in epithelium from virgin and pregnant glands, whereas claudin 8 is present only during lactation. These experiments will be described in detail elsewhere. Claudin 7 localization in other cell types Sukumar and colleagues [ 5 ] reported that claudin 7 was localized to tight junctions of the human mammary carcinoma cell line MCF-7 by using an antibody directed toward the cytoplasmic tail of the human protein, which differs by one amino acid from the mouse protein. We therefore examined claudin 7 staining in the normal mouse mammary cell line EPH4 [ 12 , 13 ]. Figure 3a shows the distribution of claudin and ZO-1 in EPH4 cells. Claudin 7 distinctly colocalizes with ZO-1 with a slightly less compact distribution, as shown by the zy image at the top of the figure as well as the xy image just below. A scattering of cytoplasmic vesicles containing claudin 7 can also be seen. This finding suggests that claudin 7 is capable of localizing to tight junctions, a conclusion supported by the distribution of claudin 7 in the epididymis (Fig. 3b ), where claudin 7 colocalized with ZO-1 at the apical borders of some cells but not others. Punctate cytoplasmic stain for claudin 7 could also be observed in some cells. Claudin 7 mRNA was reported in lung and kidney, with liver being negative [ 4 ]. We therefore examined these tissues and found that claudin 7 was present in cells lining bronchiolar structures of the lung, where its distribution was mainly cytoplasmic as observed previously [ 8 ] (Fig. 3c ). However, careful examination of a black-and-white image at the highest magnification in the luminal cells of this structure shows some claudin 7 to be colocalized with ZO-1, suggesting that it might also be associated with tight junctions in these cells. There was no stain in the pulmonary alveoli. In the renal cortex we found claudin 7 localized only to certain segments, identified as connecting tubules and cortical collecting ducts by Li and colleagues [ 7 ], where stain was distinctly basolateral (Fig. 3d ). Because our antibody was different from that used by Li and colleagues, our finding confirms this localization. At high magnification the stain is punctate, as in the mammary and airway epithelia. No specific claudin stain was observed in the liver (Fig. 3e ). These findings indicate that claudin 7 is capable of localizing to tight junctions, as in cultured mammary epithelial cells and epididymis; however, in mammary gland, airway, and kidney it is mostly or entirely confined to punctate cytoplasmic structures, often near the basolateral surfaces of the cells and possibly associated with the basolateral membranes. In no tissue was the protein observed in nuclei. Claudin 7 localization in mammary tumors We examined four types of mammary tumor: tumors arising in the transgenic mouse expressing the Erb2 receptor under the control of the mammary tumor virus promoter [ 9 ], and three transplantable tumors obtained from the laboratory of DM [ 10 ]. All tumors expressed claudin 7 mRNA at levels no more than twice that in the lactating mammary gland when normalized to ribosomal RNA (Fig. 4a ). Although tumors themselves are more closely related to the pregnant gland, we chose the lactating gland for comparison because, like the tumors, it is composed largely of epithelial cells. Interestingly, the ratio of claudin 7 to keratin 19 in the tumors ranged from 4 to 60 times that in the lactating gland, reflecting a lower expression of keratin 19, and possibly a loss of differentiation, in the tumors. In all murine tumors examined here, claudin 7 was localized to the perimembrane region, as illustrated by a section of the TM4 tumor (Fig. 4b ). None of these tumors showed ZO1 staining, indicating that they lacked tight junctions, so that the claudin 7 observed here was probably associated with membrane vesicles and possibly with basolateral membranes, as in the normal cells of the mammary gland. Sukumar and colleagues [ 5 ] observed a loss of claudin 7 expression in some human tumors, particularly lobular tumors. However, this finding was not true of the mouse mammary tumors examined. Discussion We find claudin 7 to be a constitutive component of the mammary epithelium, where it is localized to the basolateral regions of the cell. The finding that the ratio of its mRNA to that of keratin 19 is relatively constant throughout the developmental cycle suggests that the molecule might be an alternative marker to keratin for the proportion of epithelial cells in the mammary gland. It is not clear whether the more than 1000-fold increase in expression between the virgin gland and early lactation indicates that the number of epithelial cells increases in this proportion, because part of the increase could be a function of an increase in cell size as the cells differentiate. The rapid increase between the virgin gland and pregnancy day 5 is consistent with studies showing a peak of thymidine incorporation between days 2 and 5 of pregnancy, when about 25% of the cells were labeled [ 14 , 15 ]. In both studies, labeling decreased thereafter but remained about 10% almost to the end of pregnancy, consistent with the continued increase in both claudin 7 and keratin mRNA up to day P18. Our findings are consistent with images of claudin staining in the human mammary gland, where a diffuse diaminobenzidene stain from alkaline phosphatase localization was present throughout the cytoplasm [ 5 ]. However, in that study no attempt was made to localize claudin 7 stain with tight-junction components. Basolateral localization of other claudins has been observed. Claudin 1 was localized to the cytoplasm in the epididymis [ 16 ], intestine [ 17 ], and cornea [ 18 ]. Rahner and colleagues [ 19 ] observed claudins 3, 4, and 5 to be laterally distributed in various portions of the gastrointestinal track. The finding that claudin 7 is exclusively located in non-tight-junction regions of mammary and renal epithelial cells [ 7 ] suggests that claudins might have functions other than the regulation of tight-junction permeability. Our images seem to be the first to show claudin 7 stain at sufficiently high resolution to show punctate cytoplasmic stain in mammary, airway, and renal epithelial cells. Even at this resolution, obtained with digital confocal imaging with a resolution of about 200 nm, it is not possible to discern with certainty whether claudin 7 is inserted into the convoluted basolateral membranes of these cell types. If so, it is possible that the cytoplasmic spots represent vesicles en route to and from to the basolateral membranes, where claudin 7 might interact with components of the extracellular matrix. As a precedent, claudin 11, an oligodendrocyte protein, has been shown to interact with α1-integrin and to regulate the proliferation and migration of oligodendrocytes in culture [ 20 ]. Other possibilities are that vesicular claudins could regulate tight-junction permeability by sequestering tight-junction regulatory molecules away from tight junctions, or they could be involved in the stabilization of specialized vesicle compartments within the cytoplasm. Matsuda and colleagues, using time-lapse photography, have shown that claudin-containing cytoplasmic vesicles can originate from the tight junctions as the epithelial layer remodels [ 13 ]. However, because claudin 7 is never observed in association with tight junctions in mammary epithelial cells, it seems unlikely that these vesicles have an origin in the junctional complex. Our mammary epithelial cell model, EPH4 cells, does possess a complement of cytoplasmic vesicles that could allow live-cell imaging studies to determine the origin and disposition of claudin 7-containing vesicles. It is likely that the function of the vesicles can be better assessed after analysis of the composition of the claudin 7-containing cytoplasmic vesicles. Claudin 7 expression was inversely correlated with histological grade in a large series of breast tumors [ 5 ]. These same authors found, similarly to our observations with EPH4 cells, that claudin 7 colocalized with ZO1 in MCF7 breast cancer cells and could also be observed in cytoplasmic spots. It was present in luminal cells of the human breast, where the diaminobenzidine staining seemed to be localized to basolateral membranes, although it is difficult to draw a firm conclusion in the absence of an apical marker like ZO1. An image of ductal carcinoma in situ in that paper shows a distribution of stain remarkably similar to that of the TM10 tumor shown in Fig. 4b . This tumor line has been shown to have a ductal morphology [ 10 ]. Thus, we conclude that low-grade breast carcinomas show a cellular distribution of stain similar to that observed in the murine tumors. Interestingly, the murine tumors showed claudin 7 expression at the mRNA level equal to or higher than that of the lactating mammary gland, where the largest proportion of the cells are the luminal epithelial cells that give rise to mammary tumors. TM4, the most tumorigenic of these lines, had the highest claudin 7 expression, although the level was quite variable and not significantly different from the other lines examined. Interestingly, TM4 and the MMTV- neu tumor had ratios of claudin 7 to cytokeratin less than one-eighth of that in the slower-growing TM10 line. All of the tumors had claudin 7 to cytokeratin ratios significantly higher than those in the pregnant or lactating mammary gland. These observations suggest that loss of the cytoplasmic architectural stability conferred by keratin might be an early event in tumorigenesis. Together with the data from the Sukumar laboratory [ 5 ], we might speculate that loss of claudin 7, as occurs in high-grade tumors, alters cell–matrix interactions, allowing a greater degree of cell mobility and contributing to metastasis. Conclusion Claudin 7 is a constitutive component of mammary epithelium at all stages of development, maintaining a level of gene expression that is approximately proportional to the amount of epithelial tissue in the gland. The protein is not associated with tight junctions but is found in particulate structures in the cytoplasm, generally denser at basolateral borders of the cells. Its distribution is similar in mammary tumors, at least in more well-differentiated ones. Claudin 7 function in the mammary epithelium is currently unknown but might be related to vesicle trafficking or cell-matrix interactions. Abbreviations IHC = immunohistochemistry; PBS = phosphate-buffered saline. Competing interests The author(s) declare that they have no competing interests. Authors' contributions BB performed the real-time polymerase chain reaction and in situ experiments and was involved in antibody development and immunocytochemistry. She wrote the initial draft of the manuscript. TR assisted with the immunocytochemistry. SKN provided general molecular knowledge for experimental design and provided valuable input into the manuscript. DM provided the TM series of tumors. MCN conceived the study, performed the high-resolution immunocytochemical analysis and completed the manuscript for submission. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064138.xml

1064139

Long-term prognostic significance of HER-2/neu in untreated node-negative breast cancer depends on the method of testing

Introduction The prognostic significance of HER-2/ neu in breast cancer is a matter of controversy. We have performed a study in 101 node-negative breast cancer patients with long-term follow-up not treated in the adjuvant setting, and analysed the prognostic significance of immunohistochemistry (IHC) and fluorescence in situ hybridisation (FISH), both separately and in combination, in comparison with traditional prognostic factors. Methods Overexpression was classified semiquantitatively according to a score (0 to 3+) (HER-2_SCO). FISH was used to analyse HER2/ neu amplification (HER-2_AMP). Patients classified 2+ by IHC were examined with FISH for amplification (HER-2_ALG). Patients with 3+ overexpression as well as amplification of HER-2/ neu were positive for the combined variable HER2_COM. These variables were compared with tumour size, histological grade and hormone receptor status. Results HER-2_SCO was 3+ in 20% of all tumours. HER-2_ALG was positive in 22% and amplification (HER-2_AMP) was found in 17% of all tumours. Eleven percent of the tumours showed simultaneous 3+ overexpression and amplification. Only histological grade (relative risk [RR] 3.22, 95% confidence interval [CI] 1.73–5.99, P = 0.0002) and HER-2_AMP (RR 2.47, 95% CI 1.12–5.48, P = 0.026) were significant for disease-free survival in multivariate analysis. For overall survival, both histological grade (RR 3.89, 95% CI 1.77–8.55, P = 0.0007) and HER-2_AMP (RR 3.08, 95% CI 1.24–7.66, P = 0.016) retained their independent significance. Conclusion The prognostic significance of HER-2/ neu in node-negative breast cancer depends on the method of testing: only the amplification of HER-2/ neu is an independent prognostic factor for the long-term prognosis of untreated node-negative breast cancer.

Introduction Human epidermal growth factor receptor-2 is a proto-oncogene that encodes a cell-surface receptor designated HER-2/ neu or c-erbB-2. Gene amplification and/or protein overexpression occurs in 14–30% of all breast cancers [ 1 - 3 ]. Initially, the adverse prognostic impact of HER-2/ neu in breast cancer was the main focus of research. However, results from different study groups were not entirely consistent. Studies that supported the initially reported adverse prognosis in breast cancer [ 1 , 3 ] were later followed by reports that failed to show any association with prognosis [ 4 , 5 ]. Although no consensus exists concerning the prognostic value of HER-2/ neu , an increasing quantity of data indicates a predictive value for the efficacy of certain adjuvant therapies. The response of HER-2/ neu -positive breast cancer patients to tamoxifen is significantly worse than for HER-2/ neu -negative patients [ 6 , 7 ] even though this point of view is not unopposed [ 8 , 9 ]. More recently, it was shown that aromatase inhibitors might provide more benefit than tamoxifen in patients with tumours positive for erbB-1 and/or erbB-2 [ 10 ]. However, the strongest evidence for a predictive role for HER-2/ neu comes from several retrospective trials that consistently showed that HER-2/ neu -positive patients responded better to an anthracycline-based therapy than to treatment with cyclophosphamide, methotrexate and fluorouracil [ 11 - 13 ]. Because of this predictive impact of HER-2/ neu , many of the studies on the prognostic role of HER-2/ neu cannot be reliably interpreted when the patients enrolled were treated in an adjuvant setting. To avoid this potential bias, we have performed a retrospective study on the prognostic impact of HER-2/ neu in a historical cohort of node-negative T1/T2 breast cancer patients who were at that time not being treated in an adjuvant setting. A point of utmost importance when assessing the utility of HER-2/ neu as a prognostic factor is the technique of HER-2/ neu testing. Principally, gene-based assays such as Southern blot analysis or fluorecence in situ hybridisation (FISH) have to be distinguished from assays that assess the level of protein expression, such as western blot analysis or immunohistochemistry (IHC). When analysing the published studies of HER-2/ neu as a prognostic factor with regard to the technique used, Ross and colleagues [ 14 , 15 ] found that those studies that applied FISH to the assessment of gene amplification found an association of HER-2/ neu with the prognosis of the patients, whereas studies that used IHC for the assessment of protein expression gave rather ambiguous results. Reasons for the apparently worse performance of IHC than that of FISH could be differences in the sensitivity of the applied antibodies [ 16 ] or the lack of a uniform scoring system in most of the older studies that used IHC. Standardisation of HER-2/ neu testing has received considerably more attention in recent years owing to the advent of trastuzumab, a monoclonal antibody that affords prolonged survival in patients with metastatic breast cancer [ 17 ]. To improve the quality of IHC when testing for HER-2/ neu status before starting a therapy with trastuzumab, a standardised scoring protocol was developed. On the basis of these findings, an algorithm incorporating FISH only in doubtful cases (2+ in IHC) was introduced. The aim of our present study on 101 node-negative breast cancer cases was to compare gene amplification by FISH with protein expression by IHC with the standardised scoring system, the above-mentioned algorithm and the combination of HER-2/ neu overexpression and amplification in the prognostic value of HER-2/ neu . Methods Patients The study cohort consisted of 101 lymph-node-negative breast cancer patients who were treated at the Department of Obstetrics and Gynecology of the Johannes Gutenberg University Mainz between 1988 and 1993. Patients were all treated with surgery and did not receive any systemic therapy in the adjuvant setting. The established prognostic factors (tumour size, histological grade and steroid receptor status) were collected from the original pathology reports of the gynaecological pathology division within our department. Patients were treated either with modified radical mastectomy ( n = 58) or breast-conserving surgery followed by irradiation ( n = 43). Because the administration of adjuvant systemic therapy was not allowed in this study, we focused on node-negative breast cancer patients with pT1 and pT2 tumours and without any evidence of metastasis at the time of surgery. The median age of the patients at surgery was 56 years (range 29–86 years). The median time of follow-up was 131 months for the patients still alive at the time of analysis. Within this follow-up period, 31 patients relapsed, 20 patients died of breast cancer and 10 patients died of unrelated causes. The patients dying of causes other than breast cancer were censored for the survival analyses at their date of death. HER-2/ neu amplification determined by FISH FISH for HER-2/ neu gene amplification was performed with the Appligene Oncor HER-2/ neu gene amplification system (Ventana Medical Systems, Tucson, AZ, USA), following the supplier's instructions. In brief, fresh frozen slides were first treated with a protein-digesting enzyme at 37°C for 10 min, washed in 2× sodium chloride/sodium citrate (SSC) at 22°C, dehydrated in an 75–100% ethanol series and air dried. Tissue sections were than denatured for 5 min in 70% formamide, pH 7.5, at 75°C, followed by rinsing with 100% ethanol and air drying. Appligene Oncor HER-2/ neu DNA probe was prewarmed for 5 min at 37°C before application. Slides were than incubated for 24 hours at 37°C in a humidified chamber. After hybridisation, the slides were washed in 2× SSC for 5 min at 72°C; this was then followed by a wash in phosphate-buffered detergent (Oncor, Gaithersburg, MD, USA) at room temperature for 5 min. Detection was achieved with the Appligene Oncor fluorescein-labelled anti-digoxigenin antibody (Ventana Medical Systems). The slides were incubated with this antibody for 5 min in a humidified chamber at 37°C. Slides were then subjected to three washes (2 min each) in phosphate-buffered detergent at room temperature and were stored in the dark at -20°C for up to 5 days before analysis. The nuclei were counterstained with a propidium iodide/antifade solution (Oncor, Gaithersburg, MD, USA). Appropriate positive controls were included in each staining run. A serial section of each slide used for FISH was stained with haematoxylin and eosin to control for the presence of invasive tumour formations. Additionally, serial sections (6 μm thick) of formalin-fixed paraffin-embedded blocks from five randomly selected amplified and five randomly selected non-amplified tumours were deparaffinised and then subjected to the staining protocol outlined above. Interpretation of FISH results Analysis was performed with an Axioskop fluorescence microscope (Zeiss, Jena, Germany). Images were captured with an analogue camera (Leica, Bensheim, Germany). In each quadrant of the slide, the number of fluorescein signals was counted in 20 nuclei of invasive tumour cells (that is, a total of 80 tumour nuclei). Cases were considered amplified if the mean number of fluorescence signals was greater than four (HER2_AMP) [ 18 ]. Additionally, we compared tumours with low-level amplification (five or six signals per nucleus) against tumours with a higher level of amplification (more than six signals per nucleus). HER-2/ neu expression determined by IHC The immunohistochemical staining with a monoclonal antibody against HER-2/ neu was performed as described previously [ 19 ]. In brief, serial sections (4 μm thick) of formalin-fixed, paraffin-embedded blocks were first deparaffinised. They were then microwaved in 10 mM citrate buffer, pH 6.0, to unmask epitopes and treated for 10 min with 1% hydrogen peroxide to block endogenous peroxidase. The sections were incubated for 30 min at 37°C with monoclonal HER-2/ neu antibodies (clone CB-11; Novocastra, Newcastle upon Tyne, UK) diluted 1:50. The sections were then incubated with a biotin-labelled secondary antibody and streptavidin–peroxidase for 20 min each. Tissue was subsequently treated for 5 min with 0.05% 3',3-diaminobenzidine tetrahydrochloride and lightly counterstained with haematoxylin. All series included appropriate positive and negative controls. All controls gave adequate results. Interpretation of IHC results Only cases showing unequivocal staining of membranes were regarded as positive for HER-2/ neu overexpression [ 2 ]. A score was determined in accordance with the criteria used in the approval trials for trastuzumab [ 17 , 20 ]. In brief, cases showing no staining were scored 0, cases with less than 10% membrane staining 1+, cases with more than 10% weak to moderate complete membrane staining 2+, and cases with more than 10% strong complete membrane staining 3+ (HER2_SCO). Only 3+ cases were considered positive for survival analyses. Interpretation of combined IHC and FISH results Cases that were scored 2+ by IHC were considered positive for the HER-2/ neu algorithm [ 21 ] only if they showed an amplification of HER-2/neu (HER2_ALG). Finally, cases showing HER-2/neu amplification as well as overexpression with an immunohistochemical score of 3+ were considered positive for the combined HER-2/ neu evaluation (HER2_COM). Statistical analysis The concordance between two methods was assessed by using the kappa test. The sensitivity and specificity of IHC (HER2_SCO) were evaluated, with the use of FISH as reference method. Life tables were calculated in accordance with the Kaplan–Meier method. Disease-free survival (DFS) was computed from the date of diagnosis to the date of recurrence of disease. Overall survival (OS) was computed from the date of diagnosis to the date of death from breast cancer. Patients who died of an unrelated cause were censored at the date of death. Survival curves were compared with the log-rank test. Univariate Cox survival analyses were performed and multivariate analyses were done in a backward stepwise fashion with the Cox proportional hazards model. All tests were performed at a significance level of α = 0.05. All P values are two-sided. Results Distribution of traditional and HER-2/ neu -related factors In a group of 101 node-negative patients with primary breast cancer of sizes T1 and T2 without systemic treatment in the adjuvant setting, established pathological and clinical parameters (tumour size, histological grade, steroid hormone receptor status, age and menopausal status) and HER-2/neu-related parameters (HER2_AMP, HER2_SCO, HER2_ALG and HER2_COM) were assessed and are presented in Table 1 ; 36% of the patients were premenopausal and perimenopausal. Tumour size was T1 in 57% and T2 in the remaining 43%. A total of 64% of the patients had tumours with a positive steroid hormone receptor status; that is, they were oestrogen receptor and/or progesterone receptor positive. A favourable histological grade (G I) was present in 17%, G II in 58% and G III in 20%. Five percent had medullary carcinomas and were therefore not assigned a histological grade. HER-2/ neu overexpression classified as 3+ as assessed by IHC was found in 20% (HER2_SCO); 17% of tumours showed an amplification of HER-2/ neu by FISH (HER2_AMP). Of the amplified cases, five tumours showed five or six signals per nucleus, indicating low-level amplification, and 12 tumours showed more than six signals per nucleus (evidence of a higher level of amplification). Two of the patients who were scored as 2+ by IHC also exhibited an amplification of HER-2/ neu . This resulted in a total of 22% of the patients being positive for HER2_ALG. Finally, in 11% of the patients an amplification as well as a 3+ overexpression of HER-2/ neu was found (HER2_COM). The estimated DFS was 70% and the breast cancer-specific OS was 80% at 10 years for the whole group of patients. Concordance of amplification status in formalin-fixed paraffin-embedded tumour samples with fresh-frozen tumour samples All five tumours amplified for HER-2/ neu in frozen tumour samples were also amplified when formalin-fixed paraffin-embedded tumour samples were used. Similarly, complete concordance was found between frozen and formalin-fixed tissue samples when five tumours without amplification were compared pair by pair. Concordance of HER2_AMP and HER2_SCO A concordance between amplification and HER2_SCO was detected in 86 cases (85%) when only 3+ cases were considered positive for HER2_SCO. Six cases with amplification did not score 3+, whereas nine 3+ cases failed to show an amplification. This resulted in a degree of concordance (kappa) of 0.50 (95% CI 0.29–0.72). Sensitivity and specificity of HER2_SCO The sensitivity of HER2_SCO 3+ with FISH as reference method was 65% (11 of 17 amplified cases) and specificity was 89% (75 of 84). Considering also 2+ cases as positive resulted in an increase in sensitivity to 76% (13 of 17 amplified cases) with a decreased specificity of 70% (59 of 84). Breast cancer-specific DFS In univariate analysis (Table 2 ) neither age at diagnosis nor menopausal status, tumour size or steroid hormone receptor status had a significant influence on the DFS. From the classical prognostic factors only histological grade turned out to be significantly related to DFS ( P = 0.0002; RR 3.26, 95% CI 1.77–5.99) for DFS. Among the HER-2/ neu -related variables, HER2_ALG did not show an influence on DFS, whereas HER2_SCO had a borderline significance ( P = 0.059, RR 1.46, 95% CI 0.99–2.16). In contrast, both HER2_AMP ( P = 0.004, RR 3.07, 95% CI 1.44–6.57) and HER2_COM ( P = 0.006, RR 3.27, 95% CI 1.40–7.65) had a significant influence on the DFS. There was no significant difference in DFS between tumours with low-level amplification (five or six copies per nucleus) and tumours with a higher level of amplification. Three of five and 7 of 12 patients relapsed, respectively. The Kaplan–Meier estimates that yielded significant results are shown in Fig. 1 . We then conducted a multivariate Cox regression in a backward fashion. In this Cox regression only histological grade ( P = 0.0002, RR 3.22, 95% CI 1.73–5.99) and HER2_AMP ( P = 0.026, RR 2.47, 95% CI 1.12–5.48) retained an independent prognostic significance (Table 3 ). Breast cancer-specific OS In univariate analysis (Table 4 ) neither age at diagnosis nor menopausal status or tumour size had a significant influence on the OS. From the classical prognostic factors only histological grade ( P = 0.0003, RR 4.27, 95% CI 1.96–9.29) and the steroid hormone receptor status ( P = 0.012, RR 0.32, 95% CI 0.13–9.78) were significant in univariate analysis for OS. Among the HER-2/ neu -related variables, HER2_ALG did not show any influence on OS whatsoever. However, HER2_SCO ( P = 0.047, RR 1.60, 95% CI 1.01–2.53) as well as HER2_AMP ( P = 0.004, RR 3.78, 95% CI 1.54–9.26) and HER2_COM ( P = 0.004, RR 4.18, 95% CI 1.60–10.90) had a significant influence on the OS. There was no significant difference in OS between tumours with low-level amplification (five or six copies per nucleus) and tumours with a higher level of amplification. Three of five and 5 of 12 patients died of breast cancer, respectively. The Kaplan–Meier estimates that yielded significant results are shown in Fig. 2 . In a multivariate Cox regression analysis only histological grade ( P = 0.0007, RR 3.89, 95% CI 1.77–8.55) and HER2_AMP ( P = 0.016, RR 3.08, 95% CI 1.24–7.67) retained an independent prognostic significance (Table 5 ). Discussion The prognostic value of HER-2/ neu has always been controversial. Studies showing a shorter DFS and/or OS for HER-2/ neu -positive patients [ 3 , 18 , 22 , 23 ] were opposed by studies which failed to find such an association [ 4 , 5 ]. In an earlier series from our department [ 24 ] an association with survival was shown only for node-positive, but not for node-negative, breast cancer patients. This association with prognosis in node-positive patients, who are almost uniformly treated with systemic therapy, might be markedly influenced by the ability of HER-2/ neu status to affect responsiveness to systemic treatment [ 11 - 13 ]. In any case, the actual role of HER-2/ neu as a predictive marker is still a matter of debate [ 25 ]. In our present study, we could rule out any such interference of purely prognostic with treatment-related predictive effects because we examined only node-negative patients without any systemic therapy in the adjuvant setting. When examining the techniques used for the assessment of the HER-2/ neu status, gene-based assays almost uniformly confirmed the negative prognostic impact of HER-2/ neu in node-negative patients. Especially in the past few years, FISH has gained considerable interest as a reliable and valid method for determining the HER-2/ neu status, confirming its prognostic utility [ 18 , 26 ]. Compared with other gene-based assays such as Southern blotting or polymerase chain reaction, FISH is not hampered by dilutional artefacts possibly resulting from a mixture of different cell populations. Similarly to IHC it allows for the specific detection of the alteration in individual cells within the important architectural context. However, in comparison with IHC, FISH is rather time-consuming and leads to substantial costs [ 27 ]. It is nevertheless considered the gold standard for assessing HER-2/neu status. A potential advance in the practicability of in situ hybridisation could be the more recently described chromogenic in situ hybridisation, which has shown a good correlation with FISH [ 28 ] and an independent prognostic importance in patients with node-negative breast cancer [ 23 ]. When determining the amplification of HER-2/ neu with FISH in our cohort of 101 untreated node-negative breast cancer patients, we found 17% with an amplification of the gene locus. This fraction is largely in line with the literature [ 18 , 26 , 29 - 31 ]. The amplification of HER-2/neu measured by FISH showed a significant correlation with the DFS and OS. This strong association between amplification of HER-2/ neu and survival of lymph-node-negative breast cancer patients is consistent with previous studies [ 18 , 23 , 26 , 30 ]. Only one group could document this association only for OS but not for DFS [ 32 ]. Searching for an explanation for their deviating results, the authors speculated that HER-2/ neu might be more a predictive factor for treatment response than a prognostic factor per se because all patients in their study had been treated systemically after relapse. In contrast with HER-2/ neu gene amplification, overexpression of HER-2/ neu was not associated with survival in several studies [ 4 , 5 , 33 ] even though others found a prognostic impact [ 22 , 23 , 34 ]. More recently, Volpi and colleagues [ 35 ] found a prognostic significance of HER-2/ neu overexpression only for a subgroup of patients with high proliferative activity, whereas they failed to show any prognostic significance of HER-2/ neu overexpression in the overall series of node-negative breast cancer patients. These apparent discrepancies have largely prevented the widespread acceptance of HER-2/ neu as a prognostic factor in node-negative breast cancer up to now [ 36 ]. Several possible reasons could account for these controversial findings. One frequently quoted, simple but nonetheless unsatisfactory reason could be the rather small sample size of several studies. However, a strong and biologically relevant prognostic factor should eventually become evident even within a small sample size. Another possible reason is the remarkable discrepancy in sensitivity between the numerous antibodies used [ 16 ] and differences in tissue fixation and processing [ 37 ]. However, perhaps the most important reason is the lack of a standardised evaluation protocol in many of the older studies. The standardisation of IHC has become increasingly important since the successful use of trastuzumab (Herceptin™) in the treatment of HER-2/ neu -overexpressing metastatic breast cancer [ 17 , 20 , 38 ]. The United States Food and Drug Administration has approved a standardised IHC kit (HercepTest™; Dako) with a detailed published scoring system ranging from 0 to 3+. However, the HercepTest has a rather low specificity, as outlined by Jacobs and colleagues [ 39 ]. The studies mentioned above that led to the approval of trastuzumab for HER-2/ neu -positive metastatic breast cancer used a cocktail of different antibodies, one of which, the monoclonal antibody CB11, was used in our study. When we adapted the scoring system defined for the HercepTest to the CB11 antibody we used in our study, we found a borderline significant correlation with survival. To our opinion, the concordance between FISH and IHC was only moderate at best with a rate of 85%, even though we regarded only 3+ cases as overexpressing HER-2/ neu . This level of concordance is in line with previously published results [ 40 ]. However, others found higher levels of concordance between FISH and IHC, ranging from 92% up to 98% [ 39 , 41 , 42 ]. When FISH results were compared with IHC data by using computer-assisted image analysis, we showed previously that concordance rates can be significantly improved when the IHC signal is corrected by the subtraction of non-specific cytoplasmic chromogen deposition [ 43 ], suggesting that only the strictly membrane-confined IHC signal can be considered truly positive when estimating HER-2/ neu IHC slides. However, similarly to our findings others have failed to detect oncogene amplification by FISH in as many as 51% of tumours with strong 3+ staining [ 44 ]. Because overexpression of HER-2/ neu is not necessarily caused only by amplification, polysomy of chromosome 17 has to be taken into account [ 45 ]. This polysomy of chromosome 17 could well lead to an incorrect diagnosis of a low-level amplification using single-colored FISH. However, in our study tumours with five or six copies of HER-2/ neu had a survival comparable to that of tumours with a higher level of amplification. As a consequence of the moderate concordance, sensitivity and specificity for IHC in our study were lower than described by others [ 41 , 42 ]. However, our findings with the frequently used antibody CB11 should not be generalised to IHC as a whole because different antibodies show well-documented differences in terms of sensitivity and specificity [ 16 ]. Cases reported as 2+ by IHC should be reassessed with FISH in accordance with the test algorithm used in metastatic breast cancer before the start of a therapy with trastuzumab [ 21 ]. To the best of our knowledge we are the first group to use this algorithm to assess the prognosis in node-negative breast cancer patients. The use of this algorithm found 2 of 18 2+ tumours amplified and hence increased the percentage of tumours positive for HER-2/ neu to 22%. A similar percentage of HER-2/ neu -positive cases has previously been reported when performing FISH only in uncertain cases after IHC with the HercepTest [ 46 ]. For these authors, decreases in time and costs were quoted as strong arguments in favour of using this same testing algorithm for the determination of HER-2/ neu status. Because overexpression of HER-2/ neu does not necessarily mirror amplification and vice versa, and because both parameters have been correlated with a poor outcome, we investigated whether the combination of amplification and overexpression could be useful in identifying a subgroup of patients with a particularly dismal prognosis. Altogether, 11 tumours showed amplification and overexpression (defined by a score of 3+). Indeed, these patients had a significantly worse prognosis than the remainder of the study cohort. These results are comparable to the data of Sauer and colleagues [ 31 ], who also found a markedly adverse prognostic effect in this dual-positive subgroup. Nonetheless, in our multivariate analysis, the combination of both techniques did not add additional prognostic information to that obtained by the amplification alone. However, owing to the relatively small number of events, especially for HER2_COM, subtle differences might be difficult to address adequately. To assess the clinical relevance of these findings, we included the HER-2/ neu -dependent variables (HER2_SCO, HER2_AMP, HER2_ALG and HER2_COM) into a multivariate model with the variables tumour size, histological grade and steroid receptor status. These last three variables are commonly accepted for risk assessment in node-negative breast cancer [ 36 ]. The factor age (not more than 35 years at diagnosis) was ignored because only one patient in our cohort was younger than 35. In this multivariate analysis, only histological grade and HER-2/ neu amplification were identified as independent prognostic factors for DFS and OS, respectively. We therefore conclude that HER-2/ neu is an independent prognostic factor in node-negative breast cancer, even though its prognostic utility is largely influenced by the method of testing. To further validate these observations, a prospective study in patients not treated in an adjuvant setting would be desirable. However, this is largely precluded in practice, given the recent consensus recommendations for the treatment of primary breast cancer [ 36 ]. For this reason we are currently engaged in a formal meta-analysis of all published studies that have used FISH or chromogenic in situ hybridisation to determine the HER-2/ neu status and hope to clarify once and for all the controversial status of HER-2/ neu as a prognostic factor in breast cancer. Conclusions The prognostic significance of HER-2/ neu in node-negative breast cancer depends strongly on the method of testing: only the amplification of HER-2/ neu is an independent prognostic factor for the long-term prognosis of untreated node-negative breast cancer. Abbreviations CI = confidence interval; DFS = disease-free survival; FISH = fluorescence in situ hybridisation; IHC = immunohistochemistry; OS = overall survival; RR = relative risk. Competing interests The author(s) declare that they have no competing interests. Authors'contributions MS planned the study, performed the IHC and drafted the manuscript. BL performed the in situ hybridisation. CG participated in the statistical analysis. NK participated in the in situ hybridisation. ES, HP, BT and WW participated in defining the study cohort and participated in the IHC. HAL and HK participated in finally drafting the manuscript. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064139.xml

1064140

Polysomy of chromosome 17 in breast cancer tumors showing an overexpression of ERBB2: a study of 175 cases using fluorescence in situ hybridization and immunohistochemistry

Introduction One of the most common genetic aberrations associated with breast cancer is the amplification and overexpression of the ERBB2 proto-oncogene located at chromosome 17, bands q12-21. The amplification/overexpression occurs in 25 to 30% of all breast cancers. In breast cancer, aneusomy of chromosome 17, either monosomy or polysomy, is frequently observed by conventional cytogenetics and fluorescence in situ hybridization (FISH). The aim of this study was to discover whether or not numerical aberrations on chromosome 17 have a correlation to the amplification or overexpression of the ERBB2 gene and to analyze their clinical implications in subgroups showing 2+ or 3+ positive scores by immunohistochemistry (IHC). Methods We used FISH on a series of 175 formalin-fixed paraffin-embedded breast carcinomas to detect ERBB2 amplification, using a dual-probe system for the simultaneous enumeration of the ERBB2 gene and the centromeric region of chromosome 17, as well as using IHC to detect overexpression. We analyzed clinical and pathological variables in a subgroup of patients with 2+ and 3+ IHC scores (147 patients), to describe any differences in clinicopathological characteristics between polysomic and non-polysomic cases with the use of the χ 2 test. Results We found 13% of cases presenting polysomy, and three cases presented monosomy 17 (2%). According to the status of the ERBB2 gene, instances of polysomy 17 were more frequently observed in non-amplified cases than in FISH-amplified cases, suggesting that the mechanism for ERBB2 amplification is independent of polysomy 17. Polysomy 17 was detected in patients with 2+ and 3+ IHC scores. We found that nodal involvement was more frequent in polysomic than in non-polysomic cases ( P = 0.046). Conclusions The determination of the copy number of chromosome 17 should be incorporated into the assesment of ERBB2 status. It might also be helpful to differentiate a subgroup of breast cancer patients with polysomy of chromosome 17 and overexpression of ERBB2 protein that probably have genetic and clinical differences.

Introduction Proto-oncogenes and tumor suppressor genes are two classes of genes with central roles in the regulation of cell growth. One of the most common genetic alterations associated with human breast cancer is the amplification of the ERBB2 proto-oncogene [ 1 ]. The ERBB2 gene located on 17q12-q21 encodes a 185 kDa transmembrane tyrosine kinase receptor [ 2 , 3 ]. This protein is a member of the epidermal growth factor receptor family [ 4 ] that comprises four homologous receptors: HER1 (ERBB1), ERBB2 (ERBB2), HER3 and HER4. These receptors are involved in the activation of complex signaling pathways, essential for cell survival and for the regulation of normal breast growth and development [ 5 - 7 ]. Several studies performed in various laboratories have demonstrated that 25 to 30% of all breast and ovarian malignancies show the amplification and overexpression of this gene [ 8 , 9 ]. Amplification of the ERBB2 gene is found in more than 90% of cases that have ERBB2 protein overexpression [ 10 , 11 ], but in normal breast the expression of ERBB2 is due to a transcriptional activation [ 12 ]. ERBB2 overexpression in women with both node-positive [ 8 , 9 ] and node-negative [ 13 ] breast cancer is associated with a poor prognosis, and several studies have found a correlation between ERBB2 overexpression and a shorter disease-free period and shorter overall survival [ 14 , 15 ]. ERBB2 overexpression and/or gene amplification is an indication for trastuzumab (Herceptin; Genentech, South San Francisco, CA, USA) therapy in patients with metastatic breast cancer [ 16 , 17 ]. Clinical trials combining trastuzumab and chemotherapy have been initiated, based on preclinical data about potentially enhanced anti-tumor activity when anti-ERBB2 antibodies were combined with chemotherapeutic agents [ 18 - 20 ]. There are different methods available to evaluate ERBB2 status [ 21 ], although immunohistochemistry (IHC; for protein overexpression) and fluorescence in situ hybridization (FISH; for gene amplification) offer several advantages, because the aberration can be evaluated directly in malignant cells taken from archival breast cancer specimens. Reports of false-positive Herceptest cases led to suggestions that Herceptests yielding a 2+ score should also be studied by FISH [ 22 ]. Tubbs and colleagues [ 23 ], called for the US Food and Drug Administration (FDA) to mandate the retraction of the earlier accepted criteria for trastuzumab therapy, namely that of Herceptests yielding a 2+ score, unless those cases were also confirmed by FISH. The FDA-approved FISH assay, PathVysion (Vysis, Inc., Downers Grove, IL, USA), is a dual-probe system for the simultaneous enumeration of the ERBB2 gene and the centromeric region of chromosome 17, defining ERBB2 amplification as a ratio of ERBB2 gene copies per chromosome 17 centromere [ 24 ]. Another FDA-approved FISH assay, INFORM (Ventana Medical systems, Tucson, AZ, USA), defines ERBB2 amplification as a mean absolute ERBB2 gene copy number of more than four spots per nucleus, without centromere 17 correction. FISH with the PathVysion probe and immunohistochemical assay with Herceptest are highly concordant for cases showing 3+ IHC scores and for negative cases, although the 2+ IHC score group includes both ERBB2 amplified and non-amplified tumors, showing a relatively high rate of discordance [ 25 , 26 ]. The mechanisms for ERBB2 expression in non-amplified tumors scored 2+ by IHC are unclear and may involve increased gene dosage by chromosome 17 polysomy [ 27 ]. In breast carcinoma, chromosomal aneusomy – either monosomy or polysomy – is frequently observed by conventional cytogenetics and FISH [ 28 , 29 ]. Sauer and colleagues [ 30 , 31 ], found that an abnormal number of copies of chromosome 17 have a low impact on ERBB2 gene and its expression, but more studies are necessary to confirm these results. The aim of our study was to analyze the role of polysomy 17 in ERBB2 protein expression and its implication in ERBB2 gene status in a series of patients with tumors scoring 2+/3+ by IHC. Materials and methods Patients' characteristics This is a prospective analysis of 175 breast cancer patients, consecutively treated at the Hospital del Mar of Barcelona between August 2000 and August 2003. Specimens We studied specimens taken from 175 prospective cases of human breast cancer. Overexpression was determined by IHC, and amplification was studied by FISH. The breast cancer specimens used in this study were fixed with 4% buffered formalin, and the tissue was then embedded in paraffin. Sections 4 to 6 μm thick were cut from the tissue, mounted on silanized slides and then deparaffinized in a xylene series, followed by immersion in 100% ethanol. Then we serially sectioned a hematoxylin/eosin (H&E)-stained tissue section, an IHC tissue section and a FISH tissue section from each of the patient samples. IHC assay This method involved the application of primary ERBB2 antibody (rabbit anti-human ERBB2 oncoprotein; DakoCytomation, Glostrup, Denmark) at 1:200 dilution. This antibody was evaluated with the dextran/peroxidase technique (Dako Envision). We used negative and positive controls for ERBB2 overexpression in all samples. The IHC technique was applied with an automated system (Tech Mate 500) after antigen retrieval at 110°C for 1 min in a wet autoclave. When FDA approved the Herceptest kit (DakoCytomation) as a diagnostic method to detect ERBB2 overexpression, we began to use it. As a single institution, we observed an excellent correlation between these two IHC detection systems. Membrane staining was interpreted as ERBB2 oncoprotein expression. To evaluate the immunostaining for ERBB2 antibody, we considered the intensity and type of membrane expression. Expression was recorded as follows: score 0, negative staining; score 1, local positivity and incomplete membrane staining; score 2, moderate, complete membrane staining; score 3, strong, complete membrane staining. FISH assay The Pathvysion probe was used. This probe consists of two different probes, one with the centromeric α-satellite probe, specific for chromosome 17 (Spectrum green), and a locus-specific probe from the ERBB2 gene (Spectrum orange). The probe was provided denatured, in single-stranded DNA. Deparaffinized tissue sections were treated with 0.2 M HCl and then with sodium thiocyanate, to eliminate salt precipitates. Pretreated slides were incubated for 10 min in a solution of proteinase K at 37°C. The slides were then postfixed in buffered formalin. Pretreated tissue sections and probes were denatured at 78°C for 5 min and hybridized overnight at 37°C on a hotplate (Hybrite chamber; Vysis). Washes were performed for 2 min at 72°C in a solution of 2 × SSC/0.3% Nonidet P40. Tissue sections were counterstained with 10 μl of 4,6-diamino-2-phenylindole (DAPI counterstain; Vysis). Results were analyzed in a fluorescent microscope (Olympus BX51) with the Cytovysion software (Applied Imaging, Santa Clara, CA, USA). Tissue sections were scanned at low magnification (×100) with DAPI excitation to localize those areas where histopathological characteristics had been established by examining a serially sectioned H&E-stained tissue section from the same patient. ERBB2 amplification was calculated by a ratio dividing the most frequent value for ERBB2 spots per nucleus by the most frequent value of chromosome 17 centromere spots per nucleus. A minimum of 60 nuclei were scored. Amplification of ERBB2 gene was considered when the ratio was 2 or more, in accordance with the manufacturer's recommended scoring system. We considered polysomy 17 when the cells had three or more copy numbers of centromeres for chromosome 17 per cell. Statistical analysis A total of 147 patients with 2+ and 3+ IHC scores out of 175 of the complete cohort were analyzed, ensuring that all the polysomies of chromosome 17 were detected in this subgroup of patients. Table 1 describes the following clinicopathological parameters: clinical stage, nodal status, histological grade, estrogen receptor (ER), progesterone receptor (PgR), p53 protein and relapse of the disease, and age description. We performed a concordance analysis using the χ 2 test between prognostic variables and the presence or absence of polysomy 17 to detect any differences between polysomic and non-polysomic subgroups. Results To validate the FISH technique, in a first screening step we performed FISH on 50 specimens that represented all of the immunohistochemical subgroups (0, 1+, 2+ and 3+). For subsequent specimens we performed FISH on 2+ and 3+ IHC subgroups. The exclusion of the subgroups with scores of 0 and 1+ was based on our previous finding that none of those cases had ERBB2 gene amplification [ 25 ]. This selection process resulted in the under-representation of IHC-negative and 1+ specimens for the FISH assay and a high proportion of 2+/3+ cases. Among the 175 specimens studied by IHC and FISH, 22 cases showed polysomy of chromosome 17 (13%), and three cases presented monosomy (2%) (Table 2 ). Of the 175 cases, 147 showed an IHC score of 2+ or 3+. The distribution of chromosome 17 copy numbers in this 2+/3+ subgroup is illustrated in Fig. 1 . Most of the cases with polysomy 17 had four copies of chromosome 17 per cell. We observed that all cases with polysomy of chromosome 17 had an IHC score of 2+ or 3+. Thirteen of 22 cases with polysomy revealed an IHC score of 2+, and the remaining 9 cases had an IHC score of 3+. In a subgroup of 78 patients having IHC scores of 2+, 13 of 78 (17%) showed polysomy of chromosome 17 with increased ERBB2 gene copies, but without gene amplification. In a subgroup having IHC scores of 3+, 9 of 69 cases (15%) presented polysomy 17. In patients with monosomy of chromosome 17, two cases presented IHC scores of 2+, and one presented an IHC score of 3+, without ERBB2 gene amplification. According to ERBB2 gene status, all cases with polysomy and 2+ IHC scores were considered to be normal for ERBB2 gene amplification (the ratio was 2 or less). Of the cases considered normal by FISH, 15% presented polysomy 17, and 10% of the amplified cases showed amplification and polysomy 17 simultaneously (Table 2 ). Within the subgroup having scores of 3+ and polysomy, only one case presented non-amplification of the ERBB2 gene. We compared the polysomic subgroup with the non-polysomic subgroup, to determine whether there were differences in the clinicopathological features. After clinical analysis of the 147 patients who were 2+/3+ by IHC, we found that nodal involvement was significantly associated with polysomy 17 ( P = 0.046). Furthermore, patients with polysomy 17 also showed a non-statistical trend toward relapse ( P = 0.181). No statistically significant correlation was observed between histological grade, ER, PgR and p53 variables and the polysomy of chromosome 17 (Table 3 ). Discussion In our study, 175 cases diagnosed as invasive breast cancer were examined to determine the frequency of chromosome 17 polysomy in different ERBB2 IHC subgroups. In our series we found 22 specimens with polysomy 17 (13%) and three cases with monosomy 17 (2%). In a series from Wang and colleagues [ 31 ], aneusomy was very common (more than 50%), but only 10 of 189 cases (5%) showed high polysomy (at least 3.76 signals of centromere 17 per cell). Series from Watters and colleagues [ 32 ] presented a high proportion of aneusomy 17 (54%); most of those presented polysomy, but the highest proportion fell into the 2.00 to 3.00 chromosome 17 copy numbers category. In those two series [ 31 , 32 ], they considered polysomy with a mean chromosome copy number near to disomy. The reason for the discordance with our series (13% versus 50%) could be explained by the consideration of polysomy 17 when the cells had three or more copy numbers of centromeres for chromosome 17 per cell, in agreement with other authors [ 22 , 26 , 27 , 29 ]. Cases with polysomy 17 could be considered to be amplified at a low level by absolute criteria, but they had to be classified as non-amplified when the number of ERBB2 copies was corrected for the number of chromosome 17 centromeres. FISH, using the dual probe for the ERBB2 gene and the centromere of chromosome 17, offers the greatest resolution in detecting alterations of the ERBB2 gene in breast tumors because it allows true amplification to be differentiated from polysomy 17. The FISH method is less affected by tissue variables than the IHC method, and it has emerged as the gold standard for the assessment of ERBB2 status in breast cancer. We observed polysomy 17 in non-amplified cases more frequently than in FISH amplified cases (15% versus 10%), suggesting that the mechanism for amplification of the ERBB2 gene is independent of polysomy. In our series, polysomy 17 was frequently associated with a 2+ and 3+ IHC score. We had 13 of 22 polysomic cases that showed IHC 2+ scores with FISH-negative results. In contrast, other authors [ 33 , 34 ] found that the incidence of polysomy 17 in 2+ non-amplified cases was similar to the incidence of polysomy 17 in IHC-negative cases, suggesting that weak overexpression (2+) without gene amplification is not secondary to chromosome 17 polysomy. Our findings support the concept that polysomy 17 is a genetic aberration independent of IHC status but that it might have a role in ERBB2 expression in weakly positive (2+) and in strongly positive (3+) cases. It is probable that there are non-amplified cases analyzed by FISH with false positive IHC (2+) scores as a result of polysomy 17. Polysomy 17, in the absence of ERBB2 amplification, would result in an increase in protein production to such level that it might be IHC stained as a positive result (2+). We identified only one such case, IHC scored at 3+, non-amplified by FISH, and showing polysomy 17. It is reasonable that the copy number of chromosome 17 was associated with an increased level of ERBB2 protein but not with ERBB2 gene amplification. With regard to this case, we are in agreement with Lal and colleagues [ 33 ] and Varshney and colleagues [ 34 ], who suggested that the findings of strongly positive IHC staining (3+) without gene amplification might be due, in part, to the increased copy numbers of chromosome 17, resulting in ERBB2 overexpression. In our study of 147 patients having tumors with 2+/3+ IHC scores, we found a statistical association between polysomy 17 and nodal involvement, and we observed that this subgroup of patients had a trend towards relapse more frequently than the non-polysomic cases. Only a few studies have focused on the analysis of the prognostic value of chromosome 17 aneusomy, and specifically on polysomy 17. In a large series from Watters and colleagues [ 32 ], abnormalities in chromosome 17 copy numbers were associated with grade III carcinomas and ER negativity, but they were without significant impact on survival. Other studies proposed that aneusomy of chromosome 17 is associated with poor prognostic factors in breast cancer [ 30 ]. Conclusions Breast tumors scoring at 2+ or 3+ by IHC should be analyzed by FISH for ERBB2, preferably with a dual-probe system capable of detecting aberrations in chromosome 17 that could affect ERBB2 expression. The accurate measurement of chromosome 17 polysomy might also be helpful in determining subgroups of patients with genetic and clinical differences, which would permit their inclusion in future clinical trials. Abbreviations ER = estrogen receptor; FDA = US Food and Drug Administration; FISH = fluorescence in situ hybridization; H&E = hematoxylin/eosin; IHC = immunohistochemistry; PgR = progesterone receptor. Competing interests The author(s) declare that they have no competing interests. Authors' contributions MS conceived and designed the study, performed FISH analysis, interpretated the data and drafted the article. IT coordinated the study and was significantly involved in patient recruitment and the correlation of clinical data with experimental findings, also taking a role in supervising and final approval of the article. JMC contributed to the design of the study, analysed the immunohistochemistry experiments providing several data presented in the publication, and took a role in supervising and final approval of the article. MS contributed to the analysis of the data and was involved in patient recruitment. CC contributed to analysis and interpretation of the data and revised the article. BE contributed to interpretation of the data and revised the article. MB contributed to the recruitment of patients and revised the paper. XF is the head of the Oncology Department of Hospital del Mar de Barcelona. FS revised the paper, giving final approval of the version to be submitted, and is the head of the Cytogenetics Laboratory of Hospital del Mar de Barcelona. SS is the head of the Pathology Department of Hospital del Mar de Barcelona. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064140.xml

1064849

Adult Murine Skeletal Muscle Contains Cells That Can Differentiate into Beating Cardiomyocytes In Vitro

It has long been held as scientific fact that soon after birth, cardiomyocytes cease dividing, thus explaining the limited restoration of cardiac function after a heart attack. Recent demonstrations of cardiac myocyte differentiation observed in vitro or after in vivo transplantation of adult stem cells from blood, fat, skeletal muscle, or heart have challenged this view. Analysis of these studies has been complicated by the large disparity in the magnitude of effects seen by different groups and obscured by the recently appreciated process of in vivo stem-cell fusion. We now show a novel population of nonsatellite cells in adult murine skeletal muscle that progress under standard primary cell-culture conditions to autonomously beating cardiomyocytes. Their differentiation into beating cardiomyocytes is characterized here by video microscopy, confocal-detected calcium transients, electron microscopy, immunofluorescent cardiac-specific markers, and single-cell patch recordings of cardiac action potentials. Within 2 d after tail-vein injection of these marked cells into a mouse model of acute infarction, the marked cells are visible in the heart. By 6 d they begin to differentiate without fusing to recipient cardiac cells. Three months later, the tagged cells are visible as striated heart muscle restricted to the region of the cardiac infarct.

Introduction The difficulty in recovery of cardiac function after cardiomyocyte death, such as occurs with a heart attack, contrasts with injury to skeletal muscle in which myocyte numbers can increase through the recruitment of new myocytes from a local stem-cell pool called satellite cells. At present, cardiac transplant, with the intrinsic limitations of supply, immunosuppression, and organ rejection, remains the only long-term treatment for irreversible cardiac failure. Injection of fetal or embryonic stem cells into infarcted hearts holds some promise [ 1 , 2 , 3 ] but is complicated by the potential for immunologic rejection as well as by political and ethical considerations. Cell plasticity observed after in vivo transplantation of a variety of cell lineages [ 4 , 5 , 6 , 7 ] or after in vitro transformation with 5-azacytidine [ 8 ] has encouraged the study of cell-based therapy. Investigators have identified endogenous cardiomyocyte proliferation [ 9 ] and have experimented with skeletal myoblasts as well as with adult stem cells isolated from blood or heart to try to repair cardiac damage [ 9 , 10 , 11 , 12 , 13 , 14 , 15 ]. In vivo analysis of stem-cell transplantation studies has become complicated in part because of a growing understanding of the process of in vivo donor stem-cell and recipient mature-cell fusion [ 16 , 17 , 18 ]. Recent studies have shown a small population of cells in murine fat that progress to beating cells in vitro [ 19 ]. Three independent groups have isolated stem cells from the heart that show a capacity to differentiate into cardiomyocytes in vivo [ 14 , 15 , 20 ]. All three cell types are extensively passaged, after which two of these can be pushed to differentiate in vitro. In evaluating a variety of primary cell cultures of different lineages, we noticed nonadherent cells isolated from adult murine skeletal muscle that become beating floating cells. We named them “Spoc” cells (skeletal-based precursors of cardiomyocytes) for their ability to generate beating cardiomyocytes in vitro. These cells, easily obtained from skeletal muscle, provide new options for cardiac research and therapeutic interventions. When injected systemically into an acutely infarcted mouse, they migrate to the cardiac infarct region and differentiate into cardiac myocytes. Results/Discussion Characterization of Spoc Cells At initial isolation, Spoc cells are CD34 − and CD45 − (data not shown). Dissociation of 10 g of leg muscle yields approximately 2 × 10 6 Spoc cells. C-kit + cells, comprising less than 1% of Spoc cells, are present after isolation but can be removed by affinity purification without any change in the subsequent results (data not shown). Electron microscopy (EM) shows that on the day of isolation, Spoc cells are 4–8 μm in diameter, often smaller than a red blood cell, and display copious vesicles suggestive of active transport. Many lamellipodia and filopodia are present, which may in part explain the typical Velcro-like clumping of the cells that makes chemical and physical dispersion more difficult ( Figure 1 ). The CD34 − /CD45 − /c-kit − phenotype differentiates Spoc cells from other nonadherent cells previously described as derived from skeletal muscle [ 5 ]. Spoc cells are distinguished from satellite cells by the following criteria: (1) Spoc cells do not express Pax-7 [ 21 ] or the surface marker c-met [ 22 ]; (2) approximately the same number of Spoc cells are isolated from both young (less than 4-wk-old) and older (12- to 16-wk-old) mice, whereas satellite cells are difficult to isolate from normal mouse muscle after 8 wk of age without first inducing muscle injury [ 21 ]; (3) at isolation, Spoc cells remain round floating cells approximately 4 μm in diameter, whereas isolated satellite cells soon become adherent; (4) Spoc cells are CD34 − and Myf-5 − , excluding them from the class of quiescent satellite cells [ 23 ]; and (5) the three skeletal muscle super-regulatory genes myf5, myoD, and myogenin are known to be present at some time in satellite cells [ 22 ], but in Spoc cells, all three markers remain negative from the day of isolation throughout long-term culture (data not shown). Thus, Spoc cells are neither satellite cells nor further-differentiated skeletal muscle cells. Figure 1 EMs of Day 0 Spoc Cells Electron micrograph of a freshly isolated Spoc cell (A) shows its small size relative to the typical biconcave profile of the red blood cell above it (arrow). (B) shows a typical Spoc cell with a large number of vesicular structures, lamellipodia, and filopodia. (C) displays three Spoc cells so tightly clumped that their borders are difficult to distinguish. This typical clumping makes fluorescence-activated cell sorting (FACS) analysis difficult. During the first 7 d in medium containing epidermal growth factor (EGF) and fibroblast growth factor (FGF), Spoc cells undergo several rounds of division, begin to express GATA-4 (a mostly cardiac-specific transcription factor), and become clusters of floating round cardiac precursors from Spoc (CPS) cells with an increased diameter of 10–14 μm. The pattern of GATA-4 staining in these cells is unusual in that it is predominantly cytoplasmic, which is distinct from the expected nuclear staining of a transcription factor. Although rare, many transcription factors are well known to have a cytoplasmic phase. GATA-4 has been described similarly and in this case has been shown to move into the nucleus after the addition of the beta-adrenergic drug isoproterenol [ 24 , 25 ]. For these reasons, isoproterenol was added to a culture of CPS cells for 1 h, after which GATA-4 staining was observed in the nucleus of many of the cells ( Figure 2 ). These cells go on to express other cardiac-specific markers, including cardiac troponin-T, Nkx-2.5, MLC-2v, and a cardiac L-type calcium channel, detected either by immunostaining or real-time PCR ( Figure 3 A– 3 G; real-time PCR data not shown). Although Nkx-2.5 may be present in vascular smooth muscle, neither alpha- nor beta-myosin is present in smooth muscle. Beating cardiomyocytes derived from Spoc cells express both alpha- and beta-myosin at day 28, as shown by staining with polyclonal antibodies against alpha- and beta-myosin ( Figure S1 ). In later stages, connexin 43 is expressed in cell clusters ( Figure 3 H– 3 I). Even before they become adherent, some isolated cells begin to beat ( Video S1 ). Figure 2 Sublocalization of GATA-4 in Spoc Cells (A) GATA-4 is detected in the cytoplasm of day 10 Spoc cells (cytospin). (B) Nomarski image of (A), showing DAPI-stained blue nuclei. (C) Merge image showing nuclei and GATA-4 staining. (D) When Spoc cells are incubated with 20 μM isoproterenol for 1 h, the GATA-4 nuclear staining is seen. (E) Nomarski image of (D). (F) Merge of (D) and (E), showing sublocalization of GATA-4 to nuclei. A weaker GATA-4 signal is present in the cytoplasm. Figure 3 CPS Cells Stain Positive for Cardiac-Specific Proteins (A) GATA-4 in day 7 CPS cells. (B) Nuclear staining with DAPI. (C) Overlay of (A) and (B). (D) Nkx-2.5 is detected in the nuclei of round, day 21 beating cells (green). (E) Noncardiac cells (red arrowheads) do not show nuclear staining for Nkx-2.5. (F) Overlay of (D) and (E). (G) Beating cells, after 28 d in culture, stain positive for cardiac L-type Ca ++ channel. (H) Connexin 43 (green) in cluster of uninucleate day 21 beating cells in culture. (I) Nomarski light micrograph (differential interference contrast) of cell cluster in (H). Sca-1 Separation of Spoc Cells Further fractionation of Spoc cells by sorting for the Sca-1 marker (a cell-surface antigen found on murine hematopoietic stem cells [HSCs]) shows that the beating cells develop out of the Sca-1 − pool, which comprises 20%–40% of the total isolated cells. This is in contrast to recent studies in which an isolated Sca-1 + population from the heart homed to the heart after vascular injection and showed signs of cardiac differentiation [ 14 , 26 ]. When the skeletal muscle–derived Sca-1 + fraction of Spoc cells are plated under our conditions (see Materials and Methods ) separately from the Sca-1 − fraction, the former cells rapidly adhere to the plate and, with the exception of a few putative contaminating Sca-1 − cells, do not develop into beating cells. With the Sca-1 + population removed, the remaining Sca-1 − population undergoes additional divisions before beginning to differentiate. Occasionally, a small, round beating cell with no organized sarcomere is seen after 3 d, but approximately 80% of the total population of cells differentiates into beating cells after a 7- to 10-d proliferative period. These Sca-1 − –derived CPS cells remain in this immature state, i.e., round, loosely attached, and spontaneously beating, for more than 2 mo of culture. When a green fluorescent protein (GFP)–tagged Sca-1 − population of CPS cells is replated onto a nontagged adherent monolayer of the purified Sca-1 + cells, the fluorescent Sca-1 − cells adhere to the monolayer within 24 h, stretch out, and mature into beating cells. As the stretched-out GFP + cells develop a sarcomeric structure, the GFP signal diminishes. This phenomenon, which is possibly due to a down-regulation of the beta-actin promoter-driven GFP, or to the exclusion of GFP protein from the developing sarcoplasm, may compromise the detection of GFP-tagged donor cells in the mouse infarct model. Spoc Cells Are Not Derived from Bone Marrow or Adipose Tissue Spoc cells do not appear to be bone marrow cells sequestered in skeletal muscle, and because they are c-kit − , they are distinguished from the c-kit + bone marrow cells and side population cells that have been used directly or indirectly in experiments to reconstitute infarcted heart [ 11 , 13 ]. When approximately 50 mg of whole marrow or dissociated total heart is co-cultured in permeable-membrane-separated compartments in a 1:1 ratio with Spoc cells, only the Spoc cells develop into beating cardiomyocytes. Thus, bone marrow and heart do not contain a cell population that is isolatable in this manner and phenotypically similar to Spoc cells. We also performed experiments to determine whether Spoc cells are similar to adipocyte-derived stroma vascular fraction cells (SVFs), which have been derived from a mixture of murine inguinal and intrascapular fat pads [ 19 ]. Spoc cells were cultured in parallel with SVFs in our EGF/FGF–containing medium, as well as in methocult supplemented with beta-mercaptoethanol, erythropoietin, Interleukin (IL)–3, IL-6, and Stem Cell Factor, as outlined in [ 19 ], except that inguinal- and intrascapular-derived cells were kept separate from each other. In the supplemented methocult culture, Spoc cells developed into CPS cells, which subsequently fused into large (up to 1 mm in length) beating myocytes as described earlier [ 19 ]. We were not able to generate CPS cells, or any beating cells, from SVFs derived from inguinal fat alone. A result similar to culturing Spoc cells was observed with intrascapular-derived SVFs cultured in supplemented methocult, but only in cultures that had microscopic contamination with skeletal muscle fibers. Neither SVFs from the inguinal nor from the intrascapular fat pads yielded any beating cells when cultured under standard EGF/FGF Spoc cell conditions. Planat-Benard et al. [ 19 ] report an efficiency of conversion to cardiomyocytes of 0.02%–0.07%, whereas 1%–10% of total Spoc cells become beating cardiomyocytes. When pure Sca-1 − Spoc cells are plated, after the proliferation phase, approximately 80% of the cells are beating CPS cells. Unlike the case with HSCs, no colony-forming units are generated when Spoc cells are cultured in methylcellulose in the presence of erythropoietin, IL-3, IL-6, and Stem Cell Factor (data not shown). To evaluate the ability of Spoc cells to reconstitute bone marrow, standard and competitive bone marrow transplantations with Spoc cells were performed. In four mice, 3 × 10 6 bone marrow cells and 3 × 10 4 GFP + /Sca-1 − Spoc cells were injected into each of the lethally irradiated mice. All four mice survived; however, only a rare GFP + donor–derived cell was seen in the peripheral blood or bone marrow (data not shown). The marrow of six lethally irradiated mice, injected with either 1.5 × 10 5 Sca-1 − Spoc cells or 2 × 10 5 Spoc cells unfractionated for Sca-1, could not be rescued, and all six mice died within 2 wk. Thus, because only a single HSC is required to reconstitute bone marrow, no HSCs are contained in 2 × 10 5 Spoc cells. Transmission EM of CPS Cells as They Progress to Beating Cells Figure 4 shows the post-replating progression of CPS cells. Three days after the replating of floating cells (see Materials and Methods ), when increasing numbers of CPS cells show rhythmic beating, EM shows round cells with large central nuclei surrounded by copious mitochondria and myosin thick filaments ( Figure 4 A and 4 D). As the cells attach to the tissue culture plate, the central nucleus elongates and electron-dense structures are visible ( Figure 4 B; arrowhead in Figure 4 E), with myosin filaments radiating outward (box/inset, Figure 4 B). By 2 wk, the electron-dense bodies have begun to align with filaments coursing between them ( Figure 4 C). These structures are nearly identical to those seen in embryoid body–derived developing cardiomyocytes [ 27 ]. By 8 wk in culture, the electron-dense structures in EMs of beating cells have progressed to become the Z-lines of organized sarcomeres ( Figure 4 G). The beating cells also have one or two large, centrally located nuclei surrounded by mitochondria, distinguishing them from skeletal myotubes, which have many small, subsarcolemmal nuclei. Figure 4 Transmission EMs Show the Post-Replating Progression of CPS Cells (A) Round, day 3 cells contain disordered myosin filaments. Some of these cells beat while still floating (see Video S1 ) and typically have APs as shown in Figure 6 A. (B) Upper box is a blowup taken from lower panel, showing myosin filaments of characteristic 1.6-μm length radiating outward from dense body. (C) Day 14 cell with a single, central nucleus shows a stretching out of the dense bodies into an organizing sarcomere. (D) Day 3 round cells containing copious mitochondria (inset). (E) Elongated day 7 cell containing a dense body (arrowhead). (F) Uninucleate day 14 cell, same cell as in (C). (G) By day 56, a well-defined sarcomere is present, with identifiable A- and I-bands and M- and Z-lines. (H) Sarcomere from a fetal cardiomyocyte is shown for comparison. Calcium Transients and Cardiac Action Potentials in CPS Cells As the CPS cells progress to spontaneously beating cardiomyocytes in vitro, the frequency of beating ranges from 1 to 8 Hz. Cells continue to beat even after culture for 3 mo. Despite the absence of beating initially, some clusters of cells adherent to an extracellular matrix display calcium transients detected by fluo-4 (4 μM concentration) visualized with confocal microscopy ( Video S2 ). This suggests that the excitation portion of “excitation–contraction” develops before the contractile apparatus of the cell is mature enough to overcome the minimal resistance offered by an extracellular matrix. The round beating cells shown in Video S1 progress to the elliptical beating cells shown in Video S3 and subsequently to the more mature cardiomyocytes shown in Video S4 . These last cells also display calcium transients detected through the use of fluo-3 (4 μM concentration) as shown in Video S5 and Figure 5 . By day 14 after replating, 10% of the cells in a confluent dish beat spontaneously. The calcium transients indicate the existence of action potentials (APs), which have been characterized by patch recordings of single cells in culture ( Figure 6 ). A variety of cardiac APs are observed in beating and nonbeating cells ( Figure 6 A and 6 B), which both show a resting membrane potential of approximately −60 mV with a robust overshoot of 50–90 mV. The form and duration of the APs match the descriptions of adult murine cardiomyocyte APs, which lack the plateau phase seen in cardiomyocytes of other species [ 28 ]. Figure 5 Measuring Calcium Transient Frequency Graphical representation of the calcium transient in a beating CPS cell–derived cardiomyocyte (A). Fluorescent intensity is proportional to the amount of calcium binding to fluo-3 dye upon release of calcium from the sarcoplasmic reticulum. Peak intensity (B) and baseline (C) are shown. Figure 6 Whole-Cell Voltage Recordings from Spoc Cell–Derived Cardiomyocytes (A) Spontaneous AP firing in a nonbeating, teardrop-shaped cell. (B) Representative AP from recording in (A) on an expanded time scale; AP threshold is –60 mV. (C) Action potential firing in another cell is blocked upon bath perfusion with 0.5 mM cadmium chloride (horizontal bar). (D) Acceleration of AP firing upon perfusion with 25 nM isoproterenol (horizontal bar) is demonstrated, indicating the presence of adrenergic receptors on these cells. (E) Skeletal myotube APs, if present, differ in that their frequency is unaffected by Cd ++ . (F) Isoproterenol also does not affect skeletal muscle AP frequency. The beating in Spoc cell–derived cardiomyocytes differs from the sporadic twitches that have been observed in skeletal muscle because the addition of 0.5 mM cadmium chloride, a non-specific blocker of L-type Ca ++ and Na + channels, abolishes the cardiac AP [ 29 , 30 ], while having no effect on skeletal myotubes ( Figure 6 C and 6 E) [ 31 , 32 ]. Likewise, both beating and nonbeating cardiac cells have an intact adrenergic pathway [ 33 , 34 , 35 ], as shown by the increase in AP frequency with the addition of 25 nM isoproterenol ( Figure 6 D). The expected lack of effect on skeletal myotubes [ 36 ] is shown in Figure 6 F. Uninucleate myoblasts with spontaneous calcium transients and APs under standard culture conditions have not been described [ 37 ], and even when they are contrived by arresting cell fusion, they retain the electrical activity characteristic of skeletal muscle cells, such as the lack of response to cadmium chloride [ 38 ]. Transplantation of Spoc Cells into Murine Myocardial Infarction Models In order to determine whether Spoc cells engraft within a myocardial infarct (MI) and differentiate into mature cardiomyocytes, an acute infarct model in C57Bl/6J mice was created by ligation of the left coronary artery. We injected 1 × 10 5 GFP + total Spoc cells (unfractionated for Sca-1) into the peripheral circulation. After 14 wk, many donor-derived GFP + cells had engrafted ( Figure 7 B). Of donor cells that had migrated to the infarct, 8% (63/782) had developed into cardiomyocytes (arrows, Figure 7 A) displaying a phosphorylated cardiac myosin regulatory light chain (RLCP), shown by co-labeling with an antibody against GFP (green) and RLCP (red) [ 39 ]. All samples were scanned on green and red channels simultaneously. Cells were considered GFP + only if their cytoplasm fluoresced green and not red. This prevents mistaking the anti–GFP signal (green) in Figure 7 A for autofluorescence. Autofluorescence in an infarct will characteristically “bleed” through both green and red channels in the absence of exogenous fluorescent markers, causing all autofluorescence to show up as a generalized yellow, i.e., the combination of red and green [ 40 ]. In this case, only the myosin striations, which are co-stained, appear as both red and green, serving as an internal control for the technique. No labeled cells, differentiated or undifferentiated, were observed in the normal portion of the infarcted heart (data not shown). These findings suggest that Spoc cells either actively home to or are passively delivered to an area of cardiac damage where they begin to differentiate into cardiomyocytes, as observed in vitro. In order to determine if Sca-1 − Spoc cells engraft and differentiate within a MI as efficiently as Spoc cells unfractionated for Sca-1, the same acute anterior MI model described earlier was used. When 1 × 10 5 Sca-1 − /GFP + donor cells were injected via the tail vein immediately after infarction, a low level of engraftment occurred with only an occasional donor-derived cardiomyocyte found at the periphery of the infarct zone ( Figure 7 C– 7 E). This finding is consistent with our in vitro findings described earlier, in which purified Sca-1 − cells remained immature, loosely attached, round beating cells for months until co-culture with a feeder layer of immobile Sca-1 + cells allowed the Sca-1 − fraction to adhere and mature. Figure 7 In Vivo Myocardial Infarction Transplantation Studies (A) The GFP-tagged Spoc cells (green), unfractionated for Sca-1, injected into the peripheral blood of a murine acute infarct model have developed after 14 wk into cardiomyocytes (arrows) within the infarct region. Donor Spoc cell–derived GFP cardiomyocytes are characterized by single central nuclei and striations staining for RLCP (red). (B) Longitudinal fresh-frozen tissue slice showing the region of infarct from which (A) was taken (orange box) and adjacent normal endogenous cardiac tissue (RLCP, red). (C) GFP + /Sca-1 − Spoc cells were injected into the peripheral circulation of a murine acute MI model and are detected in the infarct after 4 wk by expression of GFP (green). (D) RLCP (red) is also expressed. (E) Overlay of (C) and (D). To evaluate for a similar effect in an older infarct, the same number of Spoc cells, unfractionated for Sca-1, was injected via the tail vein into two mice (8 and 14 wk after MI). Two weeks after injection, the heart of the 8-wk-old infarct model showed co-localization of GFP and GATA-4 in 3% (4/136) of donor cells that migrated to the peripheral region of the infarct. Five weeks after injection into the 14-wk-old infarct model, an increased number of GFP + /RLCP + cells (7/102 donor-derived cells) in the infarct region were evident (data not shown). Spoc cells that were partially differentiated by culturing for 7 d were injected into the hearts and tail veins of three mice with acute infarcts. No labeled cells were identified in the hearts at 7 d and 1 mo later, compared to two control mice injected with saline at the time of infarct (data not shown). This suggests that the undifferentiated cells more easily actively home to or are delivered passively to the heart. We used a Cre-recombinase (Cre) expressor/beta-galactosidase reporter system to rule out donor stem-cell fusion with host cardiomyocytes as an explanation for apparent in vivo differentiation. In the event of fusion, Cre from the donor cells will excise the floxed stop codon present in the host DNA and thus de-repress the beta-galactosidase reporter gene. Within hours of producing an acute infarct by surgical ligation of the left coronary artery in R26Rh reporter mice, tail-vein injection of 3 × 10 5 Cre + Spoc cells from an EIIa promoter/Cre donor mouse was performed. Cre + donor cells are seen in the heart as early as 2 d post infarct (data not shown). By day 7, isolated nests of donor cells that have not fused with host cells can be detected, as shown by the lack of X-gal staining in Cre + cells ( Figure 8 A– 8 F). Also by day 7, GATA-4 is co-expressed with Cre in some of the cells in these clusters ( Figure 8 G– 8 I). In order to further isolate and track donor Spoc cells, a monoclonal antibody library was generated by inoculating a rat with live Spoc cells via injection into the peripheral circulation. The library was screened for the ability to detect cell-surface antigens on total Spoc cells. One such antibody that has been generated (MSC 21) detects live Spoc cells in culture and can be used to isolate a positive and negative subpopulation. MACS separation using MSC 21 gives two populations, with the 21 + subset containing cells that develop into beating cardiomyocytes ( Video S6 ). In the Cre donor/R26Rh reporter infarct mice, some of the injected Cre + cells are co-stained by MSC 21, further supporting the donor origin of these transplanted cells ( Figure 8 J– 8 L). MSC 21 does not detect cells in normal controls or infarcted heart ( Figure 8 M). Figure 8 Cre Expressor/Beta-Galactosidase Reporter Myocardial Infarction Studies (A) Nests of Cre + cells (green) are detected 1 wk after tail-vein injection into an acute infarct model. (B) Nomarski image of (A). (C) Merged image of (A) and (B). The clusters are located near a blood vessel (arrow). (D) Infarcted tissue in a control MI model (infarction surgery but no donor-cell injection) showing a lack of staining for Cre (no green) and GATA-4 (no red). (E) Control X-gal staining of ROSA mouse heart. (F) In a sequential series of tissue sections, odd-numbered sections were immunostained for Cre, yielding the results seen in (A). These two clusters of cells were seen on five sections (sections 1, 3, 5, 7, and 9). Even-numbered sections (sections 2, 4, 6, and 8) were stained for X-gal. No X-gal + cells were found. One slide was immunostained, showing the Cre + cells present. This slide was then stained for X-gal and was found to be X-gal − . The lack of X-gal staining of the serial sections indicates that at 1 wk no fusion of donor and host cells has occurred in the infarct. (G) Cluster of Cre + donor cells detected in infarcted heart tissue of a 1-wk-old acute infarct model. Arrowheads indicate cells that also express GATA-4, as shown in (H). (H) GATA-4 (red) is mostly present in some cells in the margin of the cluster. Arrowheads indicate cells that also express Cre, as shown in (G) (I) Merged image showing co-localization (arrowheads) of Cre (green) with GATA-4 (red) in some cells of the cluster of Cre + cells. (J–L) Co-staining of donor cells with anti–Cre antibody (green) and MSC 21 (red) is apparent after 7 d in an acute infarct model. (M) There is a lack of staining with MSC 21 (no red) in the infracted tissue of mice that have not received Spoc cell injections. Concluding Remarks Our isolation and description of what we call Spoc cells was initially surprising to us in that a few simple variations in the usual technique of skeletal muscle cell culture yielded a novel finding. These adaptations included avoiding trypsin, substituting FGF and EGF for commonly used complex supplements like chick embryo extract, and discarding adherent cells while continuing to culture the nonadherent cells. The observation of isolated nonadherent beating cells in a skeletal muscle culture has probably been observed previously by others in the course of experimenting with skeletal muscle culture conditions but has anecdotally been considered to be some variant of primitive skeletal muscle cells and never been carefully investigated . This report is, to our knowledge, the first time that the phenotype of these cells has been methodically studied with single-cell patch recordings, Ca ++ transient recordings, EM, longitudinal video microscopy, and immunostaining to detect the presence of skeletal and cardiac histologic markers. The absence of skeletal and satellite cell markers in the face of the histologic and physiologic evidence of the cardiac phenotype supports our conclusion that these cells are closer to true cardiac myocytes than to some variant of skeletal muscle cells. The homing of these freshly isolated cells to the injured heart and their subsequent differentiation into a cardiac phenotype in vivo that matches the in vitro experiments is corroborating evidence. Recently, three independent groups have isolated a stem-cell population from the heart. In two studies, the stem cells were cultured extensively [ 15 , 20 ] before reintroduction directly into the heart, whereas in the other study, freshly isolated cells were injected into the vascular circulation [ 14 ]. Beltrami et al. [ 15 ] did not observe beating cells in vitro, whereas the cells isolated by Oh et al. [ 14 ] required treatment with the demethylating agent 5-azacytidine to produce beating cells, similar to the treatment that was previously used to generate beating cell lines from bone marrow stromal cells [ 41 ]. Messina et al. [ 20 ] used cardiotropin to treat round cells released from a 3-wk explant culture, generating a sphere of cells that remain primitive and proliferating on the inside while they differentiate into beating cardiac myocytes at the margins. They observed that this growth pattern is reminiscent of neurospheres. All three cardiac-derived populations appear to be distinct from one another with respect to parameters such as the presence of c-kit or the ability to beat in vitro; however, the diverse conditions make direct comparison difficult. Spoc cells do not emerge from cardiac tissue grown in parallel with the skeletal cultures. However, freshly isolated Spoc cells are observed in infarcted heart within 2 d of their systemic injection into a murine model of acute MI. Within 7 d of systemic injection, Spoc cells form a sphere-like body at the borders of the infarction near blood vessels. The Cre signal is strongly present in the center of these spheres but diminished at the borders where the GATA-4 signal emerges. This is likely due to EIIa promoter suppression in the differentiating cells. This pattern of the more proliferative cells in the center and the differentiating cells at the borders is reminiscent of neurospheres or the in vitro culture of cardiac-derived stem cells [ 20 ]. By 3 mo, many more of the cells are distributed throughout the border areas of the infarct and can be detected with an antibody to ventricular RLCP. The pattern of GATA-4 staining in early Spoc cells is particularly interesting in its unusual cytoplasmic distribution, but the shift to a nuclear distribution after the addition of the beta-adrenergic agonist isoproterenol to the cell culture illustrates the staining specificity. The importance of beta-adrenergic stimulation to the processes of differentiation and cardiac myocyte hypertrophy makes these cells particularly useful to study the Akt–GSK3 beta regulation of nuclear-expressed GATA-4 [ 25 ] as well as its beta-adrenergic–mediated interaction with the calcineurin pathway [ 24 ]. Although the beating cells in vitro contract asynchronously, local regions of confluent plates appear to beat at distinct rates. This fact, coupled with the late expression of connexin 43 in vitro (see Figure 3 H), suggests the potential for coordinated beating, a requirement for efficient cardiac contraction. Although Spoc cells appear more efficient at homing to the heart in an acute MI model, their ability to differentiate in vivo at the border of an old infarct may aid research focused on chronic heart failure or dilated cardiomyopathy. Generation of cardiomyocytes from Spoc cells derived from transgenic models may be an important tool in studying cardiac development, because the progression from undifferentiated cells to mature cardiomyocytes may be observed in vitro. Why these cardiac stem cells are sequestered in skeletal muscle and why clinically significant numbers of them are not normally recruited to salvage an infarcted heart remains a mystery. Perhaps additional cells or factors present in skeletal muscle suppress the cardiac differentiation of Spoc cells in situ or inhibit migration. Certainly, Spoc cells do not differentiate into cardiac cells while they reside in skeletal muscle. Along these lines, fractionation of Spoc cells based on Sca-1 greatly enriches for a population of cells in the Sca-1 − fraction that in vitro become beating cardiomyocytes; however, engraftment of these Sca-1 − cells in MI models seems less efficient than that of unfractionated Spoc cells. This is consistent with the increased magnitude and kinetics of cardiomyocyte differentiation observed when the Sca-1 − fraction is plated over a monolayer of the Sca-1 + fraction. Thus, there is likely a role for the Sca-1 + fraction in engraftment efficiency through any number of mechanisms, including migration, tropism, and entrapment. It remains to be determined whether Spoc-cell transplants contribute to mouse survival or electrically couple to endogenous cardiac myocytes. There is evidence, however, that Spoc cells survive more than 3 mo post transplant, during which time they mature into striated cardiomyocytes. Materials and Methods Isolation of Spoc cells To isolate cardiomyocyte precursor cells from adult mouse skeletal muscle, skeletal muscle tissue from the hind legs of 6- to 14-wk-old male C57Bl/6J mice is cut into small pieces and digested with collagenase type-2 (5 mg/ml) for 45 min at 37 o C, and then for another 45 min at 37 o C in a shaking rotator. The digested tissue is cleared of cell debris and other undigested tissue fragments by passage first through 100 μm and then through 40 μm filters. Cells are incubated in DMEM/F12 medium containing 5% FBS for 2 h at 37 o C. The cell suspension is then centrifuged at 1,500 RPM for 15 min. The pellet consists mostly of small cells, 4–10 μm in diameter. These cells are dispersed by trituration and are then sorted using anti–Sca-1 antibody on Miltenyi Biotech (Bergisch Gladbach, Germany) magnetic columns so that the Sca-1 − cells from total hind leg muscles of five to eight mice are passed sequentially through three to four columns to obtain a high degree of Sca-1 − purity (>95% by fluorescent staining). Cells are plated at a density of approximately 10 5 cells/cm 2 in regular tissue culture dishes in 50:50 DMEM/F12 supplemented with 5% FBS, 10 ng/ml each of human EGF and bFGF (Peprotech, Rocky Hill, New Jersey, United States), insulin, transferrin, selenium, ethanolamine (ITS-X; Invitrogen, Carlsbad, California, United States ), and antibiotics (gentamicin and fungizone). Within 24 h, it is clear that the Spoc cells eluted in the Sca-1 − fraction consist of mostly nonadherent round cells, whereas the Sca-1 + fraction consists of mainly adherent cells. Within 48–72 h, the Sca-1 − culture consists of a floating population of round cells and scant adherent fibroblasts. The round cells enlarge as they divide in this medium, undergoing a few rounds of cell division. The floating population of round cell clusters is gently collected after 7 d of growth in complete growth medium and replated in the same medium, minus EGF and bFGF. The cells, which have enlarged to 10–14 μm in diameter, are by now GATA-4 positive. Spontaneously beating cells are visible while they are still floating or loosely attached, but the number of beating cells increases within a few days of replating as the cells elongate and attach to the floor of the tissue culture plate. The beating cells do not undergo any more cell divisions and can be maintained in this medium for several weeks, displaying a spontaneous beating phenotype. Alternatively, the total population of Spoc cells, unfractionated for Sca-1, can be plated in growth-factor-containing medium for 3–5 d and then the nonadherent cells removed carefully and replated without growth factors. This will result in a greater population of what will become adherent cells in the replated culture. This allows the nonadherent Spoc cells to adhere and stretch out into more mature beating cardiomyocytes. The isolation and differentiation of Spoc cells from skeletal muscle is extremely reproducible and has been performed successfully in our laboratory hundreds of times. These cells have been used for characterization, as described in the text, at different times after the initiation of the culture. Transmission EM For routine transmission EM, cells were fixed in situ on petri dishes with 1.25% glutaraldehyde in 0.1M cacodylate buffer containing 1% cadmium chloride at 4 o C for 2 h. After fixation, cells were washed three times in Sabatini's solution (0.1 M cacodylate buffer containing 6.8% sucrose), and post-fixed with 1% osmium tetroxide in cacodylate buffer for 1 h. After three washes in Sabatini's solution, samples were dehydrated in alcohol and embedded in Scipoxy 812 (Energy Beam Sciences. Agawarm, Massachusetts, United States). Polymerization was carried out at 37 o C for 24 h and then at 60 o C overnight. Ultra-thin sections were cut with a Leica Ultracut UCT ultramicrotome (Leica, Wetzslar, Germany), stained with uranyl acetate and Reynolds lead citrate, and examined with a JEOL 1200 CXII transmission electron microscope (JEOL, Peabody, Massachusetts, United States). Confocal imaging and detection of calcium transients The images were collected with a Zeiss LSM-510 laser scanning confocal system and a Zeiss C-Apochromat 63x objective (1.2 NA). Fluo-3 and fluo-4 were excited at 488 nm with an argon laser, and the emission light was collected using an LP 505 filter. The pinhole was adjusted to produce a 5-μm slice to minimize axial movements with contraction that may affect viewing the calcium transients. All transmitted light images were collected simultaneously using a transmitted light detector in conjunction with the 488-nm excitation light. Data depth for the images was 8 bit. The size of the images varied from 512 × 512 pixels to 128 × 128. Whole-cell clamp recordings Current clamp recordings were carried out using the tight-seal whole-cell patch technique at room temperature in Tyrode solution containing 136 mM sodium chloride, 5.4 mM potassium chloride, 1 mM magnesium chloride, 1.8 mM cadmium chloride, 0.33 mM sodium phosphate monobasic monohydrate, 10 mM glucose, and 10 mM HEPES (adjusted to pH 7.4 with sodium hydroxide). The pipette solution contained 20 mM potassium chloride, 110 mM potassium aspartate, 1 mM magnesium chloride, 10 mM HEPES, 5 mM EGTA, 0.1 mM GTP, and 5 mM Mg 2+ /ATP (adjusted to pH 7.2 with potassium hydroxide). Voltages were filtered at 2 kHz (–3 dB; four-pole, low-pass Bessel filter). The resting membrane potential upon breaking in was –39.8 ± 1.6 mV ( n = 9), but it generally improved by 10 mV or more. In some cases, cells were hyperpolarized slightly to AP threshold. Immunostaining Cells were fixed in either 4% paraformaldehyde at 4 o C, neutral buffered 10% formalin at 4 o C, or acetone at –20 o C for 10 min and then washed in PBS for a total of 15 min (three times). Cells were permeabilized with 0.2% Triton X-100 for 10 min. Blocking was performed for 30 min at room temperature with either 3% BSA in PBS, 5% goat serum in PBS, or 10% goat serum in PBS. Incubation with primary antibody was performed at 4 o C overnight. Cells were then washed for a total of 15 min (three times) in 1X PBS. Incubation with secondary antibody was done at room temperature for 1 h. Afterwards, cells were washed three times with PBS and mounted in a fluorescent mounting medium with 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, California, United States). A cover slip was then placed over the sample. Images of cells were obtained with a laser scanning confocal fluorescence microscope (Leica TCS-4D DMIRBE) equipped with argon and argon–krypton laser sources. Excitation wavelengths of 365 nm (DAPI), 488 nm (FITC), and 568 nm (rhodamine) were used to generate fluorescence emissions in blue, green, and red, respectively. The primary antibodies used were against MyoD (Novocastra Laboratories, Newcastle-upon-Tyne, United Kingdom), GFP (Abcam, Cambridge, United Kingdom), cardiac L-type channel (United States Biological, Swampscott, Massachusetts, United States), Sca-1 (Cedarlane Laboratories, Hornby, Ontario, Canada), and CD34 and CD45 (PharMingen, San Diego, California, United States). GATA-4, Myf-5, Myogenin, connexin 43, Nkx-2.5 Pax-3/7, c-met, and c-kit antibodies were from Santa Cruz Biotechnology (Santa Cruz, California, United States). The antibody to RLCP was produced in our laboratory and has previously been described [ 39 ]. Alpha-myosin antibody has been produced and characterized by our laboratory and is unpublished. Beta-myosin antibody was produced by our laboratory and previously published [ 42 ]. Antibody MSC 21 was generated after peripheral blood injection of Spoc cells into rats (performed by Antibody Solutions, Palo Alto, California, United States). Figure S1 was taken using a Zeiss Axiovert 200M microscope in conjunction with Compix (Lake Oswego, Oregon, United States) deconvolution software. To perform GATA-4 sublocalization studies, CPS cells in culture dishes (day 10) were incubated with 20 μM isoproterenol at 37 o C for 1 h. Cells were then collected by trituration, cytospins were made, and cells were stained for GATA-4, as described earlier. Bone marrow transplantation studies Eight 1-wk-old C57Bl/6J mice were irradiated with one total dose of 850 cGy. Four hours after irradiation, mice were injected via the tail vein with either 1.5 × 10 5 GFP + /Sca-1 − Spoc cells or 2 × 10 5 unfractionated GFP + /Spoc cells. A competitive assay was performed by injecting mice with 3 × 10 6 whole bone marrow cells and 3 × 10 4 GFP + /Sca-1 − Spoc cells. Mice were evaluated for GFP + cells in the peripheral blood and bone marrow. MI studies Sixteen 1-wk-old male C57Bl/6J mice were administered Avertin (1.25%, tribromoethanol) at 0.015 ml/g body weight or pentobarbital diluted to 0.5% and administered at 100 mg/kg IP. Animals were intubated by direct visualization, and ventilation was performed. Once anesthetized, hair in the surgical field was removed, and a thoracotomy was performed. Heart was visualized after retractors were used to enlarge the incision. The left anterior descending artery was permanently ligated using 5-O silk suture. The chest was closed with sutures. For acute infarct models, mice were monitored for recovery from anesthesia and then injected (2 h post surgery) with 1 × 10 5 C57Bl/6J Spoc cells (either fractionated or unfractionated) via the tail vein. Mice were sacrificed at 14 wk; the hearts were excised and fresh frozen sections made. For chronic infarct models, infarction protocol was performed as described earlier. The mice were maintained for either 8 or 14 wk before injection of 1 × 10 5 Spoc cells into the tail vein. Mice were sacrificed at either 2 or 5 wk post injection, and the hearts were harvested for frozen sections. For Cre expressor/beta-galactosidase reporter MI studies, the infarct procedures described earlier were performed in 16-wk-old R26Rh mice. We obtained 3 × 10 5 donor cells from 8-wk-old EIIa/Cre mice and injected the cells via the tail vein on the day of surgery. Animal studies were created under National Institutes of Health protocols 2-MC-31(R) and 1-CB-2, in accordance with the guidelines set forth by the National Heart, Lung, and Blood Institute Animal Care and Use Committee. Supporting Information Figure S1 Day 28 Spoc-Derived Cardiomyocytes Express Alpha- and Beta-Myosin (A) Day 28 Spoc cell–derived cardiomyocytes express alpha-myosin, as shown by immunostaining using anti–alpha-myosin antibody. (B) Beta-myosin is also present in these cells. (531 KB TIF). Click here for additional data file. Video S1 Round Beating Cell in Culture Video microscopy of round, spontaneously beating cell about 10 d from isolation (diameter = 15 μm), at approximately the stage shown in Figure 4 D. (5.1 MB MPG). Click here for additional data file. Video S2 Calcium Transients in Beating and Nonbeating CPS Cells Calcium transients are present in a cluster of day 14 beating cells in culture as detected by fluo-4 fluorescent dye. Only some of the cells with detectable calcium transients are visibly contracting, which shows an uncoupling of excitation–contraction, presumably because of an immature contractile apparatus or decreased movement caused by a dense extracellular matrix. (7.0 MB MPG). Click here for additional data file. Video S3 More Mature Beating Cells Appear Elliptical The round beating cell shown in Video S1 has progressed by day 14 after replating to an elliptical cell (length = 30 μm) that continues to display small contractions as shown by video microscopy. (7.4 MB MPG). Click here for additional data file. Video S4 Mature Beating CPS Cell–Derived Cardiac Myocyte in Culture Beating cell in culture (length = 55 μm). Spontaneous beating is continuous (frequency 1–6 Hz) and appears to be indefinite. Cells kept at room temperature have been noted to beat continuously for at least 3 h, and 3-mo-old cultures contain beating cells. (7.6 MB MPG). Click here for additional data file. Video S5 Beating Cells Exhibit Calcium Transients Calcium transients are seen in a day 21 beating cell, as detected by fluo-3 fluorescent dye. Flashes indicate binding of calcium to fluo-3 upon release of calcium from the sarcoplasmic reticulum. (3.5 MB MPG). Click here for additional data file. Video S6 MSC 21 Selects for a Subset of Cells That Develop into Beating Cardiomyocytes A cluster of beating cardiomyocytes derived from MACS-separated MSC 21 + cells is shown here after 14 d in culture. (9.2 MB AVI). Click here for additional data file.

D:\keerthana\PMC001xxxxxx\PMC1064849.xml

1064850

Gray Wolves as Climate Change Buffers in Yellowstone

Understanding the mechanisms by which climate and predation patterns by top predators co-vary to affect community structure accrues added importance as humans exert growing influence over both climate and regional predator assemblages. In Yellowstone National Park, winter conditions and reintroduced gray wolves (Canis lupus) together determine the availability of winter carrion on which numerous scavenger species depend for survival and reproduction. As climate changes in Yellowstone, therefore, scavenger species may experience a dramatic reshuffling of food resources. As such, we analyzed 55 y of weather data from Yellowstone in order to determine trends in winter conditions. We found that winters are getting shorter, as measured by the number of days with snow on the ground, due to decreased snowfall and increased number of days with temperatures above freezing. To investigate synergistic effects of human and climatic alterations of species interactions, we used an empirically derived model to show that in the absence of wolves, early snow thaw leads to a substantial reduction in late-winter carrion, causing potential food bottlenecks for scavengers. In addition, by narrowing the window of time over which carrion is available and thereby creating a resource pulse, climate change likely favors scavengers that can quickly track food sources over great distances. Wolves, however, largely mitigate late-winter reduction in carrion due to earlier snow thaws. By buffering the effects of climate change on carrion availability, wolves allow scavengers to adapt to a changing environment over a longer time scale more commensurate with natural processes. This study illustrates the importance of restoring and maintaining intact food chains in the face of large-scale environmental perturbations such as climate change.

Introduction Average earth temperatures have increased by 0.6 °C over the last 100 years [ 1 ] and are predicted to increase by 1.4–5.8 °C over the next century [ 2 ]. Commensurate with rising global temperatures are regional changes in weather patterns affecting the quantity and timing of precipitation and moisture levels. A challenge facing ecologists is to understand how these changes in the abiotic environment will impact populations and communities of organisms. Already, studies have documented the effect of a changing climate on the phenology, range, reproductive success, and synchrony of certain plants and animals (see [ 1 ] for a comprehensive review). In addition, climate-caused community-level changes have been documented when range shifts lead to the transfer of an entire assemblage of species [ 3 ]. Given such responses by individual species, we can expect consequent shifts in trophic structure and competitive hierarchies at the community scale [ 4 ]. Studies addressing this problem have focused primarily on how species-specific responses in phenology and geographic range alter competitive balances and the timing of food availability for neonates [ 5 , 6 , 7 , 8 ]. In Britain, for instance, winter warming has precipitated disparate responses in the breeding phenology of different amphibian species, exposing frog larvae (Rana temporaria), which have shown no phenological response, to higher levels of predation from newts ( Triturus spp.) that are entering ponds earlier than before [ 5 ]. As predicted by community stability theory, the impact of climate change on communities may vary in relation to levels of species diversity [ 9 , 10 , 11 , 12 ]. Depauperate communities or those lacking keystone species [ 13 , 14 ] may be more vulnerable to the perturbing effects of climate change than more speciose communities. As such, understanding the mechanisms or pathways that confer community resistance to climate change will be important to conservationists and managers in mitigating the effects of a changing climate on shifting community patterns and local extinctions. The reintroduction of gray wolves (Canis lupus) to Yellowstone National Park (NP) in 1995 [ 15 ] provides a research opportunity for comparing the response of an ecosystem to climate change in scenarios with and without direct human alteration of species composition. Wolf restoration is already realizing a change on the Yellowstone ecosystem by altering the quantity and timing of carrion availability to scavengers [ 16 ]. Ravens (Corvus corax), bald eagles (Haliaeetus leucocephalus), golden eagles (Aquila chrysaetos), magpies (Pica pica), coyotes (Canis latrans), grizzly bears (Ursus arctos), and black bears (Ursus americanus) are each frequent visitors at wolf kills [ 17 ] and are highly reliant on winter carrion for survival and reproductive success [ 16 , 18 , 19 , 20 , 21 , 22 ]. Prior to wolf reintroduction, winter mortality of elk (Cervus elaphus), the most abundant ungulate in Yellowstone, was largely dependent on snow depth (SDTH) [ 23 ]. Deep snows lead to increased metabolic activity [ 24 ] and decreased access to food resources, thereby causing elk to weaken and die [ 25 ]. In the absence of wolves, carrion was plentiful both during severe winters and at the end of moderate winters, but more scarce in early winter or during mild winters [ 23 ]. Reintroduced wolves are now the primary cause of elk mortality throughout the year [ 26 ]. Scavengers that once relied on winter-killed elk for food now depend on kleptoparasitizing wolf-killed elk [ 16 ]. Hence carrion availability has become primarily a function of wolf pack size, with SDTH an important but secondary factor. As global temperatures rise, evidence suggests that northern latitude and high elevation areas will experience shorter winters and earlier snow melts [ 27 ]. Given the overwhelming influence of gray wolves on scavenger food webs, community-level responses to climatic changes in the absence of wolves may differ substantially from those in the presence of Yellowstone's newly restored top carnivore. As such, we analyzed over 50 y of weather data from Yellowstone's northern range for trends in winter conditions, and constructed empirically and dynamically grounded scenarios to investigate how changes in SDTH and seasonality differentially affect scavengers in the presence and absence of wolves. Results Weather Data Analysis Over the past 55 y, average monthly SDTH at the Mammoth Hot Springs weather site show a steady decline in all winter months except November [the effect is significant at p ≤ 0.05 for February through April and nearly significant for December and January ( Figure 1 )]. Furthermore, the slope of the line relating SDTH to year becomes more negative with each month, indicating a more pronounced effect of climate change in late winter. The result for April, however, is confounded by a number of zeros, which created a violation of the normality assumption for the linear regression. Average monthly SDTH at the Tower Falls weather site ( Figure 2 ) did not indicate a strong pattern in the early winter, but showed a significant decline in the late-winter months of March and April ( Figure 2 E and 2 F). Figure 1 Winter Snow Depths 1948–2003 at Mammoth Hot Springs Average monthly SDTH for November (A), December (B), January (C), February (D), March (E), and April (F) 1948–2003 at the Mammoth Hot Springs weather site. Figure 2 Winter Snow Depths 1948–2003 at Tower Falls Average monthly SDTH for November (A), December (B), January (C), February (D), March (E), and April (F) 1948–2003 at the Tower Falls weather site. Winters in Yellowstone are getting shorter. While we did not detect a difference in the date of the arrival of the first snow, we did detect a declining trend in the date of last snow on the ground ( Figure 3 A and 3 B). Figure 3 Changes in the Last Day of Snow Cover over the Last 55 Years at Mammoth Hot Springs and Tower Falls Last day of snow cover is reported as the number of days from January 1 of that year until the first day of bare ground. Changes in last day of snow cover over the last 55 y are shown for Mammoth Hot Springs (A) and Tower falls (B). The number of days from January through March that temperatures exceeded freezing at Mammoth (C) and Tower (D) are increasing with time. At both the Tower and Mammoth weather sites, the number of days that maximum temperature (TMAX) exceeded freezing for the period of January through March increased significantly ( Figure 3 C and 3 D). Furthermore, midwinter snowfall is decreasing, and late-winter minimum temperature (TMIN) and TMAX show signs of increasing in certain months ( Table 1 ). Table 1 Regression Analyses Predicting Mean Monthly SNFL, and Average Late-Winter TMIN and TMAX Included are results from regression analyses using year as the independent variable to predict dependent variables SNFL, TMIN, and TMAX for given winter months. We present results for p < 0.10 Wolf Effects Statistical model. The presence of wolves in Yellowstone significantly mitigates the reduction in late-winter carrion expected under climate change ( Figure 4 ). In the scenario without wolves, late-winter carrion availability is reduced by 27% in March and by 66% in April. In contrast, the scenario with wolves reveals a reduction in carrion availability of only 4% in March and 11% in April. There was not a significant difference in the reduction of early- to midwinter carrion (December through February) between the two scenarios. Figure 4 Reduction in Winter Carrion Available to Scavengers due to Climate Change 1950–2000: Statistical Model Shown are percent reductions (± standard error) in winter carrion available to scavengers due to climate change from 1950 to 2000 with and without wolves in our statistical model. * Significant difference between the two scenarios. Dynamic model. Percent change, z, in late-winter carrion from 1950 to 2000 was not sensitive to changes in any of the parameters in either scenario with or without wolves. Specifically, r 2 values did not exceed 0.02 for any of the parameters regressed upon z. Mean monthly percent change in carrion availability from 1950 to 2000 under scenarios with and without wolves reveals a relative reduction in late-winter carrion from 1950 to 2000 and an increase in early-winter carrion ( Figure 5 ). Note that this change in carrion availability is much less pronounced in the presence than in the absence of wolves. Figure 5 Change in Carrion Available to Scavengers due to Climate Change 1950–2000: Dynamic Model Shown is the mean monthly change (± standard error) in carrion available to scavengers due to climate change from 1950 to 2000 with and without wolves in our dynamic model. Discussion The winter period on the northern range of Yellowstone NP is shortening. Both late-winter SDTHs and the overall duration of snow cover have decreased significantly since 1948 (see Figures 1 – 3 ). There are several potential causes of reduced snow pack. Average TMIN and TMAX values are increasing in late winter, while midwinter snowfall appears to be declining ( Table 1 ). Compounding the effects of declining snowfalls on SDTH is an increase in the number of winter days with temperatures above freezing (see Figure 3 C and 3 D). Decreases in late-winter snow pack and in the date of last snow cover imply that elk will recover sooner from the detrimental stresses of winter: Smaller snow packs allow elk easier access to food and decrease energy expenditures required for movement. In addition, herbaceous plant growth usually begins within a few days to weeks of last snow cover [ 28 ], so elk may increase the quality and quantity of food intake earlier in the year, thus shortening the physiologically stressful winter period. These factors are likely to influence the timing and abundance of carrion as late-winter elk mortality declines. As we demonstrate here, climate change serves to sharply reduce the amount of late-winter carrion available to Yellowstone's scavengers (see Figure 4 ). According to our statistical and dynamic models, however, this reduction is much less pronounced in the presence of wolves. In our statistical model, for instance, we found an 11% reduction with wolves versus a 66% reduction without wolves in April (see Figure 4 ). Our dynamic model, which incorporates wolf and elk population growth, also reveals a decline in late-winter carrion, especially in the absence of wolves ( Figure 5 ). In contrast to the statistical model, our dynamic model predicts an increase in early winter carrion, but less so with wolves. As the winter period shortens, elk that normally would die in March and April will increasingly die in the early winter months, November through February. This will lead to an increasingly pulsed or seasonal carrion resource. It is important to note that our model has more detailed elk than wolf dynamics. As suitable data become available, future work can attempt to tease out such factors as the effects of SDTH and territoriality on wolf kill-rate. In both our dynamic and statistical models we find that wolves buffer the effects of climate change on carrion abundance and timing. This effect will be crucial to scavenger species in the Yellowstone area that are highly dependent on winter and spring carrion for overwinter survival and reproduction. Under scenarios without wolves, these species could face food bottlenecks in the absence of late-winter carrion. The magnitude of this effect will depend on how quickly these species adapt to a changing environment and how their other food resources respond to a shortening of the winter period. Asynchrony of organismal responses to climate change has been prevalent in other areas, leading to changes in the competitive balance between species and to food shortages at important times of year [ 1 ]. Yellowstone should prove no exception. Species that respond to weather cues, such as many herbaceous plants, will simply start growing earlier in the year in response to earlier snow melt. Species that respond primarily to day length cues, such as some hibernating species, may change less. Coyotes, for instance, are highly dependent on late-winter and early-spring carrion to carry them over until late spring, when elk calves and ground squirrels become abundant. If late-winter carrion were to disappear without a corresponding change in the timing of elk calving or ground squirrel emergence, a serious food bottleneck could develop. As carrion becomes more concentrated over a shorter window of the year, the relative access to carrion among different scavenger species may change. Highly aggregated or pulsed resources saturate local communities of scavengers, allowing species with better recruitment abilities (animals capable of covering large distances and communicating about the location of resources such as ravens and bald eagles) to dominate consumption at carcasses [ 17 ]. Resources that are more dispersed, conversely, do not saturate local scavenger communities, so that a competitive dominance hierarchy (with grizzly bears and coyotes at the top) determines which species consume the bulk of available scavenge. Our analysis suggests that winter carrion in the absence of wolves will become increasingly pulsed during winter. Consequently, areas without wolves may experience an increase in scavengers with high recruitment abilities. Actual numerical responses by scavenger species to wolf-provided carrion can now be tested in field studies by comparing areas with wolves to those without wolves in order to determine if changes in scavenger population sizes following wolf reintroduction are consistent with the predicted magnitude of the temporal subsidy due to wolves. As the climate warms, those species will persist that are able to adapt to differences in the environment. Late-winter carrion in Yellowstone will decline with or without wolves, but by buffering this reduction, wolves extend the timescale over which scavenger species can adapt to the changing environment. It is important to note that under present-day climatic conditions, we expect wolves to decrease the long-term average elk population in Yellowstone [ 29 ]. This will lead to a corresponding decrease in average yearly carrion levels, which is expected to be small, however, because declines in carrion due to a drop in elk numbers will be partly offset by a higher turnover in the elk population due to wolf predation on old animals [ 29 ]. Scenarios both with and without wolves therefore provide a meaningful and roughly equivalent (see Figure 4 in [ 29 ]) amount of carrion to scavengers. What we demonstrate here is that scavengers in areas without wolves will experience carrion as an increasingly pulsed resource under climate change, whereas in areas with wolves carrion will remain spread out over the winter months. The primary objective of this study is to understand the influence of winter climate and predation on trophic dynamics. Our analysis is retrospective, examining what would have happened to scavenge availability in scenarios with and without wolves over the last fifty years of climate change. One may ask, however, what these results imply in light of predictions for continuing global warming into the future. Elk population numbers in Yellowstone are currently constrained by the availability of winter range, where snow levels are low enough to allow for elk movement and cratering through the snow to access food resources. If snow levels in Yellowstone continue to decline in the future, winter range expansion and thus higher elk densities are likely to occur. We expect, therefore, that the wolf-elk-scavenger complex will accrue added importance in the years to come. Future studies examining climate change impacts on spring and summer rainfall, which sets forage levels for elk, will be crucial to further deciphering the effects of global change on trophic relationships in Yellowstone. We are just beginning to understand the interaction between top predators, such as wolves, and global climate patterns. On Isle Royale, trophic effects have recently been shown to be mediated by behavioral responses to climate. There, gray wolf pack size is partly controlled by climatic conditions that, in turn, affect wolf kill-rates on moose (Alces alces) and consequent herbivory levels on balsam fir (Abies balsamea) [ 30 ]. In Yellowstone, our scenarios demonstrate that wolves act to retard the effects of a changing climate on scavenger species. Together these results begin to elucidate the expected changes that may occur to boreal ecosystems as a result of climate change effects on top predators. Materials and Methods The northern range of Yellowstone NP is the wintering area of the park's largest elk herd and home to 4–6 gray wolf packs. Elevations range from 1,500 to 3,400 m, with 87% of the area between 1,500 and 2,400 m [ 25 ]. The climate is characterized by short, cool summers and long, cold winters, with most annual precipitation falling as snow. Mean annual temperature is 1.8 °C, and mean annual precipitation is 31.7 cm [ 25 ]. Large, open valleys of grass meadows and shrub steppe dominate the landscape, with coniferous forests occurring at higher elevations and on north-facing slopes. Weather data analysis Since 1948, meteorological data has been collected daily from two permanent weather stations on the northern range of Yellowstone NP. One is located in Mammoth Hot Springs at park headquarters near the northern entrance to the park. The other is located at the Tower Falls ranger station about 29 km east of Mammoth. Data for the period 01 August 1948 to 01 June 2003 were made available to us by the Western Regional Climate Center in Reno, Nevada, United States. Using linear regression, we investigated multiannual trends in monthly average SDTH over the 55 y provided in the data set. SDTH is treated as the response variable and regressed upon year. We also examined trends in the timing of the date of first bare ground. This was defined as the first day of the year for which SDTH was zero. In order to understand changing patterns in SDTH, we analyzed average monthly snowfall (SNFL), average TMIN and TMAX, and the number of days per winter that TMAX exceeded freezing. Wolf effects: Statistical model In order to compare the effects of carrion availability to scavengers under climate change in scenarios with and without wolves, we used previously published regression equations [ 23 ] relating SDTH, S, to monthly carrion availability, C p , prior to wolf reintroduction given by and relating SDTH and wolf pack size to carrion availability, C a , after wolf reintroduction [ 16 ] obtained using where K is the wolf kill-rate per wolf, P is the wolf pack size, 30 is the number of days in a month, and Q is the percent of the edible biomass of a carcass consumed by a wolf pack given by Wilmers et al. [ 16 ]. We used Monte Carlo methods, as elaborated below, to reconstruct how much carrion would have been available to scavengers during each of the winter months (November through April) in the years 1950 and 2000 under scenarios with and without wolves. Specifically, for each scenario [1950 without wolves, 2000 without wolves, 1950 with wolves, and 2000 with wolves], we drew 100 random SDTH values for each of the months, where SDTH was assumed to be normally distributed with mean and standard error for the years 1950 and 2000 given by the regression analyses of the Tower Falls weather data (see Figure 2 ). This incorporated uncertainty into our estimate of SDTH for the years 1950 and 2000, allowing us to draw random SDTH values from those years for our Monte Carlo simulation. In the scenarios without wolves, we inserted our randomly chosen monthly SDTH values for each year and each run into equation 1 to yield the amount of carrion available per month without wolves. We used the same procedure for selecting SDTH in our scenario with wolves. In order to select wolf pack size, we assumed that wolf pack sizes were normally distributed, with a mean (± standard deviation) pack size of 10.6 (± 5) representing the current distribution of Yellowstone wolves [ 31 ]. We then inserted our randomly chosen monthly SDTH values and wolf pack sizes into equation 2 to yield the amount of carrion available per month with wolves. For each run of each scenario, we recorded the reduction in monthly winter biomass available to scavengers in 2000 as a proportion of what was available in 1950. Our statistical modeling approach, although rooted empirically, is limited by the fact that it does not take into account the possible effects of wolf and elk population dynamics on carrion availability. In order to explore these effects, therefore, we used a previously published model [ 29 ] that was originally built to explore the effects of wolf and elk population dynamics on monthly carrion flow to scavengers. Wolf effects: Dynamic model The details of the model are exactly the same as in Wilmers and Getz [ 29 ], except for the following changes. In the original model, SDTH was incorporated into the elk population dynamics but was treated as a random variable. In the present study, we modified the model so that the actual progression of winter weather from 1950 to 2000 was used. We ran the model for 51 y, from 1950 to 2001. We selected SDTH, V, for the year and month in question from the Tower Falls regression equations in exactly the same manner that we describe above in the statistical model. Since the distribution of elk among age classes from 1950 is not known, we performed, as a baseline, a 50-y run of the model under average 1950 weather conditions. This is long enough for the effects of initial conditions to dissipate. We then used the numbers and age structure of the final month of the baseline run as the initial conditions of the run using observed weather data from 1950 to 2000. Sensitivity analyses were conducted using Monte Carlo methods to assess the relative effects of different parameter values on model output [ 29 , 32 ]. Since the primary goal of using the dynamic model is to assess whether late-winter carrion will be affected by elk and wolf population dynamics in the context of a changing climate, we defined an output variable, z, as the percent change in late-winter carrion from 1950 to 2000. We assigned March and April to late winter for comparison to Figure 4 , since these are the two months showing a significant effect between scenarios with and without wolves. For each scenario, we conducted 1,000 runs of the model, choosing a different set of parameter values at random from the ranges provided in Table 1 of Wilmers and Getz [ 29 ]. Each model parameter was then regressed against z to determine its effect.

D:\keerthana\PMC001xxxxxx\PMC1064850.xml

1064851

Functional Evolution of a cis-Regulatory Module

Lack of knowledge about how regulatory regions evolve in relation to their structure–function may limit the utility of comparative sequence analysis in deciphering cis- regulatory sequences. To address this we applied reverse genetics to carry out a functional genetic complementation analysis of a eukaryotic cis- regulatory module—the even-skipped stripe 2 enhancer—from four Drosophila species. The evolution of this enhancer is non-clock-like, with important functional differences between closely related species and functional convergence between distantly related species. Functional divergence is attributable to differences in activation levels rather than spatiotemporal control of gene expression. Our findings have implications for understanding enhancer structure–function, mechanisms of speciation and computational identification of regulatory modules.

Introduction The annotation of genes from comparative sequence data rests on a fundamental evolutionary dictum, first elaborated by M. Kimura, that the rate of molecular evolution will be inversely related to the level of functional constraint. But the application of this principle would not be interpretable without a corresponding understanding of gene structure and organization (i.e., the genetic code and its degeneracy, the signals for initiation and termination of translation, intron/exon junction sequences, etc.). Knowledge of equivalent scope and depth does not exist for cis- regulatory sequences. These sequences often contain docking sites for transcription factors (TFs), but the number of binding sites and the spacing between them vary, and binding-site sequences are often degenerate to the point that they can only be characterized probabilistically. Even more striking is the lack of data relating functional evolution of gene expression to cis- regulatory sequence evolution. There are good reasons to expect the two may be only weakly correlated [ 1 , 2 ]: De novo binding sites can readily evolve [ 3 ]; individual TFs often bind at multiple locations and may be exchangeable, and the spacing between binding sites can rapidly evolve. Thus, despite recent progress [ 4 , 5 ], rules have yet to be elucidated for the functional molecular evolution of this critically important component of the genome. The Drosophila gene even-skipped (eve) produces seven transverse stripes along the anterior–posterior (A–P) axis of a blastoderm embryo ( Figure 1 ). Expression of these early stripes is regulated by five distinct cis- elements ( Figure 2 A). The best studied of them, the stripe 2 enhancer (S2E), contains multiple binding sites for five TFs, the activators bicoid and hunchback, and the repressors giant, Kruppel, and sloppy-paired [ 6 , 7 , 8 ]. Maternal deposition of bicoid mRNA in the anterior pole of the egg regulates expression of the other gap genes, which are expressed in broad A–P diffusion gradients. Spatiotemporal control of eve stripe 2 expression is brought about through the integration of these graded signals by the S2E. Figure 1 Expression of eve (A–D) Embryos of four Drosophila species at early cellular blastoderm stage. EVE stained with immunoperoxidase DAB reaction enhanced by nickel. (E–H) Df(eve) D. melanogaster embryos with two copies of transgenes containing eve S2E from four species fused to D. melanogaster eve coding region (−0.9 to +1.85 kb) at blastoderm stage. Immunofluorescence-labeled EVE. The S2E ere -EVE (G) produces consistently weaker stripes than lines carrying S2Es from the other three species. (A and E) D. melanogaster, (B and F) D. yakuba, (C and G) D. erecta, and (D and H) D. pseudoobscura. Figure 2 Genetic Constructs and Rescue Scheme (A) Summary map of the eve locus and eve S2E deletion transgene (EVEΔS2E). Adam and Apple are adjacent open reading frames [ 40 ]. The late element (Auto) and early stripe enhancers are shown. (B) S2E-EVE transgenes used to rescue eve function. The rescue EVE locus used is the D. melanogaster eve flanked by 0.9 kb of 5′ and approximately 0.6 kb of 3′ of endogenous sequence. The S2E o -EVE does not have any S2E sequences and is a negative control. The known trans -factor-binding sites in the S2E from D. melanogaster : five bicoid (circles), three hunchback (ovals), six Kruppel (squares), three giant (rectangles), and one sloppy-paired (triangle) binding site. Symbols representing sites 100% conserved compared to D. melanogaster are open, while those diverged are shaded gray. Note the evolutionary gain of novel but functionally necessary [ 6 ] activator (bicoid and hunchback) binding sites (red) in D. melanogaster lineage. Full sequences are shown in Figures S1 and S2 . (C) Example of a cross between independent rescue lines and relevant offspring genotypes for the viability assay (see Materials and Methods for details). Genetic notation b: mutant black; yellow box: native eve; R13 and X'd out yellow box: eve R13 lethal mutant; P(S2EΔEVE): eve −6.4 to 8.4 kb without S2E; P(S2E A1 -EVE) and P(S2E A2 -EVE) are two independent rescue-transgene inserts with S2E from species A. We previously used a reporter transgene assay to investigate eve S2E functional evolution in three Drosophila species in addition to D. melanogaster. The sister taxa D. yakuba and D. erecta [ 9 ] are separated by approximately 5 million years ago (MYA), while the ancestor they share with D. melanogaster existed approximately 10–12 MYA. In contrast, D. pseudoobscura is a member of a different group and is believed to have split from the melanogaster clade approximately 40—60 MYA. As expected for a trait as ontogenetically important as primary pair-rule stripe formation, the temporal progression of eve stripe expression is nearly identical among the species (see Figure 1 A– 1 D). This functional conservation of gene expression, however, is not reflected in patterns of sequence conservation (see Figures 2 B, S1 , and S2 ). Instead, S2E sequences from these species are substantially diverged, including large insertions and deletions in the spacers between known factor-binding sites, single nucleotide substitutions in binding sites, and even gains or losses of binding sites for the activators bicoid and hunchback. Yet despite these evolved differences, reporter transgene analysis showed that spatiotemporal patterns of gene expression driven by S2Es of all four species are indistinguishable when placed in D. melanogaster [ 10 ], indicating that evolved changes in the enhancer have had little or undetectable impact on spatiotemporal control of gene expression. But further experiments with native and chimeric S2Es of D. melanogaster and D. pseudoobscura showed that this functional conservation required coevolved changes in the 5′ and 3′ halves of the enhancer [ 11 ], suggesting compensatory (i.e., adaptive) evolution. This functional evidence for adaptive substitution, together with indications that levels of gene expression might also differ among the four species' S2Es, raises questions about whether these orthologous enhancers are indeed functionally identical. To overcome limitations inherent in functionally interpreting the overlap of a reporter and native gene expression, here we report results of an in vivo complementation assay to investigate S2E performance. This approach allows us to put the functional equivalency hypothesis to a rigorous test. Results Strategy and Proof of Principle First, we created a fly line, EVEΔS2E, in which the native eve S2E was deleted (see Figure 2 A). We then attempted to complement, that is, rescue this lethal mutation with the introduction of a transgene, denoted S2E-EVE, containing an eve S2E from one of the four species (D. melanogaster, D. yakuba, D. erecta, or D. pseudoobscura) linked to a functional eve promoter and coding region ( Figure 2 B). This allowed us to compare both viabilities and developmental consequences among lines differing only in the evolutionary source of their S2E. By genetically manipulating rescue-transgene copy number ( Figure 2 C), effects of EVE abundance on viability and development could also be investigated. We created the eve S2E deficiency mutant by removing a 480-bp fragment corresponding to the minimal stripe 2 element (MSE; see Figure S1 ) from a 15-kb cloned copy of the eve locus [ 12 ]. A transgene containing the complete fragment is capable of rescuing eve null mutant flies to fertile adulthood [ 12 ]. EVEΔS2E is functionally a null allele for stripe 2, as evidenced by the expression of the segment polarity gene, engrailed (en). Establishment of en 14-stripe pattern is a complex process that includes involvement by eve early stripes [ 13 , 14 ]. Eve stripe 2 corresponds to parasegment 3, which is bordered by en stripes 3 and 4. We hypothesized that these en stripes might be developmental indicators of early eve stripe 2 expression. Indeed EVEΔS2E embryos lacking a functional S2E ( Figure 3 A– 3 F) produce a short parasegment 3 and vestigial en stripe 4 ( Figure 3 F). This defect alone is almost certainly a lethal condition. Figure 3 Developmental Series of EVE Abundance (A–E) Immunofluorescence labeling of time-staged early EVEΔS2E homozygous embryos. This developmental sequence, which corresponds roughly from the initialization of cellularization (A) to its completion (E), takes approximately 45 min at 25 o C in wild-type flies [ 41 ]. (F) Expression of en in same genotype at stage 10. Arrows mark third and fourth en stripes. Note the short interval between en stripes 3 and 4 (parasegment 3) and the reduced fourth stripe. (G) EVE expression in stripe 2 during the developmental series around cellularization, where times 1–5 correspond to pictures in A–E. Stage 1 is early cellularization, while the process has been completed for embryos in class 5. The series is comparable to time classes 4–8 on the FlyEx Web site ( http://flyex.ams.sunysb.edu/flyex/ ) [ 34 ]. Estimated least square means (± SE) for EVEΔS2E/Cy stock and wild-type line w1118; note the Cy/Cy homozygote is essentially wild-type. Early eve pair-rule expression is not known to be autoregulated (as occurs in postcellularization stages), and we observe a 2-fold difference in early stripe expression, with an additive component (a) of 0.62 and negligible dominance deviation (d/a) = 0.01, for the first two stages. This dosage dependency is lost after the cellularization stage (3), presumably because all embryos carry two copies of the autoregulatory element. Transgenes containing precisely orthologous S2Es from each of the four species linked to the D. melanogaster eve promoter and coding region were introduced onto the third chromosome. The fragment we chose to investigate is 692 bp in length in D. melanogaster (see Figure S1 ). It contains the central MSE, and every other previously identified TF-binding site in the S2E region. Notably, this fragment contains completely conserved sequences at its 5′ and 3′ ends in all four species, thus ensuring that we could compare precisely orthologous fragments. As expected, all four S2E-EVE transgenes express a single early eve stripe in the expected spatial location (see Figure 1 E– 1 H). Having created the EVEΔS2E chromosome line and the S2E-EVE rescue third chromosome lines, we could then produce flies carrying EVEΔS2E; S2E-EVE in a doubly balanced configuration (see Figure 2 C). Crossing this line with itself or with another line carrying an independent copy of the same S2E allowed us to estimate relative survival to adulthood of offspring carrying one or two copies of the rescue transgene. EVEΔS2E homozygotes are embryonic lethal, whereas flies carrying two copies of the D. melanogaster S2E mel -EVE transgene in an EVEΔS2E genetic background rescue approximately 34% of flies to adulthood ( Figure 4 ). This is approximately the same rescue percentages found for the same genotype ( P[EVEG84], R13), which contains the wild-type eve locus (including the native S2E) [ 12 ]. This implies that the fragment we used to drive stripe 2 eve expression is complete and that it can function normally when removed from its native context. Importantly, our negative control, S2E 0 -EVE, does not rescue, indicating that the rescue transgene requires this enhancer to drive eve stripe 2 expression. Figure 4 Rescue to Adulthood of eve Null Mutants Rescue percentages to adulthood of EVEΔS2E homozygotes with one or two copies of rescue construct from the four species, and the negative control, denoted on x-axis. Each bar represents percentages summarized over sexes and reciprocal crosses (full data in Table S1 ). Functional Equivalence of the D. melanogaster and D. pseudoobscura S2Es We evaluated the ability of S2E-EVE rescue constructs to complement the embryonic lethal EVEΔS2E deletion by estimating survival to adulthood, based on a genetic design used extensively in Drosophila evolutionary genetics [ 15 ]. Viability measurements were made by crossing two independent lines of each rescue transgene to reduce potential recessive fitness effects caused by the site of rescue-transgene insertion. Offspring with two copies of the transgene are doubly hemizygous; few deleterious effects of transgene insertion were observed in these flies (compare, for example, EVEΔS2E, R13/CyO; S2E-EVE/S2E-EVE versus EVEΔS2E, R13/CyO; S2E-EVE/TM3 survivors in Table S1 ). Rescue abilities of S2Es from different species can be compared quantitatively because the viability of each S2E-EVE transgene is calculated relative to a standard genotype present in every cross. S2Es from the four species exhibited large differences in rescue abilities that follow neither a phylogenetic trend nor net sequence divergence ( Figure 4 ). The S2E of the most distantly related species, D. pseudoobscura, is completely conserved at only three of 18 TF-binding sites identified in D. melanogaster and is missing two of them entirely (see Figures 2 B and S2 ). It is also nearly 25% longer due to insertions and deletions in the spacers between binding sites. Yet in terms of rescue ability it is indistinguishable from the D. melanogaster S2E. Functional Divergence of S2Es from Closely Related Species Given the complete functional conservation of the D. pseudoobscura S2E, we were surprised to discover the failure of the D. erecta transgene to restore viability in EVEΔS2E homozygotes (see Figure 4 ). The inability of the doubly hemizygous S2E ere -EVE genotype to rescue cannot be due to deleterious effects of transgene insertion, because the presence of each single transgene has minimal impact on viability (see Table S1 ). Two additional independent transformants were also investigated, neither of which produced viable adult flies. We conclude, therefore, that the D. erecta sequence, although precisely orthologous to the D. melanogaster and D. pseudoobscura S2E fragments, is nonfunctional when placed in a D. melanogaster embryonic context. The D. yakuba's S2E also exhibits a rescue defect in that two copies of the rescue transgene are required for robust rescue. Flies carrying a single copy of the D. yakuba rescue transgene are less than half as viable as flies carrying one copy of either the D. melanogaster or D. pseudoobscura rescue transgene. A smaller dosage effect on viability of approximately 20% is seen with the S2Es of D. melanogaster and D. pseudoobscura. Since the spatiotemporal expression of eve stripe 2 must be the same for flies carrying one or two copies of a transgene, eve stripe 2 expression level alone must have a measurable influence on fitness. As expected, embryos carrying one or two copies of either the D. melanogaster or the functionally equivalent D. pseudoobscura S2E rescue transgene exhibit a wild-type en staining pattern, indicating a normal parasegment 3 ( Figure 5 A– 5 I). In contrast, the D. erecta S2E exhibits an en pattern defect similar to the one produced in embryos lacking eve stripe 2 expression (i.e., EVEΔS2E homozygote). The inability to drive normal en expression provides further evidence that the D. erecta S2E is a weak (or nonfunctional) enhancer in the D. melanogaster genetic background. Figure 5 Effects on en Expression (A, C–I) The en pattern in homozygous EVEΔS2E and (B) wild-type (w1118) specimens at stages 9–11. All strains (except [B]) are homozygous for Df(eve) P(EVEΔS2E) second chromosomes, with the third chromosome differing only by rescue transgenes: (A) no rescue transgenes; (C) P(mel 36)/P(mel 36) is a S2E mel -EVE stock; (D) P(yak 74)/TM3 Sb and (E) P(yak 74)/P(yak 74) are S2E yak -EVE stocks; (F) P(S2E o -EVE)/P(S2E o -EVE) has no S2E; both (G) P(ere 41)/P(ere 41) and (H) P(ere 21)/P(ere 21) are S2E ere -EVE transgenic stocks; and (I) P(pse 91)/P(pse 91) is a S2E pse -EVE stock. Note the variation in distance between third and forth en stripes (arrows) and relative level of en expression in the fourth stripe. Only the first seven parasegments of the en pattern are show (except in [A]). The en protein was visualized by an immunoperoxidase DAB reaction enhanced by nickel. mel: D. melanogaster; yak: D. yakuba; ere: D. erecta; pse: D. pseudoobscura. S2E o -EVE lacks a S2E . The D. yakuba S2E also exhibits an en phenotype that correlates with its ability to rescue ( Figure 5 D and 5 E). With two copies of the enhancer present, embryos exhibit a robust en stripe 4, indistinguishable from wild-type. But with only one copy present, en stripe 4 expression is shifted anteriorly relative to its neighbors, an indication that parasegment 3 is not forming properly. Some of these embryos survive to adulthood since we do observe one-copy adults in our viability experiment, albeit at a lower than expected percentage. Although adult flies are superficially “normal,” we can observe subtle morphological defects (mouthparts and thoracic structures) in the segments corresponding to parasegment 3. Differences in eve S2E Expression Levels To test whether differential gene expression might be the critical functional difference between the S2Es, we quantified eve stripe 2 protein in early embryos. The experimental design allowed us to normalize eve stripe 2 expression in individually stained embryos relative to stripe 3, thus facilitating comparison across embryos and genotypes. We also developed a PCR method to ascertain the genotype of individually stained embryos. We validated the quantification procedure by comparing eve stripe 2 expression levels in embryos carrying zero, one, or two copies of the S2E in its native position in a wild-type eve locus—that is, EVEΔS2E/EVEΔS2E, EVEΔS2E/Cy, and Cy/Cy embryos, respectively, and a homozygous w1118 line (see Figure 3 G). The expected dose dependence is observed in response to EVEΔS2E copy number prior to cellularization, followed by a shift to dose independence as control of eve stripe expression is transferred to the late (autoregulatory) element. Unexpectedly, a weak early stripe 2 (estimated to be approximately 20% of the wild-type level) can be detected in EVEΔS2E homozygotes; we do not know what drives this stripe. Normalized stripe 2 expression in early embryos carrying S2Es from D. erecta and D. pseudoobscura is consistent with adult viability ( Figure 6 ). The D. erecta S2E-driven eve expression is too weak to observe statistically significant expression comparing embryos containing zero, one, or two copies of the rescue transgene. Note, however, that this transgene does drive weak eve stripe 2 expression in a fully eve null background (see Figure 1 G). Formally, we observe statistically significant effects of gene “dose,” S2E “species” of origin, and most notably a “dose × species” interaction on stripe 2 expression by a mixed-model analysis of variance (ANOVA) (see Tables 1 and S2 ). Therefore, the major functional evolutionary difference between these enhancers is likely to reside in their activation strengths. Figure 6 Diverged S2Es Contribute Differentially to EVE Abundance Fluorescence-labeled antibody staining of EVE in embryos with zero (A, C, and E) or two (B, D, and F) copies of rescue transgene. A dose effect is seen in D. pseudoobscura line 91, (A and B), while none is observed in D. erecta line 41 (C and D) or 21 (E and F). (G) These effects are significant when comparing EVE protein quantity (least square means ± SE) in stripe 2 (Dose × Species, F = 4.69 (2, 100.44) , p = 0.01; see Tables 1 and S2 ) D. pseudoobscura (black circles, n = 59) and D. erecta embryos (open circles, n = 71). For D. pseudoobscura the estimated additive component (a) = 0.37 and dominance deviation (d/a) = 0.17. Table 1 ANOVA on EVE Abundance in Stripe 2 from D. erecta and D. pseudoobscura S2E Rescue Constructs Mixed-model term “Dose” indicates copy number of rescue transgene; “Time” is a continuous variable of the developmental series; “Species” indicates the origin of the S2E and “DV index” reflects the dorsal–ventral orientation of each embryo. VC indicates variance components, with estimated standard errors and significance according to the z-distribution ns, not significant; p > 0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 Discussion Evolution of Enhancer Structure–Function The D. melanogaster S2E rescue transgene, and its considerably diverged D. pseudoobscura ortholog, each restore complete eve stripe 2 biological activity when placed in a genetic background lacking a native S2E. The DNA fragment we investigated, therefore, entails both the biological and evolutionary units of enhancer function. We chose this fragment based on its extensive prior characterization, including genetic, reverse genetic, and footprinting analyses [ 6 , 16 , 17 , 18 ]. In particular, Stanojevic et al.'s [ 18 ] TF footprinting data appear to have nicely delineated the functional enhancer. Our previous experiments with S2Es of these two species demonstrated that both intact enhancers, but not the chimeras between them, drive the correct spatiotemporal pattern of reporter gene expression [ 11 ]. The rescue experiments reported here extend this finding by showing that the two orthologs are in fact biologically indistinguishable. These new results reinforce our contention that the phenotypic character—early stripe 2 expression—must be under stabilizing selection. The character itself remains unchanged over evolutionary time despite substitutions in nearly all the TF-binding sites, the gain and loss of some of them, and considerable change in the spacing between sites. This suggests to us that unlike proteins, where functional conservation usually means selective constraint on important amino acids (such as the active site of an enzyme), enhancers have a more flexible architecture that allows modification, and perhaps even turnover, of their “active” sites. Dissimilarities in the structure–function of enhancers and proteins result in different emergent “rules” of molecular evolution. But the fact that the D. melanogaster and D. pseudoobscura S2Es are biologically indistinguishable does not necessarily imply that enhancer function has been evolutionary static. Rather, the similar biological activities appear to be the result of convergence. In particular, phylogenetic analysis of S2E sequences indicates that the bcd-3 binding site in D. melanogaster was acquired only recently in the lineage leading to D. melanogaster (see Figure 2 B). (There are also lineage-specific deletions in the spacers flanking both sides of the bcd-3 site in the D. melanogaster lineage, which shift the proximal and distal repressors giant and Kruppel binding sites, respectively, closer to this bicoid site. These length changes may have coevolved to enable or increase local repression of this novel activator site.) The bcd-3 site was shown by Small et al. [ 6 ] to be required for MSE stripe expression. It seems likely, therefore, that the ancestral S2E lacking this binding site would not properly activate stripe 2 expression in D. melanogaster. Perhaps the sensitivity of the enhancer (or more precisely, the fragment investigated) to activator signals has oscillated over evolutionary time, in which case the similarity between the two distantly related species' S2Es would be an example of functional convergence. The fact that the S2E fragment from D. erecta is essentially unresponsive to the D. melanogaster morphogen-gradient environment, but the precisely orthologous segment from D. melanogaster (and D. pseudoobscura ) responds properly, proves that this fragment must contain evolved differences of functional significance between the species. The lack of biological activity of the D. erecta transgene in D. melanogaster should perhaps come as no surprise, however: Its lower sensitivity to activation may represent the ancestral state of the enhancer. What is surprising is the rapidity with which these functional differences evolve. Phylogenetic footprinting of distantly related species can readily identify strongly conserved motifs [ 19 ] but runs the risk of not detecting enhancers that have retained their function but have evolved structurally. To overcome this, a technique called phylogenetic shadowing—the comparison of noncoding sequences among closely related species—has recently emerged [ 9 ]. Our results show that there is no necessary relationship between enhancer phylogenetic (or sequence) relatedness and functional similarity. Closely related species cannot be assumed to be more functionally conserved than distantly related species in enhancer structure–function. Why Is the D. erecta S2E Transgene Not Functional in D. melanogaster? D. erecta produces a native early eve stripe 2. Why then does the S2E fragment from this species not produce a robust early stripe when placed in D. melanogaster? The first possibility is that the fragment we investigated no longer contains a functional enhancer and has been replaced by an equivalent enhancer somewhere else in the eve locus. This possibility can easily be ruled out: The overall architecture of the eve locus, including all of its 5′ and 3′ enhancers, is well conserved, and there is no new cluster of the appropriate TF-binding sites that could act as a S2E. Another unlikely possibility is that the locus has been duplicated, and the fragment we investigated has become functionally inert (i.e., equivalent to a pseudogene). There is no indication of a duplicated eve locus in the D. erecta genome, and all features of the eve locus (including its S2E) are intact and do not indicate any degeneration. This leads us to conclude that the D. erecta fragment used in our experiments contains the S2E. We can consider three additional possibilities. The first is that this fragment is no longer the complete biological unit, that is, novel binding sites have evolved in this species distal or proximal to this fragment, which have become assimilated into the active enhancer by a process we call accretion. As Figure 7 shows, patser, a binding-site prediction program [ 20 ] identifies a single potential bicoid-binding site 135 bp upstream of Block-A ( Figure S1 ), the distal end of the D. erecta S2E transgene. This potential site contains an unconventional bicoid-binding motif, T CAATC CC. The next closest potential binding site is another 350 bp further upstream and also has an unconventional binding-site sequence ( A CAATC GG). So, although we cannot rule out the possibility that these are biologically active sites that contribute to S2E activity, they are relatively distant from the recognized S2E (and other bicoid sites), and their sequences do not have the consensus core motif ( TAATC ). Future experiments will allow us to formally test whether this enhancer has physically expanded. If so, this would be the first documented case for accretion, the adaptive expansion of an enhancer. Figure 7 Predicted Binding-Site Composition and Sequence Conservation in the eve 5′ Noncoding Region (A) D. melanogaster predicted binding-site composition. (B) D. yakuba predicted binding-site composition and sequence conservation with D. melanogaster. (C) D. erecta predicted binding-site composition and sequence conservation with D. melanogaster. (D) D. pseudoobscura predicted binding-site composition and sequence conservation with D. melanogaster. Coordinates of functionally characterized enhancer sequences are shown in light blue, and unannotated conserved noncoding sequences are shown in pink. The coordinates of the homologous stripe 2 sequences correspond to the constructs in Figure S1 , while the coordinates of the AR and stripe 3 enhancers have been estimated based on sequence conservation. Note that the scale of the genomic intervals plotted differs between panels (black bar = 500 bp). Binding sites are indicated by color; bicoid (blue), hunchback (red), and Kruppel (green). The second possibility is coevolution between the D. erecta S2E and its promoter region, such that it is not capable of driving transcription properly from a D. melanogaster promoter. We also view this as unlikely for several reasons. First, prior to designing these experiments, we investigated this issue with the core promoters and S2Es of D. melanogaster and D. virilis (which is an outgroup to the species studied). We could detect no difference in spatial or temporal expression of each S2E with either promoter (Ludwig, unpublished data). Second, the core promoter regions of D. melanogaster and D. erecta are highly conserved, including complete preservation of both the TATAA and the GAGA site. Indeed there are only four nucleotide differences (and no indels) between the species in a 150-bp stretch containing these sites. Finally, one might expect most functional changes in the core promoter to be pleiotropic, given the presence of more than a dozen other separable enhancers in the eve locus, and therefore to be selected against. The final possibility is that the D. erecta S2E fragment does contain the entire biological enhancer, but that the trans -acting environment—the morphogen gradients to which the enhancer responds—differ between the species, causing the enhancers to have evolved to accommodate the differences. In other words, the sensitivity, or set point, of the binary (on–off) switch function has coevolved with the trans -acting environment in order for the S2E to maintain the appropriate response to evolved activation inputs. This hypothesis implicates, in particular, evolutionary shifts in the bicoid and/or hunchback activator gradients. As noted above, there has been a lineage-specific addition of the functionally required bcd-3 binding site [ 6 ] in D. melanogaster that is not present in any of the other species. Second, there is also a lineage-specific loss of the hunchback-1 (hb-1) site in D. erecta (which may be present in its sister taxon, D. yakuba ). We propose that the lack of sites for these activators, and the presence of a species-specific six-base-pair insertion in the overlapping hb-2/kr-2 binding sites ( Figure S1 ) reduces the ability of the D. erecta enhancer to respond to the activator gradients of D. melanogaster. This hypothesis predicts stronger activator gradients in D. erecta than in D. melanogaster. Although we have not yet investigated this possibility directly, we note that native eve blastoderm stripes do not reside in the same physical locations in embryos of the two species, but rather are displaced posteriorly in D. erecta compared to D. melanogaster (compare Figure 1 A and 1 C). A similar effect can be mimicked in D. melanogaster with the addition of extra copies of bicoid gene [ 21 ], which shifts the morphogen gradient posteriorly. The possible independence of spatiotemporal and rheostat activities [ 22 ] relates to a long-standing issue in evolutionary genetics—whether developmental constraints are “tunable” [ 23 ]. One school holds that features of development are strongly canalized and that deviation from this path will be strongly selected against. An opposing school holds that these developmental constraints can always be overcome by selection. The eve S2E may exhibit elements of both an immutable developmental constraint and a smoothly evolving trait. Primary pair-rule stripes, such as those laid down by eve in developing blastoderm embryos, establish the positional landmarks that will eventually demarcate segmental identities in the fly. This complex functional network established by gap and pair-rule gene expression imposes strong constraints on spatiotemporal patterns of regulatory gene expression. The S2E must, for example, produce a stripe equidistant from stripes 1 and 3, and at a specific location with respect to other pair-rule genes. The potential for evolutionary change in the S2E set point or output, on the other hand, may be much less constrained. Genetic variability in the gain and loss of binding sites, polymorphism in binding sites leading to variation in TF binding, and modulation of TF interactions through changes in spacing between binding sites should allow for nearly continuous fine-tuning of these functions. Enhancers should be able to adapt to change in their trans -acting environment. Why might the trans -acting environment for the S2E be evolving? Residing at the head of the hierarchal cascade driving the formation of the A–P axis, perhaps the bicoid morphogen gradient is less constrained than the expression of downstream genes that are more deeply embedded in the interaction network. Or, perhaps there has been natural selection acting on traits such as egg shape, size, or ovariole morphology, leading to changes in the bicoid-diffusion gradient. Egg number and size are, after all, fundamental evolutionary trade-offs. With a properly designed genetic experiment it should be possible to test these hypotheses. Enhancer Evolution in Relation to Speciation Postmating isolation is the final step in the speciation process because it involves genetic changes between incipient species that cause hybrid inviability or infertility. Hybrid breakdown is likely to involve evolutionary changes in at least two interacting genes [ 24 , 25 ]. According to this model, the coevolution of a “speciation” gene with its partner(s) allows it to remain functional in its native background but lethal in the hybrid background. One of the mysteries of this process is the regularity and quickness with which molecular incompatibility arises. Thus, for example, exceedingly closely related species, such as D. simulans, D. sechellia and D. mauritiana, whose common ancestors occurred less than 1 MYA, nevertheless, have evolved postmating isolation involving perhaps hundreds of genes [ 26 ]. The question is what components of the genome are involved in this coevolutionary conspiracy? Cis- regulatory modules and the trans -acting TFs with which they interact provide abundant genetic substrates for coevolution leading to hybrid sterility and inviability [ 2 , 5 ]. If our experiments have captured the entire biological S2Es of D. erecta and D. yakuba, then the results suggest that this coevolution could involve changes in expression patterns or levels, rather than changes in the protein sequences of the trans -acting factors. The attractiveness of this hypothesis lies in the fact that there must be many more cis- regulatory modules in a eukaryotic genome than there are encoded proteins. Thus there is ample opportunity for the coevolution of enhancers and trans -factors to produce lethal interactions in hybrids, which may explain the abundance of lethal interactions between closely related species. The regulation of development is often modeled as a logic circuit, with cis- regulatory sequences functioning as switches controlling information flow. The long-term functional preservation of both the spatiotemporal and the activation strengths of the D. melanogaster and D. pseudoobscura S2Es speaks to the general conservation of this genetic network in fruit fly development. Our results also provide an indication that the stoichiometry of the regulatory components could matter critically for normal development, at odds with theoretical predictions [ 27 ]. Epistatic changes accompanying interspecific inviability and sterility may therefore arise more readily as a consequence of quantitative shifts in gene expression than as a result of alterations in the topology of the developmental circuits. Materials and Methods Drosophila strains Df(2R)eve : (Df[eve]) and eve R13 (R13): Df[eve] is a deficiency that includes at least three lethal complementation groups [ 28 ]. The R13 is null mutation that truncates the protein within the homeodomain [ 12 ]. These lethal mutations were balanced over marked balancer chromosome CyO P(hb-lacZ) to allow identification of mutant embryos by immunostaining for β-galactosidase or by PCR analysis for β-galactosidase gene. Analysis of embryos Histochemical staining with guinea pig polyclonal a-Eve [ 29 ] at 1:1,000 dilution, or with a-En monoclonal 4D9 [ 30 ] at 1:10 dilution was visualized using HRP-DAB enhanced by nickel [ 31 ]. Fluorescent staining of Drosophila embryos with polyclonal a-Eve [ 29 ] at 1:1,000 dilution was visualized using Alexa Fluor-594 goat antiguinea pig IgG antibodies (Molecular Probes, Eugene, Oregon, United States) at 1:400 dilution. Optical Z-sectioning (0.8-μm/step) of fluorescent embryos was carried out using motorized microscope Axioplan2 (Carl Zeiss, Thornwood, New York, United States), Hamamatsu C4742–95 camera (Hamamatsu City, Japan), and “Openlab” software (Improvision, Emeryville, California, United States). Genotyping Individual embryos were genotyped after imaging, using a PCR method (three pairs of fluor-labeled PCR primers) and Beckman Coulter CEQ ™ 8000 genetic analysis system for PCR fragment analysis (Allendale, New Jersey, United States; for detailed protocol see Supporting Information ). Three sets of fluor-labeled primers (Proligo, Boulder, Colorado, United States) were used for PCR: P(hb-lacZ): (1) marker for SM1(CyO) balancer second chromosome (156 bp), +106 adh: 5′ TCTGGGAGGCATTGGTCTGGA 3′ and −241 lac: 5′ CGGGCCTCTTCGCTATTACG3′, (2) EVEΔS2E and native eve locus: 592 bp and 113 bp markers for native S2E and S2E with 480 bp MSE deleted, respectively, +23 Df: 5′ TAACTGGCAGGAGCGAGGTATC3′ and −115 Df: 5′ CTCGCGGATCAGGGCTAAGT3′, (3) DMPROSPER, 3rd chromosome microsatellite marker for TM3 Sb balancer and rescue transgene-containing chromosome III, DMPROSRER F: 5′ CGGTACAAAGTGTGTGTTC3′ and DMPROSRER R: 5′ GACTTTTAAACATTTAAGATTAATTCC3′. S2E mutant construct (EVEΔS2E) A pCaSpeR-based vector containing wild-type eve genomic DNA from −6.4 to +8.4 kb (EVEG84) was provided by Miki Fujioka [ 12 ]. The deficiency eve S2E (EVEΔS2E) mutation was created by deleting the region from −1554 (BstEII) to −1073 kb (BssHII) relative to eve transcription start site (see Figure 2 ). S2E rescue constructs (S2E-EVE) The S2Es used in this study were cloned previously [ 10 ], and their accession numbers are given as supplemental information. The sequences employed in this study are presented in Figure S1 ; they all begin and end at conserved sequences flanking the S2E (blocks A and B). The S2E - eve rescue transgenes based on our modification of the E-eve pCaSpeR vector provided by M. Fujioka [ 32 ] were constructed as follows: 2.76-kb eve CaSpeR (negative control construct, S2E 0 -EVE ) carries the D. melanogaster wild-type eve genomic DNA fragment from −913 (FspI) to +1.85(MluI) relative to transcription start site, which includes 913 bp intact eve 5′ upstream region, protein-coding sequence, and the polyadenylation site. A unique restriction site PmeI was created instead of the FspI site, so that S2Es from different species could be cloned into the 2.76-kb eve CaSpeR vector by using unique PmeI and NotI sites in the proper orientations. The entire eve region in rescue constructs from D. melanogaster and D. erecta, including both the S2E and eve genomic DNA fragment from −913 to +1.85, were confirmed by sequencing. P-transformation P-element-mediated germline transformation was carried out as described by Rubin and Spradling [ 33 ]. A homozygous viable stock with the S2E deficiency transgene P (EVEΔS2E) on chromosome II was recombined onto eve mutant chromosomes—either Df(eve) or eve R13 —to create lines that can only drive eve expression from the EVEΔS2E transgene. These chromosomes are maintained as balanced stocks. Rescue-transgene lines (S2E-EVE) were chosen for use in the study if they were homozygous viable and were on chromosome III. Between 2 and 4 independent stable transformed lines were generated for each rescue construct and were examined for rescue ability to adulthood and for local eve and en pattern rescue. Rescue of eve function to adulthood Each transgenic rescue line was crossed into the eve R13 (R13) P (EVEΔS2E) mutant background, generating flies of the genotype w; b R13 P (EVEΔS2E)/CyO; P (S2E A1 -EVE)/TM3 Sb. These flies were reciprocally crossed with flies of the genotype w; b R13 P (EVEΔS2E)/CyO; P (S2E A2 -EVE)/TM3 Sb. A1 and A2 indicate independent transformed lines with the identical rescue construct (see Figure 2 ). Adults were scored for both a wild-type wing phenotype (non-CyO) and the black (b) phenotype (indicating R13 homozygotes). For each reciprocal cross 50 healthy virgin females from one strain were mated in a standard culture bottle with 100 healthy males from the other. Parents were transferred to fresh culture bottles every 3 d for 24 d. The emerging adult offspring were collected every day from the culture bottles for a period of 10 d for scoring. This approach ensured that mutants with slow development rates were counted. The cultures were kept at 25 °C. Viability of flies carrying one or two copies of the rescue transgene was measured relative to the number expected based on the count of flies carrying one copy of the dominantly marked second and/or third chromosome. This genetic design allowed us to estimate viability effects of transgene insertion independent of transgene rescue ability by comparing genotypes carrying one (hemizygous) or two (doubly hemizygous) copies of the rescue transgene in genotypes carrying a wild-type eve locus (i.e., EVEΔS2E, R13/CyO; S2E-EVE/TM3 versus EVEΔS2E, R13/CyO; S2E-EVE/S2E-EVE; see Table S1 ). Viabilities of rescue genotypes were calculated by comparing the number of adult survivors in EVEΔS2E homozygotes (EVEΔS2E, R13/EVEΔS2E,R13) relative to the number of survivors in the corresponding genotype with one copy of a functional eve locus (EVEΔS2E, R13/CyO; S2E-EVE/S2E-EVE). Localized rescue of eve and en expression patterns Each transgenic rescue line was crossed into the Df(eve) P( EVEΔS2E ) mutant background, generating stock w; Df(eve) P (EVEΔS2E)/CyO P (hb-lacZ) ; P (S2E-EVE)/TM3 Sb. The embryos of these stocks were stained with anti-Eve and anti-Engrailed antibodies to determine the pattern and level of genes expression. The embryos inspected for en were dissected flat, and the proctodeum and posterior midgut removed, anterior up, and viewed from the ventral side (see Figure 3 F for undissected specimen). Genotypes of the embryos were determined after images were taken, as described above. Image processing and EVE quantification eve stripes 1, 3, 4, 5, 6, and 7 are always produced from the EVEΔS2E locus in our experimental flies, whereas eve stripe 2 comes primarily from the independent S2E-EVE rescue transgene. This allowed us to compare stripe 2 expression levels in different embryos and genotypes by measuring stripe 2 expression relative to an adjacent stripe. We chose the adjacent stripe 3 as a reference, as it has similar temporal and quantitative expression, and we report the stripe 2 expression relative to stripe 3. Image stacks (0.8 μm) were deconvoluted in Huygens Essential (version 2.5.0 from Scientific Volume Imaging, Hilversum, the Netherlands) and 3–5 images in the focal plane collapsed to a single image in Image J (version 1.30, free at http://rsb.info.nih.gov/ij/ ). All were subjected to the same background subtraction and a section corresponding to stripes 1 to 4 in the middle section of the embryo cropped out and saved as a raw pixel file for analysis in Mathematica 5 ( www.wolfram.com ). The location of each stripe was estimated by fitting second order curves to local maxima/minima of smoothed data, identifying stripes 1 through 4 and the interleaving troughs. The approach worked reasonably, as of 205 embryos surveyed in the rescue experiment only 30 were rejected because the algorithm could not detect a stripe 2. The reason for rejection was in the majority of cases not the lack of stripe 2 expression in individuals homozygous for the EVEΔS2E, but rather the absence of a detectable trough between stripes 1 and 2 during early stages of stripe formation. The fitted curves were superimposed on the raw data and used as guides for the summing of the signal intensity, from the middle 25%, 50%, or 75% of the stripes and the 25% of the troughs. These percentages were derived from the total length between stripes 1 and 4, and therefore represent comparable geometric areas. This is important as we proceed to compare abundances in stripes 2 and 3. Percentages of stripes were used to account for size variation as variables extracted on the basis of absolute pixels were noisier. The measurements were summed from 5 to 11 sections per embryo, each 30-pixels high. The dependent variables analyzed were ratios of the measured fluorescence in stripe 2 (P2) over stripe 3 (P3) after adjusting the measured background in the separating trough (T2), in general form: Y = (P2 − T2)/(P3 − T2). These variables were derived for the middle 25%, 50%, and 75% of the stripes, but all yielded similar results, and we report on the 75% case. For every embryo we documented a DV index, indicating its degree of rotation along the dorsal–ventral axis (divided into five categories, dorsal, dorsal–middle, middle, middle–ventral, and ventral). In addition we placed the embryos into a series (five classes) around cellularization corresponding to approximately 45 min of development, when the EVE stripes originate and take form (see Figure 3 A– 3 E). The series is based on classes 4–8 for the 14th cleavage cycle available on the FlyEx Web site at http://flyex.ams.sunysb.edu/flyex/ [ 34 ]. ANOVA on EVE expression Mixed-model ANOVA was fitted in SAS v8.2 (2002; SAS Institute, Cary, North Carolina, United States) with the main effects of “dose,” that is, how many copies of rescue construct; “species,” designating the origin of the S2E; and a “DV index,” which accounts for orientation, along the dorsal–ventral axis, of measured embryos. Also, “time,” indicating the inferred stage of development, was used as a covariate. Finally, as each embryo was measured several times, we avoided pseudoreplication by nesting the random variable “embryo” within the fixed effects. Y = μ + Dose + Species + Time + DV index + D × S + D × T + D × DVI + D × S × DVI + Embryo (D × S × DVI) + ɛ The results were identical if the multiple recordings on each embryo were designated as repeated measures within the Proc Mixed statement. We also applied a generalized linear model to the least square mean of the phenotypic values for each embryo. Again results were in accord (see Tables 1 and S2 ). A related model, excluding “species” terms, was used to evaluate the “dose” effects in the EVEΔS2E stock. We also investigated the general capacity of the D. pseudoobscura construct to reconstitute the activity of the endogenous gene. The stock with EVEΔS2E carried over a Cy balancer was used, and embryos of all three genotypes were collected. The mixed model was constructed in the same way, adding “experiment” as a term. All stocks were constructed to have the same genetic background, and we predicted that individuals homozygous for the EVEΔS2E in both datasets would have similar expression. This was corroborated by ANOVA with a reduced mixed model, F (1, 52.54) = 1.70, p = 0.20, n = 71. Similarly, the EVE expression of the homozygous balancer could not be distinguished from a wild-type strain w1118: F (1, 63.71) = 1.00, p = 0.32, n = 76 (see Figure 3 G). Using standard quantitative genetic theory [ 35 ], we assessed the genotypic effects of the S2E on eve expression in the D. pseudoobscura transgenes and the endogenous gene. The small sample size prohibited formal tests of deviations from additivity. Sequence analysis Comparative sequence analysis was conducted with the Avid-Vista suite ( http://www-gsd.lbl.gov/vista/ , using default parameters [ 36 , 37 ]. Binding-site prediction was performed using patser [ 20 ] with identical command-line arguments and position-weight matrices used by Berman et al. [ 38 ]. Cutoffs for display of predicted binding sites were set to recover all known binding sites in the D. melanogaster S2E–ln(P) = −6.1 for bicoid, ln(P) = −8.06 for hunchback, and ln(P) = −6.65 for Kruppel. Supporting Information Figure S1 Alignment of eve S2E Regions from Four Species of Drosophila Gaps in aligned sequences are indicated by dashes. The binding sites in D. melanogaster for the trans -acting factors, bicoid (bcd), hunchback (hb), Kruppel (kr), sloppy-paired (slp), and giant (gt), are shown above the sequence. The sequences from the four species begin and end at completely conserved sequences, indicated by blocks A and B, flanking the enhancer. mel: D. melanogaster; yak: D. yakuba; ere: D. erecta; pse: D. pseudoobscura. (524 KB JPG). Click here for additional data file. Figure S2 Trans-Factor-Binding Sites in D. melanogaster S2E and Homologous Sequences from Three Other Drosophila Species Binding sites are for five proteins are designated; bicoid (bcd), hunchback (hb), Kruppel (kr), sloppy-paired (slp), and giant (gt). N/A: no homologous sequence identified. mel: D. melanogaster; yak: D. yakuba; ere: D. erecta; pse: D. pseudoobscura. The figure is expanded from Ludwig et al. [ 10 ] and Ludwig [ 2 ]. (954 KB JPG). Click here for additional data file. Protocol S1 Additional Materials and Methods (29 KB DOC). Click here for additional data file. Table S1 Viability to Adulthood Adult viability of individuals from reciprocal crosses between two independent transgenic lines with eve S2E from the four species and from a negative control (S2E 0 -EVE). See Figure 2 for constructs and crossing scheme. With Mendelian segregation of the 2nd and 3rd chromosomes, ratios of 4:2:2:1 for the genotypic classes are anticipated. This was used to calculate expected counts and rescue percentages (in brackets) for the classes with one and two copies of rescue constructs. Adjusted rescue percentages, which account for possible detrimental effects of two hemizygous transgenic inserts, are also reported. These data summarized over sexes are represented in Figure 4 . (31 KB DOC). Click here for additional data file. Table S2 ANOVAs on EVE Abundance in Stripe 2 ANOVAs on EVE quantity. (A) Analysis on amounts of EVE in stripe 2 D. erecta and D. pseudoobscura rescue data (B) on the rescue data for all four species and (C) the EVEΔS2E/Cy stock. Mixed-model ANOVA (left) fitted the multiple measures per embryo as random. Generalized linear model (right) was implemented on the least square means calculated for each embryo. “Dose” indicates the copy number of rescue transgene per embryo; “Time” indicates a continuous variable of the developmental series; “Species” indicates the origin of the S2E; “DV index” indicates orientation of each embryo along the dorsal–ventral axis. VC indicates variance components for “embryos” or residual error, with estimated standard errors and significance according to the z -distribution (VCs can only be estimated with a mixed model). ns, p > 0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. (93 KB DOC). Click here for additional data file. Accession Numbers The GenBank ( http://www.ncbi.nlm.nih.gov/ ) accession numbers for the S2Es sequences used in genetic constructs are AF042712 (D. pseudoobscura), AF042711 (D. erecta), AF042710 (D. yakuba), and AF042709 (D. melanogaster). The sequences for genomic alignments of the eve 5′ region were extracted from GenBank ( D. melanogaster genomic scaffold, AE003831, and accessions AY190939 and AY190942 for D. erecta and D. pseudoobscura, respectively [ 39 ]). The exception was the D. yakuba, sequence, which came from the 7 April 7 2004 version available on the Washington University genome page ( http://genome.wustl.edu/ ).

D:\keerthana\PMC001xxxxxx\PMC1064851.xml

1064852

Light Controls Growth and Development via a Conserved Pathway in the Fungal Kingdom

Light inhibits mating and haploid fruiting of the human fungal pathogen Cryptococcus neoformans, but the mechanisms involved were unknown. Two genes controlling light responses were discovered through candidate gene and insertional mutagenesis approaches. Deletion of candidate genes encoding a predicted opsin or phytochrome had no effect on mating, while strains mutated in the white collar 1 homolog gene BWC1 mated equally well in the light or the dark. The predicted Bwc1 protein shares identity with Neurospora crassa WC-1, but lacks the zinc finger DNA binding domain. BWC1 regulates cell fusion and repression of hyphal development after fusion in response to blue light. In addition, bwc1 mutant strains are hypersensitive to ultraviolet light. To identify other components required for responses to light, a novel self-fertile haploid strain was created and subjected to Agrobacterium -mediated insertional mutagenesis. One UV-sensitive mutant that filaments equally well in the light and the dark was identified and found to have an insertion in the BWC2 gene, whose product is structurally similar to N. crassa WC-2. The C. neoformans Bwc1 and Bwc2 proteins interact in the yeast two-hybrid assay. Deletion of BWC1 or BWC2 reduces the virulence of C. neoformans in a murine model of infection; the Bwc1-Bwc2 system thus represents a novel protein complex that influences both development and virulence in a pathogenic fungus. These results demonstrate that a role for blue/UV light in controlling development is an ancient process that predates the divergence of the fungi into the ascomycete and basidiomycete phyla.

Introduction Light is the fundamental energy source for life on earth and as such is a major environmental signal for organisms from all kingdoms of life. In the fungal kingdom, light can regulate growth, the direction of growth, asexual and sexual reproduction, and pigment formation, all of which are important aspects for the survival and dissemination of fungal species. These processes have negative implications to many aspects of human life, as the uncontrolled proliferation of fungi can lead to devastating plant disease, mold, and human disease. On the other hand, fungi are essential for recycling nutrients in the environment, in mycorrhizal interactions with plants, and as a source of food and pharmaceutical metabolites for humans. Understanding the role of environmental signals in fungal development is vital to increase the benefits and decrease the costs that fungi present. Despite the importance of light to fungal development, much has yet to be determined to illuminate the mechanisms fungi use to perceive and respond to light. The effects of light have been investigated in model fungal species. While spectral analyses and the morphological effects of light have been well characterized in genera such as Coprinus (a basidiomycete) or Phycomyces (a zygomycete), at the molecular level Neurospora crassa (an ascomycete) is best understood based on the functions of the white collar ( wc-1 and wc-2 ) genes in light sensing [ 1 , 2 , 3 ]. In N. crassa, blue light regulates induction of carotenoid pigment production, protoperithecia (sexual fruiting body) formation and phototropism of perithecial beaks, and circadian rhythm, all of which are abolished by mutations in wc-1 or wc-2 [ 4 ]. These two genes encode proteins with several conserved domains, including a zinc finger DNA binding domain found in both proteins [ 5 , 6 , 7 ]. The two proteins physically interact through PAS (conserved in Per, Arnt, Sim proteins) domains [ 8 , 9 , 10 ]. The WC-1 protein functions as the blue light receptor through a specialized PAS domain responsible for sensing light, oxygen, and voltage in other proteins (LOV domain), and, together with WC-2, acts as a transcription factor. The WC-1 protein interacts with a flavin chromophore [flavin adenine dinucleotide (FAD)] to function as the blue light receptor [ 11 , 12 ]. A small protein, VIVID, also perceives blue light through a LOV domain and modulates N. crassa sensitivity to light [ 13 ]. N. crassa has an additional four candidate photoactive protein homologs whose functions in photoperception remain elusive [ 14 , 15 ]. We set out to identify genes involved in the process by which light inhibits mating of the basidiomycete Cryptococcus neoformans . In nature, cryptococcal varieties are associated with bird excreta, soil, tree hollows, and even caves [ 16 , 17 ]. Thus, the light stimuli studied under laboratory conditions are highly relevant to the varying light signals the fungus experiences in the wild. C. neoformans exists as a haploid yeast with two bipolar mating types ( a and α). MAT a and MATα cells fuse to form a dikaryotic hypha that terminates in a basidium in which nuclear fusion and meiosis occur, producing four long chains of haploid basidiospores by mitosis and budding. A similar process, known as haploid or monokaryotic fruiting, can occur with only one mating partner that also gives rise to filaments that terminate in basidium-like structures and produce short spore chains. Spores have been implicated as an infectious propagule, further underscoring the importance of understanding the regulatory processes governing basidiospore production [ 18 , 19 ]. Mating and fruiting are controlled in the laboratory by stimuli such as the presence of potential mating partners (via pheromone signaling), nutrient limitation, desiccation, temperature, and light [ 19 ]. Many aspects of the transduction pathways for these signals have been elucidated, particularly in response to pheromones and nutrient limitation [ 20 ], but no components of light signaling had been reported to date for this important human pathogen. We identify here two genes required for C. neoformans responses to light, and demonstrate their role in blue light regulation of development and sensitivity to UV light, and their requirement for full virulence of this pathogen in a mammalian host. Results C. neoformans Expresses Three Candidate Photoreceptors That Could Regulate Development Mating and fruiting of C. neoformans can be variable during culturing. Previous work in our laboratory and others has endeavored to find environmental factors that lead to this variation. One factor identified was light; cultures wrapped in aluminum foil exhibited enhanced mating and haploid fruiting compared to cultures in the light [ 19 , 21 ]. Our assays used cardboard containers in which 9-cm 2 holes were excised from the lid and overlaid with aluminum foil or clear plastic wrap. Under these conditions, light inhibited both mating and haploid fruiting of C. neoformans, thereby ruling out any effects of plate-sealing on CO 2 levels or desiccation ( Figure 1 A and 1 B). Figure 1 Bwc1 Inhibits Filament Formation during C. neoformans Mating or Haploid Fruiting Filamentation assays were on V8 medium (pH 7) and conducted in the dark or under white fluorescent light. (A) Filamentation in crosses between wild type (WT), bwc1 mutant, and bwc1 + BWC1 reconstituted strains (48 h). Filament formation develops and is overgrown by yeast cells in wild-type or reconstituted strains crossed in the light. (B) Haploid fruiting filaments and blastospore production from yeast colonies incubated on filament agar (7 d). (C) Filament formation in wild-type (WT) or bwc1 / bwc1 mutant diploid strains (24 h). (D) Filamentation in crosses of serotype A strains wild type (WT), bwc1 mutant, and bwc1 + BWC1 reconstituted strains mated with a serotype D bwc1 mutant (48 h). In a candidate gene approach, the C. neoformans genome was searched for homologs of fungal genes implicated in light signaling or transduction: wc-1 and wc-2, opsin, phytochrome, cryptochrome, vivid, frequency, and photolyase . Unambiguous matches were obtained to opsin ( OPS1; GenBank AY882440), phytochrome ( PHY1; GenBank AY882439), and white collar 1 ( BWC1; GenBank AY882437), and transcription of these genes was confirmed by RT-PCR. Opsins are a class of seven-transmembrane proteins that bind retinal via a conserved lysine residue to form ion pumps or light receptors in animals and Archaea. Opsins have been identified in the genomes of a number of fungi, but as yet have no known function [ 15 , 22 , 23 , 24 ]. Phytochromes are histidine or serine/threonine kinase red/far-red light receptors identified in plants, and more recently in bacteria (reviewed in [ 25 ]). Two phytochrome homologs have been noted in the genome of N. crassa, but also have as yet no known function [ 14 ]. The predicted wc-1 homolog contains a LOV domain, two additional PAS domains, and a nuclear localization signal. In contrast to the N. crassa wc-1 gene, the C. neoformans homolog has no DNA binding domain; our designation of the gene as Basidiomycete White Collar 1 (BWC1) is meant to reflect this unique structure. BWC1 Mediates Inhibition of Mating by Light The three putative photoreceptor genes were mutated in a C. neoformans var. neoformans (serotype D) strain by replacing the coding region with the URA5 gene. Single-, double-, and triple-mutant strains were isolated following genetic crosses to test for possible redundant functions in light sensing. The abilities to mate and to haploid-fruit in the light and dark were examined by monitoring the production of filaments. There was no effect on mating or fruiting in strains with the opsin ops1 or phytochrome phy1 mutations, either alone or in combination (unpublished data). In contrast, bwc1 mutants were insensitive to light. All crosses in which both mating partners (bilateral crosses) carried the bwc1 deletion showed equivalent mating in the light and the dark, as assessed by the production of filaments after 24 h and 48 h, while in strains with a wild-type copy of BWC1, light reduced mating ( Figure 1 A). In unilateral crosses, i.e., in which a bwc1 mutant strain was crossed to wild type, only a very modest increase in filamentation in the light was observed relative to crosses between two wild-type parents. The bwc1 mutant strains exhibited more haploid fruiting in the presence of light compared to BWC1 wild-type strains, and a somewhat higher level of fruiting in the dark ( Figure 1 B). Reintroduction of a wild-type copy of BWC1 into a bwc1 mutant strain restored the inhibition of mating and fruiting by light ( Figure 1 ). Thus, of the three candidate photoreceptor genes identified, BWC1 functions in the control of mating and fruiting by light, whereas OPS1 and PHY1 do not. BWC1 Controls Cell Fusion and Filament Development Because filament formation is a qualitative rather than a quantitive phenotype, auxotrophic derivatives of BWC1 wild-type and bwc1 mutant strains were created, and cell fusion was assayed quantitatively; fusion of two auxotrophic parents ( ade2 or lys1 ) yields prototrophic dikaryotic or diploid strains that can grow on medium lacking adenine and lysine. By this assay, fusion of bwc1 mutant or wild-type strains was equivalent in the dark. However, light reduced fusion of wild-type strains 10- to 15-fold but had no impact on fusion of the bwc1 × bwc1 or bwc1 × wild-type crosses ( Figure 2 , and unpublished data). Figure 2 Bwc1 Regulates Fusion of C. neoformans Cells in Response to Blue Light (A) Auxotrophic strains that were wild type (WT) or bwc1 mutant were mated on V8 medium for 24 h and plated onto minimal medium to select for dikaryotic strains that result from cell fusion events. Light inhibits fusion in wild-type strains, and this inhibition is absent in bwc1 mutant strains. (B) Fusion efficiency of strains under different wavelengths of light. Fusion is reduced by white or blue light. Matings were between wild-type (+) parents, bwc1 mutant (Δ) parents, or one wild-type and one mutant parent ( bwc1 α × WT a , or WT α × bwc1 a ). Mutation of bwc1 in either or both mating partners relieves inhibition of fusion by white or blue light. Bars indicate the standard error of the mean of three replicates. Light also inhibited self-filamentous growth of a MAT a /MATα diploid strain. Stable diploid strains were selected after cell fusion by incubation of strains at 37 °C. Filamentation in the light and dark was examined using diploid wild-type or bwc1 mutant heterozygous or homozygous strains. The bwc1 / bwc1 mutant strain filamented equally well in the light and the dark, whereas filament development was reduced in the wild-type diploid strain in the light (see Figure 1 C). The filamentation of strains heterozygous at the BWC1 locus was inhibited by light, indicating that the bwc1 mutation is recessive (unpublished data). These results demonstrate that BWC1 functions in light responses at both the initial cell fusion step and during subsequent filament formation. Blue Light Inhibition of Cell Fusion Requires BWC1 To determine the approximate wavelengths of light that inhibit mating of C. neoformans, cell fusion was assayed during growth under colored filters at 24 and 48 h. In four independent experiments, blue light was sufficient to inhibit cell fusion, whereas green or red wavelengths had little or no effect (see Figure 2 B for a representative experiment). In all crosses, only one mating parent required a bwc1 mutant allele to escape the inhibition of cell fusion by light. Thus, C. neoformans Bwc1 functions similarly to the N. crassa WC-1 protein in response to blue light. BWC1 Is Required for UV Resistance To search further for a function of OPS1, PHY1, and BWC1, growth of the single- and multiple-mutant strains was examined under a variety of in vitro conditions. The mutations had no effect on previously identified attributes required for virulence in mammalian hosts, such as the production of the pigment melanin or the polysaccharide capsule, or growth at 37 °C (unpublished data). The ops1 and phy1 mutant strains were as sensitive to UV light as wild type, but the bwc1 mutants were markedly hypersensitive to UV light ( Figure 3 ). Reintroduction of a wild-type copy of the BWC1 gene into the bwc1 mutant strain restored a wild-type level of sensitivity to UV light. Based on studies in other organisms, one target gene could be that encoding photolyase. However, there is no evidence for a photolyase in C. neoformans, based on the lack of photoreactivation and the absence of a homolog in genome databases (unpublished data). We conclude that Bwc1 functions in response to blue light to inhibit mating, and to UV light to regulate resistance to UV irradiation. Figure 3 bwc1 Mutants Are Hypersensitive to UV Light Ten-fold serial dilutions of log-phase yeast cells of bwc1 mutant or wild type (WT) were plated in duplicate on YPD medium, and one plate was UV irradiated (~48 mJ/cm 2 ). Reintroduction of a wild-type copy of the BWC1 gene into the bwc1 + BWC1 mutant strain restores UV sensitivity to the wild-type level. BWC1 Function Is Conserved between Two Varieties of C. neoformans C. neoformans is a species complex of three divergent varieties or sibling species. C. neoformans var. neoformans (serotype D), utilized in the experiments described thus far, is commonly studied because of the ready availability of established congenic mating partners. However, this variety is uncommon in clinical settings (representing only 5% of cases), and we therefore re-isolated the bwc1, ops1, and phy1 mutations in the most common pathogenic type, C. neoformans var. grubii (serotype A), in which congenic strains have only recently been developed [ 26 ]. Mating of serotype A laboratory strains is less efficient than that of the congenic serotype D strains, and their growth in the light is limited on the V8 (pH 5) medium used for genetic crosses. Serotype A bwc1 bilateral crosses mated better in the dark than wild-type strains. Crosses performed in the light rarely resulted in filaments and were observed to do so only in the bwc1 × bwc1 mutant bilateral crosses (unpublished data). Using V8 (pH 7) medium and the serotype D bwc1 strains as mating type tester strains, the effects of light on mating efficiency of serotype A could be more readily established. When crossed to serotype D bwc1 strains, wild-type serotype A strains yielded fewer filaments than the bwc1 mutant strains in both the light and dark, demonstrating a role for suppression of filament formation by BWC1 under both conditions (see Figure 1 D). The serotype A bwc1 strains were also found to be hypersensitive to UV light (unpublished data). Reintroduction of the serotype A BWC1 gene complemented the mutant phenotypes of both the serotype A and D bwc1 mutants ( Figures 1 and 3 ). In summary, BWC1 function is conserved in two divergent cryptococcal varieties, and data derived from experiments on laboratory strains are also of significance to clinical isolates. Insertional Mutagenesis of a Novel Self-Filamentous Haploid Strain Identifies Other Components Required for Light Responses In contrast to the N. crassa WC-1 protein, the predicted C. neoformans Bwc1 protein has no zinc finger DNA binding domain or any other known DNA binding motif. Matches to wc-1 were obtained from other fungi and the predicted proteins examined for these domains ( Figure 4 ). The proteins share a similar structure with regard to the PAS domains, and all of the ascomycete wc-1 genes examined contained a zinc finger DNA binding domain, whereas none of the basidiomycete wc-1 homologs encode products with this domain. This suggests that the structural differences between the homologs are conserved in each phylum and are not unique to C. neoformans . Figure 4 Ascomycete White Collar 1 Homologs Contain a Zinc Finger Domain; The Basidiomycete Homologs Do Not Comparison of the structure of the predicted WC-1 proteins from the ascomycetes N. crassa (Nc), Aspergillus nidulans (An), Magnaporthe grisea (Mg), Fusarium graminarium (Fg), and Tubor borchii (Tb), and the basidiomycetes C. neoformans (Cn), Coprinus cinereus (Cc), Ustilago maydis (Um), and Phanerochaete chrysosporium (Pc). Other domains are PAS ( PER , ARNT , SIM ) and NLS (nuclear localization signal), and the specialized PAS domain that interacts with the chromophore FAD is marked. Sequences are from GenBank (Nc, X94300; An, AF515628; Tb, AJ575416), the Broad Institute (Mg, Cn, Fg, Cc, Um), or the Department of Energy (Pc). We hypothesized that Bwc1 binds to an interacting DNA binding protein, because Bwc1 would be unable to act as a transcription factor on its own. Systematic deletion of all of the C. neoformans transcription factors, assuming these were annotated, is not technically feasible at this stage. Similarly, standard insertional mutagenesis poses a problem because the filamentation phenotype requires that both the MAT a and MATα mating partners bear the same mutation. However, overexpression of a mating type-specific homeodomain protein in a haploid strain of the opposite mating type confers a self-filamentous morphology [ 27 ]. We reasoned that such a self-filametous strain could be employed to perform random insertional mutagenesis, and devised a screen to identify the hypothetical protein interacting with Bwc1. The SXI1α gene, which encodes the MATα-specific homeodomain protein [ 27 ], was introduced into the genome of a MAT a haploid strain. The resulting MAT a + SXI1α strain (AI49) exhibited self-filamentous growth that was regulated by temperature, light, and nutrients. The strain grew as a budding yeast at 37 °C and filamented at 25 °C, and, like MAT a /MATα diploids, filamentation was inhibited by light, and was most robust on V8 mating medium ( Figure 5 A). Figure 5 The BWC2 Gene also Mediates UV/Blue Light Responses in C. neoformans (A) A self-filamentous haploid strain (MAT a + SXI1α ) exhibits light-repressed filamentation. This strain was mutated by Agrobacterium -mediated T-DNA insertion, and insertional mutant strain 25F8 filaments equally well in the light and the dark. (B) Comparison of the structures of Bwc1 and Bwc2. The Bwc1 predicted protein (1,097 amino acids) has a LOV domain, two additional PAS domains and a nuclear localization signal. The Bwc2 predicted protein (392 amino acids) has a PAS domain and zinc finger DNA binding domain. (C) Bilateral mating between wild-type (WT), bwcl, bwc2, or bwc1 bwc2 double (bwc1,2) mutant strains on V8 medium in the dark and the light (48 h). Filamentation is repressed in wild-type crosses in the light, but not in crosses between mutants or in the dark. (D) Fusion efficiency of strains under different wavelengths of light. Matings were between wild-type (+) partners, bwc2 mutant (Δ) partners, or one wild-type and one mutant partner ( bwc2 α × WT a , or WT α × bwc2 a ). Mutation of bwc2 in either or both mating partners relieves inhibition of fusion by white or blue light. Bars indicate the standard error of the mean of three replicates. (E) Filament formation in wild-type or bwc2 / bwc2 mutant diploid strains (24 h). Light does not repress filament formation in bwc2 / bwc2 diploids. (F) The bwc2 and double bwc1 bwc2 ( bwc1,2 ) mutants are as hypersensitive to UV light as bwc1 mutants. Ten-fold serial dilutions of yeast cells were plated in duplicate onto YPD medium, and one set was UV irradiated (~48 mJ/cm 2 ). The self-filamentous MAT a + SXI1 α haploid strain was mutated with transfer DNA (T-DNA) containing a nourseothricin resistance cassette using the trans-kingdom DNA delivery vehicle Agrobacterium tumefaciens . Then 2,715 individual mutant strains were isolated into 96-well microtiter plates and were analyzed with a stereomicroscope to examine filament formation after 24 and 48 h of growth in the dark and light on V8 agar medium. Three strains were isolated from this mutant library with a phenotype analogous to the bwc1 mutant in that they filamented equally well in the light and dark and were UV-hypersensitive. The DNA regions flanking the T-DNA insertion were obtained by inverse PCR and compared to the C. neoformans genome database. One insertion (strain 1B4) is in a predicted gene with no database similarities (GenBank EAL21986). This gene was also identified in an independent insertion mutant with a different phenotype, and was therefore not analyzed further. The second isolate (28H3) bears an insertion in the promoter of the RUM1 a gene, which is located in the mating type locus of C. neoformans [ 28 ]. Intriguingly, the Ustilago maydis RUM1 homolog regulates transcription of the bE and bW homeodomain proteins, as well as of genes regulated by the bE/bW homeodomain complex [ 29 ]. The third isolate (25F8) contains an insertion in the promoter of a gene, designated BWC2 (for a consistent nomenclature with respect to BWC1; GenBank AY882438), which encodes a predicted protein with a PAS and a zinc finger DNA binding domain ( Figure 5 B). Importantly, the predicted structure of the Bwc2 protein is strikingly similar to that of the N. crassa WC-2 protein, which does not perceive photons directly but instead interacts physically with the light sensor WC-1 and acts as a transcription factor. We chose to examine this gene further because its structure suggested that it might physically and functionally interact with the C. neoformans Bwc1 protein. Disruption of BWC2 Results in the Same Phenotype as bwc1 Mutation A bwc2 mutant allele was isolated in a wild-type background by replacing the coding region with the nourseothricin resistance gene in a serotype D strain. MAT a bwc2 single- and bwc1 bwc2 double-mutant strains were isolated following genetic crosses. In bilateral crosses, the bwc2 mutants exhibit enhanced mating in the light, whereas wild-type mating is repressed ( Figure 5 C). As in the case of bwc1 mutants, the inhibition of cell fusion, and also of filament formation after fusion, were no longer repressed by light in bwc2 mutants ( Figure 5 D and 5 E). In addition, the bwc2 mutants were also hypersensitive to UV light ( Figure 5 F). The blind and UV-hypersensitive phenotypes of the bwc1 bwc2 double mutants are comparable to those of the bwc1 and bwc2 single-mutant strains. When a wild-type copy of the BWC2 gene was reintroduced into the bwc2 mutant strain, UV sensitivity and inhibition of mating by light were restored to the wild-type level (unpublished data). A serotype A mutant of bwc2 was also isolated, and exhibited phenotypes similar to the serotype A bwc1 mutant: enhanced mating with the serotype D bwc2 tester strain in the light and UV sensitivity (unpublished data). Thus, the bwc1, bwc2, and bwc1 bwc2 mutant strains all exhibit similar phenotypes, and the double-mutant phenotype is no more severe than that of the single mutants, supporting the hypothesis that the products of the two genes function in a common pathway. Bwc1 and Bwc2 Interact in the Yeast Two-Hybrid System A yeast two-hybrid analysis was conducted to test whether Bwc1 and Bwc2 physically interact. cDNA clones were fused to the S. cerevisiae Gal4 transcription factor activation (AD) or DNA binding (BD) domains and expressed in a S. cerevisiae strain in which the GAL promoter regulates ADE2, HIS3, and lacZ reporter genes. In S. cerevisiae strains expressing AD-Bwc1 and BD-Bwc2, or AD-Bwc2 and BD-Bwc1, the ADE2, HIS3, and lacZ reporter genes were all induced and the cells grew in the absence of adenine or histidine and expressed increased levels of β-galactosidase activity ( Figure 6 ). In contrast, S. cerevisiae strains containing single Gal4-Bwc1/2 fusions and the corresponding Gal4 domain did not. These observations indicate that Bwc1 and Bwc2 can interact with one another in vivo. There was no evidence for homodimer formation for either Bwc1 or Bwc2, and no effects of light on the reporter gene-dependent growth of the strains were observed. Attempts to demonstrate Bwc1-Bwc2 interaction in C. neoformans itself via coimmunoprecipitation of epitope-tagged forms of Bwc1 and Bwc2 have been unsuccessful so far, due to the low abundance of the proteins, cross-reactivity of the antibodies with endogenous fungal proteins, and loss-of-function of tagged proteins in strains overexpressing these proteins (unpublished data). These findings demonstrate that Bwc1 and Bwc2 can physically interact when expressed in the nucleus of another fungal species, and that they can do so in a light-independent manner. Figure 6 Bwc1 and Bwc2 Physically Interact The coding regions of the BWC1 and BWC2 genes were fused adjacent to the AD or BD of S. cerevisiae GAL4 . Plasmids were cotransformed into a S. cerevisiae strain in which Gal4 regulates the ADE2, HIS3, and lacZ genes. Growth of strains in the absence of adenine (−ade) or histidine (−his) and increased β-galactosidase activity (β-gal, ± one standard error, Miller units) indicate protein-protein interactions. Transcript Levels of BWC2 Are Regulated by BWC1 and Light In N. crassa, light regulates transcript levels of wc-1 but not wc-2 . Transcription of BWC1 and BWC2 was assayed in the light and dark on V8 solid medium ( Figure 7 A). The levels of transcript, particularly of BWC1, were very low, and therefore samples were enriched approximately 20-fold by purifying mRNA from total RNA for Northern blot analysis. BWC1 transcript levels were constant under these conditions. In contrast, BWC2 is up-regulated in the presence of light, except in the bwc1 mutant, demonstrating that BWC2 is a light-regulated gene and dependent on the presence of BWC1 . Thus, interestingly, the pattern of light induction of transcript levels is reversed between the two genes in C. neoformans compared to N. crassa . Figure 7 Transcript Analysis of BWC1 and BWC2, and Their Effects on Genes Required for Mating (A) Transcript levels of BWC2 are regulated by light, dependent on BWC1 . Wild-type, bwc1 (1 Δ ) or bwc2 (2 Δ ) strains were inoculated onto V8 agar medium and wrapped in foil. A set of plates was removed from darkness 1 h, 4 h, and 8 h before the end of a 24-h period. Messenger RNA purified from 200 μg of total RNA isolated from these cultures was separated on an agarose gel, transferred to nitrocellulose, and probed with the BWC1 , BWC2, and actin (ACT1) genes. No transcripts of BWC1 or BWC2 are observed in the bwc1 or bwc2 strains, respectively, consistent with the deletion strategy to create these strains. The transcript levels of BWC1 are constant under these conditions. In contrast, BWC2 transcript levels increase in the light, but not in strains bearing the bwc1 mutation. (B) Bwc1 and Bwc2 regulate transcript levels of pheromone MFα1 and homeodomain SXI1α genes. Crosses between wild-type (WT), bwc1, or bwc2 mutant strains were conducted on V8 pH 7 medium, incubated in the light (L) or dark (D), and cells were harvested 24 h later. RNA was size-fractionated in agarose gels and blotted to nitrocellulose, and probed to detect pheromone (MFα1) or homeodomain (SXI1α) transcription, as well as actin (ACT1) as a control for RNA loading and transfer. Transcript Levels of Genes Required for Mating Are Regulated by BWC1 and BWC2 The mating phenotype of bwc1 mutants suggested that Bwc1-Bwc2 regulates gene expression during mating. Transcript abundance of the pheromone gene MFα1 and the homeodomain gene SXI1α, both of which are required for efficient mating and are known to be induced during mating and following cell fusion [ 21 , 27 ], was examined by Northern blot analysis of mating cultures grown in the light and dark for 24 h ( Figure 7 B). In crosses with bwc1 and bwc2 mutants, transcript levels were consistently high in both the light and the dark. In contrast, in crosses with wild-type parents the transcript levels of MFα1 and SXI1α were reduced in the light compared to the dark ( Figure 7 B). These data suggest that Bwc1-Bwc2 function, directly or indirectly, to repress transcription of these two key genes that regulate mating and completion of the sexual cycle. BWC1 and BWC2 Regulate Virulence in Mammals C. neoformans is a pathogenic fungus that causes disease in humans and other animals. The wild-type, bwc1, and bwc2 mutants, as well as the bwc1 + BWC1 and bwc2 + BWC2 complemented strains, were inoculated by inhalation into ten mice each, and host fitness and survival were examined daily ( Figure 8 ). Animals infected with the wild-type or the bwc1 + BWC1 or bwc2 + BWC2 strains all died within 30 d of inoculation (average survival = 20.5 d, 24.4 d, and 21.5 d, respectively). In contrast, the mice infected with the bwc1 or bwc2 mutant strains were all healthy at 30 d after inoculation, and the first animal in these two groups that became moribund did so at day 37 (average survival = 43.2 and 45.1 d for bwc1 and bwc2 mutants, respectively). Bwc1 and Bwc2 are therefore not essential for virulence, but do contribute to the rate with which the fungus causes disease in the mammalian host. Thus, in addition to regulating development, Bwc1-Bwc2 also promote virulence. Figure 8 BWC1 and BWC2 Are Required for Full Virulence of C. neoformans in a Mammalian Host Ten mice each were infected intranasally with 1 × 10 5 cells of wild-type, bwc1 mutant, and bwc2 mutant, and reconstituted ( bwc1 + BWC1 ; bwc2 + BWC2 ) serotype A strains, and survival monitored daily. Mice infected with the wild-type and complementing strains progress to severe morbidity at the same rate, whereas mice infected with the bwc1 or bwc2 mutant strains survived twice as long. Discussion Light inhibits both mating and a related differentiation process known as haploid fruiting in C. neoformans . Two approaches were employed to identify genes regulating these responses to light. First, we examined the genome of C. neoformans to identify homologs of genes involved in light perception in other organisms. Second, we designed a novel strategy to identify genes regulating sexual differentiation, and used a self-filamentous haploid strain in an insertional mutagenesis screen to define novel genes with roles in light responses. Opsin, phytochrome, and white collar 1 homologs were found in the C. neoformans genome, and the function of these candidate photoreceptors was examined by gene disruption. No phenotypes were conferred by the ops1 or phy1 mutations, but deletion of the wc-1 homolog BWC1 abolished the inhibition of mating and haploid fruiting by light. The bwc1 mutant phenotypes in the clinical background were generally equivalent to those observed in serotype D; however, in the serotype A crosses with the bwc1 mutants, it was clear that mating inhibition occurred with a wild-type copy of BWC1 regardless of the light status. In serotype D, inhibition of mating by light was shown to occur at both the cell fusion and the hyphal developmental stages. Cell fusion assays revealed that only one parent requires a bwc1 mutation to circumvent repression by light, and the release from light repression in fusion is equivalent between unilateral ( bwc1 × wild type) and bilateral ( bwc1 × bwc1 ) crosses. This observation suggests that during mating only one cell, independent of mating type, needs to commit to fusion. Because a wild-type level of fusion was observed in unilateral crosses, rather than an expected 50% reduction, there must also be cross-talk between the two cells prior to fusion, which is probably mediated via pheromone sensing. Analysis of diploid strains showed that once cell fusion has occurred, the wild-type Bwc1 allele of the protein has sufficient activity, even in the heterozygous state, to repress filament formation in the presence of light to a level equivalent to that observed in wild-type diploid strains. In an assay for the wavelength responsible for inhibition of cell fusion, blue light (rather than green or red wavelengths) was found to reduce fusion efficiency between strains with an intact copy of BWC1 . No inhibition by white or blue light was observed during fusion of bwc1 mutant strains. These data lead us to hypothesize that Bwc1 functions as a blue light photoreceptor, as is the case for N. crassa WC-1 [ 11 , 12 ]. To test this hypothesis, we initiated efforts to analyze the photochemistry of Bwc1. However, recombinant Bwc1 or fragments of Bwc1 expressed in E. coli cells were either produced in low quantities or were largely insoluble (unpublished data), and thus formal demonstration of photoreceptor function for Bwc1 remains to be established. The bwc1 mutants were also hypersensitive to UV light, showing that the Bwc1 protein functions in response to both blue (approximately 400–500 nm) and UV light (approximately 200–400 nm) wavelengths. The ability of blue or UV light to induce carotenoid formation in N. crassa was first noted a century ago [ 30 ]. Subsequent work has focused on light in the blue wavelengths, which is sensible given that any study on UV light and its regulation of fungal development is likely to be complicated by the effects this radiation has on cell viability and media stability. Nevertheless, there is evidence that N. crassa also perceives UV light through WC-1. Prior to cloning the wc-1 gene, the spectra for inhibition of circadian rhythm and for photoinduction of carotenoid production were found to lie in both the UV and blue wavelengths [ 31 , 32 ]. Light treatment of N. crassa changes the light absorbance properties of mycelia, and the action spectrum of this response is within both the UV and blue wavelengths and closely overlaps that of flavins, with respective peaks at 360 and 470 nm [ 33 ]. In particular, the action spectra from physiological data overlap with the properties of the WC-1 protein purified from N. crassa cells, as WC-1-FAD or the chromophore FAD alone show two equal excitation peaks, one at 370 nm (UV) and one at 450 nm (blue) [ 11 , 12 ]. These data suggest that N. crassa WC-1 may also be a UV-responsive protein and function like C. neoformans in protecting the fungus from UV damage. The induction of UV-protecting carotenoids in N. crassa by light in a WC-1-dependent manner supports this hypothesis. Nevertheless, a UV-sensing function for the White collar proteins remains to be demonstrated through analysis of protein photochemistry and spectral and phenotypic analysis. The predicted Bwc1 protein lacks a DNA binding domain found in the ascomycete WC-1 homologs. We hypothesized that there must be a second protein that interacts with Bwc1, and set out to identify this component via random insertional mutagenesis. To create a haploid self-filamentous strain of C. neoformans , we expressed the MAT-specific Sxi1α homeodomain protein in a MAT a haploid cell, resulting in robust induction of filament development. The self-filamentous strain was mutated with T-DNA insertions from Agrobacterium, and three mutant C. neoformans strains with equivalent filament formation in the light and the dark, and UV hypersensitivity, were isolated. In one strain, the T-DNA insertion lies in the promoter of a gene we designated BWC2 , which has an analogous structure to the N. crassa wc-2 gene (a PAS domain and zinc finger DNA binding domain) but shares much less sequence similarity relative to that between C. neoformans BWC1 and N. crassa wc-1 . The BWC2 gene was not found in the initial candidate gene search because of this low sequence similarity and because the intron structure of C. neoformans confounded its identity. The BWC2 gene was mutated to analyze its function. The bwc2 and bwc1 bwc2 double mutants exhibit phenotypes comparable to bwc1 single mutants, and were nonresponsive to light during mating and haploid fruiting and hypersensitive to UV irradiation, suggesting they function in the same pathway. Furthermore, Bwc1 and Bwc2 interact in the yeast two-hybrid system, supporting a model in which the two proteins represent the integral components of a regulatory complex controlling light-regulated development. The mating type loci of basidiomycetes have been well studied, and comprise two distinct gene sets: those that encode pheromones and those that encode homeodomain proteins, both of which control different steps in mating [ 34 , 35 ]. We hypothesized that transcription of the C. neoformans pheromone or homeodomain genes would be controlled via Bwc1-Bwc2. We focused on the pheromone genes, because they are important cell-cell signaling molecules, and because mfα1,2,3 triple-mutant strains exhibit a reduction in fusion efficiency similar to that seen in bwc1 mutant strains [ 21 ]. In N. crassa , transcription of the pheromone genes is under control of the circadian clock and presumably wc-1 [ 36 ]. The mating type specific homeodomain protein Sxi1α of C. neoformans is important for events after cell fusion [ 27 ]. Examination of the transcription of MFα1 and SXI1α in the light and dark in wild type compared to bwc1 or bwc2 crosses at 24 h showed that the MFα1 and SXI1α genes are repressed by Bwc1-Bwc2 in the light. It is therefore likely that Bwc1-Bwc2 control mating by influencing the temporal regulation of these genes. The roles of the C. neoformans BWC1 and BWC2 genes in virulence were examined. We hypothesized that the fungus may be able to sense darkness within the mammalian host and use this as a signal (possibly via Bwc1-Bwc2) to induce virulence. We also tested virulence, because several signal transduction pathways affecting C. neoformans mating also have an impact on virulence. Disruption of both BWC1 and BWC2 reduced the ability of C. neoformans to cause disease, as mice infected with the bwc1 or bwc2 mutant strains survived twice as long as those infected with wild-type or control strains. The Bwc1-Bwc2 system represents a novel class of protein complex that is required for cellular responses to an environmental stimulus and affects both development (mating) and virulence in pathogenic fungi [ 20 ]. In contrast to the cAMP signaling and calcineurin pathways, where mutants have reduced mating efficiency and virulence, here the bwc1 and bwc2 mutants have enhanced mating yet reduced virulence. In bwc1 or bwc2 strains there was no reduction in those traits normally associated with C. neoformans virulence, such as production of melanin or capsule, or growth at 37 °C or on minimal media. Identification of the downstream targets for this complex should further elucidate the molecular basis for its role in both mating and virulence. We propose a model for Bwc1-Bwc2 function that is similar to that of WC-1-WC-2 of N. crassa but differs in several key features ( Figure 9 ). In this model, C. neoformans Bwc1-Bwc2 bind to DNA in the dark and act as weak repressors to reduce filament development. We hypothesize that photons perceived through a flavin cofactor cause a conformational change that enhances repression of filament formation and cell fusion, and activates transcription of genes required for UV resistance/DNA repair. It is also formally possible that light causes Bwc1-Bwc2 to increase transcription of a gene that functions to repress mating, and/or represses transcription of a repressor of UV sensitivity. The N. crassa model is similar but differs in several key features. Recent evidence suggests that a complex of two subunits of WC-1 and one subunit of WC-2 form in response to light [ 3 , 10 ]. The complex positively regulates transcription of genes required for conidiation, mating, and carotenoid production, in marked contrast to the negative regulation of mating observed in C. neoformans . Another difference is that wc-1 is light-regulated in N. crassa , while BWC2 is light-regulated in C. neoformans and BWC1 is not. The N. crassa complex also regulates transcription of frq, and the FRQ protein feedback inhibits the complex, thereby contributing to the wiring of the circadian clock. The roles for WC-1 and WC-2 in N. crassa photoperception also change during the day, adding to the challenge of elucidating their functions. Future studies in C. neoformans will define downstream targets of Bwc1-Bwc2, regulation of Bwc1 and Bwc2 and their complex, and the creation of alleles of Bwc1 bearing mutations in the predicted flavin interacting domain to elucidate the proposed roles of these proteins in light perception. Figure 9 A Model of How Two Fungi May Respond to Light The Bwc1-Bwc2 interaction of C. neoformans shares conserved features with the WC-1-WC-2 interaction of N. crassa but also exhibits unique functional characteristics. In this model, C. neoformans Bwc1-Bwc2 bind to DNA in the dark and act as weak repressors to reduce filament development. We hypothesize that photons perceived through a proposed flavin moiety on Bwc1 cause a conformational change that increases repression of filament formation and cell fusion, and activates transcription of genes required for UV resistance. Alternatively, UV sensitivity may be mediated through repression by Bwc1-Bwc2 of a repressor protein. The N. crassa model is simplified from [ 3 ]: a complex of two units of WC-1 forms in response to light to cause an initial up-regulation of frq transcription above the levels occurring in the dark (FRQ feedback inhibits the White collar complex). The complex also increases transcription of genes required for other processes. Components of the White collar sensing system have been identified in other fungi (see Figure 5 ), and are likely to function in light responses in these and other fungal species. Recently the Trichoderma atroviride homologs of N. crassa WC-1 and WC-2 were isolated and mutated, demonstrating a role for these genes in light-induced conidiation and the induction of photolyase [ 37 ]. A gene homologous to BWC1 was identified in the model basidiomycete C. cinereus as mutated in a strain defective in light-regulated development of the mushroom cap [ 38 ]. Developmental regulation in C. cinereus is blue-light dependent [ 39 , 40 ], and UV/blue light also regulate development of numerous other fungi (for review see [ 41 ]). A homolog of wc-1 was identified in the truffle-forming ascomycete Tubor bruchii, in which blue light inhibits hyphal growth [ 42 ]. Thus, this gene and its homologs may have applications even to cultivated, edible fungi. It will be of interest to establish whether White collar-like proteins are present in the genomes of the two other fungal phyla proposed for genome sequencing, the zygomycetes and the chytrids. The responses of the zygomycete Phycomyces to light, particularly blue and UV wavelengths, have been well characterized, and numerous mutants (e.g., ten different mad mutants) affecting responses or sensitivity to particular wavelengths have been isolated, but no photoreceptor genes have as yet been identified [ 2 ]. We predict that some of these known mutations will be found to affect White collar homologs. Finally, we speculate that White collar genes could be of major significance for terrestrial life. The discovery of the BWC1 and BWC2 genes as potential UV-blue light responsive proteins in a basidiomycete indicates that this type of protein complex is ancient in the fungal kingdom. The fossil record shows a clear divergence of the fungal kingdom into the four phyla by the Devonian [416–359 million years ago (mya)], and a Precambrian origin (prior to 542 mya) for the fungi has been suggested [ 43 , 44 , 45 , 46 ]. Margulis et al. proposed that sexual recombination and DNA repair were coselected in the Precambrian for protection against UV light [ 47 ], and genes are known that control both recombination and sensitivity to UV light. Bwc1-Bwc2 in C. neoformans regulates both UV sensitivity and sexual development, ultimately leading to recombination. During the Silurian division (416–444 mya), the UV-protection role of the WC-1 proteins could have conferred a major selective advantage to the fungi when they and plants cocolonized the continents at a time when there was no shade from solar radiation. The proteins could have been especially important at other times of global ecological change associated with elevated UV irradiation due to atmospheric and vegetation changes, such as at the end of the Permian (250 mya) or Cretaceous (65 mya), when there are spikes of fungi in the fossil record [ 48 , 49 ]. The UV-protecting ability of the WC-1 proteins is a likely selective force that has served to maintain their presence in fungi to this day. Materials and Methods Gene identification and fungal strains Candidate photoreceptors were identified in the C. neoformans genome projects. Gene transcription was tested using RT-PCR and rapid amplification of cDNA ends (RACE) with the GeneRacer Kit (Invitrogen, Carlsbad, California, United States). Three genes were mutated in serotype D strain JEC43 (MATα ura5 ) and serotype A strain JF99 (MAT a ura5 ). Mutations were made by biolistic introduction of disruption alleles generated by overlap PCR with 1.5-kb DNA on either side of the URA5 gene [ 50 , 51 ]. BWC2 was disrupted using a nourseothricin resistance cassette [ 52 , 53 ] to replace the gene in the serotype D strain JEC21 or the serotype A strain KN99α (both MATα). Gene disruption was confirmed by PCR and Southern blot analysis using standard methods. The mutant serotype D strains were crossed to the congenic strain JEC20 (MAT a ) to obtain strains with the opposite mating type. Through a series of crosses, strains with double or triple mutations in both mating types were isolated. A set of strains with auxotrophic markers (either lys1 or ade2 ) with the mutation or wild type at the BWC1 locus was also generated by crossing. The serotype A bwc1::URA5 strain was crossed to strain KN99α to obtain a MATα bwc1 strain. For reconstitution, the BWC1 or BWC2 genes were amplified with primers JOHE8744 and JOHE8745 or JOHE12641 and JOHE12642, respectively, from genomic DNA of strain H99, and ligated adjacent to a cassette conferring resistance to neomycin (G418). A linearized version of this vector was introduced into bwc1 or bwc2 mutant strains. The self-filamentous serotype D strain was obtained by introducing the SXI1α gene adjacent to URA5 into strain JEC34 (MAT a ura5 ). This strain was mutated with Agrobacterium -mediated integration of T-DNA containing the NAT gene [ 52 , 54 ]. Strains created and primers used in strain construction are listed in Tables S1 and S2 . Phenotypic analysis of mutant strains Strains were compared to each other and reference laboratory strains for the ability to mate in the presence of white fluorescent light (1,500–3,500 lux) or darkness on V8 medium at pH 5 (serotype A) or pH 7 (serotype A and D). The growth of strains was also examined at 37 °C on YPD medium, for melanin production on bird seed agar (70 g/L ground bird seed, 0.1% glucose, 0.05% Tween-20) or on low-glucose (0.1%) medium supplemented with the diphenolic molecule L -DOPA (100 mg/ml). Capsule production was assayed by growing strains in liquid medium with low levels of glucose (0.5%) and iron (20 mg/L of the chelator EDDHA), and examining exclusion of India ink particles from fungal cells. Strains were grown to logarithmic phase in liquid YPD medium and serial dilutions spotted onto YPD agar plates, which were then irradiated with UV light (0.2 min setting, approximately 48 mJ/cm 2 ; UV Stratalinker 2400, Stratagene, La Jolla California, United States) to test for UV sensitivity. Wavelength of inhibition and analysis of light inhibition during stages of mating Crosses between lys1 or ade2 auxotrophic strains with or without the bwc1 or bwc2 mutations were conducted under illumination modified with filters to provide blue, green, or red light (LE 4747 blue, LE 4758 green and LE 4725 red; Calumet, Bensenville Illinois, United States). Yeast cells (1 × 10 7 /ml) were inoculated in 5-μl drops onto V8 (pH 7) medium, and 24 or 48 h later the mating mix scraped from the surface, and cells were resuspended in sterile water and plated onto minimal medium (yeast nitrogen base; YNB) to select for prototrophs that result from fusion events. Stable diploid yeast strains were created by incubating the cells at 37 °C on YNB medium. Transcription analysis Strains (1.25 × 10 8 cells) were inoculated onto 15-cm diameter petri dishes containing V8 pH 7 medium, which induces mating. Cells were scraped from the surface 24 h later, frozen, and lyophilized. Total RNA was isolated from cells using TRIzol reagent (Invitrogen) according to manufacturer's instructions. Messenger RNA was isolated from 200 μg of total RNA using the PolyATtract isolation system (Promega, Madison, Wisconsin, United States). RNA was separated on denaturing agarose gels, blotted to nitrocellulose (Zeta-Probe, Bio-Rad, Hercules California, United States), and probed with [ 32 P]-dCTP-radiolabeled DNA fragments. Probes comprising ACT1 (encoding actin) , BWC1, BWC2, SXI1α, and MFα1 genes were amplified from genomic DNA (primers in Table S2 ). Crosses comprised wild-type strains JEC20 x JEC21, bwc1 mutant strains AI5 (MATα) × AI6 (MAT a ) , or bwc2 mutant strains AI76 (MATα) × AI78 (MAT a ) , and were maintained in constant light or dark. For analysis of BWC1 and BWC2 transcription, cultures of wild-type, bwc1, or bwc2 (strains JEC21, AI5, and AI76, respectively) were wrapped in aluminum foil and exposed to light 0 h, 1 h, 4 h, or 8 h prior to the end of a 24-h incubation. Yeast two-hybrid system cDNAs of BWC1 and BWC2 were amplified either by overlap PCR from genomic DNA or from RT-PCR from RNA, and sequenced to identify clones without errors. Products were cloned into plasmids pGBD.c1 and pGAD.c1, and the S. cerevisiae reporter strain PJ69–4A was cotransformed with plasmids using the lithium acetate/heat shock method [ 55 ]. Double transformants were selected on media lacking leucine and tryptophan. Interactions were assessed by growth in the absence of adenine or histidine (+ 5 mM 3-aminotriazole) and β-galactosidase assays [ 56 ]. C. neoformans virulence assay For murine killing assays, serotype A C. neoformans cells were used to infect 25-g female A/Jcr mice (NCI/Charles River Laboratories, Frederick, Maryland, United States) by nasal inhalation [ 57 ]. Ten mice were inoculated each with a 50-μl drop containing 1 × 10 5 yeast cells of KN99α, bwc1, bwc1 + BWC1 reconstituted, bwc2 and bwc2 + BWC2 reconstituted strains. Survival data were analyzed with a logrank test to determine statistical significance. The murine experiment protocol was approved by the Duke University Animal Use Committee. Supporting Information Table S1 Cryptococcus neoformans strains Mutant alleles were created by replacing the coding region of genes with markers to complement uracil auxotrophy (URA5) or confer resistance to nourseothricin (NAT) . Mutations were complemented by reintroduction of wild-type copies of the gene fused to a gene conferring resistance to neomycin (NEO) . (58 KB DOC). Click here for additional data file. Table S2 Oligonucleotides Primers used to create gene disruption alleles, probes, and clones for yeast two-hybrid assays. (50 KB DOC). Click here for additional data file. Accession Numbers The GenBank ( http://www.ncbi.nlm.nih.gov/ ) accession numbers of the genes discussed in this paper are Aspergillus nidulans wc-1 (AF515628), BWC1 (AY882437), BWC1 (AY882438), N. crassa wc-1 (X94300), OPS1 (AY882440), PHY1 (AY882439), and T. borchii Tbwc-1 (encodes wc-1 protein) (AJ575416). The Broad Institute ( http://www.broad.mit.edu/annotation/fungi/fgi/ ) has sequence for White collar 1 homologs from Coprinus cinereus, Fusarium graminearum , Magnaporthe grisea , and Ustilago maydis . The Department of Energy ( http://genome.jgi-psf.org/whiterot1/whiterot1.home.html ) has the sequence for the White collar 1 homolog of Phanerochaete chrysosporium .

D:\keerthana\PMC001xxxxxx\PMC1064852.xml

1064853

Two Distinct E3 Ubiquitin Ligases Have Complementary Functions in the Regulation of Delta and Serrate Signaling in Drosophila

Signaling by the Notch ligands Delta (Dl) and Serrate (Ser) regulates a wide variety of essential cell-fate decisions during animal development. Two distinct E3 ubiquitin ligases, Neuralized (Neur) and Mind bomb (Mib), have been shown to regulate Dl signaling in Drosophila melanogaster and Danio rerio, respectively. While the neur and mib genes are evolutionarily conserved, their respective roles in the context of a single organism have not yet been examined. We show here that the Drosophila mind bomb (D-mib) gene regulates a subset of Notch signaling events, including wing margin specification, leg segmentation, and vein determination, that are distinct from those events requiring neur activity. D-mib also modulates lateral inhibition, a neur - and Dl -dependent signaling event, suggesting that D-mib regulates Dl signaling. During wing development, expression of D-mib in dorsal cells appears to be necessary and sufficient for wing margin specification, indicating that D-mib also regulates Ser signaling. Moreover, the activity of the D-mib gene is required for the endocytosis of Ser in wing imaginal disc cells. Finally, ectopic expression of neur in D-mib mutant larvae rescues the wing D-mib phenotype, indicating that Neur can compensate for the lack of D-mib activity. We conclude that D-mib and Neur are two structurally distinct proteins that have similar molecular activities but distinct developmental functions in Drosophila .

Introduction Cell-to-cell signaling mediated by receptors of the Notch (N) family has been implicated in various developmental decisions in organisms ranging from nematodes to mammals [ 1 ]. N is well-known for its role in lateral inhibition, a key patterning process that organizes the regular spacing of distinct cell types within groups of equipotent cells. Additionally, N mediates inductive signaling between cells with distinct identities. In both signaling events, N signals via a conserved mechanism that involves the cleavage and release from the membrane of the N intracellular domain that acts as a transcriptional co-activator for DNA-binding proteins of the CBF1/Suppressor of Hairless/Lag-2 (CSL) family [ 2 ]. Two transmembrane ligands of N are known in Drosophila, Delta (Dl) and Serrate (Ser) [ 3 ]. Dl and Ser have distinct functions. For instance, Dl (but not Ser ) is essential for lateral inhibition during early neurogenesis in the embryo [ 4 ]. Conversely, Ser (but not Dl ) is specifically required for segmental patterning [ 5 ]. Some developmental decisions, however, require the activity of both genes: Dl and Ser are both required for the specification of wing margin cells during imaginal development [ 6 , 7 , 8 , 9 , 10 ]. These different requirements for Dl and Ser appear to primarily result from their non-overlapping expression patterns rather than from distinct signaling properties. Consistent with this interpretation, Dl and Ser have been proposed to act redundantly in the sensory bristle lineage where they are co-expressed ([ 11 ]; note, however, that results from another study have indicated a non-redundant function for Dl in the bristle lineage [ 12 ]). Furthermore, Dl and Ser appear to be partially interchangeable because the forced expression of Ser can partially rescue the Dl neurogenic phenotype [ 13 ]. Additionally, the ectopic expression of Dl can partially rescue the Ser wing phenotype [ 14 ]. The notion that Dl and Ser have similar signaling properties has, however, recently been challenged by the observation that human homologs of Dl and Ser have distinct instructive signaling activity [ 15 ]. Endocytosis has recently emerged as a key mechanism regulating the signaling activity of Dl. First, clonal analysis in Drosophila has suggested that dynamin-dependent endocytosis is required not only in signal-receiving cells but also in signal-sending cells to promote N activation [ 16 ]. Second, mutant Dl proteins that are endocytosis defective exhibit reduced signaling activity [ 17 ]. Third, two distinct E3 ubiquitin ligases, Neuralized (Neur) and Mind bomb (Mib), have recently been shown to regulate Dl endocytosis and N activation in Drosophila and Danio rerio, respectively [ 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ]. Ubiquitin is a 76-amino-acid polypeptide that is covalently linked to substrates in a multi-step process that involves a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin–protein ligase (E3). E3s recognize specific substrates and catalyze the transfer of ubiquitin to the protein substrate. Ubiquitin was first identified as a tag for proteins destined for degradation. More recently, ubiquitin has also been shown to serve as a signal for endocytosis [ 26 , 27 ]. Mib in D. rerio and Neur in Drosophila and Xenopus have been shown to associate with Dl, regulate Dl ubiquitination, and promote its endocytosis [ 18 , 19 , 20 , 22 , 25 , 28 ]. Moreover, genetic and transplantation studies have indicated that both Neur and Mib act in a non-autonomous manner [ 18 , 21 , 22 , 23 , 25 , 29 ], indicating that endocytosis of Dl is associated with increased Dl signaling activity. Finally, epsin, a regulator of endocytosis that contains a ubiquitin-interacting motif and that is known in Drosophila as Liquid facet, is essential for Dl signaling [ 30 , 31 ]. In one study, Liquid facet was proposed to target Dl to an endocytic recycling compartment, suggesting that recycling of Dl may be required for signaling. Accordingly, signaling would not be linked directly to endocytosis, but endocytosis would be prerequisite for signaling [ 30 ]. How endocytosis of Dl leads to the activation of N remains to be elucidated. Also, whether the signaling activity of Ser is similarly regulated by endocytosis is not known. Neur and Mib proteins completely differ in primary structure. Drosophila Neur is a 754-amino-acid protein that contains two conserved Neur homology repeats of unknown function and one C-terminal catalytic really interesting new gene (RING) domain. D. rerio Mib (also known as DIP-1 in the mouse [ 32 ]) is a 1,030-amino-acid protein with one ZZ zinc finger domain surrounded by two Mib/HERC2 domains, two Mib repeats, eight ankyrin repeats, two atypical RING domains, and one C-terminal catalytic RING domain. Both genes have been conserved from flies to mammals [ 18 , 19 , 33 , 34 ]. While genetic analysis has revealed that neur in Drosophila and mib in D. rerio are strictly required for N signaling, knockout studies of mouse Neur1 has indicated that NEUR1 is not strictly required for N signaling [ 33 , 34 ]. One possible explanation is functional redundancy with the mouse Neur2 gene. Conversely, the function of Drosophila mib (D-mib), the homolog of D. rerio mib gene, is not known. To establish the respective roles of these two distinct E3 ligases in the context of a single model organism, we have studied the function of the Drosophila D-mib gene. We report here that D-mib, like D. rerio Mib, appears to regulate Dl signaling during leg segmentation, wing vein formation, and lateral inhibition in the adult notum. We further show that D-mib is specifically required for Ser endocytosis and signaling during wing development, indicating for the first time, to our knowledge, that endocytosis regulates Ser signaling. Interestingly, the D-mib activity was found necessary for a subset of N signaling events that are distinct from those requiring the activity of the neur gene. Nevertheless, the ectopic expression of Neur compensates for the loss of D-mib activity in the wing, indicating that Neur and D-mib have overlapping functions. We conclude that D-mib and Neur are two structurally distinct proteins with similar molecular activities but distinct and complementary functions in Drosophila . Results Isolation of D-mib Mutations The closest Drosophila homolog of the vertebrate mib gene is the predicted gene CG5841, D-mib [ 18 ]. The D-mib mutations identified are shown in Figure 1 . A P-element inserted into the 5′ untranslated region of the D-mib gene was recently isolated ( http://flypush.imgen.bcm.tmc.edu/pscreen/ ) ( Figure 1 A). Insertion of this P-element confers late pupal lethality. Lethality was reverted by precise excision of the P-element, suggesting that insertion of this P-element is a D-mib mutation, referred to as D-mib 1 . A 13.6-kb deletion that removes the entire D-mib coding region was selected by imprecise excision of this P-element. This deletion represents a null allele of D-mib and was named D-mib 2 . This deletion also deletes the 3′ flanking RpS31 gene ( Figure 1 A). The D-mib 1 and D-mib 2 mutant alleles did not complement the l(3)72Cda J12 and l(3)72Cda I5 lethal mutations that have been mapped to the same cytological interval as the D-mib gene [ 35 ]. This indicates that these two lethal mutations are D-mib mutant alleles, and they were therefore renamed D-mib 3 and D-mib 4 , respectively. The D-mib 1 and D-mib 3 mutations behave as genetic null alleles (see Materials and Methods ). In contrast, D-mib 4 is a partial loss-of-function allele because flies trans -heterozygous for D-mib 4 and any other D-mib null alleles are viable. Figure 1 Molecular and Genetic Characterization of D-mib Mutations (A) Molecular map of the D-mib locus showing the position of the P-element inserted into the 5′ untranslated region (allele D-mib 1 ) and the 13.6 kb deletion that removes the D-mib and the RpS31 genes (allele D-mib 2 ). Transcribed regions are indicated with arrows, and exons are indicated with boxes. Open reading frames are shown in black. (B) Domain composition of D-mib and D. rerio Mib. Both proteins show identical domain organization. D-mib has an N-terminal ZZ zinc finger flanked on either side by a Mib/HERC2 (M-H) domain, followed by two Mib repeats, six ankyrin repeats, two atypical RING domains, and a C-terminal protypical RING that has been associated with catalytic E3 ubiquitin ligase activity. The D-mib 3 mutant allele is predicted to produce a truncated protein devoid of E3 ubiquitin ligase activity whereas the D-mib 4 protein carries a mutation at a conserved position in the second Mib repeat. (C and C′) Western blot analysis of D-mib (C). The endogenous D-mib protein (predicted size: 130 kDa) was detected in S2 cells (lane 2) and in imaginal discs from wild-type larvae (lane 3) but was not detectable in homozygous D-mib 1 (lane 4) and D-mib 1 /D-mib 3 (lane 5) third instar larvae. The D-mib protein produced in transfected S2 cells from the cDNA used in this study (lane 1) runs exactly as endogenous D-mib (lane 2). Panel C′ shows a Red Ponceau staining of the gel with the same protein samples as in panel C. (D–H) Wings from wild-type (D), D-mib 1 (E), Ser RX82 /Ser rev6.1 (F), D-mib 2 /D-mib 4 (G), and UAS-D-mib 2 /+; D-mib 1 /D-mib 2 flies (H). D-mib (E) and Ser (F) mutant flies showed similar wing loss phenotypes. The D-mib mutant phenotype could be almost fully rescued by a leaky UAS-D-mib transgene (H). (D′) and (G′) show high magnification views of (D) and (G), respectively, to show that D-mib 2 /D-mib 4 mutant flies (G′) exhibited ectopic sensilla (arrowheads) along vein L3. (I–N) Nota (I–K) and legs (L–N) from wild-type (I and L), D-mib 1 (J and M), and Ser RX82 /Ser rev6.1 (K and N) flies. D-mib mutant flies showed a weak neurogenic phenotype (J) that was not observed in Ser mutant flies (K). Ectopic sensory organs in D-mib mutant flies developed from ectopic sensory organ precursor cells (not shown). D-mib (M) and Ser (N) mutant legs also showed distinct growth and/or elongation defects. Arrows in (J) show ectopic macrochaetes. Arrows in (L–N) indicate the joints. Ti, tibia; t1 to t5, tarsal segments 1 to 5. These four mutations identify the CG5841 gene as D-mib by the following evidence. First, lethality of homozygous D-mib 1 pupae is associated with the insertion of a P-element into the 5′ UTR of the D-mib gene. Second, genomic sequencing of the D-mib 3 allele revealed the presence of a stop codon at position 258 ( Figure 1 B). This allele is therefore predicted to produce a truncated protein devoid of the catalytic RING domain, consistent with D-mib 3 being a null allele. Genomic sequencing of the D-mib 4 allele showed that this mutation is associated with a valine-to-methionine substitution at a conserved position in the second Mib repeat ( Figure 1 B). Third, Western blot analysis showed that the D-mib protein was not detectable in imaginal disc and brain complex extracts prepared from homozygous D-mib 1 and D-mib 1 / D-mib 2 larvae ( Figure 1 C and C′). Fourth, the leaky, GAL4-independent expression of a UAS-D-mib transgene fully rescued the lethality of D-mib 1 / D-mib 2 flies (data not shown; see also Figure 1 H). Thus, our analysis identified both complete and partial D-mib loss-of-function alleles. D-mib Regulates Dl Signaling Complete loss of zygotic D-mib activity in homozygous D-mib 1 and trans -heterozygous D-mib 2 / D-mib 3 , D-mib 1 / D-mib 3 and D-mib 1 / D-mib 2 individuals led to late pupal lethality. Mutant pupae died as pharate adults showing ectopic macrochaetes, increased microchaete density on the dorsal thorax ( Figure 1 I and 1 J), short legs lacking tarsal segmentation ( Figure 1 L and 1 M), and nearly complete loss of eye and wing tissues ( Figure 1 D and 1 E). Tissue losses were associated with a dramatic reduction in size of the eye field and of the wing pouch in mutant discs of third instar larvae ( Figure 2 A– 2 E). Hypomorphic D-mib 2 / D-mib 4 mutant flies only showed ectopic sensory organs, rough eyes, small wings, and thickened veins ( Figure 1 D, 1 D′, 1 G, and 1 G′; data not shown). Figure 2 The D-mib and neur Genes Have Distinct Functions during Wing Development (A–E) Wing imaginal discs (B–E) from wild-type (B and D), D-mib 1 (C), and D-mib 1 /D-mib 2 (E) third instar larvae stained for Cut (B and C) and wg-lacZ (D and E). D-mib mutant discs showed a dramatically reduced size of the wing pouch (see diagram in [A] showing the different regions of the wing imaginal disc; V, ventral; D, dorsal), as well as a complete loss of Cut and wg-lacZ (red arrows in [B–E]) expression at the wing margin. Expression of wg-lacZ in the hinge region (arrowheads in [D] and [E]) and the accumulation of Cut in sensory cells (small arrows in [B] and [C]) and muscle precursor cells (large arrowheads in [B] and [C]) appeared to be largely unaffected). (F and F′) Expression of Cut (red) at the wing margin was not affected by the complete loss of neur activity in neur 1F65 mutant clones (indicated by the loss of the nuclear green fluorescent protein [GFP] marker, in green). Bar is 50 μm in (B–E) and 20 μm in (F and F′). All these phenotypes may result from reduced N signaling. More specifically, the bristle and leg phenotypes are likely to result from reduced signaling by Dl (and not by Ser). Indeed, a reduction in Dl-mediated lateral inhibition can result in ectopic sensory organs and increased bristle density on the body surface. In contrast, a complete loss of Ser signaling had no effect on bristle density ( Figure 1 K). Likewise, loss of Dl signaling has been shown to result in short unsegmented legs, similar to the ones seen in the absence of D-mib activity ( Figure 1 M), whereas a complete loss of Ser activity led to the formation of elongated unsegmented legs ( Figure 1 N) [ 36 , 37 , 38 ]. Finally, the vein phenotype seen in D-mib hypomorphic flies is similar to the one seen in Dl ts mutant flies [ 39 ]. Together, these observations suggest that D-mib regulates Dl signaling in several developmental contexts. Consistent with this conclusion, we have shown that D-mib binds Dl and promotes Dl signaling and that overexpression of D-mib down-regulates the accumulation of Dl at the cell surface (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). D-mib and neur Have Distinct Functions We then studied in more detail the function of D-mib during wing development. Growth of the wing pouch depends on the activity of an organizing center located at the dorsal-ventral (D-V) boundary [ 40 , 41 ]. This boundary is established in first instar larvae and is defined by the apterous expression boundary. Apterous activates the expression of the Ser and fringe genes in dorsal cells. High levels of Ser in dorsal cells activate N in trans in ventral cells and suppress N activation in cis in dorsal cells, whereas Fringe modifies N in dorsal cells such that dorsal cells located at the D-V boundary respond to Dl. Thus, composite signaling by Ser and Dl leads to symmetric N activation in margin cells located along the D-V boundary [ 8 , 9 , 42 , 43 ]. N then regulates the expression of the vestigial and wingless (wg) genes that cooperate to promote growth of the wing pouch. N also regulates expression of the cut gene in margin cells [ 44 ]. Thus, loss of N signaling results in a reduction in size of the wing pouch accompanied by the loss of cut and wg expression along the D-V boundary. A complete loss of Cut and Wg accumulation and wg-lacZ expression was observed in the central region of third instar D-mib mutant wing discs (data not shown). Thus, the D-mib wing phenotype may result from defective N inductive signaling at the D-V boundary. We conclude that the activity of the D-mib gene is required for the specification of the wing margin and, hence, growth of the wing pouch. Interestingly, wing margin formation and expression of Cut are not affected by the complete loss of neur activity ( Figure 2 F and 2 F′) [ 45 ]. Similarly, loss of neur activity had no detectable effect on leg segmentation (data not shown) and vein determination [ 45 ], two processes shown here to depend on D-mib gene activity. We therefore conclude that D-mib and neur have distinct and complementary functions in Drosophila . D-mib Co-Localizes with Dl and Ser at the Apical Cortex We next studied the subcellular localization of D-mib ( Figure 3 ). Anti-D-mib antibodies were generated that specifically detected D-mib on Western blots (see Figure 1 C) and on fixed tissues ( Figure 3 F–F″). Using these antibodies, we found that D-mib was detected in all imaginal disc cells ( Figure 3 A and 3 B). We then examined D-mib subcellular distribution in epithelial cells located along the edge of the wing discs because cross-sectional imaging affords better resolution along the apical-basal axis. D-mib co-localized with Ser, Dl, and N at the apical cortex ( Figure 3 B– 3 D′′′). Dl and Ser were also detected in large intracellular vesicles that probably correspond to multivesicular bodies in that they also stained for hepatocyte growth factor-regulated tyrosine kinase substrate [ 46 ] ( Figure 3 B– 3 C′′′; data not shown). The intracellular dots seen with the anti-D-mib antibodies were distinct from the Dl- and Ser-positive dots and appeared to result from background staining (data not shown). The reduced cytoplasmic staining seen in D-mib mutant cells ( Figure 3 F– 3 F′′) suggests that D-mib is also present in the cytoplasm. A similar localization at the apical cortex and in the cytoplasm was seen for a functional yellow fluorescent protein (YFP)::D-mib fusion protein (see Figure 6 below). These localization data suggest that D-mib may act at the apical cortex to regulate the activity of Dl and/or Ser. Figure 3 D-mib Co-Localizes with Dl and Ser at the Apical Cell Cortex (A and A′) D-mib (green) is detected in all cells of the wing imaginal disc. In (A), Ser is in red and Discs-large (Dlg) is in blue. (B–D′′′) D-mib (green in B, B′, C, C′, D, and D′) co-localized with Ser (red in [B and B′′]), Dl (red in [C and C′′]), N (red in [D and D′′]), and E-Cadherin (E-Cad; blue in [D and D′′′]) and was found apical to Discs-large (Dlg; blue in [B, B′′′, C, and C′′′]) in notum cells located at the edges of the wing discs. (E–E′′) D-mib (green in [E and E′]) co-localized with Dl (red in [E and E′′]) at the apical cortex of wing pouch cells. (F–F′′) D-mib staining at the apical cortex (blue in [F and F′]) was not detected in D-mib 2 mutant clone (marked by loss of nuclear GFP staining; green in [F]). Loss of D-mib activity has no detectable effect on the apical accumulation of Dl (red in [F and F′′]). Bar is 50 μm for (A and A′) and 10 μm for (B–F′′). Figure 6 D-mib Is Required in Dorsal Cells for Margin Expression of Cut Large dorsal clones of D-mib 2 mutant cells (marked by the loss of nuclear GFP, in green) resulted in a complete loss of Cut (red) expression (A and B). This indicates that D-mib is required for Ser signaling by dorsal cells. In contrast, ventral clones did not prevent the expression of Cut (C and D), implying that D-mib is not strictly required for Dl signaling. Note that mutant ventral cells abutting wild-type dorsal cells expressed Cut (arrow in [D]), indicating that D-mib is not required for N signal transduction. Low-magnification views of the wing portion of the discs are shown in (A) and (C). (B) and (D) show high-magnification views of the areas boxed in (A) and (C), respectively. D-mib Regulates the Cell-Surface Level of Ser We next examined the potential role of D-mib in regulating Dl and Ser distribution in wing imaginal discs. We focused our analysis on the notum region since D-mib mutant discs have no wing pouch ( Figure 4 ). Dl and Ser co-localized both at the apical cortex and in large intracellular vesicles in wild-type cells ( Figure 4 A– 4 C′). The complete loss of D-mib activity in D-mib 1 mutant discs did not detectably change the subcellular localization of Dl ( Figure 4 C, 4 C′, 4 F, and 4 F′). In contrast, the accumulation of Ser at the apical cortex was strongly increased ( Figure 4 E) and Ser accumulation in Dl-positive vesicles was dramatically reduced ( Figure 4 E′) in D-mib 1 mutant discs. Similar results were also obtained in D-mib 2 mutant clones, which showed strongly elevated levels of cortical Ser ( Figure 4 H) whereas the amount of Dl at the apical cortex was not detectably modified (see Figures 3 F– 3 F′′ and 4 J). Of note, loss of D-mib 2 activity in clones did not block the accumulation of Ser into intracellular dots ( Figure 4 H′). Thus, trafficking of Ser towards this intracellular compartment is, at least in part, D-mib -independent. We therefore conclude that the D-mib gene is required to regulate the level of Ser at the apical cortex of wing disc cells. Figure 4 D-mib Is Required to Down-Regulate Ser at the Apical Cortex (A–F′) Distribution of Dl (green) and Ser (red) in the notum region of wild-type (A–C′) and D-mib 1 mutant (D–F′) wing imaginal discs. The boxed areas in (A) and (D) are shown at higher magnification in (B–F′). The specific loss of Ser accumulation into intracellular vesicles (compare [E′] with [B′]) correlated with the elevated levels of Ser seen at the apical cortex of D-mib mutant cells (compare [E] with [B]). (G–J′) Ser (red in [H and H′]) accumulated at the apical cortex (H) as well as in intracellular dots (H′) in D-mib 2 mutant cells (marked by the loss of nuclear GFP; green in [G]). Cut is shown in blue (G). The distribution of Dl (red in [J and J′]) was not affected by the loss of D-mib activity. Low-magnification views of the wing portion of the discs are shown in (G) and (I). (H and H′) and (J and J′) show high magnification views of the areas boxed in (G) and (I), respectively. Clone boundaries are outlined in (H and H′) and (J and J′). Bar is 40 μm for (A, D, G), 5 μm for (B–C′ and E–F′), and 10 μm for (H–J′). D-mib Is Required for Ser Endocytosis To test whether this specific increase in the level of Ser at the apical cortex resulted from reduced Ser endocytosis in D-mib mutant cells, we followed the endocytosis of Ser in living imaginal discs using an antibody uptake assay. Briefly, dissected wing discs were cultured for 15 min in the presence of antibodies that recognize the extracellular part of Ser or Dl, then washed, cultured for another 45 min in medium without antibodies, and then fixed. The uptake of anti-Ser and anti-Dl antibodies was then assessed using secondary antibodies. The results are shown in Figure 5 . Using this assay, we found that anti-Ser-and anti-Dl antibodies were internalized in wild-type epithelial cells ( Figure 5 A– 5 C′′). The complete loss of D-mib activity in D-mib 1 wing discs did not significantly change the internalization of anti-Dl antibodies ( Figure 5 D′′, 5 E′′, and 5 F′′), indicating that D-mib is not required for Dl endocytosis in this tissue. However, the loss of D-mib activity strongly inhibited the endocytosis of anti-Ser antibodies ( Figure 5 E′). Moreover, high levels of anti-Ser antibodies were seen at the apical surface ( Figure 5 D′ and 5 F′), confirming that D-mib mutant cells accumulate high levels of Ser at their surface. We therefore conclude that D-mib is specifically required for the endocytosis of Ser in wing discs. Figure 5 D-mib Is Required for Ser Endocytosis Localization of the anti-Ser (red) and anti-Dl (green) antibodies that have been internalized by wild-type (A–C′′) and D-mib 1 mutant (D–F′′) cells in the notum region of wing discs. (A–A′′) and (D–D′′) show apical sections and (B–B′′) and (E–E′′) show basal sections. (C–C′′) and (F–F′′) show confocal z-sections. The z-section axes are shown with a double-headed arrow in (A) and (D). Internalized anti-Ser and anti-Dl antibodies co-localized in wild-type cells. In contrast, high levels of anti-Ser antibodies were detected at the cell surface of D-mib mutant epithelial cells whereas anti-Dl antibodies were efficiently internalized. Bar is 10 μm for all panels. Ubiquitin-mediated endocytosis is thought to depend on monoubiquitination. Thus, by analogy with the function of Mib in D. rerio [ 18 , 28 ], we suggest that D-mib may directly monoubiquitinate Ser. Consistent with this hypothesis, we show in a companion paper that D-mib binds Ser (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). Moreover, a mutation in the C-terminal catalytic RING domain of D-mib abolished its ability to internalize Ser in transfected S2 cells (R. L. B. and F. S., unpublished data) implying that the E3 ubiquitin ligase activity of D-mib is required for Ser internalization. Biochemical analysis of the ubiquitination events regulated by D-mib will be needed to further define the mechanism by which D-mib regulates the endocytosis of Ser in vivo. D-mib Regulates Ser Signaling The regulation of Ser endocytosis by D-mib suggests that D-mib may regulate Ser signaling. Ser expression is restricted to dorsal cells in second instar wing imaginal discs [ 7 , 10 , 44 , 47 , 48 ]. Ser in dorsal cells signals across the D-V boundary to activate N in ventral cells [ 8 , 9 ]. If D-mib is required for Ser signaling during wing development, then loss of D-mib activity in dorsal cells should affect the specification of the wing margin in a non-autonomous manner. Loss of D-mib activity in large dorsal clones of D-mib 2 mutant cells resulted in a loss of Cut expression at the D-V interface ( Figure 6 A and 6 B). The lack of Cut expression in wild-type ventral cells abutting the D-V boundary indicates that D-mib is required for Ser signaling by dorsal cells and acts in a non-autonomous manner to activate N in ventral cells. Conversely, loss of D-mib activity in large ventral clones ( Figure 6 C and 6 D) did not disrupt margin specification, indicating that D-mib is not strictly required for Dl signaling by ventral cells. However, a narrowing of the Cut-positive margin was observed ( Figure 6 D), suggesting that D-mib contributes to regulating the level of Dl signaling. Of note, ventral D-mib mutant cells expressed Cut, implying that D-mib is not required for N signal transduction. We next tested whether expression of D-mib in dorsal cells is sufficient to rescue the D-mib wing phenotype. D-mib was expressed in dorsal cells of D-mib 2 / D-mib 3 mutant discs using Ser-GAL4. Similarly to the expression of the Ser gene, Ser-GAL4 expression is restricted to dorsal cells in second/early third instar larvae and is weakly expressed in ventral cells in mid/late third instar larvae, i.e., after margin cell specification [ 49 , 50 ]. Expression of D-mib in dorsal cells was sufficient to rescue growth of the wing pouch and of the expression of Cut in margin cells in D-mib mutant discs ( Figure 7 A). This result confirmed that D-mib regulates Ser signaling by dorsal cells. Figure 7 Expression of D-mib in Dorsal Cells Is Sufficient to Rescue the D-mib Mutant Phenotype (A) Expression of D-mib (green) in dorsal cells, using Ser-GAL4, rescued the growth of the wing pouch and margin Cut (red) expression in D-mib 2 / D-mib 3 mutant discs. (B–D′′′) Ser-GAL4-driven expression of YFP::D-mib (green) rescued the D-mib 2 / D-mib 3 phenotype and strongly reduced the level of Dl (blue in [B, B′, C, C′′, D, and D′′]) and Ser (red in [B, B′′, C, C′′′, D, and D′′′]) in dorsal cells. (C–D′′′) are high-magnification views (apical [C–C′′′] and basal [D–D′′′]) of the disc shown in (B–B′′). YFP::D-mib co-localized with Dl and Ser at the apical cortex in cells expressing only low levels of YFP::D-mib. Bar is 50 μm for (A–B′′) and 10 μm for (C–D′′′). A similar rescue was observed with a YFP::D-mib protein ( Figure 7 B– 7 B″), indicating that YFP::D-mib is functional. YFP::D-mib localized at the apical cortex and in the cytoplasm ( Figure 7 C– 7 D′′′), as seen for endogenous D-mib (see Figure 3 ). YFP::D-mib co-localized with Dl and Ser at the apical cortex of cells expressing low levels of YFP::D-mib. However, cells expressing high levels of YFP::D-mib showed a strong reduction in the level of both Dl and Ser at the cortex ( Figure 7 C– 7 C′′′), further indicating that D-mib down-regulates the levels of both Ser and Dl at the apical cortex (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). D-mib Acts Downstream of Ser and Upstream of Activated N The functional assay was then used to genetically position the requirement for the D-mib gene activity relative to Ser and N ( Figure 8 ). Expression of an activated version of N, Ncdc10 [ 51 ], led to the activation of Cut and promoted growth in dorsal cells of D-mib 2 / D-mib 3 mutant discs ( Figure 8 C). This indicates that D-mib acts at a step upstream of N activation. By contrast, elevated levels of Ser expression failed to restore Cut expression and growth of the wing pouch in D-mib 2 / D-mib 3 mutant larvae ( Figure 8 B). This confirms that Ser signaling requires the activity of the D-mib gene, i.e., that D-mib acts downstream of Ser. Figure 8 Expression of Neur in Dorsal Cells Is Sufficient to Rescue the D-mib Mutant Phenotype D-mib 2 / D-mib 3 mutant discs expressing GFP (A) (GFP staining not shown), Ser (B), Ncdc10 (C), or Neur (D) under the control of Ser-GAL4 were stained for Cut (red). Expression of Ser in dorsal cells did not rescue the D-mib 2 / D-mib 3 wing pouch mutant phenotype (compare [B] with [A]), consistent with D-mib being required for Ser signaling. By contrast, expression of Ncdc10, an activated version of N, led to the deregulated growth of the dorsal compartment and the expression of Cut in most dorsal cells (C), indicating that activated N acts downstream of D-mib . Expression of Neur in dorsal cells was sufficient to compensate for the loss of D-mib activity (D). Bar is 40 μm for all panels. neur and D-mib Functions Partially Overlap The different requirements for neur and D-mib gene activity may suggest that Neur and D-mib have distinct molecular activities. Alternatively, this difference may reflect a difference in gene expression. Consistent with the latter hypothesis, the neur gene is not expressed in wing pouch and wing margin cells, where it is not required, and appears to be expressed only in sensory cells [ 52 ], where it is required. By contrast, D-mib appears to be uniformly expressed in imaginal discs. To test this hypothesis, we examined whether the forced ubiquitous expression of the neur gene can suppress the D-mib loss-of-function phenotype. Expression of Neur, using actin -GAL4, restored growth of the wing pouch and formation of the wing margin (data not shown). Moreover, expression of Neur in dorsal cells, using Ser-GAL4, was sufficient to rescue growth of the wing pouch as well as the expression of Cut in margin cells in D-mib mutant discs ( Figure 8 D). We conclude that ectopic expression of Neur compensates for the loss of D-mib activity. In a converse experiment, we found that the neur -driven expression of D-mib, using neur PGAL4 , did not rescue the cuticular neurogenic phenotype of neur PGAL4 / neur 1F65 embryos. Three UAS-D-mib transgenic lines were tested, and none showed detectable rescue whereas the two UAS-neur lines used as positive controls either fully or partially rescued the cuticular neurogenic phenotype of neur PGAL4 / neur 1F65 embryos (data not shown; UAS-D-mib /+ , neur PGAL4 /+ embryos developed normally). This indicates that a key function of Neur in the embryo cannot be provided by D-mib. We therefore suggest that Neur and D-mib functions overlap but are not strictly identical. Discussion Many recent studies have revealed that endocytosis plays multiple roles in the regulation of N signaling (reviewed in [ 2 ]; see also [ 53 , 54 ]). Here, we show that the conserved E3 ubiquitin ligases Neur and D-mib have similar molecular activities in the regulation of Dl and Ser endocytosis but distinct developmental functions in Drosophila . Our analysis first establishes that D-mib regulates Ser signaling during wing development. First, clonal analysis revealed that the activity of the D-mib gene is specifically required in dorsal cells for the expression of Cut at the wing margin. Second, expression of D-mib in the dorsal Ser-signaling cells was sufficient to rescue the D-mib mutant wing phenotype. Third, results from an in vivo antibody uptake assay indicated that the endocytosis of Ser (but not of Dl) was strongly inhibited in D-mib mutant cells. This inhibition correlated with the strong accumulation of Ser (but not Dl) at the apical cortex of D-mib mutant cells. Thus, an essential function of D-mib in the wing is to regulate the endocytosis of Ser in dorsal cells to non-autonomously promote the activation of N along the D-V boundary. By analogy, the defective growth of the eye tissue may similarly result from the lack of Ser signaling and of N activation along the D-V boundary [ 55 ]. Because D-mib co-localizes with Ser at the apical cortex of wing disc cells, acts in a RING-finger-dependent manner to regulate Ser endocytosis in S2 cells (R. L. B. and F. S., unpublished results), and physically associates with Ser in co-immunoprecipitation experiments (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data), D-mib may ubiquitinate Ser and directly regulate its endocytosis. Our analysis further suggests that endocytosis of Ser is required for Ser signaling. This conclusion is consistent with observations made earlier showing that secreted versions of Ser cannot activate N but instead antagonize Ser signaling [ 56 , 57 ]. Thus, endocytosis of both N ligands appears to be strictly required for N activation in Drosophila . Different models have been proposed to explain how endocytosis of the ligand, which removes the ligand from the cell surface, results in N receptor activation (discussed in [ 17 , 20 , 21 , 30 ]). Interestingly, the strong requirement for Dl and Ser endocytosis seen in Drosophila is not conserved in Caenorhabditis elegans, in which secreted ligands have been shown to be functional [ 58 , 59 ]. Noticeably, there is no C. elegans Mib homolog, and the function of C. elegans neur (F10D7.5) is not known. We speculate that endocytosis of the ligands may have evolved as a means to ensure tight spatial regulation of the activation of N. Our analysis also establishes that the activity of the D-mib gene is required for a subset of N signaling events that are distinct from those that require the activity of the neur gene. We have shown that the D-mib gene regulates wing margin formation, leg segmentation, and vein formation, whereas none of these three processes depend on neur gene activity ([ 45 , 60 ]; this study). Conversely, the activity of the neur gene is essential for binary cell-fate decisions in the bristle lineage [ 22 ] that do not require the activity of the D-mib gene (no bristle defects were seen in D-mib mutant flies). The activity of the neur gene is also required for lateral inhibition during neurogenesis in embryos and pupae [ 4 , 45 , 61 ]. This process is largely independent of D-mib gene activity since the complete loss of D-mib function only resulted in a mild neurogenic phenotype in the notum. These data thus indicate that the neur and D-mib genes have largely distinct and complementary functions in Drosophila . Whether a similar functional relationship between Neur and D-mib exists in vertebrates awaits the study of the D. rerio neur genes and/or of the murine Mib and Neur genes. The functional differences observed between D-mib and neur cannot be simply explained by obvious differences in molecular activity and/or substrate specificity. First, both Neur and D-mib physically interact with Dl ([ 20 ]; E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data) and promote the down-regulation of Dl from the apical membrane when overexpressed (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). Furthermore, Dl signaling appears to require the activity of either Neur or D-mib, depending on the developmental contexts. We have shown here that specific aspects of the D-mib phenotype in legs and in the notum cannot simply result from loss of Ser signaling and are consistent with reduced Dl signaling, suggesting that D-mib regulates Dl signaling. Consistent with this interpretation, overexpression studies indicate that D-mib up-regulates the signaling activity of Dl, whereas a dominant-negative form of D-mib inhibits it (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). We note, however, that no clear defects in Dl subcellular localization and/or trafficking were observed in D-mib mutant cells. It is conceivable that the contribution of D-mib to the endocytosis of Dl is masked by the activity of D-mib -independent processes that may, or may not, be linked to Dl signaling. We have also shown that, reciprocally, Neur and D-mib may similarly regulate Ser. Neur and D-mib were shown to similarly promote down-regulation of Ser from the cell surface when overexpressed (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data). Moreover, D-mib binds Ser (E. C. Lai, F. Roegiers, X. Qin, R. Le Borgne, F. Schweisguth, et al., unpublished data) and regulates Ser signaling (this study). Whether endogenous Neur binds and activates Ser remains to be tested. However, the ability of Neur to rescue the D-mib mutant wing phenotype when expressed in dorsal cells strongly indicates that Neur can promote Ser signaling. Together, these data indicate that Neur and D-mib have similar molecular activities. D-mib and Neur may have identical molecular activities but distinct expression patterns, hence distinct functions at the level of the organism. Consistent with this possibility, D-mib is uniformly distributed in imaginal discs, whereas Neur is specifically detected in sensory cells [ 52 ]. Importantly, the rescue of the D-mib mutant phenotype by ectopic expression of Neur strongly supports this interpretation. This result further suggests that Neur can regulate Ser signaling. Consistent with this idea, overexpression of Neur in imaginal discs resulted in a strong reduction of Ser accumulation at the apical cortex (data not shown). Thus, despite their obvious structural differences, Neur and D-mib appear to act similarly to promote the endocytosis of Dl and Ser. Nevertheless, our observation that D-mib could not compensate for the loss of neur activity in the embryo indicates that D-mib and Neur have overlapping rather than identical molecular activities. In conclusion, Neur and D-mib appear to have similar molecular activities in the regulation of Dl and Ser endocytosis but distinct developmental functions in Drosophila . The conservation from Drosophila to mammals of these two structurally distinct but functionally similar E3 ubiquitin ligases is likely to reflect a combination of evolutionary advantages associated with: (i) specialized expression pattern, as evidenced by the cell-specific expression of the neur gene in sensory organ precursor cells [ 52 ]; (ii) specialized function, as suggested by the role of murine MIB in TNFα signaling [ 32 ]; (iii) regulation of protein stability, localization, and/or activity. For instance, Neur, but not D-mib, localizes asymmetrically during asymmetric sensory organ precursor cell divisions [ 22 ]. Materials and Methods Flies The D-mib 1 mutation corresponds to the EY97600 P-element insertion generated by the Gene Disruption Project ( http://flypush.imgen.bcm.tmc.edu/pscreen/ ). The D-mib 2 allele was selected as w − D-mib mutant derivative by imprecise excision of the EY97600 P-element. The precise breakpoints of the D-mib 2 deletion were determined by sequencing a PCR fragment amplified from genomic DNA prepared from D-mib 2 homozygous larvae. The l(3)72Cda J12 and l(3)72Cda I5 alleles originally isolated by [ 35 ] failed to complement the D-mib 1 and D-mib 2 mutations and were renamed D-mib 3 and D-mib 4 . The D-mib 1 , D-mib 2 , and D-mib 3 alleles appear to be genetically null alleles since the phenotypes of D-mib 1 / D-mib 1 and D-mib 1 / D-mib 3 mutant pupae are indistinguishable from the ones seen in D-mib 1 / D-mib 2 and D-mib 2 / D-mib 3 pupae. Sequence analysis of the D-mib 3 and D-mib 4 alleles was carried on PCR products prepared from genomic DNA prepared from D-mib 3 /D-mib 2 and D-mib 4 / D-mib 2 mutant pupae. Genomic DNA from l(3)72Cda/D-mib 2 mutant pupae was used as control for polymorphism. D-mib 2 mutant clones were generated in y w hs-flp;FRT2A D-mib 2 /FRT2A M(3) i55 ubi-nlsGFP larvae. neur 1F65 mutant clones were generated as previously described [ 22 ]. UAS-D-mib and UAS-YFP::D-mib lines were generated via standard P-element transformation. These constructs were derived from the SD05267 cDNA obtained from ResGen (Invitrogen, Carlsbad, California, United States). Cloning details for these constructs are available upon request. UAS-Dl (gift from M. Muskavith), UAS-Ser (gift from R. Fleming), UAS-Neur (gift from C. Delidakis), UAS-Ncdc10 (gift of T. Klein), Ser-GAL4 lines, and Ser mutant alleles are described in FlyBase ( http://flybase.bio.indiana.edu/ ). Antibodies Dissected imaginal discs were fixed in 4% paraformaldehyde (15 min) and incubated with antibodies at room temperature in PBS 1X with 0.1% TritonX-100. Rabbit polyclonal anti-D-mib antibodies were raised against the CYNERKTDDSELPGN peptide (CovalAb, Lyon, France). Immunopurified anti-D-mib antibodies (rabbit 541) were used (immunofluorescence, 1:100; Western blot, 1:1,000). Other primary antibodies were mouse anti-Cut (2B10; Developmental Studies Hybridoma Bank [DSHB, Iowa City, Iowa, United States]; 1:500); rat anti-DE-Cadherin (gift from T. Uemura; 1:50); guinea pig anti-Discs-large (gift from P. Bryant; 1:3,000); anti-β-galactosidase (Cappel [MP Biomedicals, Irvine, California, United States]; 1:1,000); mouse anti-DeltaECD (C594.9B; DSHB; 1:1,000); mouse anti-NotchECD (C548.2H; DSHB; 1:1,000); rat anti-Ser (gift from K. Irvine; 1:2,000); rat anti-Ser (gift from S. Cohen; 1:200); rabbit anti-Ser (gift from E. Knust; 1:10); and guinea pig anti-Senseless (gift from H. Bellen; 1:3,000). Cy2-, Cy3-, and Cy5-coupled secondary antibodies were from Jackson Laboratory (Bar Harbor, Maine, United States). Alexa488-coupled secondary antibodies and phalloidin were from Molecular Probes (Eugene, Oregon, United States). Images were acquired on a Leica (Wetzlar, Germany) SP2 microscope and assembled using Adobe Photoshop (Adobe Systems, San Jose, California, United States). Endocytosis assay Third instar larvae wing discs were dissected in Schneider's Drosophila medium (Gibco BRL, San Diego, California, United States) containing 10% fetal calf serum (Gibco BRL). Wing discs were cut between the wing pouch and the thorax to facilitate antibody diffusion. Wing discs were cultured for 15 min with mouse anti-Dl (C594–9B at 1:100) and rat anti-Ser antibody (1:500; from K. Irvine). Following three medium changes and a 45-min chase period, wing discs were fixed and incubated with secondary antibodies. Supporting Information Accession Numbers The FlyBase accession numbers for the gene products discussed in this paper are Dl (FBgn0000463), N (FBgn0004647), Neur (FBgn0002932), P-element inserted into the 5′ untranslated region of the D-mib gene (FBgn0036558), and Ser (FBgn0004197). The NCBI Entrez Protein ( http://www.ncbi.nlm.nih.gov/entrez/ ) accession number for D. rerio Mib is NP_779353.

D:\keerthana\PMC001xxxxxx\PMC1064853.xml

1064854

PGC-1α Deficiency Causes Multi-System Energy Metabolic Derangements: Muscle Dysfunction, Abnormal Weight Control and Hepatic Steatosis

The gene encoding the transcriptional coactivator peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) was targeted in mice. PGC-1α null (PGC-1α −/− ) mice were viable. However, extensive phenotyping revealed multi-system abnormalities indicative of an abnormal energy metabolic phenotype. The postnatal growth of heart and slow-twitch skeletal muscle, organs with high mitochondrial energy demands, is blunted in PGC-1α −/− mice. With age, the PGC-1α −/− mice develop abnormally increased body fat, a phenotype that is more severe in females. Mitochondrial number and respiratory capacity is diminished in slow-twitch skeletal muscle of PGC-1α −/− mice, leading to reduced muscle performance and exercise capacity. PGC-1α −/− mice exhibit a modest diminution in cardiac function related largely to abnormal control of heart rate. The PGC-1α −/− mice were unable to maintain core body temperature following exposure to cold, consistent with an altered thermogenic response. Following short-term starvation, PGC-1α −/− mice develop hepatic steatosis due to a combination of reduced mitochondrial respiratory capacity and an increased expression of lipogenic genes. Surprisingly, PGC-1α −/− mice were less susceptible to diet-induced insulin resistance than wild-type controls. Lastly, vacuolar lesions were detected in the central nervous system of PGC-1α −/− mice. These results demonstrate that PGC-1α is necessary for appropriate adaptation to the metabolic and physiologic stressors of postnatal life.

Introduction Mitochondrial functional capacity is dynamically regulated to meet the diverse energy demands imposed on the mammalian organism following birth. Postnatal mitochondrial biogenesis involves multiple signaling and transcriptional regulatory pathways that control the coordinate expression of nuclear and mitochondrial genes involved in mitochondrial structure, metabolism, and proliferation [ 1 ]. Recent evidence points toward a transcriptional coactivator, peroxisome proliferator-activated receptor-γ (PPARγ) coactivator-1α (PGC-1α), as an integrator of the molecular regulatory circuitry involved in the transcriptional control of cellular energy metabolism, including mitochondrial function and biogenesis [ 1 , 2 ]. PGC-1α was discovered in a yeast two-hybrid screen for brown adipose-specific factors that interact with the adipogenic nuclear receptor PPARγ [ 2 ]. Subsequently, two additional PGC-1 family members were identified, PGC-1 related coactivator (PRC) [ 3 ] and PGC-1β [ 4 , 5 ]. PGC-1α serves as a direct transcriptional coactivator of nuclear and nonnuclear receptor transcription factors involved in cellular energy metabolism [ 6 ]. PGC-1α is distinct among most coactivators in that it exhibits a tissue-enriched expression pattern and is highly inducible by physiologic conditions known to increase the demand for mitochondrial ATP or heat production [ 2 , 6 , 7 ]. PGC-1α is enriched in brown adipose tissue (BAT), heart, slow-twitch skeletal muscle, and kidney—all tissues with high-capacity mitochondrial systems. The expression of the gene encoding PGC-1α is rapidly induced by cold exposure, short-term exercise, and fasting [ 2 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 ]. These latter observations suggest that PGC-1α is involved in the physiologic control of energy metabolism. Several lines of evidence, based on the results of overexpression studies, indicate that PGC-1α is sufficient to promote mitochondrial biogenesis and regulate mitochondrial respiratory capacity. First, PGC-1α activates the transcription of mitochondrial uncoupling protein-1 (UCP-1) in BAT through interactions with the nuclear hormone receptors PPARγ and thyroid receptor [ 2 ]. Second, forced expression studies in adipogenic and myogenic mammalian cell lines demonstrated that PGC-1α activates mitochondrial biogenesis through a group of transcription factor targets including nuclear respiratory factors 1 and 2 (NRF-1 and -2) and mitochondrial transcription factor A (Tfam), key transcriptional regulators of mitochondrial DNA transcription and replication [ 8 ]. Third, studies in primary cardiac myocytes in culture and in the hearts of transgenic mice have demonstrated that overexpression of PGC-1α promotes mitochondrial biogenesis [ 10 , 16 ]. Lastly, forced expression of PGC-1α in skeletal muscle of transgenic mice triggers mitochondrial proliferation and the formation of mitochondrial-rich type I, oxidative (“slow-twitch”) muscle fibers [ 17 ]. Collectively, these results indicate that PGC-1α is sufficient to drive mitochondrial biogenesis. Recent evidence also implicates PGC-1α in the homeostatic control of systemic energy metabolism. PGC-1α has been shown to regulate several key hepatic gluconeogenic genes [ 18 , 19 , 20 , 21 ]. Recent studies have also shown altered expression of PGC-1α and downstream mitochondrial target pathways in skeletal muscle of humans with insulin resistance and diabetes [ 22 , 23 , 24 ]. In addition, single nucleotide polymorphisms within the human PGC-1α gene have been shown to be associated with obesity, hypertension, and diabetes [ 25 , 26 , 27 , 28 , 29 , 30 ]. The gain-of-function studies described to date provide compelling evidence that PGC-1α is capable of regulating postnatal energy metabolism. However, the necessity of PGC-1α for energy metabolic homeostasis, mitochondrial biogenesis, development, and growth can only be addressed using loss-of-function strategies. To this end, we have established and characterized mice with targeted deletion of the PGC-1α gene. Our studies of PGC-1α −/− mice demonstrate that PGC-1α is not absolutely required for prenatal viability including mitochondrial biogenesis. However, our findings indicate that the coactivator PGC-1α serves a critical role in the normal metabolic function of multiple organs and for appropriate adaptation to physiologic stress during postnatal life. Results Disruption of the PGC-1α Gene in Mice A neomycin-based gene targeting vector was generated to delete exons 4 and 5 of the murine PGC-1α gene. The targeting event resulted in a 3′ homologous recombination with insertion of the remainder of the construct ( Figure 1 A). The insertion/recombination event was confirmed by Southern blotting and DNA sequencing. The insertion caused an exon 3 duplication between exons 5 and 6 that creates a coding region frameshift resulting in a premature termination at amino acid 255. Germline transmission of the mutant allele was confirmed using Southern blotting ( Figure 1 B) and PCR (unpublished data). The PGC-1α gene disruption resulted in an unstable transcript that could not be detected by RNA blot analysis in heart and other tissues in PGC-1α −/− mice ( Figure 1 C and unpublished data). Quantitative RT-PCR was utilized to further evaluate the efficacy of the gene targeting. For these studies, PCR primers were designed to amplify a region of the PGC-1α gene transcript containing the exon 5–6 border (predicted to be absent in PGC-1α −/− mice) or the exon 5–3 border (predicted to be present only in the PGC-1α −/− mice). The exon 5–6 amplicon was detected in heart and BAT of wild-type (WT) but not PGC-1α −/− mice ( Figure 1 D). Conversely, the exon 5–3 product was present only in PGC-1α −/− mice ( Figure 1 D). An exon 10–11 border amplicon (predicted to be present in both genotypes) was detected in WT and PGC-1α −/− mice, but was greatly diminished in the PGC-1α −/− mice, indicating that the mutant transcript is unstable. PGC-1α protein was not detected in whole cell ( Figure 1 E) or nuclear protein extracts (unpublished data) isolated from BAT of PGC-1α −/− mice under basal conditions or in response to cold exposure, a condition known to markedly induce the expression of PGC-1α in BAT. Smaller mutant PGC-1α proteins were also not detected by Western blot analysis (unpublished data). Lastly, expression of the genes encoding the other known PGC-1 family members, PGC-1β and PRC, was not significantly altered in heart of PGC-1α −/− mice ( Figure 1 C). Taken together, these results support the conclusion that the gene targeting event resulted in a PGC-1α null allele. Figure 1 Deletion of the PGC-1α Gene (A) Schematic of the gene targeting strategy. A region of the murine PGC-1α gene containing exons 3–6 is shown schematically at the top. Relevant restriction endonuclease sites are also shown. The targeting construct containing a neomycin (Neo) cassette is shown below the PGC-1α gene with dashed lines indicating the regions targeted for homologous recombination. Homologous recombination between the 3′ end of the targeting vector and the PGC-1α gene is indicated by the solid lines. The targeting construct inserted into the PGC-1α gene resulting in a duplication of exon 3 and incorporation of the targeting construct DNA into the final recombinant as shown. Probes used for the Southern blot studies and relevant restriction fragments predicted by digestion of the recombinant are also shown. (B) Southern blot analysis of embryonic stem cells (ESC) (digested with Xba1) and tail DNA (digested with Pst1) is shown. The blots were hybridized with the probes shown in Figure 1 A. Results for PGC-1α +/+ (+/+), PGC-1α +/− (+/−), and PGC-1α −/− (−/−) genotypes are shown as denoted at the bottom. (C) Northern blot analysis using RNA isolated from the hearts of the three relevant PGC-1α genotypes (as in Figure 1 B) is shown using PGC-1α cDNA as a probe. In addition, PGC-1β and PRC cDNA probes were used as shown. Ethidium bromide staining of 18S ribosomal RNA is shown at the bottom. (D) Quantitative real time RT-PCR (Sybr green) was used to detect PGC-1α transcripts using primer sets crossing different exon borders as denoted. The exon 5–3 primer set detects only the mutant transcript (−/−), whereas the exon 5–6 primer set detects only the WT transcript. The values represent arbitrary units for RNA isolated from the tissues shown for the three amplicons comparing PGC-1α +/+ and PGC-1α −/− . N.D. = not detectable. The exon 10–11 amplicon was evaluated to assess levels of mutant versus WT transcripts. The values represent arbitrary units normalized to actin control. (E) Western blot analysis using whole-cell protein extracts prepared from BAT under basal conditions and following exposure to 4 °C for 8 h. The signal shown was obtained with polyclonal anti-PGC-1α antibody [ 10 ]. An epitope-tagged PGC-1α, overexpressed in neonatal cardiac myocytes using an adenoviral vector (Ad-PGC-1α), is shown as a positive control. The Ponceau S stain of the protein gel is shown at the bottom. General Characteristics of the PGC-1α −/− Mice: Age- and Sex-Dependent Obesity Heterozygous (PGC-1α +/− ) mice were bred to generate PGC-1α −/− offspring. The observed genotype ratios of the offspring were consistent with the expected Mendelian ratios (unpublished data). Unexpected deaths of the offspring were not observed, and PGC-1α +/− and PGC-1α −/− offspring appeared normal. Total body weights obtained 1 wk after birth revealed a 15%–20% reduction in total body mass for male and female PGC-1α −/− mice relative to sex-matched PGC-1α +/+ littermates ( Figure 2 ). The weight decrement between PGC-1α −/− and PGC-1α +/+ littermates disappeared by 3 wk of age ( Figure 2 A). At 18 wk of age, body weight was modestly but significantly greater in male and female PGC-1α −/− mice compared to sex-matched PGC-1α +/+ controls ( Figure 2 A). This weight difference was also significant for female PGC-1α −/− mice at 24 wk of age ( Figure 2 A). The abnormal weight gain in PGC-1α −/− mice was not associated with differences in food intake (unpublished data) or alterations in general activity as monitored for 48 h ( Figure S1 ). Percent body fat, as determined by dual-energy X-ray absorption (DEXA), was greater in 18- and 24-wk-old female PGC-1α −/− mice compared to age-matched female PGC-1α +/+ counterparts, indicating that the body weight difference was due, at least in part, to increased body fat ( Figure 2 A). Lean mass was not significantly different between the genotypes (unpublished data). Although DEXA did not detect excess body fat in male PGC-1α −/− mice at 18 or 24 wk of age, older male mutant mice (over 7 mo of age) accumulated more body fat than male WT controls ( Figure 2 A and unpublished data). Figure 2 Evidence for Tissue-Specific Growth Abnormalities and Mild Sex-Limited, Age-Dependent Obesity in PGC-1α −/− Mice (A) The bars represent total body weight for the ages indicated for male (left graph) and female (center graph) PGC-1α +/+ and PGC-1α −/− mice. The body weight (BW) of the 1-wk-old PGC-1α −/− mice was normalized to that of PGC-1α +/+ littermates, which was assigned a value of 100 (left axis). For the 3-, 18-, and 24-wk time points, absolute weights of PGC-1α −/− mice were compared to age-matched controls (right axis). Percent fat as determined by DEXA scanning for PGC-1α +/+ and PGC-1α −/− mice (right graph). The results represent n = 4 (males) and n ≥ 11 (females) for each genotype at 24 wk. * p < 0.05 compared to corresponding PGC-1α +/+ mice. (B) The bars represent organ weights corrected to body weight for 3-wk-old male and female PGC-1α +/+ and PGC-1α −/− mice. The error bars represent ± SEM. Results represent n ≥ 14 for each group. * p < 0.05 compared to corresponding PGC-1α +/+ mice. Individual organ weights were assessed, given the importance of mitochondrial energy metabolism for postnatal growth in certain organs. The weights of heart and slow-twitch fiber-enriched skeletal muscles, including gastrocnemius and soleus, but not the less oxidative tibialis anterior, were significantly lower in male and female PGC-1α −/− mice compared with age and sex-matched PGC-1α +/+ controls at 3 and 8 wk of age ( Figure 2 B and unpublished data). In contrast, the weights of brain, liver, kidney, and BAT were not significantly different between the genotypes at the 3-wk time point ( Figure 2 B). Thus, certain tissues with high mitochondrial energy requirements, such as heart and slow-twitch skeletal muscle, exhibit modest growth defects in PGC-1α −/− mice. Abnormal Muscle Mitochondrial Phenotype in PGC-1α −/− Mice General histologic analyses were performed to begin to evaluate the mild growth defect found in postnatal heart and skeletal muscle of the PGC-1α −/− mice. There were no obvious abnormalities in cellularity, cell size, or extracellular matrix in the tissues of 1–2-mo-old PGC-1α −/− mice (unpublished data). Given the important role of PGC-1α in mitochondrial function and biogenesis, we examined mitochondrial ultrastructure in the relevant tissues. Electron microscopic analysis revealed fewer and smaller mitochondria in soleus muscle of PGC-1α −/− mice compared to sex- and age- matched PGC-1α +/+ controls ( Figure 3 A). Quantitative morphometry of the electron micrographs confirmed that the cellular volume density of soleus mitochondria was significantly lower in PGC-1α −/− mice compared to PGC-1α +/+ controls independent of changes in the myofibrillar component ( Figure 3 B). Consistent with a defect in mitochondrial biogenesis, we found a reduction in the expression of nuclear genes encoding proteins involved in mitochondrial electron transport (cytochrome c and cytochrome oxidase IV) and oxidative phosphorylation (beta subunit of ATP synthase) in soleus muscle of PGC-1α −/− mice compared with PGC-1α +/+ controls. In addition, the expression of Tfam, a known PGC-1α target involved in mitochondrial DNA replication/transcription, was diminished in PGC-1α −/− soleus, providing one potential mechanism for defective mitochondrial biogenesis ( Figure 3 C). In contrast to the results with soleus, no significant differences in mitochondrial ultrastructure or volume density were noted in heart or BAT of PGC-1α −/− mice (unpublished data). Figure 3 Abnormal Mitochondrial Phenotype in Slow-Twitch Skeletal Muscle of PGC-1α −/− Mice (A) Representative electron micrograph of soleus muscle from 1-mo-old female PGC-1α +/+ and PGC-1α −/− mice. (B) Quantitative morphometric measurements of the cellular volume density for the mitochondrial (Mito) and myofibrillar (Myo) fractions based on analysis of electron micrographs (three sections from three animals per group). The bars represent mean ± SEM. * p < 0.05 compared to corresponding PGC-1α +/+ values. (C) Gene expression data. The results of real-time PCR analysis of nuclear and mitochondrial genes involved in various components of mitochondrial metabolism and mitochondrial biogenesis: cytochrome C (Cyto c), ATP synthase β, Tfam, and cytochrome oxidase IV (COX IV). Eight littermate pairs were used for analysis at 1–2 mo of age and normalized to the WT value, which was assigned a value of 100, in each case. * p < 0.05 compared to corresponding PGC-1α +/+ values. (D) Mitochondrial respiration rates as determined by oxygen consumption (VO 2 ) performed on saponin-permeabilized muscle strips prepared from soleus of PGC-1α +/+ and PGC-1α −/− mice (as described in Materials and Methods ). The results are based on six female animals in each group, using succinate as a substrate in the presence of rotenone. Mean values (± SEM) are shown for state 2 (basal), state 3 (ADP-stimulated), and state 4 respiration (presence of oligomycin). To determine whether mitochondrial function was altered in the skeletal muscle of PGC-1α −/− mice, mitochondrial respiration rates were measured using tissue strips prepared from soleus muscle. In soleus of PGC-1α −/− mice, a significant defect in state 3 (ADP-stimulated) respiration, but not state 2 (basal), was detected using succinate as the substrate ( Figure 3 D). State 4 respiration rates (in the presence of oligomycin) were also similar between the genotypes, indicating that the coupling of respiration to ATP production was not significantly altered in PGC-1α −/− mice. These results are consistent with the modest but significant reduction in mitochondrial volume density. Altered Skeletal Muscle Function in PGC-1α −/− Mice The abnormality in mitochondrial number and respiratory function in skeletal muscle led us to further evaluate the skeletal muscle phenotype. As an initial step, we measured locomotor activity levels over a 1-h period using a high-resolution photobeam system. PGC-1α −/− male mice exhibited a significantly lower mean number of ambulations and rearings during the hour compared to the PGC-1α +/+ age-matched controls ( Figure 4 ). However, an analysis of exploratory behavior showed that the PGC-1α −/− mice were reluctant to go into the center of the “field” compared to controls. Specifically, PGC-1α −/− mice made significantly fewer entries into, spent significantly less time in, and traveled a significantly shorter distance in the central area of the “field,” although differences in distance traveled in the peripheral zone of the “field” was not significantly different between groups ( Figure S2 ). These data suggest that the general activity level may have been affected by the reluctance of the PGC-1α −/− mice to go into the central area of the field and thus remain in the periphery (thigmotaxis), possibly reflecting altered emotionality such as increased fear. Figure 4 PGC-1α −/− Mice Exhibit an Abnormal Skeletal Muscle Functional Phenotype (A) Measures of general activity and muscle strength. General activity levels were measured in 3.5-mo-old male PGC-1α +/+ ( n = 8) and PGC-1α −/− ( n = 11) mice using a photobeam system as described in Materials and Methods . Total ambulations (left graph), and rearings (center graph) provide a general measure of locomotor activity. Time spent on an inverted screen (right graph) represents a general measure of extremity muscle strength. The results from two trials are shown. * p < 0.05 compared to corresponding PGC-1α +/+ . (B) Exercise studies. Male 6–8-mo-old PGC-1α −/− and PGC-1α +/+ mice were subjected to a run-to-exhaustion protocol on a motorized treadmill (left graph) as described in Materials and Methods . * p < 0.001 compared to corresponding PGC-1α +/+ values. VO 2max measurements were determined for 2-mo-old male mice for each genotype using a motorized treadmill at an elevation of 150 m and indirect calorimetry set-up (right graph) as described in Materials and Methods . * p < 0.05 compared to corresponding PGC-1α +/+ values. (C) Time course of fatigue following repeated stimulation of soleus muscle is shown for 4-mo-old male PGC-1α −/− ( n = 5) and PGC-1α +/+ ( n = 5) mice (left graph). The mean percent force remaining at 2 min (Fatigue Resistance Index) is shown (right graph). * p < 0.05 compared to corresponding PGC-1α +/+ values. A battery of tests was performed to further evaluate the general sensorimotor phenotype of the PGC-1α −/− mice. No differences were found between PGC-1α −/− mice and PGC-1α +/+ controls on the ledge, platform, walking initiation, and 60° and 90° inclined screen tests (unpublished data), suggesting that several sensorimotor functions were intact in the PGC-1α −/− mice. However, the PGC-1α −/− mice were unable to remain on an inverted screen for as long as the PGC-1α +/+ controls ( Figure 4 A). Since the groups did not differ on the times it took to turn around and climb to the top of 60° and 90° inclined screens, the differences on the inverted screen test suggest that impaired strength rather than deficits in coordination were responsible for these differences. To further evaluate the skeletal muscle phenotype, exercise capacity was assessed in the PGC-1α −/− mice. To this end, the PGC-1α −/− mice were exercised on a motorized treadmill apparatus using a run-to-exhaustion format. PGC-1α −/− mice (6–8 mo of age) exhibited a markedly reduced capacity to sustain running exercise (PGC-1α −/− mice, 64 ± 6 s; age-matched PGC-1α +/+ mice, 586 ± 104 s; Figure 4 B). The same result was obtained with younger PGC-1α −/− mice, i.e., at 3.5 mo of age (unpublished data). To quantify aerobic exercise capacity, VO 2max (maximum oxygen consumption, measured in milliliters of oxygen per kilogram of body weight per minute) was measured with the treadmill-running protocol using indirect calorimetry. VO 2max was significantly lower for the PGC-1α −/− mice (120.9 ± 2.0 ml O 2 · kg −1 · min −1 ) compared to PGC-1α +/+ controls (141.6 ± 2.1 ml O 2 · kg −1 · min −1 ) ( Figure 4 B). To directly evaluate muscle fatigability, the force response to repetitive stimulation of isolated soleus muscle was determined. The capacity to generate force following a series of tetani is dependent upon mitochondrial ATP production. During the initial phase of the stimulation period, there was no difference in force generation in muscles isolated from PGC-1α −/− mice and PGC-1α +/+ controls. However, fatigue resistance index, defined as the percent of initial force generated following a 2-min series of fatiguing contractions, was significantly lower in the PGC-1α −/− mice (14.6 ± 1.5%) compared to PGC-1α +/+ controls (24.8 ± 2.9%) ( Figure 4 C). These results, together with the observed abnormalities in skeletal muscle mitochondrial structure and function, indicate that PGC-1α is necessary for functional adaptation of skeletal muscle to physiologic demands. Functional Abnormalities in Hearts of PGC-1α −/− Mice PGC-1α expression is enriched in heart, a tissue that relies heavily on mitochondrial energy metabolism to maintain pump function throughout the postnatal life of the mammalian organism. Echocardiographic screening studies of PGC-1α −/− mice at ages 4–6 mo did not reveal any significant differences in chamber sizes or ventricular function compared to WT controls (unpublished data). Cardiac functional and metabolic reserve was evaluated using exercise echocardiographic stress testing (EST). Given that the exercise capacity of PGC-1α −/− mice is diminished, a series of preliminary treadmill exercise studies were performed to define a reasonable exercise duration for run-to-exhaustion to be used as a target duration for the EST. Based on the results of these studies, an EST regimen was performed in which PGC-1α +/+ control animals were exercised for a duration of 60 s to match the predicted average for the PGC-1α −/− mice (ages 6–8 mo). Echocardiographic images were obtained immediately following 60 s of exercise for the PGC-1α +/+ controls or at the point of exhaustion for PGC-1α −/− mice (mean 60 ± 6.1 s, range 45–90 s). Echocardiographic-determined left ventricular fractional shortening and heart rate were monitored for the 10-min period immediately post exercise. The mean heart rate of the PGC-1α −/− mice exhibited an inappropriate decline during the post exercise period ( Figure 5 A). In addition, echocardiographically determined left ventricular fractional shortening was decreased in the PGC-1α −/− mice, but not the PGC-1α +/+ mice during the first 4 min of the post exercise period ( Figure 5 A). Figure 5 Abnormal Cardiac Response to Physiologic Stress in PGC-1α −/− Mice (A) Exercise echocardiographic studies. PGC-1α +/+ ( n = 4) and PGC-1α −/− ( n = 8) female mice aged 6–8 mo were subjected to an exercise protocol on a motorized treadmill. This protocol was designed such that the PGC-1α −/− mice ran to exhaustion based on the results of the exercise studies shown in Figure 4 . Accordingly, an exercise regimen of 60 s was used for both groups. The graphs depict the heart rate (left graph) and echocardiographically-determined ventricular fractional shortening (FS) as a percent (right graph). Responses were monitored for 10 min immediately post exercise. (B) In vivo hemodynamic response to the β 1 ,α 1 -adrenergic agonist dobutamine. Male and female PGC-1α +/+ ( n = 6) and PGC-1α −/− ( n = 6) mice at 10–12 wk of age were anesthetized and a 1.4-French Millar catheter was placed through the carotid artery into the left ventricle as described in Materials and Methods . Heart rate (left graph) and a measurement of ventricular systolic performance, dP/dt (right graph), were measured following infusion of dobutamine. * p < 0.05. The results of the EST did not distinguish between a primary cardiac abnormality versus effects secondary to the exhaustion caused by reduced exercise tolerance related to skeletal muscle dysfunction. To directly assess cardiac function, the hearts of PGC-1α −/− and PGC-1α +/+ mice were isolated and perfused in the working mode. Hearts isolated from PGC-1α −/− mice generated lower cardiac work (cardiac output multiplied by peak systolic pressure) compared to PGC-1α +/+ mice at identical loading conditions ( Table 1 ). This reduction in cardiac work was due to a reduced cardiac output ( Table 1 ). The relative contribution of heart rate and stroke volume to diminished cardiac output in the PGC-1α −/− mice could not be delineated, because both were decreased but neither to a significant degree ( Table 1 ). To further distinguish between abnormalities in heart rate and ventricular function, in vivo hemodynamic response to the β 1 ,α 1 -adrenergic-selective agonist dobutamine was evaluated using a miniaturized Millar catheter. The ventricular functional response to dobutamine was similar in PGC-1α +/+ and PGC-1α −/− mice ( Figure 5 B, right graph). However, PGC-1α −/− mice exhibited a significantly blunted heart rate response to β-adrenergic stimulation ( Figure 5 B, left graph). Taken together with the EST, these results strongly suggest that the PGC-1α −/− hearts are unable to mount an appropriate chronotropic response to exercise and other physiologic stimuli that activate β-adrenergic input to the heart. However, our results did not reveal evidence for contractile dysfunction. Table 1 Cardiac Hemodynamics Measured in Isolated Working Hearts of PGC-1α +/+ and PGC-1α −/− Mice Values represent mean ± SEM a p < 0.05 PGC-1α −/− Mice Exhibit an Abnormal Thermogenic Response PGC-1α has been implicated as an inducible regulator of mitochondrial respiratory uncoupling, an important source of heat production in BAT [ 2 ]. To determine whether PGC-1α is necessary for an appropriate thermogenic response, PGC-1α +/+ and PGC-1α −/− mice were subjected to cold exposure (4 °C) for a 5-h period while core body temperature was monitored. PGC-1α −/− mice exhibited a markedly abnormal drop in core temperature compared to the WT controls ( Figure 6 A). Specifically, the mean decline in core temperature was greater than 12 °C at the 5-h time point in PGC-1α −/− mice, compared to an approximately 3 °C decrement in PGC-1α +/+ controls. Although this thermogenic phenotype was consistently present in mice aged 28–37 d, it was absent in older mice (unpublished data). Figure 6 PGC-1α −/− Mice Exhibit an Abnormal Thermogenic Response (A) PGC-1α +/+ ( n = 15) and PGC-1α −/− ( n = 21) mice aged 28–37 d were subjected to cold (4 °C). Core rectal temperature was monitored over a 5-h period. The change in core temperature ± SEM is shown in the graph (left) as a function of time. * p < 0.05. (B) Representative Northern blot analysis (blot and gel at top) performed with RNA isolated from BAT to detect UCP-1 transcript at baseline (RT) and after 5 h of exposure to cold (4 °C) (UCP1). Ethidium bromide (Eth Br) staining of ribosomal RNA is shown as a control. Quantitative real-time RT-PCR for UCP-1 transcript is shown on the graph at the bottom. The values represent mean arbitrary units normalized to a 36B4 transcript (control). (C) Altered response to β 3 -adrenergic agonist. To evaluate the oxygen consumption (VO 2 ) in response to the stimulation of BAT uncoupled respiration, the β 3 -adrenergic agonist BRL 37344 was administered to littermate PGC-1α +/+ ( n = 5) and PGC-1α −/− ( n = 5) female mice followed by measurement of VO 2 by indirect calorimetry. Mean ± SEM VO 2 is shown. * p < 0.05. The histologic appearance and neutral lipid stores of BAT were assessed as an initial step to characterize the thermogenic phenotype exhibited by PGC-1α −/− mice. Histologic and lipid quantification studies were performed. Electron microscopic analyses indicated that the mitochondrial ultrastructure was similar in BAT isolated from PGC-1α −/− and PGC-1α +/+ mice before and after cold exposure (unpublished data). In addition, levels of BAT triglyceride were similar between the two genotypes (unpublished data). UCP-1 is a cold-inducible protein involved in mitochondrial respiratory uncoupling to generate heat in BAT. UCP-1 gene transcription is known to be activated by PGC-1α [ 2 ]. Surprisingly, basal and cold-induced BAT UCP-1 mRNA levels were similar in PGC-1α +/+ and PGC-1α −/− mice ( Figure 6 B). These results suggest that PGC-1α is not necessary for the induction of the expression of UCP-1 with cold exposure, and that other factors, such as reduced capacity for mitochondrial respiration, likely contribute to the abnormal thermogenic response in the PGC-1α −/− mice. Thermogenesis in rodents related to mitochondrial uncoupling is under the control of β 3 -adrenergic receptor coupled signaling. Accordingly, the in vivo oxygen consumption response to β 3 -adrenergic stimulation was examined in PGC-1α −/− mice. For these experiments, VO 2 (oxygen consumption) was measured following administration of the β 3 -agonist BRL 37344 using indirect calorimetry. VO 2 was significantly increased in response to BRL 3744 in PGC-1α +/+ but not PGC-1α −/− mice ( Figure 6 C). These results indicate that the metabolic response of BAT to an acute stimulus such as cold and/or β 3 -adrenergic stimulation is altered in the PGC-1α null mice, likely related to reduced capacity for mitochondrial respiratory uncoupling. Fasting-Induced Hepatic Steatosis in PGC-1α −/− Mice Previous studies have implicated PGC-1α in several hepatic metabolic functions including fatty acid oxidation and gluconeogenesis [ 18 , 19 , 20 , 21 ]. Accordingly, the hepatic phenotype was evaluated under basal conditions and following a 24-h fast, a stimulus known to induce fatty acid oxidation and gluconeogenic rates in liver. Under basal fed conditions, the livers of the PGC-1α −/− mice appeared grossly normal and did not exhibit histologic abnormalities (unpublished data). However, following a 24 h-fast, the PGC-1α −/− mice exhibited marked hepatic steatosis as determined by gross inspection, oil red O staining, electron microscopy, and measurements of liver triglyceride (TAG) levels ( Figure 7 ). There were no differences in plasma triglycerides or free fatty acids between the genotypes in fed or fasted states (unpublished data). To further investigate the mechanisms involved in the fasting-induced hepatic steatotic response, hepatocytes were isolated from PGC-1α −/− mice and WT controls. Oleate loading experiments revealed that the PGC-1α −/− hepatocytes accumulated neutral lipid to a significantly greater extent than the WT cells ( Figure 8 A). 3 H-palmitate oxidation rates were significantly lower in PGC-1α −/− hepatocytes compared to PGC-1α +/+ hepatocytes under basal conditions and following exposure to oleate ( Figure 8 B). Taken together, these latter results indicate a cell-autonomous defect in PGC-1α −/− hepatocytes that results in an inability to maintain cellular lipid balance in the context of increased delivery of lipid such as occurs with fasting. Figure 7 Fasting-Induced Hepatic Steatosis Develops in PGC-1α −/− Mice (A) The photograph depicts the development of a pale liver in PGC-1α −/− mice subjected to a 24-h fast. (B) Oil red O staining of histologic sections of liver taken from PGC-1α −/− mice under fed and 24 h fasted conditions. The red staining indicates neutral lipid. (C) Representative electron micrographs of the liver from PGC-1α +/+ and PGC-1α −/− mice following a 24-h fast. The droplets are indicative of neutral lipid accumulation. (D) Mean liver TAG levels in PGC-1α +/+ ( n = 5) and PGC-1α −/− ( n = 5) mice under fed and 24-h fasted conditions. * p < 0.05. Figure 8 Hepatocytes Isolated from PGC-1α −/− Mice Exhibit Reduced Oxidative Capacity (A) Oil red O staining of isolated hepatocytes exposed to BSA alone (BSA) or 50 μM oleate complexed to BSA (oleate). (B) 3 H-palmitate oxidation rates. 3 H-palmitate oxidation rates determined in hepatocytes isolated from PGC-1α +/+ and PGC-1α −/− mice under cell culture conditions containing BSA or BSA + 50 μM oleate (2:1 oleate/BSA ratio). Values were derived from ten sets of triplicates for each group using hepatocytes from 5 mice of each genotype. The bars represent mean oxidation rates ( n = 100) normalized to the condition of PGC-1α +/+ in BSA alone. * p < 0.05 compared to the corresponding PGC-1α +/+ mice. † p < 0.05 compared to PGC-1α +/+ with BSA treatment. (C) State 2 and 3 respiration rates determined for hepatocytes isolated from PGC-1α +/+ ( n = 3) and PGC-1α −/− ( n = 3) mice using succinate/rotenone as a substrate. * p < 0.05 compared to corresponding PGC-1α +/+ . (D) TAG synthesis rates in isolated hepatocytes. The bars represent mean TAG synthesis rates (glycerol incorporation, see Materials and Methods ) for hepatocytes isolated from PGC-1α +/+ ( n = 6) and PGC-1α −/− ( n = 6) mice. * p < 0.05 compared to the corresponding PGC-1α +/+ condition. PPAR, a known regulator of hepatic mitochondrial fatty acid oxidation enzyme gene expression, is a target for coactivation by PGC-1α [ 31 ]. Therefore, we sought to determine whether the steatotic phenotype of the PGC-1α −/− mice related to reduced expression of PPAR target genes. To this end, a survey of candidate genes and gene expression profiling experiments were performed. Surprisingly, the hepatic expression of PPAR target genes involved in cellular fatty acid and oxidation (MCPT and MCAD) were not significantly different between the genotypes under fed or fasted conditions ( Table 2 ). Next, we performed experiments to determine whether the reduced capacity for fat oxidation in the hepatocytes of the PGC-1α −/− mice was related to altered mitochondrial respiratory function. Compared to the WT controls, PGC-1α −/− hepatocytes exhibited a modest but significant reduction in both state 2 and state 3 respiration rates ( Figure 8 C). These results identify one potential mechanism responsible for the fasting-induced hepatic steatosis: reduced capacity for fat oxidation due to mitochondrial respiratory dysfunction. Table 2 Metabolic Gene Expression in Liver of Fed and Fasted PGC-1α +/+ and PGC-1α −/− Mice Values represent mean (± SEM) ( n ≥ 6 for each group) mRNA levels as determined by real-time RT-PCR corrected for GAPDH signal intensity and normalized (to 1.0) to the value of fed PGC-1α +/+ mice a p < 0.05 versus fed mice of the same genotype b p < 0.05 versus PGC-1α +/+ mice of the same dietary treatment L-CPT I, liver-type carnitine palmitoyltransferase; MCAD, medium-chain acyl-CoA dehydrogenase; SREBP, sterol regulatory element binding protein; SCD, steroyl-CoA desaturase; FAS, fatty acid synthase; GPAT, glycerol-3-phosphate acyltransferase; DGAT, diacylglycerol acyltransferase Although the liver gene expression profiling studies did not reveal abnormalities in the fatty acid oxidation pathway in the PGC-1α −/− mice, several interesting differences in the activity of the sterol regulatory element binding protein-1c (SREBP-1c) pathway were noted. Specifically, the fasting-mediated down-regulation of SREBP-1c and its target gene stearoyl-CoA desaturase (SCD1), was abolished in PGC-1α −/− mice ( Table 2 ). Furthermore, expression of the gene encoding diglyceride acyltransferase (DGAT), which catalyzes the last step in TAG synthesis, was activated at baseline and induced by fasting to a greater level in PGC-1α −/− mice ( Table 2 ). These results suggest that, in addition to a defect in oxidation, components of the TAG synthesis pathway are activated in the PGC-1α −/− mice. To evaluate this possibility directly, rates of 3 H-glycerol incorporation into TAG were determined in isolated hepatocytes. 3 H-TAG incorporation was increased nearly 50% in hepatocytes isolated from PGC-1α −/− mice compared to PGC-1α +/+ controls ( Figure 8 D), confirming that TAG synthesis rates are increased in PGC-1α null hepatocytes, identifying a second potential mechanism contributing to the fasting-induced hepatic steatosis. Despite a Mild Obese Phenotype, Female PGC-1α −/− Mice Do Not Exhibit Insulin Resistance Recent studies have suggested that specific PGC-1α single nucleotide polymorphisms and haplotypes may influence the development of insulin resistance and diabetes [ 27 , 30 ] and that PGC-1 activity is diminished in insulin-resistant and diabetic muscle [ 22 , 23 ]. Accordingly, peripheral glucose disposal and insulin responsiveness were examined in PGC-1α −/− mice. Glucose tolerance testing of 2-mo-old male and female mice revealed no significant difference in blood glucose excursion between PGC-1α +/+ and PGC-1α −/− groups (unpublished data). Given that older female PGC-1α −/− mice develop an increase in body fat stores, glucose tolerance and insulin responsiveness were further evaluated in this group. Glucose tolerance testing in 4.5-mo-old female PGC-1α −/− mice revealed that, despite increased body weight [mean ± standard error of the mean (SEM) weight of PGC-1α +/+ mice = 22.4 ± 0.79 g; PGC-1α −/− mice = 25.2 ± 1.04 g), PGC-1α −/− mice exhibited similar levels of glucose tolerance compared to WT mice on standard rodent chow ( Figure 9 ). To examine glucose homeostasis in response to high-fat diet, female PGC-1α +/+ and PGC-1α −/− mice were placed on high-fat chow (43% calories from fat) for 6 wk starting at 8 wk of age. The weight gained on the high-fat diet was similar for the PGC-1α +/+ and PGC-1α −/− groups ( Figure S3 ). Surprisingly, the PGC-1α −/− mice on a high-fat diet were significantly more glucose-tolerant and insulin-sensitive compared to the PGC-1α +/+ mice ( Figure 9 B). Taken together, these results indicate that, despite excess body fat under standard conditions, the female PGC-1α −/− mice do not exhibit insulin resistance. Moreover, the PGC-1α −/− mice are more glucose-tolerant and insulin-sensitive than WT mice on a high-fat diet. Figure 9 Female PGC-1α −/− Mice Are More Glucose Tolerant and Insulin Sensitive Compared to PGC-1α +/+ on High-Fat Diet (A) At 4.5 mo of age, glucose tolerance testing (GTT) was performed on female PGC-1α +/+ ( n = 6) and PGC-1α −/− ( n = 6) mice maintained on standard chow. (B) At 8 wk of age, PGC-1α +/+ ( n = 8) and PGC-1α −/− ( n = 11) mice were provided a diet containing 43% of its calories from fat (HF chow). The graphs depict blood glucose levels ± SEM in PGC-1α −/− mice during GTT (left graph) and ITT (right graph) studies. Studies were performed 5 wk (GTT) and 6 wk (ITT) after the initiation of the high-fat diet. * p < 0.05 compared to PGC-1α +/+ mice at the same time point. Structural Abnormalities of the Central Nervous System in PGC-1α −/− Mice In surveying the tissues of the PGC-1α −/− mice, structural abnormalities of the brain were observed. Light microscopic examination of PGC-1α −/− brain tissue samples demonstrated a well-preserved cerebral cortical neuronal complement, a result confirmed by measurement of neuron density in sections of the parietal lobe (PGC-1α +/+ , 1,261 ± 91 neurons/mm 2 versus PGC-1α −/− , 1,299 ± 82 neurons/mm 2 ; not significant). Patchy areas of microvacuolation involving the neuropil and individual pyramidal neurons of the deep layers of the cerebral cortex were noted in the PGC-1α −/− mice but not the PGC-1α +/+ mice ( Figure 10 A). Immunolocalization of an astrocytic marker, glial fibrillary acidic protein, failed to show an increase in numbers of astrocytic processes in PGC-1α −/− mouse cerebral cortex (unpublished data). The hippocampus also showed neuronal microvacuolation, albeit to a lesser degree than the parietal cortex. Microvacuolation of the neuropil and neurons of the PGC-1α −/− basal ganglia (caudate and putamen) was also noted in association with a patchy increase in the number and intensity of glial fibrillary acidic protein-immunoreactive astrocytic processes (unpublished data). Areas of microvacuolation also involved multiple brainstem regions. Only rare vacuolated Purkinje and granule cell neurons were identified in the PGC-1α −/− cerebellar cortex. Neither microglial proliferation nor perivascular lymphocytic inflammatory infiltrates were noted in the PGC-1α −/− CNS. Figure 10 Neuropathology of the Central Nervous System of PGC-1α −/− Mice (A) Light microscopic appearance of representative cerebral cortex of 2-mo-old PGC-1α −/− mice demonstrates marked vacuolation of the neuropil (arrows) and scattered neuronal perikarya, which are absent in PGC-1α +/+ mice (hematoxylin and eosin). The scale bar shown is applicable to all sections. (B) Ultrastructural appearance of typical vacuoles containing membranous debris, denoted by the arrow, in the cerebral cortex of a representative PGC-1α −/− mouse in comparison to PGC-1α +/+ (magnification 4000×). Ultrastructural examination of the PGC-1α −/− parietal cerebral cortex confirmed the presence of microvacuolated neurons and neuropil ( Figure 10 B). Vacuoles containing aggregates of membranous material were present in a subset of cortical neurons. Subcellular localization of the vacuoles was difficult to establish; some may represent vacuolated elements of the neuropil, material in phagocytic cells, presynaptic nerve terminals compressing the soma, or genuine intraperikaryal deposits. Discussion Previous studies using gain-of-function strategies have shown that the coactivator PGC-1α is capable of coactivating an array of transcription factors involved in energy metabolic processes including fatty acid oxidation, electron transport, and oxidative phosphorylation [ 6 ]. Forced expression of PGC-1α triggers mitochondrial biogenesis by activating a complex circuitry of factors including NRF-1, NRF-2, and the orphan nuclear receptor estrogen-related receptor α [ 23 , 32 ]. However, gain-of-function strategies cannot determine whether PGC-1α is essential for critical energy metabolic processes including mitochondrial biogenesis and function. Using targeted gene deletion in mice, we show here that PGC-1α is not essential for normal embryologic development or the fundamental events of mitochondrial biogenesis. However, several lines of evidence support the conclusion that PGC-1α is necessary for the programs that regulate postnatal mitochondrial function and cellular energy metabolism, processes that equip the organism for the energy metabolic rigors of the postnatal environment. First, mitochondrial volume density is diminished in slow-twitch skeletal muscle of PGC-1α −/− mice. Second, mitochondrial respiratory capacity is modestly but significantly altered in skeletal muscle and liver of PGC-1α −/− mice. Third, the growth of heart and soleus muscle, tissues with high reliance on mitochondrial energy production, is blunted. Fourth, control of body fat mass is abnormal in the PGC-1α −/− mice. Finally, PGC-1α −/− mice do not respond normally to a variety of physiologic and dietary stresses known to increase oxidative energy demands. Taken together, these results strongly suggest that PGC-1α is necessary for the terminal stages of mitochondrial maturation necessary to meet the energy demands of the postnatal environment. Extensive phenotypic analyses demonstrated that mice lacking PGC-1α are unable to cope with physiologic stressors relevant to postnatal survival. For example, a skeletal muscle phenotype was unveiled in PGC-1α −/− mice under conditions in which energy supply becomes limiting. This was most clearly demonstrated by the profound abnormalities exhibited by PGC-1α −/− mice with exercise-to-exhaustion and repetitive muscle stimulation studies. Similarly, cardiac performance of PGC-1α −/− mice was compromised following severe exertion. This effect was largely due to an abnormal heart rate response. The basis for the observed abnormalities of cardiac heart rate, including a blunted response to β-adrenergic stimulation, is unknown, but could be related to the effects of late-stage growth arrest and corresponding derangements in energy metabolism on sinus node function. PGC-1α was first identified as a coactivator in BAT [ 2 ]. Indeed, we found that exposure of the PGC-1α −/− mice to cold, another relevant physiologic stress, resulted in an untoward drop in core body temperature consistent with an abnormality in thermogenesis despite normal cold induction of UCP-1 mRNA in BAT. Studies with a β 3 -adrenergic agonist confirmed that the peak oxygen consumption rate in thermogenic tissue is diminished in PGC-1α −/− mice. We propose that the thermogenic phenotype is related to reduced capacity for mitochondrial respiration in BAT. Interestingly, this phenotype was only evident during a rather narrow window of postnatal life. Animals at an older age did not exhibit cold intolerance, possibly due to the insulating properties of increased body mass. Collectively, these results demonstrate the importance of PGC-1α as a key transducer of physiologic stimuli to the control of energy metabolism. The observation of fasting-induced hepatic steatosis is another example of the inability of PGC-1α −/− mice to respond to postnatal environmental metabolic demands. Following short-term starvation, we found that the PGC-1α −/− mice developed marked hepatocyte triglyceride accumulation. Further analysis revealed that palmitate oxidation rates were reduced in hepatocytes isolated from the PGC-1α −/− mice, which would predispose to lipid accumulation. Surprisingly, the reduction in fatty acid oxidation rates in PGC-1α null hepatocytes was not due to altered expression of PGC-1α/PPAR target genes involved in mitochondrial fatty acid oxidation. However, mitochondrial respiratory rates were diminished. In addition, we found that triglyceride synthesis was abnormally activated, and the expression of genes encoding SREBP-1c and SCD-1, key proteins in the hepatic lipogenic pathway, failed to be appropriately down-regulated in fasted PGC-1α −/− mice. The mechanism involved in this latter finding is unknown. Indeed, the relative contribution of increased triglyceride synthesis rates to the steatotic phenotype cannot be fully discerned from our data, given that this response could reflect the direct effects of PGC-1α deficiency on target genes or a secondary compensatory response to hepatocyte fatty acid accumulation. Consistent with the former possibility, recent evidence indicates that PGC-1α coactivates the nuclear receptor FXR, a negative regulator of SREBP-1c expression and triglyceride synthesis [ 33 ]. We conclude that reduced hepatocyte mitochondrial respiratory capacity, and possibly activation of lipogenic programs, result in hepatocyte triglyceride accumulation in the context of increased hepatic delivery of fatty acids such as occurs with fasting. We found that after 18 wk of age, female PGC-1α −/− mice exhibit a mild but significantly abnormal weight increase associated with increased fat stores. Lean mass was unchanged at the time points examined. With further aging, a modest but significant increase in body fat was also noted in male PGC-1α −/− mice (unpublished data). The basis for the observed abnormalities in weight control is unknown. We did not find differences in food intake or activity levels in female PGC-1α −/− mice. It is possible that a reduction in systemic energy utilization, related to the mitochondrial dysfunction, leads to increased fat mass and weight gain in the PGC-1α −/− mice. Interestingly, an association between PGC-1α gene polymorphisms and obesity in humans has been recently reported [ 26 ]. Clearly, future studies of male and female PGC-1α −/− mice in pure-strain backgrounds over a range of ages will be necessary to fully investigate the observed abnormalities in weight control and fat distribution. We did not find evidence for glucose intolerance or insulin resistance in the PGC-1α −/− animals on standard chow. Moreover, female PGC-1α −/− mice were more glucose-tolerant and insulin-sensitive than PGC-1α +/+ controls when consuming a high-fat diet. These findings are surprising, given the results of several recent studies demonstrating reduced expression of PGC-1α in human diabetic skeletal muscle [ 24 , 34 ]. It is certainly possible that compensatory metabolic regulatory mechanisms have been activated in the PGC-1α-deficient mice, accounting for this observation. Alternatively, PGC-1α could serve as a coactivator of factors that mediate diet-induced insulin resistance. Consistent with this notion, we and others have shown that mice lacking the PGC-1α target PPAR exhibit resistance to diet-induced glucose intolerance [ 21 , 35 , 36 ]. Histologic surveys of the PGC-1α −/− mice revealed ultrastructural abnormalities in the central nervous system. Inspection of sections prepared from the brains of PGC-1α −/− mice revealed patchy areas of microvacuolation in the pyramidal neurons of the cerebral cortex, accompanied by a mild increase in the number of astrocytes in the basal ganglia. The basis for this interesting but relatively nonspecific finding is unknown. It is possible that PGC-1α plays an important role in lipid metabolism related to membrane synthesis. Alternatively, the normal process of cellular debris turnover could be altered due to a defect in the energetics of the microglial component of the central nervous system. Although overt neurologic dysfunction was not apparent in PGC-1α −/− mice during the first 6 mo of life (no group differences were found on five of six sensorimotor tests), the PGC-1α −/− mice showed clear deficits on the inverted screen test. These deficits are likely due to impaired muscle strength in the PGC-1α −/− mice, but contributions by peripheral or central nervous system determinants (or both) could be contributory. Moreover, evidence of altered emotionality from the 1-h locomotor activity test also suggests the possibility of altered brain function in PGC-1α −/− mice. It will be of interest to determine whether the neurologic abnormalities contribute to the systemic metabolic abnormalities of the PGC-1α null mice. During the preparation of this manuscript, Lin et al. reported an independent mouse line in which the PGC-1α gene was targeted [ 37 ]. Phenotypic comparison of the our PGC-1α-deficient line with that of Lin et al. reveals a number of similarities and several interesting differences. Both PGC-1α-deficient lines exhibit cold intolerance, reduced hepatocyte respiration rates, and neurologic lesions. However, a number of interesting differences are notable. First, in contrast to Lin et al., the PGC-1α −/− mice described here do not exhibit any postnatal mortality. Second, we did not find evidence for a defect in gluconeogenesis based on fasting blood glucose levels (unpublished data). In addition, whereas Lin et al. found an abnormal expression profile for CCAAT-enhancer-binding protein β and δ and the gluconeogenic genes encoding phosphoenolpyruvate carboxykinase and glucose-6-phosphatase at baseline and with fasting in the PGC-1α −/− mice, we did not (unpublished data). Third, we found evidence for an age-related increase in body fat in PGC-1α −/− mice (females earlier than males), whereas Lin et al. identified a male-specific resistance to diet-induced obesity and insulin resistance. We have also found that male PGC-1α-deficient mice are somewhat protected against diet-induced obesity ( Figure S4 ). However, we observed that the insulin-sensitive phenotype of the female PGC-1α −/− mice occurred in the context of normal weight gain with high-fat diet. These latter results indicate that the insulin-sensitive phenotype of PGC-1α −/− mice cannot be fully explained by a lean phenotype. Of interest, mice lacking the nuclear receptor estrogen-related receptor α, a known target of PGC-1α, exhibit resistance to diet-induced obesity similar to that of male PGC-1α null mice [ 38 ]. Fourth, the PGC-1α −/− mice described here exhibit a dramatic fasting-induced hepatic steatotic phenotype, whereas the Lin et al. mouse does not. Fifth, Lin et al. found a neurologic phenotype in males characterized by hyperactivity, whereas the PGC-1α −/− mice described here show reduced locomotor activity. However, it should be noted that we did not study activity levels over an extended period of time in males as did Lin et al., so it is possible that our findings reflect an emotional disturbance that manifests only when the animals are placed in a new environment. Finally, we report significant skeletal muscle and cardiac functional abnormalities (although the report by Lin et al. did not address these phenotypes, so this may not represent a true difference). The reasons for the interesting differences between the two PGC-1α-deficient mouse lines are not clear. It is possible that distinct genetic backgrounds related to hybrid strains confer different degrees of secondary compensatory responses. In addition, the incompletely penetrant postnatal mortality noted in the PGC-1α −/− mice reported by Lin et al. could have resulted in a selection bias toward greater levels of compensatory responses in liver and other tissues in the surviving group. It is also possible that the method of gene targeting led to different phenotypes. Lin et al. generated PGC-1α −/− mice by Cre recombinase-mediated excision of exons 3–5 in oocytes. The PGC-1α −/− mice described here were generated by a targeting event that involved a 3′ homologous recombination leading to an insertion of the targeting vector including an extra exon 3 between exons 5 and 6. The exon 3 insertion, which was confirmed by RT-PCR, results in a mutant transcript that encodes a truncated protein. We were unable to detect normal transcript containing an exon 5–6 border, indicating that the targeting was accurate and complete. In addition, we could not detect full-length or smaller PGC-1α proteins by Western blotting. However, we cannot exclude the possibility that the sensitivity of the immunoblotting was not high enough to pick up a small amount of mutant (truncated) PGC-1α protein that could have some activity, given that it would contain nuclear receptor-interacting domains and the amino-terminal activation domain. If small amounts of PGC-1α activity are present in the mice reported here, it could explain some of the observed differences between the models. However, the bulk of data presented here support the conclusion that the PGC-1α −/− mice described are completely deficient in PGC-1α. Future direct comparison of the two mouse lines in pure background strains will be of interest. In summary, this body of work provides evidence that PGC-lα is critical for the adaptive responses necessary to meet postnatal energy demands. Our results also suggest a broader role for inducible transcriptional coactivators such as PGC-1α in transducing cellular signals triggered by physiologic and developmental cues to the transcriptional control of energy metabolism and other dynamic cellular processes. In this regard, the inducible coactivator PGC-1α serves as a transcriptional “booster” to augment the capacity of downstream metabolic pathways critical for metabolic maturation and postnatal growth. Indeed, although PGC-1α null mice survive in the protected environment of the laboratory, our results indicate that in the rigors of a typical external environment, PGC-1α would be necessary for survival. Lastly, we propose that the PGC-1α −/− mice should serve as a useful murine model to investigate the role of altered energy metabolism in obesity, diabetes, hepatic steatosis, and diseases of the heart, skeletal muscle, and central nervous system. Materials and Methods Targeting the PGC-1α gene in mice A BAC genomic clone containing the murine PGC-1α gene, isolated from an Sv129 genomic library, was obtained from Incyte Genomics (Palo Alto, California, United States). A 3-kb region spanning exon 3 was amplified from the genomic clone. A 5′ primer was designed to amplify a fragment with the 5′ end beginning 732 bp upstream of exon 3 just upstream of an endogenous Kpn1 restriction site (5′- AGTTTCCTTAGCAACTTCATA-3′). The 3′ primer contained a BamH1 site engineered by mutating the bases shown in lowercase (5′- AAGGATTTTAgGATcc CAGTAC-3′). A second fragment downstream of exon 5 was amplified. In this latter amplicon, Not1 and Xho1 sites were engineered into the 5′ and 3′ primers, respectively (5′- TGGAGTgc GGCCGCTGGGA-3′ and 5′- AAAGAGTCTCgAg AATAGTTTCT-3′). The fragments were cloned into p1339-PGK-Neomycin targeting vector. The construct was linearized with Xho1 and electroporated into RW4 ES cells (Sv129 derived) using G418 selection. The electroporation was performed by the Siteman Cancer Center ES Cell Core at Washington University (St. Louis, Missouri, United States). The clones were screened by Southern blot using an Xba1 digest (see Figure 1 A and 1 B). One clone out of approximately 400 screened was positive for the homologous recombination on the 3′ end and an insertion on the 5′ end. This clone was injected into a C57BL6/J blastocyst. Chimeras were mated to C57BL6/J mice and germline transmission was confirmed by Southern blotting of tail DNA (see Figure 1 B). All experiments were performed using sex- and age-matched or littermate controls as noted. General animal studies All animal experiments and euthanasia protocols were conducted in strict accordance with the National Institutes of Health guidelines for humane treatment of animals and were reviewed and approved by the Animal Care Committee of Washington University. Animals were weighed at different time points. Male and female 3- to 8-wk-old PGC-1α +/+ and PGC-1α −/− mice were euthanized, and tissues were dissected and weighed on an analytical balance. Tissue weights were corrected for total body weight before comparison. DEXA studies were performed as previously described [ 39 ] using a Lunar PIXIMUS DEXA system at 10, 18, and 24 wk in male and female PGC-1α +/+ and PGC-1α −/− mice. For cold exposure experiments, male and female PGC-1α +/+ and PGC-1α −/− mice were singly housed and placed at 4 °C for 5 h without food. Core body temperatures were monitored by rectal probe at baseline and every hour thereafter. Mice were monitored at least every 30 min to check for lethargy. At the end of 5 h, mice were sacrificed and tissues harvested for RNA and protein extraction. For fasting studies, animals were singly housed and given water ad libitum. Food was removed from cages in the morning and tissues harvested at 24 h for RNA and histology. Photography of the mice was performed by MedPic at Washington University School of Medicine. 48-h activity monitoring was performed by JAX Services (The Jackson Laboratory, West Sacramento, California, United States) using a Comprehensive Laboratory Animal Monitoring System (CLAMS, Columbus Instruments, Columbus, Ohio, United States). Briefly, 3-mo-old female mice were acclimated for 17 h before data collection. Data were collected every 30 min. Total beam breaks in the XY direction were tabulated for the 12-h light and dark cycles and compared across genotypes. RNA, DNA, and protein analyses Total RNA was isolated by the RNAzol method (Tel-Test, Friendswood, Texas) and Northern blotting was performed as previously described [ 40 ]. The PGC-1β and PRC cDNAs were generous gifts from Bruce Spiegelman and Richard Scarpulla, respectively. The UCP-1 cDNA was a gift from Daniel Ricquier. RT-PCR was performed as described [ 41 ]. In brief, total RNA isolated from soleus muscle, BAT, and heart of 1–2-mo-old PGC-1α +/+ or PGCα −/− mice was reverse transcribed with Taqman reverse transcription reagents (Applied Biosystems, Foster City, California, United States). Reactions were performed in triplicate in 96-well format using Taqman core reagents and a Prism 7500 Sequence Detector (Applied Biosystems). The mouse-specific primer-probe sets used to detect specific gene expression can be found in Table S1 . The primers for UCP-1 have been previously described [ 42 ]. Actin primer-probe set (Applied Biosystems) was included in a separate well and used to normalize the soleus, BAT, and heart gene expression data. GAPDH Rodent primers (Applied Biosystems) were used in the same well to normalize the liver gene expression data. For Southern blot studies, 5 μg of genomic DNA was digested with Pst1 or Xba1, electrophoresed on a 0.8% TAE gel and transferred to a Gene Screen (Perkin Elmer, Wellesley, California, United States) membrane for hybridization. Western blotting was performed as described [ 43 ] using the Enhanced Chemiluminescence detection system (Amersham Pharmacia Biotech, Piscataway, New Jersey, United States). Ponceau S (Sigma, St. Louis, Missouri, United States) staining of the membrane was used as a control. Mitochondrial respiration studies Mitochondrial respiration was assessed in saponin-skinned soleus fibers with succinate as substrate and in the presence of rotenone as previously described [ 44 ]. In brief, 3-mo-old female mice were anesthetized with chloral hydrate (400 mg/kg of body weight). Soleus fibers were separated and then transferred to a buffer (2.77 mM K 2 Ca-EGTA, 7.23 mM K 2 EGTA, 6.56 mM MgCl 2 , 20 mM imidazole, 53.3 mM K-MES, 20 mM taurine, 5.3 mM ATP, 15 mM PCr, 3 mM KH 2 PO 4 , 0.5 mM DTT [pH 7.1]) supplemented with 50 μg/ml saponin and permeabilized for 30 min at 4 °C with gentle stirring. Fibers were then washed twice for 10 min each (2.77 mM K 2 Ca-EGTA, 7.23 mM K 2 EGTA, 1.38 mM MgCl 2 , 20 mM imidazole, 100 mM K-MES, 20 mM taurine, 3 mM KH 2 PO 4 , 0.5 mM DTT, 2 mg/ml BSA [pH 7.1]). Respiration was measured at 25 °C using an optical probe (Oxygen FOXY Probe, Ocean Optics, Dunedin, Florida, United States). Following measurement of basal state 2 respiration, maximal (ADP-stimulated) state 3 respiration was determined by exposing fibers to 1 mM ADP. The integrity of the outer mitochondrial membrane was assessed by adding 8 μM exogenous cytochrome c to ADP-stimulated fibers. State 4 respiration (uncoupled) was evaluated following addition of oligomycin (1 μg/ml). The solubility of oxygen in the respiration buffer at 25 °C was taken as 246.87 nmol O 2 · ml −1 . Respiration rates were expressed as nmol O 2 · min −1 · mgdw −1 . Insulin and glucose tolerance tests Glucose and Insulin tolerance tests were performed as described [ 35 ]. Prior to studies, mice were fasted overnight (GTT) or 6 h (ITT). In GTT studies, mice were injected with a 10% solution of D-glucose (1 g/kg). For ITT, mice received an IP injection of human regular insulin (Eli Lilly, Indianapolis, Indiana, United States) at a dose of 0.75 units/kg of body weight. Tail blood glucose was determined at 0, 30, 60, and 120 min after challenge using a B-GLUCOSE Analyzer (Hemacue AB, Angelholm, Sweden). Indirect calorimetry Oxygen consumption rates (VO 2 ) of 5-wk-old female mice were measured using a Columbus Instruments Oxymax System. Resting baseline oxygen consumption rates were assessed for at least 1.0 h. For β 3 -adrenergic stimulation studies, BRL 37344 (Sigma) was dissolved in sterile saline and injected IP (2 μg/g of body weight) [ 45 ]. Postagonist assessment of oxygen consumption was recorded for an additional 1.0 h, with data collected at the 40-min time point. Histology and electron microscopy Soleus muscle and liver were dissected and fixed overnight in 2% glutaraldehyde, 1% paraformaldehyde, and 0.08% sodium cacodylate buffer. The tissues were postfixed in 1% osmium tetroxide, dehydrated in graded ethanol, embedded in Poly Bed plastic resin, and sectioned for electron microscopy. Cardiac and skeletal muscle mitochondrial and myofibrillar volume densities were determined from electron micrographs as described previously [ 10 ]. For each animal, three different fields at the magnification of 7500× were quantified in blinded fashion. Data were expressed as mean volume density of mitochondria or myofibrils in each field. For electron microscopic analysis of the brain, tissue was prepared as previously described [ 46 ]. Ultrathin sections of cortex were cut onto formvar-coated slot grids stained with uranyl acetate and lead citrate and examined with a JEOL 1200 electron microscope. For H&E staining, sections of brain, including cerebral cortex, brainstem, and cerebellum, were dehydrated in graded concentrations of alcohol and embedded in paraffin from which 5-μm sections were prepared. Primary mouse hepatocyte studies Primary cultures of mouse hepatocytes were prepared from male PGC-1α +/+ and PGC-1α −/− mice essentially as described [ 47 ]. Fatty acid oxidation and triglyceride synthesis experiments were commenced 2–3 h after the cells were plated. Triglyceride synthesis studies, were performed as previously described [ 47 ]. Palmitate oxidation rates were quantified using [9,10- 3 H]-palmitic acid as described [ 48 ] and corrected for total cellular protein content. For respiration studies, cells were spun down prior to plating and resuspended in a permeabilization buffer (described above) containing 50 μg/ml saponin. Respiration studies were performed in the presence of 5 mM succinate in the presence of 10 μM rotenone. Respiration rates were expressed as nmol O 2 · min −1 · mg of protein −1 . Evaluation of locomotor activity, sensorimotor capabilities, and muscle function To evaluate general activity levels and muscle use, mice were evaluated over a 1-h period in transparent (47.6 cm × 25.4 cm × 20.6 cm) polystyrene enclosures as previously described [ 49 ] using a high-resolution photobeam system (Motor Monitor, Hamilton-Kinder, Poway, California, United States). Each enclosure was surrounded by a frame containing a 4 × 8 matrix of photocell pairs, the output of which was fed to an on-line computer. The system software (Hamilton-Kinder) was used to define a 33 cm × 11 cm central zone and a peripheral or surrounding zone that was 5.5 cm wide with the sides of the cage being the outermost boundary. This peripheral area extended along the entire perimeter of the cage. Variables that were analyzed included the total number of ambulations, as well as the number of entries, the time spent, and the distance traveled in the center area as well as the distance traveled in the periphery surrounding the center. The total number of ambulations and rearings were recorded. For the inverted screen test, mice were placed on a wire mesh grid (16 squares per 10 cm) and the screen was inverted to 180°. A maximum score of 60 s was given if an animal did not fall. The tests included in the sensorimotor battery [ 50 ] and accompanying protocols were designed as follows. (1) Inclined screen and inverted screen tests: For the inclined screen tests, each mouse was placed on top of an elevated (47 cm above the floor) wire mesh grid (16 squares per 10 cm) that was inclined to 60° or 90°. Each animal was placed in the middle of the screen with its head oriented down and was timed for how long it remained on the screen and how long it took to climb to the top of the screen. For the inverted screen test, mice were placed as above and then the screen was inverted to 180°. A maximum score of 60 s was given if an animal did not fall; (2) Platform test: Each mouse was timed for how long it remained on an elevated (47 cm above the floor) circular platform (1.0 cm thick and 3.0 cm in diameter). A maximum score of 60 s was assigned if the mouse remained on the platform for the maximum amount of time or if it could climb down on a very thin pole that supported the platform, without falling; (3) Ledge test: Each mouse was timed for how long it could maintain its balance on a 0.75-cm wide Plexiglas ledge without falling (60 s maximum). A score of 60 s was also assigned if the mouse traversed the entire length (51 cm) of the Plexiglas ledge and returned to the starting place in less than 60 s without falling; (4) Walking initiation test: Each mouse was placed in the middle of a square outlined by white cloth tape (21 cm × 21 cm) on a smooth black surface of a large tabletop. The time it took each mouse to leave the square (place all four paws outside of the tape) was recorded. The maximum time allowed was 60 s. 6–8-mo-old PGC-1α +/+ ( n = 4) and PGC-1α −/− ( n = 8) mice were run to exhaustion employing a motorized, speed controlled, modular treadmill system (Columbus Instruments). The treadmill was equipped with an electric shock stimulus and an adjustable inclination angle. Running velocity was set at 35 m/min, with a level inclination angle. VO 2max studies VO 2max was determined while the mice were running on a treadmill using an open flow system (Columbus Instruments Oxymax System). All measurements of oxygen consumption took place at an elevation of 150 m (ambient P BAR = 745 torr). Animals were placed into the metabolic chamber for 3–5 min to allow the system to equilibrate. Mice were then induced to run up an 18° incline at a speed of 40 m/min using a shock grid in the rear of the chamber. The speed was increased by 5 m/min every 2 min until the animals were unable to continue. Maximal effort was determined when oxygen uptake did not increase with power output and subsequently the mouse failed to maintain effort. VO 2max was calculated using the averaged values over 1 min during which the animal's O 2 consumption reached a plateau. Isolated muscle stimulation studies Animals were anesthetized with ketamine/xylazine and the soleus muscle was removed from one leg. Upon removal, the muscle was suspended in a Krebs solution aerated with 95% O 2 and 5% CO 2 . The muscle and Krebs solution were suspended within a water bath maintained at 37 °C, and the muscle was anchored to a Grass (West Warwick, Rhode Island, United States) isometric force transducer (model FTO3C). Muscles were stimulated to contract with a Grass stimulator (model S88) generating a field stimulus through electrodes located at both ends of the muscle. Force-voltage (maximal force at about 100 V) and length-tension relationships were determined using single twitch stimuli. The stimulator then delivered repeating trains of stimuli at one per second at 40 Hz for 2 min. Each train lasted 330 ms, and were digitally recorded using MacLab (AD Instruments, Colorado Springs, Colorado, United States). Fatigue resistance was calculated as the ratio of the force generated by the last tetanus divided by the highest force generated multiplied by 100 to give the percent of force generation that remained after the fatigue protocol. Exercise echocardiography Adult female mice (6–8 mo old) were exercised on the motorized treadmill using the run-to-exhaustion settings described above for 60 s or until exhaustion, whichever came first. Immediately following its treadmill run, the mouse was subjected to serial echocardiography using an Acuson Sequoia Echocardiography System performed as previously described [ 51 ]. In vivo cardiac hemodynamic studies Hemodynamic studies were performed as previously described with some modifications [ 52 ]. In brief, adult mice (10–12 wk) were anesthetized intraperitoneally (IP) with thiopental sodium (60 mg/kg). The mice were intubated and ventilated with a Harvard ventilator. The right carotid artery was isolated in the region of the trachea and cannulated with a 1.4-French high-fidelity micromanometer catheter (Millar Instruments, Houston, Texas, United States), which was inserted into the left ventricle retrograde across the aortic valve. Hemodynamic measurements were recorded at baseline and 3 min following continuous infusion of incremental doses of dobutamine (β 1 , β 2 , and α 1 -adrenergic agonist) up to 32 ng · gBW −1 · min −1 [ 53 ]. Continuous pressure-volume data were acquired and digitized with the BioBench computer software data acquisition system (National Instruments, Austin, Texas, United States). Isolated working mouse heart perfusion Isolated working mouse heart perfusion was based on a previously described procedure [ 54 ]. Adult mice (4–7 mo old) were heparinized (100 units IP) 10 min prior to anesthesia. Animals were then deeply anesthetized with 5–10 mg of sodium pentobarbital IP. Hearts were excised and placed in an ice-cold Krebs-Henseleit bicarbonate (KHB) solution [118 mM NaCl, 25 mM NaHCO 3 , 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 2.5 mM CaCl 2 , 5.0 mM glucose, and 100 units/L insulin (pH 7.4)]. Hearts were cannulated first via the aorta and perfused retrogradely by the Langendorff method. Following left atrial cannulation, perfusion was switched to the working mode with KHB solution containing 1.2 mM palmitate bound to 3% fatty acid-free BSA with a preload pressure of 11.5 mm Hg and an afterload pressure of 50 mm Hg for 60 min with oxygenated buffer solution. Functional measurements, namely cardiac output, aortic flows, peak systolic pressure, and heart rate were acquired every 10 min using inline flow probes (Transonic Systems, Ithaca, New York, United States), a pressure transducer (TSD 104A, BIOPAC Systems, Santa Barbara, California, United States) and data acquired with the MP100 system from AcqKnowledge (BIOPAC Systems). Cardiac work was calculated as the product of peak systolic pressure and cardiac output. Statistics Data were analyzed using T-tests or ANOVAs (measures of general activity and sensorimotor battery). The level of significance was set at p < 0.05 in all cases. Data are reported as mean values ± SEM, unless otherwise noted. The ANOVA model used to analyze each sensorimotor test included one between-subjects variable (genotype), and one within-subjects variable (trials). When ANOVAs with repeated measures were conducted, the Huynh-Feldt (H-F) adjustment of alpha levels was used for all within-subjects effects containing more than two levels, in order to protect against violations of the sphericity/compound symmetry assumptions underlying this ANOVA model. In addition, Bonferroni correction was used when appropriate to help maintain prescribed alpha levels (e.g., p < 0.05) when multiple comparisons were conducted. Supporting Information Figure S1 Activity Levels in Female PGC-1α −/− Mice Is Unchanged Using a CLAMS system, PGC-1α +/+ ( n = 4) and PGC-1α −/− ( n = 3) female mice were monitored for 48 h after a 17-h period of acclimation. XY beam breaks were tabulated over the 12-h light and dark cycles as denoted on the bottom. The bars represent mean (± SEM) beam breaks per each 12-h cycle. (342 KB EPS). Click here for additional data file. Figure S2 Altered Emotionality in PGC-1α −/− Mice An analysis of exploratory behavior included the number of entries into the center of the cage (upper left), the time spent in the center of the cage in seconds (sec) (upper right), the distance traveled in the center of the cage in meters (m) (lower left) as well as the distance traveled in the periphery (lower right). * p < 0.05 compared to the PGC-1α +/+ mice. (620 KB EPS). Click here for additional data file. Figure S3 No Difference in Weight Gain on a High-Fat Diet in Female PGC-1α −/− Mice Compared to WT Controls 8-wk-old female mice were fed a diet high in fat (43% calories from fat) for 6 wk. The change in body weight (grams) after 6 wk on a high-fat diet is shown for PGC-1α +/+ ( n = 8) and PGC-1α −/− ( n = 11) mice. NS, not significant. (264 KB EPS). Click here for additional data file. Figure S4 Male PGC-1α −/− Mice Are Somewhat Resistant to Diet-Induced Obesity Male and female PGC-1α +/+ ( n ≥ 6) and PGC-1α −/− ( n ≥ 6) mice were fed a high-fat diet (43% calories from fat) beginning at 4 wk of age. Body weight was monitored weekly as shown on the graph on the left. The mean (± SEM) change in body weight is shown in the bar graph on the right. *, significant difference compared to the PGC-1α +/+ controls, p < 0.05. (794 KB EPS). Click here for additional data file. Table S1 Probes and Primers Sequences of mouse-specific probes and primers used for real-time RT-PCR. (25 KB DOC). Click here for additional data file. Accession Numbers The GenBank ( http://www.ncbi.nlm.nih.gov/ ) accession numbers of the vector discussed in this paper is p1339-PGK-Neomycin targeting vector (AF335420).

D:\keerthana\PMC001xxxxxx\PMC1064854.xml

1064855

Heart Repair Gets New Muscle

When you think of the cell as the fundamental unit of life, it's not surprising that some organs deal with injury better than others. A flesh wound or muscle tear might hurt, but, assuming you are otherwise healthy, both will heal. The prognosis for a heart attack, on the other hand, is not so clear-cut. What accounts for the difference? Skin cells reproduce regularly to replace dead cells, and simply increase production in the event of injury. Skeletal muscles recruit new muscle cells from a type of precursor cell within the muscle, called satellite cells, to repair a tear. Cardiac cells (cardiomyocytes), it has long been thought, appear to lack this capacity for self-renewal and repair, impeding the chances of a full recovery. That's why therapies derived from stem cells—which retain a unique ability to morph into any of the body's 200-plus cell types—hold such promise. But stem cells are a hot-button issue in the United States, complicating efforts to explore this promise. Recent evidence suggests that the heart might harbor stem cells after all and that such cells can be transformed into cardiomyocytes. In a new study, Neal Epstein and colleagues report that cells isolated from the skeletal muscle of adult mice can turn into beating cardiomyocytes in a test tube within days of isolation—and without the addition of gene-altering drugs or special cardiac factors. When freshly isolated cells (called skeletal precursors of cardiomyocytes, or Spoc cells) are injected into the tail veins of mice with heart damage, they migrate to the damaged tissue and differentiate into cardiac muscle cells. What distinguishes a cardiomyocyte from a skeletal muscle cell? Specialized cells produce unique proteins, allowing scientists to use those proteins as identifying markers. The so-called Spoc cells do not express any of the usual markers associated with either skeletal muscle satellite cells or partially differentiated skeletal muscle cells. By day 7 in culture, Spoc cells have undergone several rounds of cell division and have begun to express a (mostly) cardiac-specific protein, and have formed clusters of cardiac precursor cells, some of which beat. These precursors in turn express other cardiac-specific proteins. Epstein and colleagues further divided Spoc-derived precursor cells into two groups based on whether or not they expressed another protein marker (Sca-1, a common marker found on blood stem cells). About 80% of cells without this protein differentiated into immature beating cells after proliferating for seven to ten days. They remained in an immature state (round and loosely attached) for over two months in culture, but differentiated into mature beating heart cells (elongated and adherent) when mixed with Sca-1 cells. The authors use video microscopy to track the cells' progression to beating cells, complete with contraction-generating thick myosin filaments that are “nearly identical” to those seen in developing cardiomyocytes. Epstein and colleagues also demonstrate that the Spoc cells are distinct from stem cells cultured out of bone marrow, heart, or fat tissue—sources of beating cells in other studies. The authors also injected these Spoc cells into mice with acute heart lesions to test the cells' ability to integrate into the damaged tissue. Many cells successfully migrated to and engrafted into the site of injury; some of these cells developed into cardiomyocytes. The cells showed a similar, though less robust, response to an older heart injury. Spoc cells can help repair a damaged heart Epstein and colleagues argue that Spoc cells are more likely to be precursors to cardiomyocytes than to be some other type of skeletal muscle stem cell. This is based on an absence of protein markers for skeletal muscle or skeletal satellite cells in Spoc cells, as well as the fact that Spoc-derived cells display spontaneous rhythmic beating and express cardiac markers, whether they are grown in a test tube or have migrated to injured hearts in study mice. The authors can't say why skeletal muscle would harbor cardiac stem cells or why so few of these cells pitch in to repair a cardiac injury. But for now, the Spoc cells provide a valuable tool for studying heart cell differentiation. And with time, they might prove an important resource for developing cell-based therapies for heart disease. See also the Primer “Alchemy and the New Age of Cardiac Muscle Cell Biology” (DOI: 10.1371/journal.pbio.0030131 ).

D:\keerthana\PMC001xxxxxx\PMC1064855.xml

1064856

Gray Wolves Help Scavengers Ride Out Climate Change

Average earth temperatures rose 0.6 °C over the last century, according to the latest Intergovernmental Panel on Climate Change. But that increase pales in comparison to the 1.4–5.8 °C expected increase over this century. As temperatures climb, climate models predict that high-latitude, high-altitude regions like Yellowstone National Park will experience shorter winters and earlier snow melts. How these environmental shifts will impact species and ecosystems remains to be seen. The effects of climate change are already evident at the species level, with disruptions in range, reproductive success, and seasonal phenomena like migration, and the decoupling of evolutionarily paired events like new births and food availability. Both experimental and data-driven modeling studies predict that climate change may well precipitate shifts in the structure of ecosystems as well. In a new study, Christopher Wilmers and Wayne Getz investigated the effects of climate change on ecosystem dynamics by studying a keystone species in Yellowstone, the gray wolf ( Canis lupus ). Gray wolves inhabited most of North America until US extirpation campaigns nearly eradicated them by the 1930s. In 1995, the US Fish and Wildlife Service reintroduced the persecuted predator into Yellowstone. Wilmers and Getz used data from the past 50 years at two weather stations in the park's northern range (where elk over winter and four to six wolf packs now live) to establish winter trends and model wolves' impact on the fate of resident scavengers faced with a changing climate. Not surprisingly, their models show that this top predator exerts significant influence over animals at lower levels in the food chain: wolf kills temper the potentially devastating effects of climate-related carrion shortages on scavengers. Unlike mountain lions and grizzly bears, wolves abandon their prey (usually elk or moose) once sated, leaving much-coveted leftovers for ravens, eagles, coyotes, bears, and other scavengers. These findings indicate that individual species stand a better chance of adapting to climate change in an ecosystem with an intact food chain. Wilmers and Getz's weather data analysis found that both late-winter snow depth and snow-cover duration have decreased significantly since 1948—winters in Yellowstone are getting shorter. That's good news for elk—navigating deep snow taxes stamina and reduces access to forage—but bad news for scavengers that rely on elk carcasses to carry them through the winter. The authors generated two sets of models to estimate the effects of shorter winters on the wolf–elk–scavenger dynamics. In the first, late-winter carrion availability drops by 66% without wolves but by only 11% when the predators are present. The second model examines the impact of elk and wolf population dynamics on carrion availability. This analysis predicts that more elk will die in early winter than in late winter, a scenario that favors eagles and ravens—which can cover a lot of ground quickly—over bears and coyotes. Altogether, these modeling studies show that shorter winters without wolves will create intermittent food supplies that no longer track the needs of local scavengers. With or without wolves, late-winter carrion abundance will decline with shorter winters. But wolf kills buffer these shortages, providing meals that could determine whether scavengers will be able to survive and reproduce. Reintroduced wolves do their part: an intact food chain buffers the impact of deteriorating environmental conditions (Photo: Dan Hartman) It seems clear that wolves have the potential to provide a safety net for scavengers, extending the time they need to adapt to a changing environment. Thanks to a rebounding wolf population, field researchers can measure the magnitude of this predicted buffer effect. The models described here can guide their efforts and help species adjust to major environmental shifts like climate change. As a young US ranger “full of trigger-itch,” Aldo Leopold killed his share of wolves under the federal eradication policy—until he “watched a fierce green fire dying” in the eyes of a slain mother flush with pups and realized he had not understood the wolf's ecological role. Wilmers and Getz's study shows that a robust food chain—including this still embattled top predator—may be even more important as ecological conditions deteriorate.

D:\keerthana\PMC001xxxxxx\PMC1064856.xml

1064857

A New Role for a Protein Involved in Energy Metabolism

Adjusting to life after birth takes a lot of energy. One way cells meet increased demand is by ramping up synthesis of mitochondria, the cells' power generators This ability to increase mitochondria becomes limited in a variety of diseases including diabetes and heart failure. Therefore, it is important to identify the factors that control mitochondrial function. One way researchers have searched for candidate proteins that play a role in this process is by overexpressing proteins in targeted cells to see what happens. That's how several previous studies concluded that a protein called PGC-1α triggers pathways that promote mitochondrial synthesis and regulate both mitochondrial activity and energy metabolism. In a new study, Daniel Kelly and colleagues took a different approach. Rather than increasing the protein's activity, they blocked it. To do that, Kelly and colleagues engineered “knockout” mice that lack functional copies of the PGC-1 α gene. PGC-1α, they found, isn't absolutely required for mitochondrial biogenesis but plays a vital role later in life by “boosting” the ability of cells to increase mitochondrial function in response to the shifting energy demands and physiological stresses encountered after birth. Though leaner than the control mice soon after birth, by 18 weeks the female knockouts were slightly heavier and had more body fat, even though their food intake and activity levels matched the controls. Knockout mice had observable growth defects in skeletal and heart muscle—tissues with high mitochondrial energy requirements—were less active and more easily fatigued than the controls, and had abnormal heart rates after physical exertion. And their livers showed a propensity to accumulate fat because of abnormal mitochondria. PGC-1α deficient mice can't keep pace (Photo: MedPic, Washington University) Altogether, these results demonstrate PGC-1α's critical role in regulating the adaptive metabolic responses required by the increasing energy demands and changing physiological stimuli associated with a growing organism. The increased fat stores and weight gain in the knockout mice, the authors propose, could result from a systemic reduction in energy use, related to defective mitochondria. Given the recently reported link between PGC-1α mutations and human obesity and diabetes, this connection will likely trigger further investigations. And given the pivotal role mitochondria play in a wide range of organs, this mouse model could help shed light on metabolic defects associated with a wide range of diseases. Interestingly, another group, led by Bruce Spiegelman, reported on a PGC-1α knockout model last year. Their mice share traits with the mice described here, but also exhibit a number of contrasting traits, including hyperactive, lean males, which the Spiegelman group attributed to a neurological defect. Kelly and colleagues speculate on possible causes for the differences in the results of the two studies, but only direct comparison of both mouse models will explain the inconsistencies.

D:\keerthana\PMC001xxxxxx\PMC1064857.xml

1064858

Plus Ça Change: Gene Enhancers Upset Evolutionary Assumption

A standard evolutionary assumption is that the DNA of closely related species should be more similar in both structure and function than that of more distantly related ones. This parsimonious rule of thumb holds true across wide expanses of time and among widely divergent species, but it has exceptions. In this issue, Michael Ludwig and colleagues show that one exception is in an enhancer region of a key developmental gene in fruitflies of the genus Drosophila . Here, the enhancer is functionally similar enough in two species that diverged 60 million years ago that switching them produces normal development, but different enough in two species separated by only 10 million years that exchanging them between the two flies aborts development. The gene in question is even-skipped , a patterning gene that creates seven transverse stripes along the anterior–posterior axis in the fruitfly embryo. Its expression is regulated by five enhancing elements located upstream from the promoter. The best characterized of these, the stripe 2 enhancer (S2E), binds five different transcription factors at multiple locations. Ludwig et al. deleted S2E in D. melanogaster , and then added back S2E from one of four Drosophila species: D. melanogaster itself; D. yakuba or D. erecta , both of which have been separated from D. melanogaster for 10–12 million years; or D. pseudoobscura , which split from the D. melanogaster line 40–60 million years ago. Despite serving identical functional roles in each species, the structures of these enhancers differ, with large deletions and insertions between transcription factor binding sites, as well as other changes. Nonetheless, within each species, the spatiotemporal pattern of expression induced by S2E is essentially identical, suggesting that, despite their structural differences, they might be functionally interchangeable. Since loss of S2E is lethal, the viability of the embryos that resulted from these experiments gives a measure of each enhancer's ability to function in its new environment. The authors found that while viability of D. melanogaster with the D. pseudoobscura S2E was identical to that with D. melanogaster 's own, viability with S2E from the more closely related D. erecta was almost zero, essentially the same as not having the enhancer at all. S2E from D. yakuba impaired the viability of D. melanogaster as well, although not as much as that from D. erecta . Viability was closely correlated with the level of stripe 2 expression induced by each S2E, with very low levels induced by the D. erecta enhancer and normal levels by that of D. melanogaster and D. pseudoobscura . Why did the D. erecta enhancer fail to respond in the D. melanogaster environment? Ludwig et al. suggest it may be due to a change in the sensitivity of the “set point” in D. erecta 's enhancer, which acts like an on–off switch governing gene expression, making the enhancer unresponsive to the gradients of transcription factors found in D. melanogaster . This change may be due to relatively small differences in the two species' enhancers that have accumulated since their evolutionary split. The results of this study indicate an important caveat about interpreting the evolution of gene regulatory regions. As a complex functional unit that integrates a host of signals, the S2E is likely to be under strong stabilizing selection, maintaining its output within narrow limits. Thus, the phenotypic result of the enhancer—the location and timing of stripe formation it induces in its native environment—remains conserved among the four species. However, unlike an enzyme or structural protein, in which structural changes are tightly constrained by their effects on function, the structure of any particular enhancer need not be so rigidly preserved. As long as the consequences of change in one region, such as loss of a transcription factor binding site, are matched by compensatory changes in another, such as gain of one, or, as Ludwig et al. speculate, by complementary changes in genetic background, the final output of the enhancer can remain the same. Thus, the utility of structural similarities in understanding evolutionary relationships is likely to be less for gene regulatory regions than for structural genes or the proteins they encode.

D:\keerthana\PMC001xxxxxx\PMC1064858.xml

1064859

Crucial Roles in Drosophila Development for Little-Known Protein

In the intricate dance of animal development, several critical steps are choreographed by the Notch receptor and its ligands, including Delta (Dl) and Serrate (Ser). When these ligands bind, an intracellular segment of the receptor is cleaved off and becomes a transcriptional regulator within the nucleus. While Dl and Ser both occur in Drosophila , and appear to have similar interactions with Notch, less is known about exactly how they trigger Notch activity. In particular, few details are known about the activators of Dl and Ser, which begin the cascade ultimately leading to Notch signaling. In this issue, François Schweisguth and colleagues show that two activators, Drosophila mind bomb (D-mib) and Neuralized (Neur), while sharing a similar molecular activity, have distinct roles in Notch-related Drosophila development, and elucidate several important activities of the heretofore mysterious D-mib. D-mib and Neur are ubiquitin ligases, enzymes that attach the small protein ubiquitin onto a target protein. While ubiquitination was first appreciated as a tag for protein degradation, more recently, it has been recognized to be a signal for endocytosis, a process that brings substances outside a cell into the cell. Endocytosis, in turn, has recently been found to be a key step in Dl activation of Notch; its importance to Ser signaling has not been previously identified. In Drosophila , Neur targets Dl, and Dl is endocytosed. In zebrafish, Dl is a target for a separate ubiquitin ligase, Mind bomb. But the function of the Drosophila homolog, D-mib, has not been elucidated. Thus, the essential questions in this study were, what role does D-mib play in Drosophila , and how is Ser signaling regulated? Schweisguth and colleagues studied development in a D-mib mutant. The pattern of phenotypic changes seen was in keeping with a loss of Dl signaling, indicating that D-mib interacts with Dl, just as Neur does. But D-mib also interacts with the other Notch ligand, Ser, as shown by the aberrant distribution of Ser in the absence of D-mib. Furthermore, endocytosis of Ser occurred normally in the presence of D-mib, and was strongly inhibited in its absence. Acting through Notch, Ser activates a downstream gene that codes for a protein called Cut, whose absence leads to a particular pattern of wing defects. This pattern was seen in D-mib mutants, and could not be rescued by overexpressing Ser, thus indicating that D-mib not only prompts endocytosis of Ser, but allows it to trigger Notch signaling. It's unclear whether D-mib ubiquitinates Ser, as this study did not specifically address that question. While Neur and D-mib differ structurally, they share the same molecular function, ubiquitination of Notch ligands, and the authors show that, when ectopically expressed, each can at least partially compensate for the absence of the other. Nonetheless, they normally have distinct developmental functions, owing largely to the fact that they are expressed in different locations and at different times during development, and thus mutations in the two lead to different patterns of developmental aberrations.

D:\keerthana\PMC001xxxxxx\PMC1064859.xml

1064860

White Collar Proteins Help Fungi Do It in the Dark

Fungi live mainly in the dark, and they like it that way—out of the sunlight, they can avoid desiccation and damage from ultraviolet rays. The ability to sense light, therefore, is adaptive for fungi of all kinds. A pair of light-sensing proteins had been identified in the model fungus Neurospora crassa , an ascomycete (one of the three fungal subgroups, defined by the production of sexual spores within sac-like structures), but little was known about mechanisms in other fungal phyla. In a new study, Alexander Idnurm and Joseph Heitman show that the basidiomycete (a subgroup defined by the production of sexual spores on the ends of club-like structures) Cryptococcus neoformans employs a similar protein pair, which regulate mating, growth, and virulence of this human fungal pathogen. In N. crassa , blue light is sensed by the protein White collar 1, which interacts with a flavin (light-absorbing pigment) tuned to photons in the blue region of the spectrum. White collar 1 then binds to White collar 2, and the complex serves as a transcription factor. In this study, the authors searched the C. neoformans genome for genes with similar evolutionary origins to these two genes (called homologs), as well as others implicated in light sensing, and identified Basidiomycete white collar 1 , or BWC1 , along with other light-sensor candidates, including an opsin and a phytochrome homolog. Mutations of BWC1 , but not the other candidate photoreceptors, rendered C. neoformans insensitive to light. While mating and fruiting in the wild-type fungus is suppressed by exposure to blue light, bwc1 mutants were unaffected by light. Interestingly, the mating process was released from light inhibition when either one of the two mating strains were mutated, suggesting that the cell fusion process at the heart of fungal mating requires only one cell to commit to fusion. In addition, bwc1 mutants were extremely sensitive to ultraviolet radiation. No homolog of photolyase, a protein that uses light to repair DNA damage, was identified in the genome. Future studies will be necessary to understand how Bwc1 functions in ultraviolet resistance, but these findings suggest the protein could sense photons in both the ultraviolet and blue wavelengths. To identify other proteins with which the Bwc1 protein functionally interacts, the authors examined nearly 3,000 mutant strains, yielding three with a phenotype similar to the bwc1 mutant. They found that the gene for one of these, dubbed BWC2 , is a homolog of N. crassa White collar 2, and that its protein binds to Bwc1. Together, the two influence transcript levels of two key genes required for C. neoformans mating, further strengthening the case that the pair function as a transcription factor as do their homologs in N. crassa . Interestingly, mutants of either BWC1 or BWC2 were less virulent than the wild-type strain of the fungus, revealing a novel environmental signaling pathway involved in C. neoformans virulence. The functional and structural similarities of the ascomycote and basidiomycote White collar proteins indicate that they arose prior to the split of these two lineages more than 500 million years ago. The fungal kingdom contains an estimated one million species. The authors suggest that the ultraviolet protection afforded by the White collar system may have been crucial to the evolutionary diversification of this kingdom, in particular when ultraviolet radiation on the earth's surface was higher than it is today, such as when life emerged from the sea and colonized the barren continents. They also note that the same proteins are found in clinical isolates of C. neoformans , and that the mitigation of virulence by bwc1 and bwc2 mutations will be useful in the identification of new genes required for disease development in this important pathogen.

D:\keerthana\PMC001xxxxxx\PMC1064860.xml

1064861

Collagen type II (CII)-specific antibodies induce arthritis in the absence of T or B cells but the arthritis progression is enhanced by CII-reactive T cells

Antibodies against type II collagen (anti-CII) are arthritogenic and have a crucial role in the initiation of collagen-induced arthritis. Here, we have determined the dependence of T and B cells in collagen-antibody-induced arthritis (CAIA) during different phases of arthritis. Mice deficient for B and/or T cells were susceptible to the CAIA, showing that the antibodies induce arthritis even in the absence of an adaptive immune system. To determine whether CII-reactive T cells could have a role in enhancing arthritis development at the effector level of arthritis pathogenesis, we established a T cell line reactive with CII. This T cell line was oligoclonal and responded to different post-translational forms of the major CII epitope at position 260–270 bound to the A q class II molecule. Importantly, it cross-reacted with the mouse peptide although it is bound with lower affinity to the A q molecule than the corresponding rat peptide. The T cell line could not induce clinical arthritis per se in A q -expressing mice even if these mice expressed the major heterologous CII epitope in cartilage, as in the transgenic MMC (mutated mouse collagen) mouse. However, a combined treatment with anti-CII monoclonal antibodies and CII-reactive T cells enhanced the progression of severe arthritis.

Introduction Collagen-induced arthritis (CIA) is a widely used animal model for rheumatoid arthritis (RA). Immunization with native collagen type II (CII) in adjuvant induces autoimmune polyarthritis in susceptible rodents and primates [ 1 ]. The separate roles of T cells and B cells in both the initial and the progression phases of arthritis in this model are still undefined. Clearly, immunization with heterologous CII activates both CII-reactive T cells and B cells. The T cell response is dominated by reactivity to CII used for immunization, and T cells do not readily cross-react with mouse CII [ 2 ]. In contrast, B cells produce high levels of autoreactive and arthritogenic IgG antibodies reactive with both heterologous and homologous CII. The most likely scenario is that the heteroreactive T cells give help to autoreactive B cells that cross-react with mouse CII. Molecular identification of the relevant epitopes supports this interpretation because there is a critical difference in the T cell epitope but not in the major B cell epitopes between mouse CII and heterologous CII. Furthermore, depletion of T cells with anti-CD4 or anti-T-cell receptor (anti-TCR) antibodies is more effective if given before immunization than if given afterwards [ 3 , 4 ]. Finally, severe arthritis is readily induced with anti-CII antibodies [ 5 ], whereas transfer with T cells induces only synovitis and not clinical arthritis [ 6 ]. However, it is unlikely that CIA pathogenesis can be reduced to mediation by anti-CII antibodies alone. The question is whether autoreactive T cells might have an additional role in CIA, in particular whether they have a role in the further progression of arthritis and during the chronic relapsing disease course that follows the initial arthritis in some mouse strains. This possibility has also been highlighted by the finding that many heteroreactive T cells are most probably potentially autoreactive to CII in vivo , because a major difference is the binding of the peptide to the MHC rather than interaction with TCR [ 2 , 7 ]. The difference between the mouse and the heterologous immunodominant peptide is dependent on differences in binding to the MHC class II molecule A q . Thus, they recognize the same peptide but different densities of the peptide are presented depending on whether the CII is of mouse or of heterologous origin. Interestingly, immunization with mouse CII induces arthritis in a smaller number of mice but gives a more chronic disease course than immunization with heterologous CII [ 8 , 9 ]. Furthermore, in the mutated mouse collagen (MMC) mouse, which expresses a mutated CII with the heterologous CII – namely mutated at position 266, changing Asp to Glu – the heterologous CII is expressed in the joints. In this mouse T cells are partly tolerized and the development of arthritis is differently genetically controlled [ 10 , 11 ]. The development of arthritis after injection of collagen antibodies (collagen-antibody-induced arthritis; CAIA) is thus likely to be different from the development of arthritis in CIA, although the resulting clinical arthritis shares many common characteristics [ 5 ]. CAIA is known to develop independently of MHC alleles, whereas CIA is crucially dependent on MHC alleles, with the A q molecule as one of the most susceptible alleles. This suggests that CAIA develops independently of MHC-restricted T cells, and thereby also independently of T cell-dependent B cells. To confirm this assumption directly we investigated mice deficient in B cells and T cells on backgrounds susceptible to CIA. Interestingly, such mice not only developed CAIA but had a more severe disease, suggesting that these cells have a modifying role in this model. We also readdressed the role of transferred T cells by using a T cell line reactive with the major CII epitope 260–270 but with oligoclonal reactivity to the various post-translational modifications. As expected, these T cells could not induce clinical arthritis in either wild-type or MMC mice. However, the transferred T cells enhanced the CII-antibody-induced arthritis into a more prolonged disease course. Materials and methods Animals Male B10.Q and QD ([B10.Q × DBA/1]F 1 ) mice at 4–6 months of age were used in the present study. The B10.Q strain was obtained from Professor Jan Klein (Tübingen, Germany), and DBA/1 mice were from Jackson Laboratories (Ban Harbor, ME, USA). B cell-deficient mice (μMT mice kindly provided by Dr Werner Muller [Cologne, Germany]) on the (C57Bl/6 × 129)F 1 background were backcrossed to B10.Q background (12 n ) and T cell-deficient mice (lacking αβ T cells as a result of targeted germline mutation in their TCRβ gene, obtained from Jackson Laboratories) on the (C57Bl/6 × 129)F 1 background were backcrossed to B10.Q background (6 n ). To obtain mice deficient in both B and T cells, heterozygous female mice deficient in B cells and T cells were crossed with doubly deficient males, and offspring were investigated for the absence of B cells and T cells by cytofluorimetric analysis. Blood cells were stained with anti-CD45Ra (B220 coupled to fluorescein isothiocyanate) and anti-TcR (145-2C11 coupled to phycoerythrin) before analysis. MMC transgenic mice (previously named MMC-1), which originated on the C3H.Q background as described previously [ 10 ], were backcrossed for eight generations onto the B10.Q background. The transgene MMC is a mutated mouse CII gene in which position 266 has a been changed from aspartic acid (D) to glutamic acid (E), thereby expressing the rat CII260–270 epitope in a CII-restricted fashion. All mice were kept in a conventional but barrier animal facility (as defined in ) with a climate-controlled environment having 12 hours light/12 hours dark cycles in polystyrene cages containing wood shavings; the mice were fed with standard rodent chow and water ad libitum . All animal experiments had been approved by the local animal welfare authorities. CII-specific monoclonal antibodies The CII-specific hybridomas were generated and characterized as described in detail elsewhere [ 12 - 14 ]. From the panel of monoclonal antibodies generated, a combination of an IgG2b antibody of the clone M2139 binding to the J1 epitope (amino acids 551–564) and an IgG2a antibody of the clone CIIC1 binding to the C1 I epitope (amino acids 358–363) was found to be more arthritogenic [ 5 ], whereas CIIF4 monoclonal antibody binding to F4 epitope (amino acids 926–936) was found to be inhibitory [ 15 ]. Recent studies in vitro also emphasize that these arthritogenic monoclonal antibodies M2139 and CIIC1 suppressed the self-assembly of CII into fibrils, whereas CIIF4 was found to be inert [ 16 ]. Figure 1 illustrates the B cell and T cell epitopes present in the CII α-chain recognized by the monoclonal antibodies and the T cell line used in this study. Monoclonal antibodies were generated as culture supernatants and purified by affinity chromatography with γ-bind plus affinity gel matrix (Pharmacia, Uppsala, Sweden). The IgG content was determined by freeze-drying. The antibody solutions were filter-sterilized using syringe filters with a pore size of 0.2 μm (Dynagard; Spectrum Laboratories, CA, USA), aliquoted and stored at – 70°C until use. The amount of endotoxin in the antibody solutions prepared was found to be in the range 0.02–0.08 EU/mg of protein as analysed with the Limulus amebocyte lysate (Pyrochrome) method (Cape Cod Inc., Falmouth, MA, USA). Passive transfer of antibodies The cocktail of M2139 and C1 monoclonal antibodies was prepared by mixing equal concentrations of each of the sterile filtered antibody solutions to get a final amount of 9 mg. Mice were injected intravenously twice with 0.25–0.4 ml of antibody solution, with a minimum interval of 3 hours. As a control, groups of mice received equal volumes of PBS. On day 5, lipopolysaccharide (25 μg per mouse) was injected intraperitoneally in all mice. A pair of irrelevant antibodies of the same subclass (mouse anti-human HLA-DRα, IgG2a [L243] and mouse anti-human parathyroid epithelial cells, IgG2b [G11]) did not induce arthritis in the most susceptible strain, BALB/c mice [ 5 ]. Characteristics of CII-specific T cell line A T cell line specific for rat CII was established as described previously [ 17 ]. In brief, draining lymph nodes from rat CII-immunized QD mice (on day 8) were stimulated in vitro with rat CII for 4 days. These cells were allowed to rest for a week in the presence of interleukin-2 (IL-2) without antigen-presenting cells. T cells were subsequently re-stimulated with irradiated (3000 rad) syngenic splenocytes and rat CII for 3 days (5 × 10 5 T cells/ml, 5 × 10 6 antigen-presenting cells/ml, 10 μg/ml rat CII) followed by 2 weeks of resting in medium containing IL-2. At the time of re-stimulation, an aliquot of the cell line was tested for antigen specificity. Lathyritic CII was used for the first in vitro re-stimulation, to avoid contamination of pepsin-reactive T cells. The cell line responded towards denatured CII, the non-modified CII 256–270 peptide and the glycosylated CII 256–270 peptide with proliferation and interferon-γ (IFN-γ) production (Fig. 2 ), but no response towards pepsin was observed. IFN-γ was measured by enzyme-linked immunosorbent assay as described previously [ 18 ]. Arthritis development Development of clinical arthritis was followed by means of visual scoring of the mice. Mice were examined daily or on alternate days for arthritis development until the end of the experiment. Arthritis was scored with an extended scoring protocol ranging from 1 to 15 for each paw, with a maximum score of 60 per mouse, based on the number of inflamed joints in each paw, inflammation being defined by swelling and redness. Each arthritic toe and knuckle was scored as 1, with a maximum of 10 per paw, and an arthritic ankle or mid-paw was given a score of 5. Statistics Arthritis incidence and severity were analysed by χ 2 analysis and the Mann–Whitney U -test respectively. P ≤ 0.05 was considered as significant. Results Antibody-mediated arthritis in mice deficient in B and/or T cells To understand the role of B and T cells in antibody-mediated inflammation, mice deficient in either B cells or T cells or both were injected with a combination of two different monoclonal antibodies against CII. The antibodies have been shown to bind to cartilage surfaces shortly after intravenous injection [ 19 ] and the epitopes recognized are depicted in Fig. 1 . As shown in Fig. 3a , most of the B cell-deficient (71%) and T cell-deficient (87%) mice developed severe arthritis; 50% of the mice deficient in both B and T cells also developed arthritis, whereas only 25% of littermate controls developed the disease (B cell-deficient versus controls, cumulative incidence P ≤ 0.0354; T cell-deficient versus controls, cumulative incidence P ≤ 0.0117). Mice deficient in either B or T cells developed more severe arthritis than mice deficient in both populations or than the control littermates (Fig. 3b ); however, the difference in arthritis severity between the groups on different days was not significant. These data show that neither T cells nor B cells are necessary for CAIA development. Furthermore, the observed enhancement in the frequency of arthritis in the T cell-deficient mice and the B cell-deficient mice suggest that these cells might play regulatory roles in the initiation of disease. Effect of T cells transfer on CAIA To ascertain whether a transfer of CII-specific T cells after antibody injection induced more susceptibility or prolonged the disease period, we established a rat CII-specific T cell line. The line was established from rat CII-immunized QD mice and re-stimulated in vitro four or five times with rat CII before transfer. The established T cell line was A q -restricted and oligoclonal because it responded to both the non-modified and hydroxylated, as well as the glycosylated, versions of the major CII peptide 260–270 containing various post-translational modifications at the major T cell recognition site on lysine 264 (Figs 1 and 2 ). The strongest reactivity was seen to the galactosylated peptide, but the hydroxylated peptide also mounted a strong response. Interestingly, the T cell line cross-reacted to the glycosylated mouse peptide, and the lower reactivity to the mouse glycopeptide than to the rat glycopeptide is most probably dependent on both the lower affinity of the mouse peptide for A q and also a different reactivity of the clonally distinct glycopeptide-reactive T cells [ 7 ]. To investigate the role of T cells in the acute effector stage of clinical arthritis, newly activated T cells were injected into QD mice intravenously 1 day after the antibody transfer. As expected, injection of the antibodies alone was sufficient to induce arthritis, but co-transfer of T cells did not enhance the initiation of arthritis (Fig. 4 ). However, transfer of both antibodies and T cells did result in persistent disease activity. As the T cell line was mainly considered as heteroreactive when transferred into wild-type mice (Fig. 2 ), co-transfer of T cells and antibodies was also performed in MMC mice (which express the rat CII epitope in cartilage) to see whether the presence of truly autoreactive T cells could have a different effect on the acute phase of arthritis. However, T cells again did not affect the initiation phase of the disease; instead, the effect was noted in the more chronic phase of the disease. Still, co-transfer of T cells into MMC mice resulted in an even more pronounced and significant progression of arthritis than in mice that had antibodies transferred. In contrast, wild-type (MMC-negative) mice that had both T cells and antibodies transferred did not show a significant difference from antibody-transferred mice (Fig. 4 ). Furthermore, an ovalbumin-specific T cell line failed to enhance and perpetuate the arthritis induced by anti-CII antibodies (data not shown). Discussion As we show in the present study, anti-CII monoclonal antibodies are capable of initiating disease independently of B and T cells during the effector phase of arthritis. This is not an unexpected finding because CAIA is induced in naive mice by preformed anti-CII antibodies. Interestingly, however, immune cells might have a regulatory role in CAIA because mice deficient in both T cells and B cells are more susceptible to arthritis than their control littermates. There are several possible explanations for this observation. Clearly, B cells could be regulatory owing to the secretion of a cytokine such as IL-10 [ 20 ], the expression of inhibitory receptors such as FcgRIIb [ 21 ] or the secretion of regulatory antibodies such as anti-CII antibodies [ 15 , 22 ]. Similarly, there are several ways in which T cells might be regulatory in an effector state like this: for example, they might control bone destruction through interaction with the osteoprotegrin system [ 23 ] or through the regulation of cytokines such as IFN-β, tumour necrosis factor-α or IL-4 [ 24 - 26 ]. However, a surprising finding was that mice deficient in both cell types were not as susceptible as the respective single-cell deficient mice. In the doubly deficient mice a complete absence of the adaptive immune response could have led to a more predominant role for the innate immune system in the regulation of the antibody-mediated inflammatory response. In addition, we have shown here that already activated CII-reactive T cells reactive to glycosylated CII could prolong the disease initiated by antibodies, a finding that is highly relevant for comparison with the CIA model. As in RA, susceptibility to CIA is linked to the expression of certain class II MHC alleles, explaining the crucial role depicted to T cells. The predominant role of T cells in CIA development was demonstrated by using anti-CD4 or anti-TCRαβ monoclonal antibodies and T cell-deficient mice [ 3 , 4 , 27 ]. Mice deficient in the co-stimulatory molecule CD28 were found to be resistant to CIA [ 28 ]. Similarly, administration of CTLA4Ig at the time of immunization prevented the development of CIA [ 29 ]. These studies demonstrate the importance of T cell activation in CIA pathogenesis. Depletion of CD4 + T cells has a major influence during the priming phase of arthritis [ 3 ] and suppressed the adoptive transfer of disease to severe combined immunodeficient mice using spleen cells from CII-immunized mice [ 30 ]. Partial protection of CD4-deficient B10.Q mice and significantly reduced incidence in CD8-deficient mice from CIA suggested an initiating role for the T cells during the priming phase of CIA [ 31 ]. However, T cell reactivity alone could not explain the disease pathology in CIA. Transfer of synovitis but not clinical arthritis using CII-specific T cells has previously been shown. In contrast, the high incidence of arthritis induced by native but not denatured collagen indicated the importance of B cells in CIA. It has been shown that mice pre-sensitized with heat-denatured collagen developed progressive arthritis after the transfer of anti-CII antibodies. In addition, adoptive transfer of lymphoid cells together with antibody in the T cell-depleted mice was shown to induce arthritis, and the effector cells were identified as Thy-1 + and L3T4 + Lyt-2 - [ 32 ]. The recognition of CII by T cells is critical to the establishment of autoimmune arthritis in CIA. However, it is debatable whether T cells are capable of recognizing tissue antigens such as insoluble CII in the cartilage tissue. It therefore becomes all the more important to understand the role of antigen-specific T cells in arthritis pathogenesis. Antigen-specific T cells might have important roles either during the initiation phase of arthritis or in the perpetuation and exacerbation of the disease after the onset, or they might simply maintain immunity to CII and perpetuate antibody production. Results from the present study demonstrate that CII-specific T cells could have a role in the perpetuation and exacerbation of already established disease rather than having any direct influence on the initiation phase of arthritis. It is possible that the pro-inflammatory cytokines induced and/or secreted by the co-transferred CII-specific cells could provide a constant cytokine milieu in or near the joints for exacerbating the events induced by the formation of collagen–IgG immune complexes. It should also be noted that the ovalbumin-specific T cell line failed to enhance and perpetuate the arthritis induced by anti-CII antibodies. With the use of CII-specific monoclonal antibodies, it has been shown that IL-1 and tumour necrosis factor-α are the important cytokines for disease development [ 33 ], similar to anti-glucose-6-phosphate isomerase antibody-induced disease [ 34 ]. The observed enhancement of arthritis in the T cell and B cell singly deficient mice also suggests that these cells might have regulatory roles in the initiation of disease by modulating the cytokine environment. Despite a prolongation of arthritis, co-transfer of the CII-specific T cell line with the monoclonal antibodies did not alter the acute phase of antibody-mediated disease into a chronic disease course, suggesting the importance of other cellular mediators in the pathogenesis of arthritis. However, experiments to understand the factor(s) and cells involved during the progression of arthritis from the initiation stage will provide tools for effective intervention in arthritis progression in patients with RA. Conclusions We demonstrated that anti-CII monoclonal antibodies are capable of initiating arthritis independently of B and T cells during the effector phase of arthritis. Already activated CII-reactive T cells, especially reactive to glycosylated CII, could prolong the disease initiated by antibodies, a finding that is highly relevant for comparison with the CIA model. Experiments to understand the factor(s) and cells involved during the progression of arthritis from the initiation stage could therefore provide tools for effective intervention in arthritis progression in patients with RA. Competing interests None declared. Abbreviations CAIA = collagen-antibody-induced arthritis; CIA = collagen-induced arthritis; CII = type II collagen; IFN-γ = interferon-γ ; IL = interleukin; MMC = mutated mouse collagen; RA = rheumatoid arthritis; TCR = T cell receptor.

D:\keerthana\PMC001xxxxxx\PMC1064861.xml

1064864

Use of HLA-B27 tetramers to identify low-frequency antigen-specific T cells in Chlamydia-triggered reactive arthritis

Reports of the use of HLA-B27/peptide tetrameric complexes to study peptide-specific CD8 + T cells in HLA-B27 + -related diseases are rare. To establish HLA-B27 tetramers we first compared the function of HLA-B27 tetramers with HLA-A2 tetramers by using viral epitopes. HLA-B27 and HLA-A2 tetramers loaded with immunodominant peptides from Epstein–Barr virus were generated with comparable yields and both molecules detected antigen-specific CD8 + T cells. The application of HLA-B27 tetramers in HLA-B27-related diseases was performed with nine recently described Chlamydia -derived peptides in synovial fluid and peripheral blood, to examine the CD8 + T cell response against Chlamydia trachomatis antigens in nine patients with Chlamydia -triggered reactive arthritis (Ct-ReA). Four of six HLA-B27 + Ct-ReA patients had specific synovial T cell binding to at least one HLA-B27/ Chlamydia peptide tetramer. The HLA-B27/ Chlamydia peptide 195 tetramer bound to synovial T cells from three of six patients and HLA-B27/ Chlamydia peptide 133 tetramer to synovial T cells from two patients. However, the frequency of these cells was low (0.02–0.09%). Moreover, we demonstrate two methods to generate HLA-B27-restricted T cell lines. First, HLA-B27 tetramers and magnetic beads were used to sort antigen-specific CD8 + T cells. Second, Chlamydia -infected dendritic cells were used to stimulate CD8 + T cells ex vivo . Highly pure CD8 T cell lines could be generated ex vivo by magnetic sorting by using HLA-B27 tetramers loaded with an EBV peptide. The frequency of Chlamydia -specific, HLA-B27 tetramer-binding CD8 + T cells could be increased by stimulating CD8 + T cells ex vivo with Chlamydia -infected dendritic cells. We conclude that HLA-B27 tetramers are a useful tool for the detection and expansion of HLA-B27-restricted CD8 + T cells. T cells specific for one or more of three Chlamydia -derived peptides were found at low frequency in synovial fluid from HLA-B27 + patients with Ct-ReA. These cells can be expanded ex vivo , suggesting that they are immunologically functional.

Introduction Chlamydia -triggered reactive arthritis (Ct-ReA) is strongly associated with HLA-B27 like other spondylarthropathies, and especially ankylosing spondylitis [ 1 , 2 ]. ReA occurs 1 to 4 weeks after urogenital infection with Chlamydia trachomatis or gastroenteral infection with enterobacteria such as Yersinia enterocolitica [ 3 ]. After acute onset, most patients have a self-limiting course, but up to 20% suffer from a disease duration of more than 1 year [ 4 ]. Of HLA-B27 + -reactive arthritis patients, 20–40% move on to ankylosing spondylitis after 10–20 years, suggesting that the ReA-associated bacteria can cause ankylosing spondylitis [ 5 ] and that immune mechanisms triggering the disease are induced by T cell responses to microbial antigens. The main hypothesis advanced for the association between HLA-B27 and spondylarthropathies is the arthritogenic peptide theory. It states that some HLA-B27 subtype alleles, owing to their unique amino acid residues, bind a specific arthritogenic peptide that is recognized by CD8 + T cells [ 6 - 9 ]. Recently we and several other groups have reported on Chlamydia -specific CD8 + T cells capable of lysing target cells primed with Chlamydia antigens [ 10 - 12 ]. CD8 + T cell responses in spondylarthropathies other than Ct-ReA have also been described [ 13 - 15 ]. Recently a new method for antigen-specific T cell recognition has been established by using multimerized MHC/peptide molecules [ 16 ]. These molecules are called tetramers because they contain four soluble and biotinylated MHC molecules linked to labelled streptavidin that specifically bind with high avidity to T cell receptors. In comparison with intracellular cytokine staining, the major advantage of tetramer technology is the identification of antigen-specific T cells independently of their cytokine secretion profile, the possibility of sorting unstimulated T cells and of having a tool for the antigen-specific detection of T cells in experiments in situ [ 17 ]. In humans, MHC class I tetramers are widely used, and HLA-A2 tetramers in particular are an important tool in tumour immunology [ 18 ]. However, the use of HLA-B27 tetramers in HLA-B27-related diseases is rare [ 10 , 19 ]. The rarity of their use might be related to heavy protein aggregation during the refolding procedure of the recombinant HLA-B27 monomer [ 19 , 20 ]. To determine optimised conditions for the refolding procedure of soluble HLA-B27 monomers with bacteria-derived epitopes we first used HLA-B27 tetramers with a well-described HLA-B27-restricted viral epitope from Epstein–Barr virus (EBV). We analysed the refolding rate of HLA-B27 monomers and compared our results with refolding gained with an HLA-A2 molecule loaded with a viral epitope from EBV [ 21 ]. On the basis of these results we applied the HLA-B27 tetramer technology to specify the HLA-B27-restricted CD8 + T cell response to Chlamydia -derived peptides in patients with Ct-ReA. This is the first report of a systematic use of HLA-B27 tetramers in humans in an HLA-B27-related disease. Methods Patients We analysed six HLA-B27 + and three HLA-B27 - patients with ReA after infection with Chlamydia trachomatis (Table 1 ). We diagnosed ReA if patients had a prior urogenital infection, which was confirmed by the detection of Chlamydia trachomatis in the morning urine by polymerase chain reaction. An additional criterion was the detection of Chlamydia -specific antibodies [ 6 ] at the beginning of the disease or highest synovial T cell proliferation against Chlamydia trachomatis [ 22 ] in proliferation assays with whole Chlamydia antigen. The results were compared with tetramer staining in six HLA-B27 + healthy blood donors. We also examined synovial T cells from three HLA-B27 + patients with ReA after gastroenteritis and having highest synovial proliferation against enterobacteria. We also tested the synovial fluid of three patients with rheumatoid arthritis. In addition we used HLA-B27 + and HLA-A2 + blood donors with previous EBV infection for experiments comparing HLA-B27 and HLA-A2 tetramers. The ethical committee of the Benjamin Franklin Medical Centre gave ethical approval for this study. Search for peptide binding affinity The quantification of HLA-B27 binding affinity was conducted with two different programs that analyse HLA-peptide binding motifs, one called SYFPEITHI described by Rammensee and colleagues [ 23 ] and the other called BioInformatics and Molecular Analysis Section (BIMAS; ). Peptide synthesis Nonamer peptides were synthesized by standard 9-fluorenyl-methyloxy-carbonyl solid-phase synthesis methods on a Syro-Synthesizer (MultiSyn Tech, Witten, Germany), purified by high-performance liquid chromatography (Shimadzu LC-10; Shimadzu Scientific Instruments, Duisburg, Germany) and identified by mass spectroscopy (LCQ, ion trap; Thermoquest, Eberbach, Germany). The purity of the peptides was more than 95%. Peptides were dissolved in dimethyl sulphoxide. For T cell stimulation and fluorescence-activated cell sorting (FACS) analysis of intracellular cytokine staining, the peptides were further diluted with serum-free medium at a concentration of 5 mg/ml and frozen at -80°C. FACS analysis of antigen-specific T cells with HLA-B27 tetramers HLA-B27 tetramers were generated as described previously [ 19 ], with some modifications. The expression vector pLM1-HLA-B27 was modified by tagging with the BirA recognition sequence as described previously and by mutating the cysteine residue at position 67 to serine. After being refolded, the recombinant protein was concentrated and centrifuged at 13,000 rpm (16,060 g ; Haereus Biofuge Pico; Kendro Laboratories, Langenselbold, Germany) followed by biotinylation and gel filtration with a Superose 12 column (Pharmacia) on an Äkta Basic system (Pharmacia). Correct folding and biotinylation were analysed by gel filtration (Äkta Basic, Pharmacia) and gel electrophoresis (Bio-Rad). Tetramers were generated by adding phycoerythrin (PE)-labelled streptavidin (Molecular Probes) at a ratio of 1.5:1. We generated HLA-B27 tetramers with the EBV EBNA peptide (residues 258–266) [ 24 ]. For the detection of Chlamydia -peptide-specific CD8 + T cells we used the previously described immunodominant peptides 8, 68, 80, 131, 133, 138, 144, 145, 146, 194, 195 and 196 [ 10 ] (Table 2 ). Peptides 144 and 194 caused heavy aggregation during refolding procedure and were excluded from tetramer staining; peptide 146 was excluded because of high background staining in more than 50% of the patients. HLA-A2 monomers with the EBV peptide [ 21 ] were generated with an HLA-A2 heavy chain (gift from Dr KH Lee, Berlin, Germany) with the same protocol. For FACS analysis, frozen mononuclear cells (MNCs) from synovial fluid or peripheral blood were incubated with tetramer and PerCP-labelled anti-human CD8 antibody (BD Pharmingen, San Diego, USA) in parallel for 30 min at room temperature (20°C) followed by washing twice with phosphate-buffered saline (PBS)/2% bovine serum albumin (BSA) and incubation with Cy5-labelled anti-human CD3 antibody for 30 min at room temperature. Cells were washed twice in PBS/2% BSA and resuspended in Annexin V buffer (Molecular Probes) and 2.5 μl of Alexa 488-labelled Annexin V (Molecular Probes) was added. CD8 + and tetramer-positive T cells were analysed after gates were set on CD3 + and Annexin V-negative cells. Depending on the availability of additional synovial lymphocytes we repeated the staining experiments, which was true for the synovial fluid of patient no. 6. T cell lines from magnetic activated cell sorting (MACS)-sorted HLA-B27 tetramer-positive CD8 + T cells Peripheral MNCs were incubated for 30 min with Cy5-labelled anti-CD8 antibody (BD) and 5 μg/ml PE-labelled HLA-B27/EBV EBNA (258–266) tetramer at room temperature. Cells were washed twice and incubated for 15 min at 4°C with anti-PE-labelled MACS beads (Miltenyi) at a ratio of 20 μl of beads to 80 μl of cell suspension. Labelled cells were loaded on an LS MACS column (Miltenyi) and eluted after the column had been washed three times with washing buffer including PBS, EDTA and BSA. MACS-sorted tetramer-positive and CD8 + T cells were further separated by FACS sorting. Sorted cells (1000) were incubated with 500,000 autologous antigen-presenting cells in the presence of 20 U/ml interleukin (IL)-2, 10 ng/ml IL-7 and 10 ng/ml IL-15 added every 3–4 days. Determination of the refolding rate of recombinant HLA-B27 monomers The refolding rate of recombinant HLA-B27 monomer was analysed by gel filtration and by determining the relative amount of soluble HLA-B27 monomer eluted at 13.7 ml in comparison with precipitated protein eluted earlier in a Superose 12 column (Pharmacia). An Akta basic system (Pharmacia) was used. The elution profile was analysed by using Unicorn (version 4) software (Pharmacia). Refolding was defined as ++ when more than 75% of proteins loaded on the gel filtration column after refolding, biotinylation and sharp centrifugation was soluble HLA-B27 monomer molecule; + for more than 50% soluble HLA-B27 monomer, (+) for more than 10% soluble HLA-B27 monomer, and - for less than 10% soluble HLA-B27 monomer (Table 2 ). FACS analysis of intracellular cytokine staining Intracellular cytokine staining was used after antigen-specific T cell stimulation. Synovial MNCs and peripheral MNCs were stimulated for 6 hours in 1 ml of culture medium with anti-CD28 antibody (1 μg/ml) plus single peptides (10 μg/ml) or without antigenic peptide as a negative control. Brefeldin A was added after 2 hours to stop the stimulation, and cells were harvested after a further 4 hours and then stained with 5 μg/ml anti-CD69-PE antibody (BD Pharmingen) and 1 μg/ml anti-CD8-PerCP (BD Pharmingen). Cells were then fixed in 2% formalin and resuspended in saponin buffer, followed by incubation with 1 μg/ml Cy5-conjugated anti-human interferon-γ antibody (IFN-γ; BD). Gated CD8 + T cells that were positive for early activation marker CD69 and for intracellular IFN-γ were counted as antigen-specific. Analysis was performed with a BD Biosciences FACScan flow cytometer with CellQuest software. Infection of peripheral-blood-derived dendritic cells in vitro with viable Chlamydia trachomatis CD14 + cells from peripheral blood were incubated for 1 hour with anti-CD14-conjugated magnetic beads (Miltenyi) and sorted by MACS. The purity of separated cells was confirmed by FACS analysis. Cells (500,000) were cultured for 7 days in 24-well plates at 37°C at 5% CO 2 in 1 ml of RPMI culture medium supplemented with 10% fetal calf serum, 2 mM L-glutamine and 50 ng/ml granulocyte/macrophage colony-stimulating factor and 10 ng/ml IL-4 to induce transformation to dendritic cells (DCs). Cells were washed and harvested and incubated for 24 hours with infectious elementary bodies of Chlamydia trachomatis at a ratio of 1:50. DCs were analysed by FACS with the use of anti-CD80, anti-CD86, anti-HLA-DR, anti-CD14 (BD Pharmingen) and anti- Chlamydia trachomatis lipopolysaccharide antibodies (Dako) before and after infection with viable Chlamydia trachomatis . Expansion of Chlamydia -specific CD8 + T cells in vitro with Chlamydia -infected peripheral-blood-derived dendritic cells We stimulated CD8 + T cells from peripheral blood with Chlamydia trachomatis -infected peripheral-blood-derived DCs at a ratio of 50:1 in RPMI culture medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. Recombinant IL-7 (10 ng/ml) and IL-15 (10 ng/ml) were added on both days 2 and 7. T cells were analysed by FACS on day 14. Results MHC class I tetramer staining with HLA-A2/EBV peptide-specific and HLA-B27/EBV peptide-specific tetramers To determine optimal conditions for the refolding procedure of soluble HLA-B27 monomers we used HLA-A2 tetramers and HLA-B27 tetramers with well-described HLA-B27 and HLA-A2 restricted viral epitopes from EBV. The binding scores of the two immunodominant EBV peptides to the HLA-A2 [ 21 ] (score 29) and HLA-B27 [ 24 ] (score 28) receptor were almost identical in the SYFPEITHI program. By generating the HLA-B27 tetramer (Fig. 1a,1c ; lanes 1 and 2) and the HLA-A2 tetramer (Fig. 1b,1c ; lanes 3 and 4) the percentages of protein aggregates eluted between 7 and 13 ml in gel filtration, and the refolded monomer eluted at 13.7 ml in gel filtration, were also similar. The large peak at 16 ml most probably contained reagents from the biotinylation reaction because we did not detect any proteins with a molecular mass of more than 5 kDa by SDS-PAGE. This peak was excluded when the relative amount of soluble HLA-B27 monomers was estimated. In FACS analysis, tetramer-positive antigen-specific T cells could be detected with both tetramers (Fig. 1d ), although HLA-A2 tetramers stained with greater intensity (log 0.8 more) than the HLA-B27. On the basis of these results we generated HLA-B27/ Chlamydia peptide tetramers. Generation of antigen-specific CD8 + T cell lines after MACS sorting of HLA-B27 tetramer-positive T cells To determine whether tetramer-binding CD8 + T cells could be sorted and further cultured we stained peripheral MNCs with HLA-B27/EBV EBNA (258–266) tetramer. Before MACS, 0.22% of peripheral MNCs were CD8 + and tetramer-positive. After MACS, EBV EBNA (258–266)-specific T cells were enriched to 41.4%. MACS-sorted cells were further separated by FACS sorting and cultured for 4 weeks in the presence of IL-2. The purity of antigen-specific CD8 + was increased to 95.0%, as shown by tetramer staining (Fig. 2a ). In parallel we performed intracellular cytokine staining after peptide-specific stimulation of peripheral MNCs and of the tetramer-sorted T cell line after 4 weeks of culture. In comparison with HLA-B27 tetramer staining, only 68.3% of these antigen-specific T cells were detected by intracellular cytokine staining of IFN-γ (Fig. 2b ). Generating HLA-B27/ Chlamydia peptide tetramers The generation of HLA-B27 tetramers with Chlamydia -derived peptides strongly indicated that the yield of refolded and soluble HLA-B27/ Chlamydia peptide monomers depended on the binding affinity of the peptide for HLA-B27. Gel-filtration analysis showed that Chlamydia peptide 133 (Table 2 ; binding score 25 in [ 23 ]) (Fig. 3a ) induced significantly more protein aggregation, seen by protein elution between 7 and 13 ml, than Chlamydia peptide 8 (Table 2 ; binding score 26 in [ 23 ] but 10,000 in BIMAS) (Fig. 3b ). In SDS-PAGE analysis the large quantity of aggregated proteins is also shown by numerous bands of higher molecular mass (Fig. 3a ). After the addition of streptavidin, the major band with biotinylated HLA-B27 molecule could be captured to become a tetramer (Fig. 3a,3b ; SDS-PAGE). This phenomenon of protein aggregation depending on the affinity between peptide and HLA-B27 could also be observed with the other Chlamydia -derived peptides. The refolding rate of all HLA-B27 tetramers used in this manuscript are summarized in Table 2 . HLA-B27/ Chlamydia peptide tetramer staining of synovial T cells On the basis of our recently identified Chlamydia -derived immunodominant peptides in Ct-ReA [ 10 ] we successfully synthesized nine HLA-B27 Chlamydia peptide tetramers and used them to stain MNCs from the synovial fluid of nine patients (six HLA-B27 + , three HLA-B27 - ) with Ct-ReA. Four of the six HLA-B27 + patients had a specific T cell binding to at least one HLA-B27/ Chlamydia peptide tetramer. The results of tetramer staining in all patients are summarized in Table 3 ; HLA-B27/ Chlamydia peptide 195 tetramer bound to the synovial T cells of three (patient nos 2, 3 and 5) of these four patients. Two patients (nos 5 and 6) showed a T cell response to Chlamydia peptide 133 as detected by tetramer staining, and one (patient no. 3) had a T cell response to Chlamydia peptide 68. The results of three patients are illustrated in Figs 4 and 5 . Figure 4a shows that T cells specific for Chlamydia peptides 195 and 68 were detected with HLA-B27/ Chlamydia peptide tetramers in patient no. 3: 0.09% of CD8 + T cells were positive for peptide 195 and 0.06% were positive for peptide 68. All other HLA-B27 tetramers with Chlamydia -derived peptides such as peptide 138 were negative (data not shown). In patient no. 2 we detected 0.06% HLA-B27/ Chlamydia peptide 195 tetramer-positive T cells (Fig. 4b ). All other HLA-B27 tetramers such as HLA-B27/ Chlamydia peptide 138 were negative (data not shown). We did not analyse the cytokine secretion profile of CD8 + T cells in response to chlamydial peptides in these two patients. The example of patient no. 6 is shown in Fig. 5a , with 0.02% of HLA-B27/ Chlamydia peptide 133 tetramer binding to CD8 + T cells but no binding to any of the other HLA-B27 Chlamydia peptide tetramers. In patient no. 6 we were able to repeat this experiment and obtained a similar result, with 0.02% of tetramer binding to CD8 + T cells. To confirm the specificity of the T cell response to peptide 133, two further experiments were performed in this patient. First, synovial T cells were expanded by Chlamydia peptide 133-specific T cell stimulation for 1 week ex vivo , which revealed 0.22% tetramer-positive CD8 + T cells (Fig. 5a ). Second, when FACS analysis of IFN-γ secretion after peptide-specific stimulation was done in the same patient, only peptide 133 induced this cytokine secretion (Fig. 5b ), again confirming the specificity of this response. We also analysed CD8 + T cells from peripheral blood of patient nos 2, 3, 5 and 6, who were responders when synovial fluid was tested for HLA-B27/ Chlamydia peptide binding, but we could not detect any specific binding (data not shown). The HLA-B27 - patients with Ct-ReA and all six HLA-B27 + healthy controls had no HLA-B27-restricted, Chlamydia -peptide-specific T cell response (data not shown). Tetramer staining of synovial T cells from three HLA-B27 + patients with enterobacteria-triggered ReA and from three patients with rheumatoid arthritis revealed no specific staining of CD8 + T cells with 0–0.01% tetramer binding to CD8 + T cells. Expansion of Chlamydia -specific CD8 + T cells after stimulation with Chlamydia -infected dendritic cells Because the frequency of Chlamydia -specific CD8 + T cells in these patients is low in synovial fluid and absent in peripheral blood with both methods (tetramer staining and intracellular cytokine staining), we investigated whether enrichment of these cells could be achieved by short-term stimulation with autologous Chlamydia -infected DCs. By doing this we intended to obtain a higher frequency of tetramer-positive CD8 + T cells, to underline the specificity of tetramer staining. We generated DCs from CD14 + monocytes from peripheral blood of patient no. 5; they were separated by MACS first. After 7 days of cultivation in vitro , the cells turned into DCs, as indicated by the loss of CD14 receptors and the upregulation of HLA-DR, CD80 and CD86 receptors (data not shown). We infected these DCs with viable Chlamydia trachomatis and confirmed infection by using an anti- Chlamydia trachomatis lipopolysaccharide antibody and by quantification of Chlamydia -positive cells by FACS analysis (data not shown). We revealed at least 41.3% Chlamydia -infected DCs. Peripheral MNCs from the same patient were stimulated with these Chlamydia -infected DCs for 2 weeks in the presence of IL-7 and IL-15. Subsequently, FACS analysis for intracellular cytokine staining for IFN-γ performed after restimulation of this cell line with Chlamydia -infected DCs revealed 0.11% IFN-γ-secreting CD8 + T cells, and stimulation with different peptide pools including the nine relevant peptides revealed between 0.07% and 0.21% antigen-specific IFN-γ-secreting CD8 + T cells (data not shown). When the cell line was analysed with HLA-B27/ Chlamydia peptide tetramers we found a similar quantity of expanded CD8 + T cells with significant tetramer staining of CD8 + T cells specific for Chlamydia peptides 8 (0.09%), 68 (0.10%), 133 (0.17%), 138 (0.08%), 195 (0.23%) and 196 (0.06%) (Fig. 6 ) and a weaker response to the other Chlamydia -derived peptides. HLA-B27 tetramer staining with peptides 133 and 195 showed some unusual bright staining, which was also frequently observed with the HLA-B27/EBV EBNA (258–266) tetramer and might have been caused by aggregated tetramers. Staining of untreated peripheral MNCs from the same patient did not reveal any tetramer binding (data not shown); staining with an HLA-B27/EBV peptide tetramer was performed as a positive control. We repeated this procedure in an HLA-B27 - patient with Ct-ReA (Fig. 7 ) and in an HLA-B27 + healthy blood donor (data not shown). In neither case could we observe staining with any of the HLA-B27/ Chlamydia peptide tetramers even after stimulation with Chlamydia -infected DCs (patient no. 9; Fig. 7 ). Discussion The arthritogenic peptide theory states that some HLA-B27 subtype alleles, owing to their unique amino acid residues, bind one or more specific arthritogenic peptides that are recognized by CD8 + T cells [ 6 - 9 ]. To test this theory it is of great importance to establish methods to identify the peptide specificity of such CD8 + T cells in human beings with HLA-B27-associated arthritis. The use of MHC class I tetramers to detect antigen-specific CD8 + T cells is well established [ 16 ]. However, surprisingly few publications present data with HLA-B27 tetramers. We have reported preliminary experiments with HLA-B27 tetramers in single patients with Ct-ReA [ 10 ]. HLA-B27 tetramers were also used to determine critical T cell receptor binding regions in HLA-B27-restricted T cells specific for an immunodominant peptide from influenza virus [ 19 ]. The biochemical features of the protein might be the limiting factor for using this molecule as frequently as other MHC class I molecules such as HLA-A2 tetramers. During the refolding process of recombinant HLA-B27, which is expressed in inclusion bodies, significant amounts of aggregated proteins occur [ 19 , 20 ]. The free cysteine residue at position 67 in the HLA-B27 α-chain is chemically highly reactive, causing homodimerization and protein aggregation [ 19 , 20 , 25 - 27 ]. It was therefore a reasonable strategy to generate HLA-B27 tetramers by substituting serine for cysteine at position 67 [ 10 , 19 ]. The mutated HLA-B27 heavy chain was also used in these experiments. However, even with the mutated HLA-B27 molecule we experienced significant protein aggregation when HLA-B27 molecules were generated with Chlamydia -derived peptides, especially with those with a low binding affinity for HLA-B27. We addressed the question of whether this finding was related to the protocols we used or whether it was specifically related to HLA-B27. We generated an HLA-B27 tetramer with a well-described immunodominant peptide from EBV and compared the results with those for an HLA-A2 molecule also loaded with an immunodominant peptide from EBV having almost the same binding affinity. The refolding rate of both molecules was almost the same, and we obtained comparable results when these molecules were used in FACS analysis. From this we concluded that the use of HLA-B27 tetramers is limited if the binding affinity of a peptide is too low for the molecule to remain stable. We therefore excluded peptides causing heavy protein aggregation and high background staining from further experiments. Here we have also demonstrated another useful property of HLA-B27 tetramers as a 'proof of principle'. We sorted antigen-specific tetramer-positive CD8 + T cells and generated highly specific T cell lines. After 4 weeks of non-specific stimulation, 95% of T cells were antigen-specific, which could be detected by tetramers but not by intracellular cytokine staining. The latter experiments detected only 68.3% IFN-γ secreting CD8 + T cells after antigen-specific stimulation. These results show clearly that HLA-B27 tetramers have the advantage of detecting antigen-specific T cells independently of their cytokine-secreting profile. Tetramers are also capable of detecting resting antigen-specific T cells, which probably constitute most non-IFN-γ-secreting CD8 + T cells of the T cell line in Fig. 4B (CD69 - and IFN-γ - ). Using HLA-B27-restricted Chlamydia peptides with higher binding scores, previously defined as immunodominant in Ct-ReA [ 10 ], we have generated tetramers and identified Chlamydia -peptide-specific T cell responses in four of six patients with Ct-ReA. These experiments suggest that Chlamydia peptides 195 and 133 are immunologically important epitopes, because we could detect CD8 + T cells with such specificity in three and two, respectively, out of six HLA-B27 + patients. We identified these antigen-specific T cells at a frequency of 0.02–0.09% in the synovial fluid of these patients, which is concordant with previous results [ 10 ]. The low frequency of Chlamydia -peptide-specific CD8 + T cells detected with HLA-B27 tetramers sometimes makes discrimination from non-specific staining difficult. We confirmed our tetramer staining result in one patient by expansion of peptide-specific CD8 + T cells followed by tetramer staining with increased amounts of tetramer-binding CD8 + T cells; even more importantly, peptide-specific CD8 + T cells were also detected by intracellular IFN-γ staining after peptide-specific stimulation. However, for future experiments it would be useful to confirm such findings with antigen-specific T cell expansion as shown here (Figs 2 , 5a and 6a ) and also in collaboration with other authors [ 28 ]. To underline further the specificity of HLA-B27 tetramer staining we performed peptide-specific expansion of CD8 + T cells specific for Chlamydia . For this, we generated Chlamydia -infected DCs, which are assumed to be excellent antigen-presenting cells for both CD4 + and CD8 + T cells [ 29 ], for Chlamydia -antigen-specific CD8 + T cell stimulation; we obtained antigen-specific T cell expansion. This Chlamydia -specific CD8 + T cell line showed an increased response to HLA-B27/ Chlamydia peptides 8, 68, 133, 138, 195 and 196 tetramers and a weaker response to the other HLA-B27/ Chlamydia peptide tetramers. The generation of CD8 + T cell lines with Chlamydia -infected DCs has recently been described, but without defining the MHC restriction and peptide specificity of such T cells [ 30 ]. Because we could not detect any Chlamydia -peptide-specific CD8 + T cells from peripheral blood with either method (tetramer staining and intracellular cytokine staining) without prior stimulation, we assume that the frequency of CD8 + T cells with such specificity in the peripheral blood is below the sensitivity of both methods. In contrast, low frequencies of Chlamydia -derived peptide-specific CD8 + T cells in the peripheral blood were observed by another group by using HLA-A2 tetramers [ 31 ]. These researchers detected Chlamydia trachomatis major outer membrane protein (MOMP) 258 peptide-specific and MOMP 249 peptide-specific CD8 + T cells in patients with acute urogenital tract infection. They found 0.01–0.2% MOMP-specific CD8 + T cells in the peripheral blood of these individuals with acute infection, who had no clinical symptoms of Ct-ReA. Conclusion We conclude that HLA-B27 tetramers are useful tools for the study of HLA-B27/peptide-specific T cells in HLA-B27-associated diseases. Although Chlamydia -specific HLA-B27-restricted CD8 T cells were detected in the synovial fluid of four of six HLA-B27 + patients with Chlamydia -induced reactive arthritis, their frequency was low, arguing against a major role in fighting Chlamydia and in the pathogenesis of arthritis. However, these antigens might be able to induce a cross-reactive T cell response to self-antigens, implying that the 'arthritogenic peptide' is not necessarily identical with the immunodominant peptide that is capable of inducing the T cell response to eliminate the microbe. Nevertheless, their frequency could be significantly expanded after stimulation in vitro with Chlamydia -infected autologous DCs, suggesting that these cells have full replicative capacity. Competing interests None declared. Abbreviations BIMAS = BioInformatics and Molecular Analysis Section; BSA = bovine serum albumin; Ct-ReA = Chlamydia -triggered reactive arthritis; DC = dendritic cell; EBV = Epstein–Barr virus; FACS = fluorescence-activated cell sorting; IFN-γ = interferon-γ ; IL = interleukin; MACS = magnetic activated cell sorting; MNC = mononuclear cell; MOMP = major outer membrane protein; PBS = phosphate-buffered saline; PE = phycoerythrin.

D:\keerthana\PMC001xxxxxx\PMC1064864.xml

1064865

Expression analysis of three isoforms of hyaluronan synthase and hyaluronidase in the synovium of knees in osteoarthritis and rheumatoid arthritis by quantitative real-time reverse transcriptase polymerase chain reaction

Hyaluronan is a major molecule in joint fluid and plays a crucial role in joint motion and the maintenance of joint homeostasis. The concentration and average molecular weight of hyaluronan in the joint fluids are reduced in osteoarthritis and rheumatoid arthritis. To elucidate the underlying mechanism, we analyzed the message expression of three isoforms of hyaluronan synthase and hyaluronidase from knee synovium, using real-time reverse transcriptase polymerase chain reaction. Synovia were obtained from 17 patients with osteoarthritis, 14 patients with rheumatoid arthritis, and 20 healthy control donors. The message expression of hyaluronan synthase-1 and -2 in the synovium of both types of arthritis was significantly less than in the control synovium, whereas that of hyaluronidase-2 in the synovium of both arthritides was significantly greater than in the control synovium. The decreased expression of the messages for hyaluronan synthase-1 and -2 and/or the increased expression of the message for hyaluronidase-2 may be reflected in the reduced concentration and decreased average molecular weight of hyaluronan in the joint fluids of patients with osteoarthritis and rheumatoid arthritis.

Introduction High-molecular-weight (HMW) hyaluronan (average molecular weight 6–7 × 10 6 Da) is a major component of synovial joint fluids [ 1 - 5 ]. It physically acts as a viscous lubricant for slow joint movements, such as walking, and as an elastic shock absorber during rapid movements, such as running [ 6 ]. HMW hyaluronan has a variety of biologic effects on cells in vitro , such as: the inhibition of prostaglandin E 2 synthesis and the release of arachidonic acid induced by interleukin-1 from cultured fibroblasts [ 7 , 8 ]; protection against proteoglycan depletion and cytotoxicity induced by oxygen-derived free radicals, interleukin-1, and mononuclear-cell-conditioned medium [ 9 , 10 ]; and the suppression of phagocytosis, of locomotion, and of enzyme release by leukocytes and macrophages [ 11 - 14 ]. HMW hyaluronan has been shown to suppress the degradation of cartilage matrix induced by fibronectin fragments [ 15 , 16 ] and cytokines [ 17 ]. Moreover, it has been shown to relieve joint pain by masking free nerve ending organelles in animal experiments [ 18 , 19 ]. Hence, it is suggested that HMW hyaluronan is an indispensable component in the maintenance of articular joint homeostasis. Reductions in the concentration and average molecular weight of hyaluronan in knee synovial fluids from patients with osteoarthritis (OA) or rheumatoid arthritis (RA) have been reported [ 2 , 3 , 20 - 25 ]. These reductions indicate hyaluronan's involvement in the pathogenesis of these joint disorders and are reflected in the pathological changes of hyaluronan metabolism. Hyaluronan is synthesized by hyaluronan synthases (HASs) located at the plasma membrane of cells [ 26 ]. Three HAS isoforms, encoded by three distinct genes, are expressed in human knee synovium [ 27 ]. It is believed that joint fluid hyaluronan is mainly supplied from type B cells – proper synoviocytes – of the synovial lining [ 2 - 5 , 28 ]. Little is known about hyaluronan catabolism in synovial fluid. It is thought that hyaluronan is eliminated by the lymphatic or vascular system after fragmentation by an unknown process [ 29 ] or that macrophagic type A cells of the synovial lining absorb and digest hyaluronan, because type A cells have many vesicles and vacuoles containing lysosomal enzyme – such as nonspecific esterase, acid phosphatase, and cathepsins B, D, and L – and type A cells are active in the uptake of substances in synovial fluids [ 28 ]. Hyaluronidase, which specifically degrades hyaluronan, is a lysosomal enzyme. Among five homologous isozymes in humans, hyaluronidase-1, -2, and -3 are thought to be expressed in synovium and involved in hyaluronan degradation, since hyaluronidase-4 is a chondroitinase and hyaluronidase-5, the sperm-specific enzyme PH-20, is specifically expressed in sperm [ 30 ]. The messages of hyaluronidase-1, -2, and -3 are expressed in chondrocyte monolayer cultures and in extracts of fresh human cartilage [ 31 ]. In the present study, we investigated message expression levels for three isoforms of HAS and hyaluronidase in knee synovium obtained from control donors and patients with OA or RA, by quantitative reverse transcriptase polymerase chain reaction (RT-PCR), in order to confirm whether message levels differed. Materials and methods Materials An RNeasy kit was purchased from QIAGEN KK (Tokyo, Japan). Primer Express computer software, gene-specific primer pairs and probes, TaqMan Gold RT-PCR reagents without controls, Pre-Developed TaqMan assay reagents of endogenous control human beta-actin, and a 7700 sequence detector were purchased from Perkin-Elmer Corp (Norwalk, CT, USA). The Hyaluronate-Chugai test kit was from Chugai Pharmaceutical Corp (Tokyo, Japan). Patients and controls Baseline data for patients with OA or RA and for control donors from whom synovial samples were obtained are summarized in Table 1 . Two of us (MY and SS), both physicians, clinically diagnosed OA and diagnosed RA according to the criteria of the American Rheumatism Association. Pharmacological treatment before sampling was limited to analgesics or nonsteroidal anti-inflammatory drugs in all study subjects. Rheumatoid arthritis patients were classified in stage II or stage III, and class II grade according to the Steinbrocker classification. The radiographic grades of all knee joints were determined on frontal views of the tibiofemoral joints according to the radiographic atlas recommended by the Osteoarthritis Research Society [ 32 ]. Grade B radiographic appearance, corresponding to grade 1 of the Larsen grading system, is defined by the presence of grade 1 joint space narrowing combined with osteophytes, or of grade 2 or 3 joint space narrowing. Control synovial samples were obtained from donors who had no intra-articular pathologic findings under arthroscopy at second-look observations after partial meniscectomy or from donors who complained of knee pain of unknown etiology but who had no intra-articular pathologic findings under arthroscopy on routine examination. The control synovium donors were significantly younger than the patients with OA or RA ( P < 0.01). Sampling of synovial tissues and isolation of total ribonucleic acid We obtained informed consent from all the study subjects and approval by the university ethical committee and the institutional review board. Synovial tissue samples were obtained from the central area of the suprapatellar pouches of the knees during arthroscopic examination, arthroscopic surgery, or open surgery performed in a hospital of the Jikei University School of Medicine. After subsynovial or fatty tissues were macroscopically resected from the obtained samples, all synovial samples were immediately frozen with liquid nitrogen and stored at -80°C. The total RNA of each sample was isolated using an RNeasy kit. Analysis of hyaluronan in joint fluid Joint fluid was aspirated from the knee immediately before an examination or surgery and was stored at -80°C. Joint fluid was obtained from 10 healthy control donors, 10 patients with OA, and 10 with RA. Hyaluronan concentration in joint fluid was measured by a sandwich binding protein assay using a Hyaluronate-Chugai test kit [ 33 ]. The molecular weight of hyaluronan was calculated from the intrinsic viscosity of hyaluronan in fluid, which was measured with a capillary viscometer [ 34 ] after pronase treatment. This method was chosen because it is more precise than HPLC analysis for the measurement of the average molecular weight of HMW hyaluronan. Analysis of message expression by quantitative real-time RT-PCR Message expression in the synovium of knees and the relative differences in message levels between the control group and patients with OA or RA were determined by real-time RT-PCR in accordance with the manufacturer's instructions and reported methods [ 35 - 39 ]. The gene-specific PCR oligonucleotide primer pairs and gene-specific oligonucleotide probes labelled with a reporter fluorescent dye (FAM) at the 5'-end and a quencher fluorescent dye (TAMURA) at the 3'-end were designed using the Primer Express computer software for HAS-1, -2, and -3 and hyaluronidase-1, -2, and -3 genes. For the HAS-2 probe, a minor groove binder probe was used to achieve an optimal melting temperature, because a suitable site for the regular probe was not found in the DNA sequences of HAS-2 (Table 2 ). A minor groove binder is an enhancer of the probe's melting temperature. Total RNA (200 ng for each) was added to a 50 μL RT-PCR reaction buffer containing 0.2 mmol/L deoxynucleotide triphosphates, 1.5 mmol/L MgSO 4 , 2.5 μmol/L random hexamers, 0.1 U/mL MultiScribe reverse transcriptase, 0.1 U/μL AmpliTaq Gold DNA polymerase, 900 nmol/L concentration of PCR primer pairs, 200 nmol/L concentration of the corresponding probe, and 2.5 μL Pre-Developed TaqMan assay reagents of endogenous control human β-actin, which contained β-actin-specific primers and probes labeled with a different reporter fluorescent dye (VIC). RT-PCR was carried out in a 96-well plate under the following conditions: one cycle at 50°C for 2 minutes to activate the uracil N -glycosylase, one cycle at 60°C for 30 minutes, one cycle at 95°C for 5 minutes, and 50 cycles at 95°C for 20 seconds and 60°C for 1 minute. The fluorescence energy emitted from the reporter dye without a quencher was monitored directly by a 7700 sequence detector in real time when the annealed probes were broken by DNA polymerases during the polymerization period. The threshold cycle numbers ( C T ), from which the logarithmic amplification phase of the PCR reaction started, were determined simultaneously for the messages of both target gene and β-actin gene in the same sample tube when the intensity of the reporter fluorescent signal reached 10 times the standard deviation of the baseline of fluorescent signal intensity. The C T value of the β-actin message was used as an internal standard. When the target messages were detected in both the control samples and the OA or RA samples by RT-PCR, the ratio for the amount of the message expressed in control samples to the amount of the message expressed in OA or RA samples was determined as a relative expression level. Relative expression levels of the target messages were calculated as follows: the Δ C T of each target message was obtained by subtracting the C T of β-actin message from the C T of each target message in the same RNA sample. The Δ C T values of the same target message between the control and OA or RA groups were analyzed statistically. When these Δ C T values were significantly different ( P < 0.01), the averageΔ C T value of each target message was calculated from all the Δ C T values. The ΔaverageΔ C T value of each message was obtained by subtracting the averageΔ C T of control samples from the averageΔ C T of OA or RA samples. Finally, the relative expression level of each target message was determined using the formula: relative expression level = 2 -ΔaverageΔ C T Statistical analysis Statistical analysis was with Wilcoxon's matched-pairs signed rank test. A probability value of <0.01 was considered statistically significant. Results Concentration and average molecular weight of hyaluronan in synovial fluid The concentration and average molecular weight of hyaluronan in the synovial fluid of OA or RA patients were significantly lower than those of control donors (Table 3 ). Expression profile of hyaluronan synthase isoform messages Expressed messages for all three HAS isoforms were detected in all synovial samples. The expression of the messages for HAS-1 and HAS-2 was significantly less in the synovium of OA than in the control synovium (83% and 48% of the respective control values), whereas no significant difference was observed for HAS-3 message expression. HAS-1 and HAS-2 message expression in RA synovium was significantly less than in control synovium (30% and 77% of the respective control values), while the expression of HAS-3 message was significantly greater than that in the control synovium (250% of the control value) (Fig. 1 ). Expression profile of hyaluronidase isozyme messages Message expression of all three hyaluronidase isozymes was detected in all synovial samples. Message expression for hyaluronidase-2 in OA synovium was significantly increased (to 430% of that in control synovium), while no significant differences were observed for hyaluronidase-1 and hyaluronidase-3 message expression. The expression of the message for hyaluronidase-2 was significantly greater (400% of the control value), while the expression of messages for hyaluronidase-1 and hyaluronidase-3 was significantly decreased in RA synovium (to 40% and 3% of the respective control values) (Fig. 1 ). Discussion The present study showed that HAS-1 and HAS-2 message expression was decreased in OA and RA synovium. This finding suggests that the protein expression of HAS-1 and -2 is decreased, as it has been reported that message levels are correlated with HAS protein levels and with the production of hyaluronan in cultured cells [ 40 ]. It has been suggested that the expression level of HAS proteins and their synthetic activities regulate the total volume of hyaluronan produced by cells, because detergent-purified HAS proteins alone can synthesize hyaluronan and no associated proteins or components are necessary for hyaluronan synthesis in vitro [ 41 ]. HAS activity of stable transfectants of HAS-2 is approximately 1.2 times that of HAS-1 or HAS-3 [ 42 ]. Stable transfectants of HAS-1 and HAS-3 produce hyaluronan with a broad size distribution (molecular weights of 2 × 10 5 Da to approximately 2 × 10 6 Da), whereas stable transfectants of HAS-2 produce hyaluronan with a broad size distribution that ranges higher (average molecular weight of >2 × 10 6 Da) [ 41 ]. Among HAS isoforms, the predominant message expressed in human knee synovium is HAS-1 [ 27 ]. Therefore, synovial production of hyaluronan, including HMW hyaluronan, may be decreased in OA or RA. A reduced production of HMW hyaluronan may be involved in the pathogenesis of these joint disorders, since HMW hyaluronan has important physical and biologic functions, as described in the Introduction. An age-associated change in synoviocyte population revealed that the number of type B cells was significantly decreased in older animals, although this was not confirmed in humans [ 28 ]. The message levels of all three HAS isoforms were not uniformly decreased in the knee synovium of OA or RA patients, even though the patients were significantly older than the control donors. Hence, it is unclear whether the different expression profiles of HAS messages in the controls, OA and RA patients are attributable to age-associated change, to physical senility, or to a pathologic factor specific for arthritic joint disorders. Hyaluronidase activity was detected in human knee synovial fluid of OA or RA patients when the assay was performed at the acidic pH of 4.5, but not at pH 5.0–7.0 [ 43 ]. Hyaluronidase-1 may be present in the fluids, because it is a major isozyme in plasma and urine and is unable to bind hyaluronan at neutral pH [ 30 ]. We suggest that soluble forms of hyaluronidases in synovial fluids are not involved in the direct digestion of hyaluronan in joint fluids, because a neutral pH is maintained in synovial fluids, and so hyaluronidase-1 may function only within lysosomes. Hyaluronidase-2 is linked to the outer cell membrane by a glycosylphosphatidyl-inositol (GPI) anchor and it digests hyaluronan to intermediate-sized fragments of approximately 20 kDa, while hyaluronidase-1 digests hyaluronan to tetrasaccharides [ 30 ]. A process of hyaluronan catabolism in somatic cells proposed in the review literature [ 30 ] is that hyaluronan is taken up into unique endocytic vesicles by an unknown mechanism and is digested into 20-kDa fragments by hyaluronidase-2 located in vesicles at an acidic pH; subsequently, the fragments are transported into lysosomes, where hyaluronidase-1 and two exoglycosidases digest hyaluronan into monosaccharides. The present study showed that the message expression of hyaluronidase-2 in the synovium of OA and RA was approximately four times that in the control synovium. This finding suggests that in OA and RA, the protein expression of hyaluronidase-2 in the synovium is increased and the hyaluronan digestion by hyaluronidase-2 is accelerated. Little is known about hyaluronidase-3. Strong hybridization expression patterns are found in mammalian testis and bone marrow [ 30 ]. Hyaluronidase-3 message expression was detected in synovium in the present study. This isozyme may work only in the lysosomes, as does hyaluronidase-1 [ 30 ]. The expression level in RA synovium was significantly lower than in OA or control synovium. The reduction in message expression may be due to the different cellular populations found in OA versus RA, since many inflammatory cells such as granulocytes or lymphocytes appeared in RA synovium. Joint fluid hyaluronan concentration is determined by the production volume of hyaluronan, the elimination volume of hyaluronan from the joint, and the total volume of joint fluid. The production of hyaluronan in OA or RA may be decreased because of the reduced expression of HAS-1 and -2 messages. The elimination volume of hyaluronan may be increased by the elevated expression of hyaluronidase-2, because hyaluronidase-2 digests hyaluronan in the endosome after uptake of hyaluronan into cells [ 30 ]. Hence, it is thought that the decreased expression of HAS-1 and -2 and/or the increased expression of hyaluronidase-2 are among the causes leading to the reduced hyaluronan concentration in OA or RA synovial fluid. The average molecular weight of hyaluronan in synovial fluid is determined by the molecular weights of hyaluronan produced and hyaluronan digested in the fluid. The average molecular weight of newly produced hyaluronan may be reduced by the decreased expression of HAS-2, because, of the three HAS isoenzymes, HAS-2 synthesizes the highest-molecular-weight hyaluronan [ 42 ]. The decreased expression of HAS-2 may be one of the causes for the reduced average molecular weight of hyaluronan in joint fluid. Moreover, there may be a mechanism whereby HMW hyaluronan is digested into low-molecular-weight hyaluronans in synovial fluid, since the average molecular weight of hyaluronan in OA or RA fluid is lower than that of hyaluronan synthesized by HAS-1 or -3. HAS-3 message expression was increased in RA synovium, although hyaluronan concentration was reduced. The increased expression of HAS-3 message may be due to the increased number of inflammatory cells invading the pannus tissue (which is inflammatory and proliferative granular synovial tissue specific for RA), since a high expression level of HAS-3 message in inflammatory cells was observed in another study by two of us (NI and KK). It is supposed that the hyaluronan produced by inflammatory cells does not diffuse into the joint cavity and that it surrounds cells, protecting them or aiding their migration, because it has been reported that pannus tissue with inflammatory cells contains a greater amount of hyaluronan than is found in OA or traumatic injury [ 44 ]. Conclusion Message expression for three isoforms of hyaluronan synthase and hyaluronidase in knee synovium differs in OA or RA from that in healthy controls. Differential expression of hyaluronan synthases and/or hyaluronidases may be reflected in the pathological metabolism of hyaluronan in the knee synovial fluid of patients with OA or RA. Competing interests None declared. Abbreviations HAS-1/-2/-3 = hyaluronan synthase-1/-2/-3; HMW = high-molecular-weight; HPLC = high-performance liquid chromatography; OA = osteoarthritis; PCR = polymerase chain reaction; RA = rheumatoid arthritis; RT-PCR = reverse transcriptase polymerase chain reaction; SD = standard deviation.

D:\keerthana\PMC001xxxxxx\PMC1064865.xml

1064866

Analysis of HLA DR, HLA DQ, C4A, FcγRIIa, FcγRIIIa, MBL, and IL-1Ra allelic variants in Caucasian systemic lupus erythematosus patients suggests an effect of the combined FcγRIIa R/R and IL-1Ra 2/2 genotypes on disease susceptibility

Dysfunction in various parts of immune defence, such as immune response, immune complex clearance, and inflammation, has an impact on pathogenesis in systemic lupus erythematosus (SLE). We hypothesised that combinations of common variants of genes involved in these immune functions are associated with susceptibility to SLE. The following variants were analysed: HLA DR3, HLA DQ2, C4AQ0, Fcγ receptor IIa (FcγRIIa) genotype R/R, Fcγ receptor IIIa (FcRγIIIa) genotype F/F, mannan-binding lectin (MBL) genotype conferring a low serum concentration of MBL (MBL-low), and interleukin-1 receptor antagonist (IL-1Ra) genotype 2/2. Polymorphisms were analysed in 143 Caucasian patients with SLE and 200 healthy controls. HLA DR3 in SLE patients was in 90% part of the haplotype HLA DR3-DQ2-C4AQ0, which was strongly associated with SLE (odds ratio [OR] 2.8, 95% CI 1.7–4.5). Analysis of combinations of gene variants revealed that the strong association with SLE for HLA DR3-DQ2-C4AQ0 remained after combination with FcγRIIa R/R, FcγRIIIa F/F, and MBL-low (OR>2). Furthermore, the combination of the FcγRIIa R/R and IL-1Ra 2/2 genotypes yielded a strong correlation with SLE (OR 11.8, 95% CI 1.5–95.4). This study demonstrates that certain combinations of gene variants may increase susceptibility to SLE, suggesting this approach for future studies. It also confirms earlier findings regarding the HLA DR3-DQ2-C4AQ0 haplotype.

Introduction The genetic contribution to the aetiology of systemic lupus erythematosus (SLE) is high, as is indicated by familial aggregation and a higher concordance rate in monozygotic than dizygotic twins [ 1 ]. The major histocompatibility complex (MHC) haplotype HLA DR3-DQ2-C4AQ0 is strongly associated with SLE in Caucasians [ 2 , 3 ]. The IgG Fc receptors appear to be important in the pathogenesis of SLE, as recently reviewed by Salmon and Pricop [ 4 ]. With the allelic variant of R (arginine) instead of H (histidine) on amino acid position 131, the ability of Fcγ receptor IIa (FcγRIIa) to bind IgG 2 is diminished [ 5 ]. Similarly, an amino acid substitution in position 158 (phenylalanine [F] instead of valine [V]) in the Fcγ receptor IIIa (FcγRIIIa) reduces the IgG 1 -, IgG 3 -, and IgG 4 -binding capacity of the receptor [ 6 ]. These variants can result in suboptimal clearance of immune complexes from the circulation, which might contribute to the pathogenesis of immune-complex-mediated manifestations [ 7 ]. Mannan-binding lectin (MBL) is structurally similar to C1q and has the ability to activate the complement cascade through the lectin pathway. Point mutations are found in the structural gene that affect the MBL serum concentration and the stability of MBL complex formation required for efficient complement activation [ 8 ]. In the promoter regions, there are two polymorphisms that influence serum concentration, with LX conferring the lowest MBL level, LY a medium level, and HY the highest [ 8 - 11 ]. MBL variant alleles have been suggested as a minor risk factor in susceptibility to SLE in several populations [ 8 , 10 , 12 ]. Interleukin-1 receptor antagonist (IL-1Ra) is a naturally occurring competitive inhibitor of IL-1. The IL-1Ra gene contains a polymorphism in intron 2 consisting of a variable number of copies of an 86-base-pair repeat sequence (two, three, four, five, or six copies) [ 13 ]. An association has been found between the IL-1Ra 2 allele and SLE [ 13 , 14 ]. Multiple genes are involved in the development of SLE, and the relative importance of these genes may vary between populations and with environmental exposure. We investigated common variant alleles involved in the immune response, immune complex clearance, and regulation of inflammation, with the hypothesis that combinations of polymorphic candidate genes could have synergistic effects on disease susceptibility. Therefore, we have analysed polymorphisms in the genes HLA DR, HLA DQ, C4A, FcγRIIa, FcγRIIIa, MBL, and IL-1Ra and their association with the development of SLE. Materials and methods Patients The study population comprised 124 female and 14 male Caucasian SLE patients, and 200 blood donors (100 men, 100 women) were used as controls. One hundred thirty-eight patients fulfilled four or more criteria of the American College of Rheumatology (ACR) classification for SLE [ 15 ]. Five patients with a clinical SLE diagnosis were included in the study even though they fulfilled only three ACR classification criteria; these five patients had multisystemic disease with an immunologic disorder, i.e. presence of anitnuclear antibodies and symptoms characteristic of SLE such as arthritis, photosensitivity, serositis, nephritis, thrombocytopenia, and leucopenia [ 16 ]. A breakdown of the ACR criteria is shown in Table 1 . There were 129 families with a single case of SLE and 14 families in which multiple cases were recorded. However, from each multicase family, only the first family member with SLE diagnosis, the index case, was included in the statistical analysis. The mean age at diagnosis of the patients was 40 years (range 10–83) and the mean disease duration was 16 years (range 1–42). The mean Systemic Lupus International Collaborating Clinics/ACR-Damage Index score was 1.9 (range 0–9) [ 17 ]. The study was approved by the local ethics committee at Lund University. Genetic analyses DNA was extracted by the salting-out method described by Miller and colleagues [ 18 ]. Analysis of genetic polymorphism was predominantly performed by polymerase chain reaction (PCR). HLA HLA DR and DQ alleles were determined with PCR (Olerup SSP™ DQ-DR SSP Combi Tray, Olerup SSP AB, Stockholm, Sweden). However, a minority of the patients had previously been typed with a lymphocytotoxicity test or by restriction fragment length polymorphism as described before [ 2 ]. C4A gene deletion was determined by PCR as described by Grant and colleagues [ 19 ], or in a few cases by analysis of restriction fragment length polymorphism and determination of MHC haplotypes [ 2 ]. With the presence of a DR3 allele together with a DQ2 and a C4AQ0 allele, due to C4A gene deletion, the subject was considered to have the haplotype HLA DR3-DQ2-C4AQ0, although family studies were not uniformly performed to confirm this assumption. FcγRIIa gene polymorphism The genetic polymorphism resulting in amino acid R or H in amino acid position 131 was determined as previously described [ 20 ]. Analysis of FcγRIIIa gene polymorphism The analysis of the F/V polymorphism was performed essentially as previously described [ 21 ]. MBL gene polymorphism Variants of MBL due to mutations at codon 52 (D), 54 (B), and 57 (C) in exon 1 of the MBL gene and promotor variants at position -550 (H/L) and -221 (X/Y) were determined by allele-specific PCR amplification, essentially as described before [ 9 ]. The wild-type structural allele is designated A, while 0 is a description of the mutant alleles B, C, and D. Based on previously described associations between MBL genotype and serum concentrations, which were confirmed in our 200 healthy controls, the MBL genotypes were divided into three groups. Group 1 (MBL-low) consisted of patients with two structural mutant alleles (0/0) or on one haplotype a structural mutant allele together with another haplotype containing an LX promoter and the wild-type structural allele (ALX/0). Group 2 (MBL-intermediate) consisted of patients with the promoters LX conferring low serum MBL on both haplotypes but with normal structural alleles (ALX/ALX), or, alternatively, haplotypes with one mutant and one wild-type structural allele with a non-LX promoter together with the wild-type allele. Group 3 (MBL-high) included patients with the A/A genotype and at least one non-LX promoter. IL-1Ra gene polymorphism Genetic polymorphism in the IL-1Ra gene was determined with a PCR essentially as previously described [ 13 , 22 ], although one primer was modified. Primers: 5'-CTC AGC AAC ACT CCT AT-3' 5'-TTC CAC CAC ATG GAA C-3' The amplified fragment size depends on the number of repeats (two repeats, designated allele 2; three, allele 4; four, allele 1; five, allele 3; six, allele 5). Statistics Two group comparison tests were performed using the Fisher exact test. Comparisons between multiple groups were made using the χ 2 multiple comparison test. Significance was considered when P <0.05. Correction for multiple comparisons was not applied to the results, because the study design consisted in hypothesis testing. The presence of synergistic interaction between genetic variants was investigated by calculating relative excess risk due to interaction (RERI) [ 23 ]. Results A strong association between the HLA DR3-DQ2-C4AQ0 haplotype and SLE was found, although this haplotype also was common among the controls. HLA DR2 was present in 50 of the 143 SLE patients and 72 of the 200 controls, while DR4 frequencies were 45/143 and 72/200, respectively. In the SLE group, HLA DQ2 was present in 80 of 143 cases, while DQ3 and DQ6 was recorded in 60 of 143 and 85 of 143 cases, respectively. The corresponding numbers in the control group were for DQ2, 73/200; for DQ3, 100/200; and for DQ6, 112/200. Other DR and DQ variants were less common. Ninety percent of the SLE patients with HLA DR3 displayed the haplotype DR3-DQ2-C4AQ0, compared with 86% of the controls. The frequencies of the FcγRIIa, FcγRIIIa, MBL, and IL-1Ra genotypes are displayed in Fig. 1 . The FcγRIIa R/R, FcγRIIIa F/F, IL-1Ra 2/2, and MBL-low genotypes were not individually associated with SLE. Additionally, the combination of genetic variants and susceptibility to SLE was tested (Table 2 ). HLA DR3-DQ2-C4AQ0 in combination with FcγRIIa R/R, FcγRIIIa F/F, or MBL-low was still associated with SLE but did not significantly increase the odds ratio (OR) in comparison with HLA DR3-DQ2-C4AQ0 alone. A combination of FcγRIIa R/R and IL-1Ra 2/2 yielded a strong association with SLE (OR 11.8), although the confidence interval was wide (1.5–95.4). Testing of RERI did not confirm the hypothesis that this interaction was synergistic (RERI 11.1, 95% CI -13.8 – 36.1, P = 0.38). A combined analysis of carriage rates for the R allele and the 2 allele (i.e. the patient should have at least one R allele and one 2 allele) was also performed, but no significant difference was detected between the SLE and the control group. No other combination displayed any association with SLE. Discussion The increasing number of reports on polymorphic genes involved in susceptibility to SLE prompted us to investigate whether a combination of polymorphic candidate genes, tentatively thought to be involved in the pathogenesis of SLE, could further elucidate the genetic basis of the disease. In the present study we found that the combination of the FcγRIIa R/R genotype with the IL-1Ra 2/2 genotype was strongly associated with SLE. Although only a few of the patients had this particular genetic background, the results indicate that certain combinations of susceptibility genes can be of crucial importance. Furthermore, a strong association between the haplotype HLA DR3-DQ2-C4AQ0 and susceptibility to SLE was seen in this study, which is in concordance with the findings of previous studies [ 2 , 22 , 24 , 25 ]. The patients and controls studied were all from a homogeneous Caucasian population, although a possible bias exists in the fact that the controls used were blood donors, which principally include only healthy individuals, instead of age-matched controls from the normal population. The distributions of the polymorphic variants in the controls were in agreement with data published by others [ 13 , 26 , 27 ]. There have been ample studies on the association between FcγRIIa and SLE [ 24 , 28 - 30 ]. However, the results are somewhat conflicting regarding whether or not the R allele is associated with increased susceptibility to SLE in general or for SLE glomerulonephritis or other clinical manifestations of SLE. In our study, there was no association between either the R allele or the R/R genotype and susceptibility to SLE, with a glomerulonephritis frequency of 27%. The MBL genotype did not seem to be involved in susceptibility to SLE in our Caucasian cohort. This differs from a finding of a recent meta-analysis in which MBL variant alleles were found to be associated with SLE [ 27 ]. Furthermore, in that study the conclusion was drawn that several studies are too small to detect an increased SLE susceptibility dependent on MBL risk alleles, which could also explain the lack of association in our study. An increased carriage rate of the 2 allele of the IL-1Ra gene has been shown for SLE patients [ 13 , 14 ]. In our study, the 2/2 genotype in conjunction with the FcγRIIa R/R genotype was associated with SLE. This IL-1Ra genotype is associated with higher IL-1 beta concentrations as well as higher serum IL-1Ra levels [ 31 , 32 ]. Furthermore, immune complex binding to Fc receptors can influence the production of IL-1Ra [ 33 ], which provides a possibility for a pathogenetic mechanism concordant with the genetic interaction seen in our study. Analyses of disease phenotypes were beyond the scope of this study and will be addressed in future studies. However, there were no apparent associations between the various genotypes and clinical subsets of SLE. Because of the low number of patients included in the study, the results must be interpreted cautiously, and independent confirmation is needed. Conclusion Our findings suggest that the combination of the FcγRIIa R/R and IL-1Ra 2/2 genotypes is associated with SLE in Caucasian patients, whereas individually these genotypes do not increase susceptibility to the disease. This finding illustrates that combinations of polymorphic genes may act in concert in the pathogenesis of SLE, a concept that may be instrumental in the analysis of the genetics of SLE as well as providing hypotheses for pathways in the pathogenesis of lupus. Competing interests None declared. Author contributions AJ was responsible for data analysis and interpretation and wrote the report. AAB contributed to the data analysis and interpretation. GS and LT were both responsible for the planning of the work and contributed to data analysis, interpretation, and write-up. Abbreviations ACR = American College of Rheumatology; F = phenylalanine; FcγRIIa = Fcγ receptor IIa; FcγRIIIa = Fcγ receptor IIIa; H = histidine; IL-1Ra = interleukin-1 receptor antagonist; MBL = mannan-binding lectin; MBL-low/-intermediate/-high = MBL genotype conferring a low/intermediate/high serum concentration of MBL; MHC = major histocompatibility complex; OR = odds ratio; PCR = polymerase chain reaction; R = arginine; RERI = relative excess risk due to interaction; SLE = systemic lupus erythematosus; V = valine.

D:\keerthana\PMC001xxxxxx\PMC1064866.xml

1064867

Intrathecal levels of matrix metalloproteinases in systemic lupus erythematosus with central nervous system engagement

Symptoms originating from the central nervous system (CNS) occur frequently in patients with systemic lupus erythematosus (SLE), and CNS involvement in lupus is associated with increased morbidity and mortality. We recently showed that neurones and astrocytes are continuously damaged during the course of CNS lupus. The matrix metalloproteinases (MMPs) are a group of tissue degrading enzymes that may be involved in this ongoing brain destruction. The aim of this study was to examine endogenous levels of free, enzymatically active MMP-2 and MMP-9 in cerebrospinal fluid from patients with SLE. A total of 123 patients with SLE were evaluated clinically, with magnetic resonance imaging of brain and cerebrospinal fluid (CSF) analyses. Levels of free MMP-2 and MMP-9 were determined in CSF using an enzymatic activity assay. CSF samples from another 22 cerebrally healthy individuals were used as a control. Intrathecal MMP-9 levels were significantly increased in patients with neuropsychiatric SLE as compared with SLE patients without CNS involvement ( P < 0.05) and healthy control individuals ( P = 0.0012). Interestingly, significant correlations between MMP-9 and intrathecal levels of neuronal and glial degradation products were noted, indicating ongoing intrathecal degeneration in the brains of lupus patients expressing MMP-9. In addition, intrathecal levels of IL-6 and IL-8 – two cytokines that are known to upregulate MMP-9 – both exhibited significant correlation with MMP-9 levels in CSF ( P < 0.0001), suggesting a potential MMP-9 activation pathway. Our findings suggest that proinflammatory cytokine induced MMP-9 production leads to brain damage in patients with CNS lupus.

Introduction Central nervous system (CNS) involvement has been reported to occur in 14–75% of all systemic lupus erythematosus (SLE) patients [ 1 - 3 ]. However, frequency rates vary considerably, depending on the diagnostic criteria applied. CNS lupus can occur at any time during the course of SLE, and its symptoms are diverse. The features of this condition can include seizures, stroke, depression, psychoses and disordered mentation. Neuropsychiatric involvement in SLE (NPSLE) has been shown to predict a high frequency of flares, and it is considered a major cause of long-standing functional impairment as well as a cause of mortality [ 4 ]. Over the past decade CNS lupus has been treated with cytotoxic drugs, which improve disease outcome [ 5 , 6 ]. Because of the multiple pathogenic mechanisms that cause manifestations of CNS lupus, no single confirmatory diagnostic test is available. Several clinical, laboratory and radiographic test findings are reported to be abnormal in some but not all patients with CNS lupus. Magnetic resonance imaging (MRI) of brain has been shown to be valuable in detecting even minor lesions caused by CNS lupus, and these correlate with CNS manifestations in SLE [ 7 ]. Pleocytosis and elevated protein levels are found in some but not all patients with CNS lupus [ 8 , 9 ]. Elevated concentrations of IgG in cerebrospinal fluid (CSF), IgG:albumin ratio and IgG index, and the presence of oligoclonal bands have all been described with varying frequencies in patients with NPSLE [ 10 - 12 ]. Few studies have demonstrated elevated IL-6 levels in CSF from patients with CNS lupus [ 13 - 17 ]. Some other reports have described increased levels of IL-1 [ 13 ], IL-8 [ 15 ] and interferon-γ [ 18 ] in CSF from patients with CNS lupus. We recently reported neuronal damage, astrogliosis and formation of toxic metabolic products such as Aβ42 in patients with NPSLE [ 19 ]. One of the prototypical destructive events in the human brain, initiated by the release of inflammatory cytokines and ending with tissue destruction, is production of matrix metalloproteinases (MMPs). The MMPs are a family of endopeptidases produced by a variety of inflammatory cells [ 20 ]. All of the cell types that exist in the CNS are potential sources of MMPs. In vitro , neurones, astrocytes, microglia [ 21 , 22 ] and oligodendrocytes [ 23 ] express various MMP family members, and production of MMPs by neuronal cells can be upregulated by several inflammatory cytokines. MMPs have a multitude of regulatory functions, including control of influx of inflammatory mononuclear cells into the CNS, participation in myelin destruction and disruption of the integrity of the blood–brain barrier [ 24 ]. With respect to MMP-9, it was recently shown that CSF levels of this enzyme increase during bacterial meningitis and that it is associated with brain damage [ 25 , 26 ]. The aim of the present study was to measure levels of free active MMP-2 and MMP-9 in CSF of SLE patients with CNS lupus, and to relate these data to clinical and laboratory measures of brain parenchymal degradation. Our results suggest that MMP-9 but not MMP-2 actively participates in brain destruction in CNS lupus. Methods Participants A total of 122 patients fulfilled four or more of the American Rheumatism Association 1987 revised criteria for the classification of SLE [ 27 ]. The patients (106 females and 16 males, aged 17–91 years [mean age ± standard deviation 42 ± 14 years]) were being treated at the Department of Rheumatology at Sahlgrenska University Hospital. Disease duration varied between 0 and 49 years (mean duration 8 ± 9 years). The patients were consecutively included in the study. They underwent a thorough clinical examination by experienced staff rheumatologist, neurologist and neuropsychologist. Examination of CNS signs and symptoms included lumbar puncture, neuropsychological tests and MRI of the brain. The proposed definition of CNS lupus in the American Rheumatism Association criteria for SLE [ 27 ] appears inadequate, given that only two elements – psychosis and seizures – are included. As previously described [ 28 ], we defined CNS lupus as the presence of at least two of the following seven items in association with clinical evidence of disease progression: recent onset psychosis, transverse myelitis, aseptic meningitis, seizures, pathological brain MRI, severely abnormal neuropsychiatric test findings [ 29 ] and oligoclonal IgG bands in CSF. The pathogenesis of antiphospholipid antibody mediated brain damage is a thrombotic rather than inflammatory complication of SLE, and so we decided to exclude this condition from the definition of CNS lupus. Non-SLE causes of neurological events (e.g. cerebral infections) were also ruled out. Based on the criteria above, patients were divided into three distinct groups: patients with CNS lupus ( n = 43); patients with SLE but without any signs of CNS involvement ( n = 71); and patients with SLE complicated by antiphospholipid syndrome or patients who met the CNS lupus criteria but were ruled out because of non-SLE origin of neurological events ( n = 9). The frequencies of various CNS manifestations and patient treatments in the SLE patients are summarized in Tables 1 and 2 , respectively. The study was approved by the ethical committee of the University of Göteborg. Control individuals CSF from 22 healthy individuals (mean age ± standard deviation: 38 ± 11 years; 12 females and 10 males), without previous history of neurological disorder and with normal neurological status, served as control individuals. There were no significant differences between males and females with respect to intrathecal levels of MMP-9 (0.17 ± 0.17 pg/ml versus 0.30 ± 0.30 pg/ml; not significant) or intrathecal levels of MMP-2 (388 ± 48 pg/ml versus 379 ± 48 pg/ml; not significant). Cerebrospinal fluid analyses Levels of MMP-2 and MMP-9 were determined using an activity assay system (Amersham Pharmacia Biotech, Buckinghamshire, UK), which was constructed to measure enzymatically active forms of MMP-2 and MMP-9. The detection level was 190 pg/ml for MMP-2 and 125 pg/ml for MMP-9. All values below the detection levels were considered negative. Paired serum and CSF samples were analyzed for albumin and IgG levels using nephelometry. As an indicator of blood–brain barrier function, the quotient of CSF albumin × 10 3 /serum albumin was analyzed (normal values <6.8 [<45 years of age] and <10.2 [>45 years of age]) [ 30 ]. The CSF/serum IgG index was used as a measure of intrathecal IgG production and calculated by using the following formula (normal value <0.7): (CSF IgG × 10 3 )/([CSF albumin × 10 3 ]/serum albumin). All CSF samples were also analyzed by isoelectric focusing to permit detection of oligoclonal IgG bands. Magnetic resonance imaging analyses Neuroimaging was performed to evaluate the extent and localization of brain lesions. The neuroimaging technique used was multiplanar MRI. The MRI examinations (Philips Gyroscan T5-II, Einhoven, The Netherlands) were performed with axial proton density and T 2 -weighted images of the brain. MRI abnormalities were seen in 72% of patients with CNS lupus and in 33% of SLE cases classified as cerebrally healthy. Statistical analysis Statistical comparisons were made using the nonparametric Mann–Whitney U-test or, in case of follow-up data, the Wilcoxon's test for paired data. Results are presented as means ± standard error of the mean. P < 0.05 was considered statistically significant. Spearman-rank correlation was used for calculation of correlation. The statistical analyses were conducted using the Statview ® (SAS, Cary, NC, USA) program. Results A total of 123 patients met criteria for SLE. Forty-three patients were found to have NPSLE (in accordance with the criteria presented in the Methods section above), and nine patients either were found to have phospholipid antibody syndrome or met the CNS lupus criteria but were excluded because of non-SLE origin of neurological events. The remaining 71 SLE patients were considered to be cerebrally healthy. Mild pleocytosis was seen in patients with CNS lupus (9 × 10 6 ± 6 × 10 6 cells/l) as compared with cerebrally healthy SLE patients (2 × 10 6 ± 0.5 × 10 6 cells/l; not significant). As previously validated [ 31 ], we found an increased number of oligoclonal bands in CSF from the CNS lupus group (1.7 ± 0.4) as compared with SLE patients without CNS involvement (0.5 ± 0.1; P < 0.05). The mean level of CSF:serum albumin ratio was not increased in patients with NPSLE (6.1 ± 0.7 mg/dl) as compared with cerebrally healthy SLE patients (5.2 ± 0.3 mg/dl; not significant). There were no significant differences in levels of serum antibodies specific for native DNA or in complement levels (C3 and C4) between SLE patients with and those without CNS involvement. Intrathecal MMP-9 levels were significantly increased in all SLE patients as compared with cerebrally healthy control individuals (153 ± 27 versus 0.28 ± 0.16 pg/ml; P = 0.016). In CNS lupus patients MMP-9 levels were significantly increased, both compared with cerebrally healthy SLE patients (240 ± 60 versus 100 ± 20 pg/ml; P < 0.05; Fig. 1 ) and compared with cerebrally healthy control individuals (240 ± 60 versus 0.28 ± 0.16 pg/ml; P = 0.0012). On stratification of all SLE patients into two groups with respect to the presence or absence of brain MRI pathology, we found increased CSF MMP-9 levels in SLE patients with MRI pathology as compared with those without (160 ± 50 versus 140 ± 30 pg/ml; not significant). There were no statistically significant differences in CSF MMP-2 levels between NPSLE and patients without CNS lupus (499 ± 70 versus 555 ± 70 pg/ml; not significant) or compared with cerebrally healthy control individuals (499 ± 70 versus 384 ± 33 pg/ml; not significant; Fig. 2 ). CSF levels of IL-6 (47 ± 25 pg/ml versus 15 ± 3 pg/ml; P < 0.008) and IL-8 (91 ± 23 pg/ml versus 45 ± 6 pg/ml; P < 0.05) were both significantly increased in CSF from patients with CNS lupus as compared with cerebrally healthy SLE patients, supporting our previous findings [ 15 , 19 ]. Importantly, intrathecal levels of IL-6 and IL-8 significantly correlated with those of MMP-9 (r = 0.30 [ P < 0.002] and r = 0.47 [ P < 0.0001], respectively; Table 3 and Fig. 3 ). A neuronal degeneration marker (protein tau) and an astrocytic degeneration marker (glial fibrillary acidic protein) were both significantly increased in CSF from NPSLE patients as compared with CSF from SLE patients who were clinically free from CNS involvement (311 ± 78 pg/ml versus 178 ± 16 pg/ml [ P < 0.05] and 1288 ± 708 pg/ml versus 396 ± 30 pg/ml [ P < 0.009]), in concordance with previous findings [ 28 , 32 ]. Importantly, a significant correlation was noted between intrathecal MMP-9 and levels of tau and glial fibrillary acidic protein ( P < 0.05 in both cases; Table 3 ). Discussion We previously reported that patients with CNS lupus express increased intrathecal levels of proinflammatory cytokines IL-6, IL-8 and interferon-γ. Furthermore, we recently demonstrated that inflammation in CNS during SLE leads to neurodegeneration, which manifests as increased levels of neuronal and astrocytic degradation products [ 28 , 32 ]. However, the mechanisms responsible for the brain damage occurring during CNS lupus have never been clarified. It was previously demonstrated [ 33 , 34 ] that proinflammatory cytokines are able to trigger production and release of tissue-derived MMPs. It is also established that most of the resident cells in human brain have the capacity to produce MMP-2 and MMP-9 – enzymes that are known to participate in brain damage (e.g. in case of infectious meningitis) [ 35 - 37 ]. In the present study we found significantly higher levels of MMP-9 in CSF from NPSLE patients than in CSF from SLE patients without CNS lupus and healthy control individuals. This finding may be of clinical significance because of correlation between MMP-9 and levels of neuronal/astrocytic degradation products in CSF, reflecting the potential of MMP-9 to damage brain parenchyma. Findings of a relationship between MMP-9 and brain damage in SLE are new and should be scrutinized critically. We believe that effort should be invested in analyzing free (i.e. non-tissue inhibitor of metalloproteinase bound) and metabolically active enzyme to ascertain the validity of our conclusions. The method used in the present study fulfils both of these requirements. MMP-9 plays an essential role in the breakdown of extracellular matrix molecules at the blood–brain barrier. This, together with locally elevated levels of the chemokine IL-8, greatly facilitates the transmigration of activated inflammatory cells across the endothelium. Indeed, a significant correlation was observed between the intrathecal levels of MMP-9 and IL-8. A broad range of cytokines that are expressed in viral meningitis are able to regulate the production of MMPs and tissue inhibitors of metalloproteinase [ 38 , 39 ]. IL-1α and IL-1β, as well as IL-6, induce or upregulate the transcription of MMP genes [ 40 ]. This, along with increased IL-6 levels in CSF of patients with CNS lupus, may provide an explanation for the local synthesis of MMP-9. CSF levels of enzymatically active MMP-2 were not increased in NPSLE patients as compared with SLE patients without CNS involvement or compared with cerebrally healthy control individuals. This finding is consistent with a previous study of viral meningitis, in which increased expression of MMP-9 but constitutive expression of MMP-2 was found [ 37 ]. The reason for this discrepancy, seen in the present study as well as in the previous one, might be differential regulation of gene promotors for MMP-9 and MMP-2. Indeed, whereas the promotor regions of MMP-9, which is regulated by cytokines, harbor a nuclear factor-κB responsive element, those of MMP-2 lack a TATA box and contain two SP-1 sites, both characteristic of house keeping genes [ 40 , 41 ]. Conclusion Our present and previous findings indicate that the following chain of events takes place during NPSLE: intrathecal release of inflammatory cytokines, leading to synthesis and release of MMP-9, potentially culminating in insult to brain parenchyma, resulting in release of neuronal and astrocytic degradation products and culminating in MRI verifiable lesions and clinical states of brain deficiency. Interestingly, a recent study [ 42 ] indicated that IgG dose-dependently downregulates secretion of MMP-9 by macrophages, suggesting a possible early therapeutic manipulation in NPSLE. Competing interests None declared. Abbreviations CNS = central nervous system; CSF = cerebrospinal fluid; IL = interleukin; MMP = matrix metalloproteinase; MRI = magnetic resonance imaging; NPSLE = neuropsychiatric involvement in systemic lupus erythematosus; SLE = systemic lupus erythematosus.

D:\keerthana\PMC001xxxxxx\PMC1064867.xml

1064868

Infliximab therapy in rheumatoid arthritis and ankylosing spondylitis-induced specific antinuclear and antiphospholipid autoantibodies without autoimmune clinical manifestations: a two-year prospective study

Treatment of rheumatoid arthritis (RA) with infliximab (Remicade ® ) has been associated with the induction of antinuclear autoantibodies (ANA) and anti-double-stranded DNA (anti-dsDNA) autoantibodies. In the present study we investigated the humoral immune response induced by infliximab against organ-specific or non-organ-specific antigens not only in RA patients but also in patients with ankylosing spondylitis (AS) during a two-year followup. The association between the presence of autoantibodies and clinical manifestations was then examined. The occurrence of the various autoantibodies was analyzed in 24 RA and 15 AS patients all treated with infliximab and in 30 RA patients receiving methotrexate but not infliximab, using the appropriate methods of detection. Infliximab led to a significant induction of ANA and anti-dsDNA autoantibodies in 86.7% and 57% of RA patients and in 85% and 31% of AS patients, respectively. The incidence of antiphospholipid (aPL) autoantibodies was significantly higher in both RA patients (21%) and AS patients (27%) than in the control group. Most anti-dsDNA and aPL autoantibodies were of IgM isotype and were not associated with infusion side effects, lupus-like manifestations or infectious disease. No other autoantibodies were shown to be induced by the treatment. Our results confirmed the occurrence of ANA and anti-dsDNA autoantibodies and demonstrated that the induction of ANA, anti-dsDNA and aPL autoantibodies is related to infliximab treatment in both RA and AS, with no significant relationship to clinical manifestations.

Introduction Clinical trials in rheumatoid arthritis (RA) have demonstrated that antibodies directed against tumor necrosis factor α(TNF-α) (adalimumab, infliximab [Remicade ® ]) are highly beneficial for most patients who are refractory to classic treatment with disease-modifying anti-rheumatic drugs, methotrexate or steroid therapy [ 1 - 4 ]. These anti-inflammatory effects of infliximab have led to their use in other inflammatory diseases such as Crohn's disease [ 5 ] and ankylosing spondylitis (AS), with a similar efficacy to that in RA [ 6 - 8 ]. The side effects of these treatments are acknowledged to be very infrequent, with the exception of opportunistic intracellular infection, due particularly to the reactivation of latent Mycobacterium tuberculosis . The other major side effects are an exacerbation of demyelinating disorders and the induction of severe neutropenia and thrombocytopenia [ 1 , 2 , 4 , 9 - 11 ]. Infusion reactions have also been observed and have been correlated with the induction of anti-chimeric antibodies against infliximab [ 12 ]. The development of autoantibodies that are usually associated with systemic lupus erythematosus (SLE), namely antinuclear (ANA) and anti-double-stranded DNA (anti-dsDNA) autoantibodies, has also been observed after infliximab treatment in 63.8% and 13% of RA patients and in 49.1% and 21.5% of Crohn's disease patients, respectively [ 13 - 15 ]. Among the sera that were positive for anti-dsDNA autoantibodies, 9% were also positive for anti-Sm autoantibodies, which are specific for SLE [ 13 ]. However, only a few cases of SLE-like syndrome have been reported in infliximab-treated patients [ 9 , 13 , 16 - 18 ]. As yet, the occurrence of other autoantibodies has not been clearly demonstrated, such as antiphospholipid (aPL) autoantibodies and anti-β 2 -glycoprotein I (anti-β 2 GPI) autoantibodies, which are often associated with SLE [ 19 , 20 ], or autoantibodies associated with vasculitis, autoimmune hepatitis or autoimmune endocrine diseases, which have been reported in therapy that interferes with cytokine balance [ 21 ]. In the present study we investigate the prevalence of such autoantibodies during 2 years of follow-up in patients with RA or AS successfully treated with infliximab. The aim of the study was to discover whether the humoral response induced by infliximab is restricted to non-organ specific autoantibodies and to identify any associated clinical presentations, with the aim of monitoring their occurrence by detecting these autoantibodies. Concurrently, 30 patients whose RA was controlled only by methotrexate were analyzed at 1-year intervals as controls for autoantibody production. Materials and methods Patient sera Twenty-four patients with RA and 15 patients with AS, fulfilling the ACR criteria [ 22 ] and the modified New York criteria [ 23 ], respectively, were monitored for autoantibody production over a 2-year period during which they were good responders, as defined by the modified disease activity scores [ 24 ], to a combination of methotrexate and infliximab. Concurrently, 30 RA patients well controlled by methotrexate for 6–15 years (mean 12 years) gave blood samples at 1-year intervals as controls for autoantibody production. Demographic and clinical statuses are presented in Table 1 . Patients were followed clinically by the same physician during this period at regular intervals and in particular when they were receiving infliximab infusions. Clinical assessment (painful and swollen joint count, spine stiffness, careful examination of side effects, significant concomitant clinical features suggestive of infections or autoimmune disorders) were recorded accurately (Table 1 ). Nine patients discontinued infliximab treatment before the end of the study, between 3 and 18 months, because of adverse events, treatment inefficacy or severe infectious disease. Further details are given in Table 1 . Treatment protocol Twenty-four RA and 15 AS patients were treated with infliximab (Centocor, Malvern, PA, USA). In RA patients, infliximab was administered in accordance with the schedule of the ATTRACT phase III clinical trials [ 4 ]. Patients were given infliximab at a dose of 3 mg/kg at 0, 2, 4 and 6 weeks and thereafter every 8 weeks. In AS patients, after the initial 6-week protocol with 5 mg/kg, infliximab was delivered every 6 or 8 weeks, depending on the clinical response. When AS patients presented a remission, the timing of infusions was dictated by disease relapse [ 25 ]. Follow-up of autoantibodies Tests for autoantibodies were performed at baseline before the start of infliximab treatment and during the 24-month duration of infliximab treatment as indicated below. The sera of the 30 control RA patients were analyzed twice with a 1-year interval. Detection of ANA Tests for ANA were performed at the start of infliximab treatment and at 6, 12, 18 and 24 months, by an indirect immunofluorescence technique (IIF) using HEp2 cells (Bio-Rad, Marnes-la-Coquette, France). Sera were diluted 1:80 and the conjugate was a goat anti-human F(ab') 2 IgG, A, M (H+L) antibody conjugated to fluorescein isothiocyanate (diluted 1:100) (Bio-Rad). Classic titration of each ANA positive at a titer of 1:80 was performed by serial dilutions to 1:5120. A titer equal to or greater than 1:160 was interpreted as a positive result. For positive sera that had nuclear granular or cytoplasmic staining, the identification of autoantibodies against (ENA) was further investigated by enzyme-linked immunosorbent assay (ELISA) with an anti-human IgG (H+L) conjugate (Biomedical Diagnostics, Marne-la-Vallée, France). Detection of anti-dsDNA autoantibodies Tests for anti-dsDNA autoantibodies were performed at the start of infliximab treatment and, depending on the formation of ANA, at 6, 12, 18 and 24 months of treatment with the use of a radioimmunological test (Dade Behring, Paris, France) in accordance with the manufacturer's instructions. A titer equal or greater than 5 IU/ml was interpreted as a positive result. For positive sera, the anti-dsDNA autoantibody isotype was determined by ELISA (Pharmacia, Freiburg, Germany) with an anti-human IgG (H+L) (Bio-Rad) or an anti-human IgM (H+L) (Dako) conjugate. Detection of anti-smooth muscle (SMA), anti-mitochondrial (AMA), anti-liver kidney microsomes (LKM), anti-thyroid peroxidase (TPO), anti-thyroglobulin (TG) and anti-adrenal (ADA) autoantibodies Tests for SMA, AMA, LKM, TPO, TG and ADA autoantibodies were performed at the start of infliximab treatment and then at 3, 6, 12, 18 and 24 months. The sera of the 30 control RA patients were analyzed twice with a 1-year interval. For SMA, AMA, LKM and ADA, sera diluted 1:30 were tested on mouse stomach, kidney, liver or adrenal sections (Biomedical Diagnostics), with the same technique as described for ANA. For TPO and TG autoantibodies, an ELISA technique was performed in accordance with the manufacturer's instructions with an anti-human IgG (H+L) conjugate (Pharmacia, Saint-Quentin-en-Yvelines, France). Detection of anti-neutrophil cytoplasmic autoantibodies (ANCA) Tests for ANCA were performed at the start of infliximab treatment and then after 6 and 12 months. The sera of the 30 control RA patients were analyzed twice with a 1-year interval. The sera diluted 1:20 were tested by IIF on human neutrophils fixed in ethanol (Menarini Diagnostics, Antony, France) with the same technique as for ANA. Positive sera were further tested for reactivity against myeloperoxidase and proteinase 3 by using an ELISA with an anti-human IgG (H+L) conjugate (Bioadvance, Emerainville, France). Titers were considered positive when they were 20 arbitrary units (AU)/ml or more. Detection of aPL and anti-β 2 GPI autoantibodies Investigation of aPL autoantibodies was performed by the detection of anticardiolipin autoantibodies (ACL). ACL and anti-β 2 GPI autoantibodies were evaluated at baseline and at 6, 12 and 24 months after the start of infliximab treatment. The sera of the 30 control RA patients were analyzed twice with a 1-year interval. ACL were detected with an ELISA in accordance with the manufacturer's instructions by using anti-human IgG (H+L) or IgM (H+L) conjugates. Values were expressed as arbitrary G phospholipid (GPL) or M phospholipid (MPL) units. Positive results were graded as low positivity (IgG 11–23 GPL, IgM 6–10 MPL), moderate positivity (IgG 24–39 GPL, IgM 11–29 MPL) or high positivity (IgG ≥ 40 GPL, IgM ≥ 30 MPL). The anti-β 2 GPI autoantibodies were detected by using a home-made assay previously described [ 26 ] with serum samples diluted 1:50 and peroxidase-conjugated anti-human IgG (H+L) or IgM (H+L) (Cappell, ICN Biomedicals, OH, USA) diluted 1:400. Positive results were graded as low positivity for a value from a ratio of 1.2–1.9 AU/ml (attenuance ['optical density'] of the sample divided by attenuance of the cut-off), moderate positive for a value of ratio between 2 and 3 AU/ml and high positive for a value superior to a ratio of 3 AU/ml. The cut-off was determined by using the mean plus 5 standard deviations of attenuance of 100 sera from blood donors (data not shown). Statistics Statistical analysis (95% and 99% confidence interval) was performed with the χ 2 test when applicable and with Fisher's exact test in other conditions. Ethics Written informed consent was obtained from all patients and the study was approved by the Research and Ethics Committee of the Hospices Civils de Lyon. Results Occurrence of ANA and anti-dsDNA autoantibodies in RA and AS patients At baseline, 9 of 24 (37.5%) infliximab-treated RA patients, 2 of 15 (13.3%) AS patients and 5 of 30 (16.7%) control RA patients were tested positive for ANA (Table 2 ). After 12 months of therapy, the induction of ANA was observed in 12 infliximab-treated RA patients, 8 AS patients and 4 control RA patients. At that time, the total number of positive ANA patients was 21 of 24 (87.5%) for infliximab-treated RA patients, 10 of 15 (66.7%) AS patients and 9 of 30 (30%) control RA patients. The difference between the number of induced ANA compared with the number of positive ANA at baseline was statistically significant ( P < 0.0001) for infliximab-treated RA and AS patients, whereas the difference was not significant for the RA control group. The difference in induction was also significant for the infliximab-treated RA patients ( P < 0.0001) comparing the two RA groups. After 2 years of infliximab therapy, ANA became positive in one other infliximab-treated RA patient and three more AS patients, giving a total induction of 87% in RA and 85% in AS. The induction of ANA appeared between 3 and 18 months (mean 6.35 months) for RA and between 3 and 24 months (mean 10.6 months) for AS. Except in two RA patients, all the induced ANA were still positive at the end of the study, including in eight of nine patients who discontinued the treatment. One RA became negative 3 months after the end of the treatment. Furthermore, in six of the nine sera of infliximab-treated RA patients positive at baseline, the ANA titer increased up to twofold (data not shown). The titer of ANA showed a higher level between the positive ANA at the baseline compared with the titer of induced ANA, but the difference was not significant. In most ANA-positive sera during infliximab treatment, the pattern of staining was homogeneous. Two of the 22 infliximab-treated RA ANA-positive sera had granular nuclear staining characteristic of ENA. The specific target could not be identified with the use of the classic ELISA kit for ENA detection. For anti-dsDNA autoantibodies, 1 of 24 infliximab-treated RA patients (4.2%), 2 of 15 AS patients (13.3%) and none of the control RA patients were positive at baseline (Table 2 ). After 12 months of treatment, induction of anti-dsDNA autoantibodies was observed in 10 of 23 (46.5%) infliximab-treated RA patients, 3 of 13 (23%) AS patients and 2 of 30 (6.7%) control patients. The induction was observed between 3 and 12 months. Three further infliximab-treated RA patients and one further AS patient became positive at 18 and 24 months, respectively, giving a total induction of 57% in RA and 31% in AS. After the 2-year follow-up, the total number of positive patients was 14 of 24 (58.33%) for infliximab-treated RA patients, 6 of 15 (40%) for AS patients and 2 of 30 (6.7%) for control RA patients. All patients who became positive for anti-dsDNA autoantibodies were also positive for ANA. All the induced anti-dsDNA autoantibodies remained positive until the end of the study, including in three of four positive patients who discontinued the treatment. One RA patient became negative 3 months after the end of the treatment. The difference between the number of induced anti-dsDNA autoantibodies and the number of positive anti-dsDNA autoantibodies at baseline was statistically significant for the infliximab-treated RA patients ( P < 0.0001) and for the infliximab-treated AS patients ( P < 0.02) compared with the RA control group. Comparing the two RA groups, the difference in induction was also significant for the infliximab-treated RA patients ( P < 0.0001). The titer of anti-dsDNA autoantibodies showed a higher level between the positive autoantibodies at baseline compared with the induced autoantibodies. The formation of ANA and anti-dsDNA autoantibodies was not linked to clinical events, namely infectious side effects, allergy or lack of efficacy. Occurrence of aPL/ACL and anti-β 2 GPI autoantibodies in RA and AS patients At baseline, no RA or AS patients were positive for ACL or for anti-β 2 GPI autoantibodies. At the end of the study, significant levels of ACL were found in infliximab-treated RA patients (5 of 24, P < 0.01) and in infliximab-treated AS patients (4 of 15, P < 0.01) compared with the RA control group (Table 3 ). Induction of anti-β 2 GPI autoantibodies was observed in 2 of 24 infliximab-treated RA patients and in none of the control RA patients (Table 3 ). The difference was not significant between the two RA populations nor within the RA and AS group, comparing the number of induced autoantibodies at baseline and after treatment. In infliximab-treated RA patients, sera were positive for ACL or anti-β 2 GPI autoantibodies (Table 3 ). In the group of AS patients, the two induced sera were positive for both ACL and anti-β 2 GPI autoantibodies. All ACL and anti-β 2 GPI autoantibodies were from patients positive for ANA. Five of 11 ACL or anti-β 2 GPI autoantibody-positive sera were positive for anti-dsDNA autoantibodies. Induction did not occur simultaneously and did not seem to be determined by clinical events. Two AS sera positive for anti-dsDNA autoantibodies at baseline became positive for ACL autoantibodies 6 and 10 months after the beginning of treatment. For the other sera, anti-dsDNA, ACL and anti-β 2 GPI autoantibodies developed between 6 and 12 months after the start of treatment. No correlation was found between the occurrence of side effects (including infections), clinical status (including lupus-like symptoms, thrombopenia or thrombosis) and anti-β 2 GPI or ACL autoantibodies. Isotypes of induced anti-dsDNA, aPL/ACL and anti-β 2 GPI autoantibodies in RA and AS patients Most of the anti-dsDNA autoantibodies detected during infliximab treatment of RA patients, of AS patients and in the control RA population were of IgM isotype (11 of 13 [85%], 4 of 4 [100%] and 2 of 2 [100%] respectively). One of the sera from infliximab-treated RA patients and one from AS patients were positive for both IgG and IgM. Two sera in the infliximab-treated RA group were positive only for IgG. The presence of the IgG isotype was not associated with any particular clinical pattern such as infections, lupus-like syndrome or side effects of infusion. Among the ACL, five of five infliximab-treated RA patients and one of four AS patients were of IgM isotype. Three AS patients were of IgG isotype. The isotypes of the positive anti-β 2 GPI autoantibodies were IgM and IgG (one case), IgG (one case) for RA and IgM or IgG for AS. As for the IgG isotype in induced anti-dsDNA autoantibodies, no significant clinical association was observed in patients presenting the IgG ACL and/or anti-β 2 GPI autoantibody profile. Occurrence of TPO, TG, AMA, LKM, SMA, ADA and ANCA autoantibodies Three of the 24 (12.5%) infliximab-treated RA patients, 6 of 30 (20%) control RA patients and no AS patients had TPO or TG autoantibodies at baseline. Patients with RA remained positive during infliximab treatment and at the 1-year intervals of methotrexate treatment. Only one patient (1 of 21, 4.8%) in the infliximab-treated RA group developed both TPO and TG autoantibody positivity after 12 months of treatment. One of 24 infliximab-treated RA patients (4.2%) and 2 of 15 AS patients (13.3%) who were negative at baseline became positive for ANCA as determined by IIF. The target of these ANCA was identified by ELISA as proteinase 3 for two sera (25 and 40 AU/ml) and both myeloperoxidase and proteinase 3 for one serum (45 and 30 AU/ml). For the RA control group, ANCA were observed in two patients at baseline and remained positive at 1 year. The target of these ANCA was neither proteinase 3 nor myeloperoxidase. No other patient developed such autoantibodies after the 1-year interval analysis. Three of the 24 infliximab-treated RA patients (12.5%) and 2 of 30 RA controls (6.7%) were SMA positive at baseline. Three of 21 infliximab-treated RA sera (14.3%) and 2 of 15 AS sera (13.3%) that were negative at baseline became positive for SMA autoantibodies at 1.5, 3, 6, 3 and 6 months respectively. These SMA autoantibodies were not antiactin autoantibodies, the only autoantibodies that are specific for autoimmune hepatitis. Neither RA nor AS patients developed AMA, LKM or ADA autoantibodies. The occurrence of TPO, TG, ANCA, AMA, LKM, SMA or ADA autoantibodies during infliximab therapy was not statistically significant. Discussion The occurrence of a large panel of autoantibodies that are considered as biological markers of various autoimmune diseases has been investigated in a population of RA and AS patients treated with infliximab for 2 years, the longest period described so far for this kind of management. To avoid the bias of spontaneous autoantibody production under methotrexate, a control population of RA patients treated only with methotrexate was analyzed in parallel at 1-year intervals. ANA, anti-dsDNA and aPL were the only autoantibodies to be significantly induced by infliximab treatment in RA and AS patients. This induction has already been described for ANA and anti-dsDNA autoantibodies [ 13 , 27 ] but our study demonstrates for the first time that infliximab treatment can also induce aPL autoantibodies in both RA and AS patients. Our observation of ANA in up to 91.7% and 86.7% of RA and AS patients, respectively, after infliximab therapy is consistent with recent data published during the course of the present study [ 27 ]. However, the occurrence of anti-dsDNA autoantibodies was higher in our study for both RA and AS [ 27 , 28 ]. These discrepant results may be due to the longer period of our analysis. Indeed, the previous study analyzed this occurrence for 8.5 months after the initiation of infliximab treatment [ 27 ]; we found that anti-dsDNA autoantibodies can be induced after this period, the latest induction being found 24 months after the onset of infliximab treatment. Clinical monitoring of the patients did not show any symptoms characteristic of SLE in the subgroup that was positive for ANA/anti-dsDNA autoantibodies. Antiphospholipid autoantibodies were induced in 21% (5 of 24) and 27% (4 of 15) of our RA and AS patients, respectively. Anti-β 2 GPI autoantibodies were induced in 8% and 13% of our RA and AS patients, respectively. Until now, the induction of such autoantibodies has not been described in patients treated with infliximab therapy. However, it has been demonstrated in a single study in 5 of 8 (63%) RA patients treated with etanercept [ 29 ]. In that study, the presence of aPL autoantibodies along with anti-dsDNA autoantibodies was concomitant with several infections [ 29 ]. In our study, we also found an association between anti-dsDNA and aPL autoantibodies but clinical monitoring of the patients did not show any relationship between a particular serological profile and the occurrence of infection, thrombosis or thrombocytopenia for the aPL autoantibody-positive subgroup. In contrast with other studies showing aPL autoantibodies in RA and AS populations, we found no aPL autoantibodies at baseline [ 30 - 32 ]. Most of the studies reporting a high frequency of aPL autoantibodies were conducted with non-standard tests; furthermore, the titers of most of the positive sera were very low. The difference in sensitivity might also be due to the choice of a different cut-off of positivity for the aPL autoantibody test and to the different clinical characteristics of the patients analyzed. Thus, the ACL test is not specific with low-positive results, so we chose a high cut-off for this test [ 33 ]. The induction of ANA and aPL autoantibodies was clearly due to infliximab, especially in the RA group, because no such induction was observed in the control RA group treated with methotrexate alone. The mechanisms that underlie autoantibody development during infliximab treatment are intriguing. These autoantibodies do in fact occur in a variety of disorders, such as RA, AS and Crohn's disease, which are characterized by different physiopathological mechanisms and different doses of infliximab. One can then postulate that this particular induction is due to the partial blockage of TNF-α induced by infliximab therapies. The role of the disturbance of the cytokine network in such induction has already been demonstrated for another cytokine, interferon γ, inducing the development of autoantibodies in patients with hepatitis C viral infection or RA [ 21 , 34 , 35 ]. Induction of autoantibodies could be a predictable consequence of anti-TNF-α blockade because this blockade could promote humoral autoimmunity by inhibiting the induction of cytotoxic T lymphocyte response, which normally suppresses autoreactive B-cells [ 36 ]. Infliximab might also act by neutralizing the biological activity of TNF-α by binding the soluble forms of TNF-α, thereby preventing the interaction of TNF-α with its cellular receptors, p55 and p75. Infliximab also binds the transmembrane form of TNF-α and could induce antibody-dependent or complement-dependent cellular cytotoxicity of the cells expressing the cytokine [ 37 ]. Furthermore, infliximab has been shown to increase the number of apoptotic T lymphocytes in the lamina propria [ 38 ] and apoptotic monocytes in peripheral blood in Crohn's disease [ 39 ]. In this case, one hypothesis concerning the development of autoimmune diseases such as SLE is that an increased apoptotic process could promote the release of numerous autoantigens, leading to the development of autoantibodies against cytoplasmic and nuclear compounds such as ANA and dsDNA [ 40 ], especially if production of these autoantibodies is no longer suppressed by the action of infliximab on the suppressor T cell population. This apoptotic process might not occur in organ-specific cells because these cells, namely thyrocytes, do not harbor TNF-α receptor, thus shedding some light on the findings concerning the absence of organ-specific autoantibodies associated with autoimmune vasculitis, hepatitis or endocrine diseases. Like Charles and colleagues [ 13 ], we demonstrated that most of the anti-dsDNA autoantibodies detected during the treatment of RA with infliximab were of IgM isotype. Furthermore, we showed that most of the detected aPL autoantibodies were also of IgM isotype. The role of these IgM in the development of autoimmune diseases remains to be elucidated. Natural autoreactive IgM autoantibodies might suppress autoimmunity by inducing B cell tolerance and thus by participating in the negative selection of autoreactive B cells. The larger pool of autoantibodies of IgM isotype observed during infliximab treatment might be the consequence of a higher production of natural autoreactive IgM, but might also be an induced population that can further switch to IgG with a well-known pathogenic effect. A high frequency of IgM might also result from TNF blockade, as it was demonstrated in a murine model of collagen-induced arthritis that anti-TNF-α monoclonal antibodies reduce isotype switching to IgG in the local draining lymph node [ 41 ]. Conclusion Our results show that infliximab induces ANA, anti-dsDNA and aPL autoantibodies at various times after the start of treatment. However, it seems that the development of such autoantibodies is not predictive of the development of SLE-like syndrome, because during the 2-year follow-up of infliximab therapy no APL syndrome or SLE syndrome appeared. Nevertheless, these findings do not exclude the possibility that such pathology might develop after a longer period of infliximab treatment. They underline the need to monitor the humoral response, namely autoantibodies and clinical manifestations, in patients treated with infliximab over a longer period. Competing interests None declared. Abbreviations ACL = anticardiolipin; ADA = anti-adrenal autoantibodies; AMA = anti-mitochondrial autoantibodies; ANA = antinuclear autoantibodies; ANCA = anti-neutrophil cytoplasmic autoantibodies; aPL = antiphospholipid; AS = ankylosing spondylitis; dsDNA = double-stranded DNA; ELISA = enzyme-linked immunosorbent assay; ENA = anti-extractible nuclear antigen; β 2 GPI = β 2 -glycoprotein I autoantibodies; GPL = G phospholipid; IIF = indirect immunofluorescence; LKM = anti-liver kidney microsomes; MPL = M phospholipid; RA = rheumatoid arthritis; SLE = systemic lupus erythematosus; SMA = anti-smooth muscle; TG = anti-thyroglobulin; TNF-α = tumor necrosis factor α; TPO = thyroid peroxidase.

D:\keerthana\PMC001xxxxxx\PMC1064868.xml

1064871

Apolipoprotein A-I infiltration in rheumatoid arthritis synovial tissue: a control mechanism of cytokine production?

The production of tumor necrosis factor α (TNF-α) and interleukin-1β (IL-1β) by monocytes is strongly induced by direct contact with stimulated T lymphocytes, and this mechanism may be critical in the pathogenesis of rheumatoid arthritis (RA). Apolipoprotein A-I (apoA-I) blocks contact-mediated activation of monocytes, causing inhibition of TNF-α and IL-1β production. This study examined the hypothesis that apoA-I may have a regulatory role at sites of macrophage activation by T lymphocytes in inflamed RA synovial tissue. Synovial tissue samples were obtained after arthroscopy from patients with early untreated RA or treated RA and from normal subjects. As determined by immunohistochemistry, apoA-I was consistently present in inflamed synovial tissue that contained infiltrating T cells and macrophages, but it was absent from noninflamed tissue samples obtained from treated patients and from normal subjects. ApoA-I staining was abundant in the perivascular areas and extended in a halo-like pattern to the surrounding cellular infiltrate. C-reactive protein and serum amyloid A were not detected in the same perivascular areas of inflamed tissues. The abundant presence of apoA-I in the perivascular cellular infiltrates of inflamed RA synovial tissue extends the observations in vitro that showed that apoA-I can modify contact-mediated macrophage production of TNF-α and IL-1β. ApoA-I was not present in synovium from patients in apparent remission, suggesting that it has a specific role during phases of disease activity. These findings support the suggestion that the biologic properties of apoA-I, about which knowledge is newly emerging, include anti-inflammatory activities and therefore have important implications for the treatment of chronic inflammatory diseases.

Introduction Inflammation is a critical host-defense mechanism. One of its functions is to direct plasma factors and immunoinflammatory cells to lesions in order to eradicate infection and facilitate tissue repair. In many chronic inflammatory diseases, infiltration of the target tissue by blood-derived cells precedes tissue damage. For example, it is believed that in rheumatoid arthritis (RA), the initial cellular event in synovial tissue is proliferation of fibroblast-like synoviocytes, which release chemokines that contribute to the recruitment of inflammatory cells, including monocytes and lymphocytes [ 1 ]. It has been proposed that the first cells to infiltrate synovial tissue are T lymphocytes, suggesting that they have an important role in pathogenesis. We previously showed that stimulated T cells induced pathological effects through direct cellular contact with monocyte–macrophages, causing the abundant production of interleukin-1β (IL-1β) and tumor necrosis factor α (TNF-α). This observation has been confirmed by others (for review see [ 2 ]). The unregulated production of IL-1β and TNF-α in RA has been recognized for several years, and their role in the pathophysiology has been confirmed by the demonstration that targeted blockade improves patients' clinical status [ 3 , 4 ]. We therefore postulate that contact-mediated cytokine production is highly relevant to the pathogenesis and the maintenance of chronic inflammation in diseases such as RA. Regulating a potent mechanism that induces both IL-1β and TNF-α may be important in maintaining a low level of monocyte activation within the bloodstream. We recently identified apolipoprotein A-I (apoA-I) as a specific inhibitor of contact-mediated activation of monocytes [ 5 ]. ApoA-I is a 'negative acute-phase protein' and the principal protein of high-density lipoproteins (HDLs). Variations of apoA-I concentration have been observed in several inflammatory diseases. In RA, the levels of circulating apoA-I and HDL cholesterol in untreated patients are lower than in normal controls [ 6 - 8 ]. In contrast, apoA-I levels were increased in the synovial fluid of patients with RA [ 9 ], although these were still only one-tenth those in plasma. The elevation of apoA-I levels in the synovial fluid of untreated patients with RA was accompanied by increased cholesterol levels, suggesting infiltration of HDL particles in the inflamed joint. In this study, we examined synovial tissue from patients with active RA in order to determine if apoA-I infiltration had occurred at sites of contact between T lymphocytes and macrophages. Materials and methods Synovial tissue samples Synovial biopsies were obtained from the knee joints after arthroscopy in patients diagnosed with RA, who had all given their informed consent. Normal synovium was obtained from a patient without arthritis who was having a leg amputated. Arthroscopy and biopsy were performed under local anesthesia using a 2.7-mm Storz arthroscope and a 1.5-mm grasping forceps. The sampled tissue was immediately embedded in Tissue-Tek ® OCT compound (Sakura, Zoeterwoude, the Netherlands) and snap frozen in liquid nitrogen. Monoclonal antibodies All antibodies used were murine antihuman monoclonal antibodies (antibodies were diluted in PBS; anti-apoA-I contained 0.6 M sodium chloride); anti-apoA-I, type 2 (Calbiochem-Novabiochem Corporation, Darmstadt, Germany), was used at 1/3000 dilution; anti-C-reactive protein (CRP), clone CRP-8 (Sigma Chemicals, St Louis, MO, USA), at 1/200 dilution; anti-Von Willebrand factor/factor VIII-related antigen (FVIII), clone F8/86 (DAKO, Glostrup, Denmark), at 1/50 dilution; and anti-acute-phase serum amyloid A (A-SAA) (gift from Dr AS Whitehead, Philadelphia, PA, USA), at 1/1200. Isotype-matched murine IgG1 (DAKO) was used at the same concentration as each of the primary antibodies. Immunohistochemistry Synovial tissue sections were cut at 7 μm and mounted on slides coated with 3-aminopropyltriethoxy-silane (Sigma). Slides were air-dried overnight, wrapped in foil, and stored at -80°C. A standard three-stage immunoperoxidase technique was used, with a Peroxidase VECTASTAIN ® Elite ABC kit (Vector Laboratories, Burlingame, CA, USA). Slides were removed from the -80°C freezer and allowed to thaw at room temperature for 20 minutes. Sections were fixed in acetone for 10 minutes and with normal horse serum (VECTASTAIN ® Elite ABC kit) for 15 minutes. The relevant primary antibody was added to sections for 1 hour at room temperature. Sections were washed and incubated with PBS for 5 minutes. Anti-mouse IgG secondary antibody (VECTASTAIN ® Elite ABC kit) was added for 30 minutes and the ABC solution (VECTASTAIN ® Elite ABC kit) was added to sections for 30 minutes. Sections were treated with 3% hydrogen peroxide for 7 minutes, washed in distilled water for 1 minute, and incubated in PBS for 5 minutes, followed by the addition of 3,3'-diaminobenzidine (Sigma) for 12 minutes. The chromogenic reaction was stopped by immersion in water. Sections were counterstained in Mayer's hemalum, dehydrated in alcohol, cleared in xylene, and mounted in DPX (BDH, Poole, UK). Results The demographic and clinical details of the patients studied are outlined in Table 1 . Synovial tissue samples from eight patients with active RA were selected. The mean duration of disease was 19 (range 1–48) months, he mean swollen joint count was 20 (range 10–36), and the mean CRP level was 12.3 (range <3 to 22)mg/L. Six patients were receiving nonsteroidal anti-inflammatory drugs at the time of synovial biopsy. Two were receiving a disease-modifying anti-rheumatic drug, methotrexate, 15 mg/week in both cases. Two were receiving prednisolone, 10 mg/day. None had received an intra-articular corticosteroid injection to the biopsied knee joint. Synovial tissue was also obtained from two patients with quiescent RA (no swollen joints, CRP <3 mg/L) and from one patient who was unaffected by arthritis. Both patients with quiescent RA were receiving methotrexate, 7.5 mg/week. Table 1 Demographic and clinical details of patients with active rheumatoid arthritis Total no. of patients 8 Mean duration of disease (range) 19 (1–48) months Mean no. of swollen joints (range) 20 (10–36) Mean C-reactive protein (range) 12.3 (0–22)mg/dL No. of patients receiving: NSAIDs 6 DMARDs (MTX 15 mg/wk) 2 Prednisolone 2 DMARD, disease-modifying antirheumatic drug; MTX, methotrexate; NSAID, nonsteroidal anti-inflammatory drug. All synovial tissue sections from the eight patients with active RA showed prominent blood vessels and perivascular cellular infiltration. Specific apoA-I staining was present in all samples. The immunohistologic appearances were consistent, and included prominent endothelial apoA-I staining of most blood vessels (Fig. 1a ). The vessels were surrounded by a confined area of intense staining that was consistent with extravasation of apoA-I within the perivascular cell infiltrate. No staining was observed in the negative control tissue sections (Fig. 1b ). In tissue samples obtained from patients with RA that were in apparent remission, only faint vascular and perivascular apoA-I staining was present (Fig. 1e ), even though the sections contained blood vessels that were easily identified (Fig. 1f ). As expected, the cellular infiltrate in these sections was less intense. There was no perivascular apoA-I staining in the synovial tissue sample obtained from the knee joint unaffected by arthritis (Fig. 1c ). Contrary to the abundant presence of perivascular apoA-I staining in tissue sections obtained from patients with active RA, there was no evidence of perivascular CRP or A-SAA. Tissue samples from three patients were studied for the presence of perivascular CRP. The serum CRP levels were elevated in all three at the time of biopsy (11–20 mg/L). Faint CRP staining of endothelial cells was observed (Fig. 1g ). Tissue samples from five patients were studied for the presence of perivascular A-SAA. As expected, A-SAA staining was demonstrated in lining layer cells but not within the perivascular infiltrate (Fig. 1h ). Figure 1 Apolipoprotein A-I (apoA-I) is localized in the perivascular region of the inflamed synovium. (a) Active rheumatoid arthritis (RA) synovium stained with anti-apoA-I; (b) active RA synovium stained with isotype-matched negative control; (c) normal synovium stained with anti-apoA-I; (d) normal synovium stained with anti-factor VIII; (e) remission RA synovium stained with anti-apoA-I; (f) remission RA synovium stained with anti-factor VIII; (g) active RA synovium stained with anti-C-reactive protein; (h) active RA synovium stained with antibody against serum amyloid A. Discussion The most salient observation from this study was apoA-I infiltration in inflamed synovial tissue and its retention in perivascular regions, where T lymphocytes and macrophages accumulate. The localization of positive acute-phase proteins, such as CRP and A-SAA, was different from that of apoA-I: only faint staining, limited to vascular endothelium, was observed for CRP, and A-SAA was observed in lining layer cells, which are a known source of local synthesis [ 10 ]. We have previously shown apoA-I to inhibit the production of both IL-1β and TNF-α in monocytes activated by direct contact with stimulated T cells. This mechanism may have a role in regulating monocyte activation in the bloodstream [ 5 ]. This study demonstrated that apoA-I infiltrated perivascular regions of the synovium where A-SAA, which can dissociate apoA-I from HDLs [ 11 ], and CRP were absent. The perivascular localization of apoA-I suggests that it could have an inhibitory role in zones where T lymphocytes are in close contact with monocyte–macrophages, with a tendency to form 'lymphoid microstructures' [ 12 ]. The absence of A-SAA suggests that it is unlikely to restrict the inhibitory activity of apoA-I in the contact-mediated induction of IL-1β and TNF-α production in tissue [ 13 ]. To overcome apoA-I inhibition, A-SAA would be expected to localize in the same area. Since apoA-I is virtually absent from the synovial tissue of patients with inactive RA (Fig. 1c ), its presence in actively inflamed tissue suggests that its infiltration during a flare-up may represent a physiologic mechanism that inhibits proinflammatory cytokine production and limits disease recurrence. The transient infiltration of apoA-I may also explain why RA, like many other chronic inflammatory diseases, characteristically presents as a relapsing–remitting disease in many patients. During phases of RA associated with joint damage, the inhibitory effects of apoA-I on the destructive mechanisms may not be sufficiently potent. In RA, variations of apoA-I concentrations were observed in plasma, where it was decreased, and in synovial fluid, where it was increased [ 6 - 9 ]. The elevation of apoA-I levels in synovial fluid of RA patients correlated with a rise in cholesterol, suggesting infiltration of HDL particles into the inflamed joint. Similarly, active juvenile RA was associated with reduced HDL blood levels and a significant decrease in apoA-I concentration in plasma [ 14 ]. These studies suggest that variations of apoA-I levels may inversely correlate with disease activity. The observation that apoA-I can infiltrate and be retained at the inflammatory site suggests that apoA-I may inhibit the local triggering of IL-1β and TNF-α release by monocyte–macrophages that are in direct contact with stimulated T cells in these areas [ 15 ]. Conclusion In conclusion, the localization of apoA-I in inflamed synovium suggests that it can locally inhibit the production of proinflammatory cytokines by monocyte–macrophages upon direct contact with stimulated T cells. Thus, it is possible that after immune cell infiltration, formation of lymphoid-like microstructures, and the proliferation of blood vessels that resemble high-endothelial venules, inhibitory plasma components may infiltrate the developing inflammatory lesion. ApoA-I that binds surface factors on stimulated T cells is retained in the perivascular regions, where it may limit contact-mediated cytokine induction in monocyte–macrophages [ 5 ] and inhibit critical pathways associated with disease exacerbation. The alterations in apoA-I infiltration may also explain fluctuations of disease activity. The finding that apoA-I can infiltrate inflamed tissue, together with its newly emerging anti-inflammatory properties, may have important implications for treatment in chronic inflammatory diseases. Competing interests The authors declare that they have no competing interests. Author contributions BB and OF cared for the patients included in this study and supervised arthroscopy and biopsy procedures. MG carried out the histochemical study. BB, JMD, and DB conceived of the study and participated in its design and coordination. All authors read and approved the final manuscript. Abbreviations apo A-I = apolipoprotein A-I; A-SAA = acute-phase serum amyloid A; CRP = C-reactive protein; HDL = high-density lipoprotein; IL-1β = interleukin-1β; PBS = phosphate-buffered saline; RA = rheumatoid arthritis; TNF-α = tumor necrosis factor α.

D:\keerthana\PMC001xxxxxx\PMC1064871.xml

1064873

Resistance to IL-10 inhibition of interferon gamma production and expression of suppressor of cytokine signaling 1 in CD4+ T cells from patients with rheumatoid arthritis

IL-10 has been shown to block the antigen-specific T-cell cytokine response by inhibiting the CD28 signaling pathway. We found that peripheral blood CD4 + T cells from patients with active rheumatoid arthritis (RA) were able to produce greater amounts of interferon gamma after CD3 and CD28 costimulation in the presence of 1 ng/ml IL-10 than were normal control CD4 + T cells, although their surface expression of the type 1 IL-10 receptor was increased. The phosphorylation of signal transducer and activator of transcription 3 was sustained in both blood and synovial tissue CD4 + T cells of RA, but it was not augmented by the presence of 1 ng/ml IL-10. Sera from RA patients induced signal transducer and activator of transcription 3 phosphorylation in normal CD4 + T cells, which was mostly abolished by neutralizing anti-IL-6 antibody. Preincubation of normal CD4 + T cells with IL-6 reduced IL-10-mediated inhibition of interferon gamma production. Blood CD4 + T cells from RA patients contained higher levels of suppressor of cytokine signaling 1 but lower levels of suppressor of cytokine signaling 3 mRNA compared with control CD4 + T cells, as determined by real-time PCR. These results indicate that RA CD4 + T cells become resistant to the immunosuppressive effect of IL-10 before migration into synovial tissue, and this impaired IL-10 signaling may be associated with sustained signal transducer and activator of transcription 3 activation and suppressor of cytokine signaling 1 induction.

Introduction IL-10 is a key cytokine in regulating inflammatory responses, mainly by inhibiting the production and function of proinflammatory cytokines. IL-10 binds to the IL-10 receptor (IL-10R) complex that is composed of two subunits, the primary ligand-binding component type 1 IL-10R (IL-10R1) and the accessory component type 2 IL-10R [ 1 ]. The interaction of IL-10 and IL-10R engages the Janus kinase (JAK) family tyrosine kinases Jak1 and Tyk2, which are constitutively associated with IL-10R1 and type 2 IL-10R, respectively [ 2 ]. IL-10 induces tyrosine phosphorylation and activation of the latent transcriptional factors signal transducer and activator of transcription (STAT) 3 and STAT1 [ 3 ]. Upon phosphorylation, STAT1 and STAT3 proteins form homodimers or heterodimers, rapidly translocate into the nucleus, and modulate gene transcription. Intriguingly, STAT3 is indispensable for both IL-10-derived anti-inflammatory and IL-6-derived proinflammatory responses [ 4 ]. Studies of cell-type-specific STAT3-deficient mice have shown that STAT3 activation is essential for IL-10-mediated anti-inflammatory reactions in macrophages and neutrophils [ 5 ], but is responsible for IL-6-mediated prevention of apoptosis in T cells [ 6 ]. The suppressor of cytokine signaling (SOCS) proteins have been identified as a family of endogenous JAK kinase inhibitors that can act in classic feedback inhibition loops, but their roles as the mediators of crosstalk inhibition by opposing cytokine signaling pathways have been clarified [ 7 ]. Recent studies indicate that SOCS3 plays a key role in regulating the divergent action of IL-10 and IL-6, by specifically blocking STAT3 activation induced by IL-6 but not that induced by IL-10 [ 8 , 9 ]. The synovial membrane of rheumatoid arthritis (RA) is characterized by an infiltrate of a variety of inflammatory cells, such as lymphocytes, macrophages, and dendritic cells, together with proliferation of synovial fibroblast-like cells. Numerous cytokines are overproduced in the inflamed joint, and macrophages and synovial fibroblasts are an important source of proinflammatory cytokines. Tumor necrosis factor alpha (TNF-α) and IL-1, two major macrophage products, are crucial in the process of chronic inflammation and joint destruction, and they give rise to effector components, including other inflammatory cytokines, chemokines, growth factors, matrix proteases, nitric oxide, and reactive oxygen species [ 10 ]. IL-6 is a pleiotropic cytokine produced substantially by activated fibroblasts, and its proinflammatory actions include simulating the acute-phase response, B-cell maturation into plasma cells, T-cell functions, and hematopoietic precursor cell differentiation [ 11 ]. However, anti-inflammatory cytokines and cytokine inhibitors are also present in large quantities in RA joints. IL-10, produced by macrophages and partly by T cells in the synovial tissue (ST), is best known as a negative regulator for macrophage and Th1 cells, but the expression level is insufficient to counterbalance the cascade of proinflammatory events [ 12 ]. In addition, the anti-inflammatory action of IL-10 appears to be modulated at the level of signal transduction during chronic inflammation. IL-10 signaling is impaired in macrophages upon chronic exposure to proinflammatory cytokines such as TNF-α and IL-1 and immune complexes [ 13 , 14 ]. Cell surface expression of IL-10R1 is decreased in synovial fluid dendritic cells due to the presence of TNF-α, IL-1, and granulocyte–macrophage colony-stimulating factor [ 15 ]. CD4 + T cells may be activated by arthritogenic antigens, in conjunction with CD28-mediated costimulatory signaling, in RA. The significance of this autoimmune process has been supported by the linkage of the MHC class II antigens HLA-DRB1*0404 and HLA-DRB1*0401 with disease susceptibility and severity [ 16 , 17 ], and by the high-level expression of MHC class II molecules and both CD28 ligands, CD80 and CD86, in the inflamed ST [ 18 - 20 ]. The continuing emergence of activated CD4 + T cells, even though few in number, may be crucial in sustaining the activation of macrophages and synovial fibroblasts through cell surface signaling by means of cell surface CD69 and CD11, as well as the release of proinflammatory Th1 cytokines such as interferon gamma (IFN)-γ and IL-17 [ 21 , 22 ]. In addition, CD4 + T cells could stimulate B-cell production of autoantibodies such as rheumatoid factor and osteoclast-mediated bone destruction. Their obligatory role in RA synovitis was recently proved by successful treatment of active disease by selective inhibition of T-cell activation with fusion protein of cytotoxic T-cell-associated antigen 4 (CD152)-IgG, which can block the engagement of CD28 on T cells by binding to CD80 and CD86 with high avidity [ 23 ]. IL-10 efficiently blocks the antigen-specific T-cell cytokine response by inhibiting the CD28 signaling pathway [ 24 ], as well as indirectly by downregulating the function of antigen-presenting cells. To elucidate the resistance of CD4 + T cells to this direct inhibition in RA, we investigated the production of IFN-γ after CD3 and CD28 costimulation in the presence of IL-10, the induction of STAT1 and STAT3 phosphorylation by IL-10, and the expression of SOCS1 and SOCS3 mRNA in peripheral blood (PB) CD4 + T cells from RA patients. Materials and methods Patients and samples The total patient population consisted of 32 patients with RA (25 women and seven men; mean ± standard deviation age, 52.8 ± 12.4 years) diagnosed according to the revised 1987 criteria of the American College of Rheumatology (formally, the American Rheumatism Association) [ 25 ]. All patients were receiving prednisolone (≤ 7.5 mg/day) and disease-modifying antirheumatic drugs. Clinical parameters in the study patients were as follows (mean ± standard deviation): erythrocyte sedimentation rate, 55.9 ± 35.4 mm/hour; serum C-reactive protein (CRP) level, 32.0 ± 32.0 mg/l; and IgM class rheumatoid factor titer, 142 ± 158 U/ml. Patients were divided into two groups: 24 patients with active disease, who had multiple tender and/or swollen joints and elevated serum CRP level (≥ 10 mg/l); and eight patients with inactive disease, who satisfied the American College of Rheumatology preliminary criteria for clinical remission [ 26 ]. Sixteen healthy volunteers (11 women and five men; age, 45.8 ± 11.2 years) served as controls. ST samples were obtained from three RA patients undergoing total knee replacement. All patients gave informed consent. Isolation of CD4 + T cells Peripheral blood mononuclear cells (PBMC) were prepared from heparinized blood samples by centrifugation over Ficoll-Hypaque density gradients (Pharmacia, Uppsala, Sweden). CD4 + T cells were purified from PBMC by positive selection using anti-CD4 mAb-coated magnetic beads (Miltenyi Biotec, Gladbach, Germany), according to the manufacturer's instructions. CD4 + T cells were isolated from ST samples, as previously described [ 27 ]. Briefly, fresh ST samples were fragmented and digested with collagenase and DNase for 1 hour at 37°C. After removing tissue debris, ST cell suspensions in culture medium (RPMI 1640 medium; Life Technologies, Gaithersburg, MD, USA) supplemented with 25 mM HEPES (2 mM L-glutamine, 2% nonessential amino acids, 100 IU/ml penicillin, and 100 mg/ml streptomycin; Life Technologies) with 10% heat-inactivated FCS (Life Technologies) were incubated at 37°C in six-well plates (Coster, Cambridge, MA, USA) for 45 min. Non-adherent cells were harvested and CD4 + T cells were purified by positive selection as already described. Culture of CD4 + T cells PB CD4 + T-cell populations were resuspended at a density of 1 × 10 6 cells/ml in culture medium with 10% FCS, and 0.5 ml cell suspensions were dispensed into the wells of 24-well microtiter plates (Coster) coated with 1 μg/ml anti-CD3 mAb (Immunotech, Marseille, France). The cells were incubated with 1 μg/ml anti-CD28 mAb (Immunotech) in the presence or absence of the indicated concentrations of IL-10 (Becton Dickinson, San Jose, CA, USA) at 37°C in a humidified atmosphere containing 5% CO 2 [ 28 ]. Culture supernatants were collected 36 hours later and cell-free samples were stored at -30°C until cytokine assay. To examine the effect of IL-6 on T-cell responsiveness to IL-10, CD4 + T cells from healthy controls were incubated in culture medium with 10% FCS in the presence or absence of 10 ng/ml IL-6 (Becton Dickinson) for 36 hours. Cells were then stimulated for 36 hours with anti-CD3 mAb and anti-CD28 mAb in the presence or absence of 1 ng/ml IL-10. Culture supernatants were measured for IFN-γ concentrations. Flow cytometric analysis for IL-10R1 expression A sample of 5 × 10 5 cells of PBMC was resuspended in PBS with 1% FCS. PBMC were incubated with saturating concentrations of anti-IL-10R1 mAb (IgG 1 ; R&D systems, Minneapolis, MN, USA) or with isotype-matched control mAb (Immunotech), followed by incubation with FITC-conjugated goat anti-mouse IgG 1 polyclonal antibody (Santa Cruz Biotechnologies, Santa Cruz, CA, USA). Cells were then incubated with phycoerythrin-conjugated anti-CD4 mAb (Becton Dickinson). Cells were washed well with 1% FCS/PBS between incubations. Analysis was performed on a FACScan flow cytometer (Becton Dickinson). Immunoassay for IFN-γ and IL-2 Concentrations of IFN-γ and IL-2 in culture supernatants of CD4 + T cells were measured in duplicate by the quantitative sandwich ELISA using cytokine-specific capture with biotinylated detection mAb and recombinant cytokine proteins (all from Becton Dickinson), according to the manufacturer's protocol. The detection limits for IFN-γ and IL-2 were 15 pg/ml. Isolation of mRNA and real-time PCR Total cellular RNA was extracted from PB CD4 + T cells using an RNA isolation kit (RNeasy Mini kit; Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. cDNA was synthesized from total RNA with Molony murine leukemia virus reverse transcriptase (US Biochemical, Cleveland, OH, USA) and oligo-(dT) 15 primers (Promega, Madison, WI, USA). Real-time PCR was performed with the LightCycler Instrument (Roche Diagnostics, Penzberg, Germany) in glass capillaries. The reaction mix containing Taq DNA polymerase and DNA double-strand-specific SYBR Green I dye (Lightcycler FastStart DNA Master SYBR Green I; Roche Diagnostics) and specific primers were added to cDNA dilutions. The cDNA samples were denatured at 95° C for 10 min, and were then amplified for 40–50 cycles: at 95° C (10 s), at 65° C (15 s), and 72° C (22 s) for β-actin; at 95° C (10 s), at 62° C (15 s), and at 72° C (10 s) for SOCS1; and at 96° C (10 s), at 68° C (15 s), and at 72° C (15 s) for SOCS3. Amplification curves of the fluorescence values versus cycle number were obtained, and a melting curve analysis was then performed. The levels of SOCS1 and SOCS3 expression were determined by normalizing relative to β-actin expression. The forward and reverse primers were as follows: for β-actin, 5'-GTGGGGCGCCCCAGGCACCA-3' and 5'-CTCCTTAATGTCACGCACGATTTC-3' ; for SOCS1, 5'-AGACCCCTTCTCACCTCTTG-3' and 5'-GCACAGCAGAAAAATAAAGC-3' ; and for SOCS3, 5'-CCCGCCGGCACCTTTCTG-3' and 5'-AGGGGCCGGCTCAACACC-3'. Western blot analysis CD4 + T cells were stimulated for 20 min by the indicated concentrations of IL-10 and IL-6 at a density of 5 × 10 5 cells in 0.5 ml culture medium with 10% FCS. To examine the effect of serum IL-6 on STAT phosphorylation, normal CD4 + T cells were stimulated for 20 min with 30% active RA serum in culture medium with 40 μg/ml neutralizing goat anti-IL-6 polyclonal antibody (IgG; Techne, Princeton, NJ, USA) or control goat IgG (Techne). Whole cell lysates were prepared by placing cells in 100 μl SDS lysing buffer (62.5 mM Tris–HCl [pH 6.8], 2% SDS, 10% glycerol, 50 mM dithiothreitol, 0.1% bromphenol blue). Then 20 μl protein samples were fractionated on 10% SDS-polyacrylamide gels and were transferred to nitrocellulose membranes (Amersham, Buckinghamshire, UK), and the membrane was blocked with 5% skim milk in Tris-buffered saline with 0.1% Tween 20. Tyrosine phosphorylation of STAT1 and STAT3 was detected using commercial available kits (Cell Signaling Technology, Beverly, MA, USA) according to the manufacturer's instructions. Briefly, the membrane was incubated with the antibodies (rabbit IgG) anti-STAT1 antibody, anti-phosphorylated tyrosine 701 of STAT1 antibody, anti-STAT3 antibody, and anti-phosphorylated tyrosine 705 of STAT3 antibody, diluted as recommended at 1/2000 with Tris-buffered saline with 0.1% Tween 20 with 5% BSA. Antibody binding was detected by horseradish peroxidase-conjugated anti-rabbit IgG antibody diluted at 1/4000 with Tris-buffered saline with 0.1% Tween 20 with 5% BSA, and was revealed using the chemiluminescence system. Protein bands were quantified by densitometry using NIH-Image analysis, and STAT phosphorylation was compared with the total amount of STAT protein. IFN-γ-stimulated Hela cells were used as a positive control for STAT1 phosophorylation. Statistical analysis Data are expressed as the mean value ± standard error of the mean or box plots. The statistical significance of differences between two groups was determined by the Mann–Whitney U test or the Wilcoxon signed rank test. P < 0.05 was considered significant. Results Resistance to IL-10 inhibition of IFN-γ production in RA CD4 + T cells The CD28 costimulatory pathway is crucial for effective antigen-specific T-cell cytokine production, and IL-10 can directly suppress this response by inhibiting CD28 tyrosine phosphorylation and binding of phosphatidylinositol 3-kinase [ 24 ]. To evaluate the responsiveness of RA CD4 + T cells to IL-10, purified PB CD4 + T cells from three patients with active RA and from three healthy controls were stimulated by immobilized anti-CD3 antibody and anti-CD28 antibody with or without diluted concentrations of IL-10 for 36 hours, and IFN-γ production was measured by ELISA. As shown in Fig. 1 , IFN-γ production by activated normal CD4 + T cells was mostly inhibited at concentrations as low as 1 ng/ml IL-10. However, RA CD4 + T cells were able to produce significant amounts of IFN-γ in the presence of 1 ng/ml IL-10, and the maximal but not complete inhibition by IL-10 was obtained at 10–100 ng/ml. We thus compared the levels of IFN-γ production by CD4 + T cells after CD3 and CD28 costimulation in the presence of 1 ng/ml IL-10 in RA patients with active disease (multiple inflammatory joints, CRP level ≥ 10 mg/l) and inactive disease (in remission, CRP level < 10 mg/l) [ 26 ] and in healthy controls. There were no statistically significant differences in IFN-γ production without IL-10 among these three groups (Fig. 2a ), but the inhibitory effect of IL-10 on IFN-γ production was significantly limited in the active RA group as compared with the inactive RA group and healthy controls (percentage decrease: active RA, 2.9 ± 14.4%; inactive RA, 45.6 ± 14.4%; controls, 65.8 ± 7.9%) (Fig. 2b ). As a consequence, CD4 + T cells from active RA patients produced higher levels of IFN-γ in the presence of 1 ng/ml IL-10 than did normal CD4 + T cells (Fig. 2a ). In addition, we compared IL-2 production by CD4 + T cells after CD3 and CD28 costimulation in the presence of IL-10 in active RA patients and in healthy controls. Similarly, IL-2 production was not affected by 1 ng/ml IL-10 in RA patients (percentage decrease, -2.1 ± 13.8%), while it was significantly reduced in healthy controls (61.1 ± 13.7%; P < 0.05). Taken together, these results indicate that RA CD4 + T cells become less susceptible to the immunoregulatory effect of IL-10 during the active phase. Increased expression of cell surface IL-10R1 on RA CD4 + T cells The functional receptor complex of IL-10 consists of two subunits, the primary ligand-binding component IL-10R1 and the accessory component type 2 IL-10R [ 1 ]. IL-10R1 expression plays a critical role in cellular responses to IL-10 [ 29 ]. To examine whether the resistance to IL-10 inhibition in RA CD4 + T cells was due to limited receptor expression, the cell surface expression of IL-10R1 on PB CD4 + T cells from active RA patients and from healthy controls was determined by flow cytometric analysis. As shown in Fig. 3a,3b , the intensity of IL-10R1 expression on CD4 + T cells was significantly increased in RA patients compared with in healthy controls. These results suggest that the intracellular signal transduction pathway of IL-10 may be impaired in CD4 + T cells of active RA. Defective IL-10-mediated STAT3 phosphorylation in RA CD4 + T cells The interaction of IL-10R with IL-10 induces tyrosine phosphorylation and activation of the latent transcription factors STAT1 and STAT3 [ 3 ]. Macrophage-specific STAT3-deficient mice demonstrated that STAT3 plays a dominant role in IL-10-mediated anti-inflammatory responses [ 5 ], which has recently been confirmed in human macrophages by studies of dominant-negative STAT3 overexpression [ 30 ]. The induction of STAT1 and STAT3 phosphorylation by IL-10 in PB CD4 + T cells from active RA patients and from healthy controls was examined using western blotting. STAT3 phosphorylation was dose-dependently induced after IL-10 activation for 20 min in normal CD4 + T cells (Fig. 4a,4b ). In contrast, STAT3 was phosphorylated in freshly isolated PB CD4 + cells from RA patients and this STAT3 phosphorylation was detectable for up to 6 hours. STAT3 phosphorylation was augmented only when activated by as much as 10 ng/ml IL-10. Both sustained STAT3 phosphorylation and defective IL-10-induced STAT3 phosphorylation were found in RA ST CD4 + T cells (Fig. 4c ). On the other hand, IL-10-induced STAT1 phosphorylation was not detected in either RA CD4 + T cells or normal CD4 + T cells (Fig. 4a ). These results indicate that STAT3 is the major IL-10-activated STAT in CD4 + T cells, and IL-10-induced STAT3 activation may be diminished in active RA, in association with sustained STAT3 phosphorylation. IL-6-mediated STAT3 phosphorylation and inhibition of IL-10 effect in normal CD4 + T cells STAT3 is activated by many cytokines and growth factors such as the IL-6 family of cytokines (IL-6, IL-11, leukemia inhibitory factor, and oncostatin M), platelet-derived growth factor, and epidermal growth factor, in addition to IL-10 [ 4 ], but previous studies have demonstrated that IL-6 is the major factor in RA synovial fluid that induces constitutive activation of STAT3 in mononuclear cells [ 31 ]. Since IL-6 is also abundant in sera of active RA patients, frequently detected at > 1 ng/ml [ 27 ], we examined whether persistent exposure of CD4 + T cells to high concentrations of IL-6 in the blood circulation was responsible for their sustained STAT3 activation and resistance to IL-10 inhibition in active RA. Both STAT1 and STAT3 phosphorylation was activated by IL-6 in normal CD4 + T cells (data not shown), in agreement with previous observations [ 4 ]. Normal CD4 + T cells were thus incubated for 20 min with culture medium containing 30% serum from active RA patients and neutralizing anti-IL-6 antibody or control antibody, and STAT phosphorylation was examined by western blot analysis. RA serum was able to induce tyrosine phosphorylation of STAT3 but not STAT1, and this STAT3 activation was mostly abolished by neutralization of IL-6 activity (Fig. 5a ). These results indicate that IL-6 is the dominant STAT3-activating factor contained in sera of active RA patients. The lack of STAT1 activation by RA serum suggests that much higher concentrations of IL-6 may be required for STAT1 activation as compared with STAT3 activation, or that inhibitors of STAT1 signaling may be present in RA serum. We next examined whether IL-6 could suppress the effect of IL-10 to inhibit IFN-γ production by CD4 + T cells. After preincubation with or without 10 ng/ml IL-6 for 36 hours, normal CD4 + T cells were stimulated by CD3 and CD28 costimulation in the presence or absence of 1 ng/ml IL-10 for 36 hours, and the IFN-γ production was measured by ELISA. IL-6 pretreatment of normal cells reduced IL-10-mediated inhibition of IFN-γ production (Fig. 5b ), indicating that high concentrations of IL-6 could modulate T-cell responsiveness to IL-10. Taken together, these findings suggest that persistent exposure to serum IL-6 may have a role in both the induction of STAT3 activation and the resistance to the inhibitory effect of IL-10 in RA CD4 + T cells. High expression of SOCS1 mRNA in RA CD4 + T cells IL-6 induces two potent inhibitors of JAKs (SOCS1 and SOCS3 proteins) that not only act as mediators of negative feedback inhibition, but also play a major role in crosstalk inhibition by opposing other cytokine-signaling pathways [ 7 ]. SOCS3 has recently been shown to specifically inhibit STAT3 activation induced by IL-6 but not by IL-10, thereby regulating the divergent action of IL-6 and IL-10 [ 8 , 9 ]. On the contrary, SOCS1 is able to partially inhibit IL-10-mediated STAT3 activation and cellular responses, as well as IFN-γ-mediated STAT1 activation [ 32 ]. To determine whether SOCSs were involved in the defective IL-10-induced STAT3 activation of RA CD4 + T cells, the levels of SOCS1 and SOCS3 mRNA expression in PB CD4 + T cells from active RA patients and from healthy controls were compared by semiquantitative real-time PCR. The RA CD4 + T cells contained higher levels of SOCS1 but lower levels of SOCS3 transcripts than did control CD4 + T cells (Fig. 6a ). Constitutive expression of SOCS1 mRNA in RA CD4 + T cells was comparable with the expression in normal CD4 + T cells stimulated by 10 ng/ml IL-6 (Fig. 6b ), supporting its functional significance. Defective IL-10-induced STAT3 activation therefore appears to be due at least in part to an abundance of SOCS1 in RA CD4 + T cells. Discussion CD4 + T cells orchestrate the Th1-type cell-mediated immune response in RA [ 22 ]. Activated CD4 + T cells stimulate macrophages, synovial fibroblasts, B cells, and osteoclasts through the expression of cell surface molecules and Th1 cytokines, thereby contributing to both the chronic inflammation and the joint destruction. CD4 + T cells require two signals to be activated; antigen receptor occupancy and CD28-mediated costimulation. In the ST lesion, the CD28 ligands, both CD80 and CD86, together with MHC class II antigens, are substantially expressed by antigen-presenting cells such as macrophages and dendritic cells [ 18 - 20 ]. The significance of CD28 engagement in the T-cell-mediated disease process has recently been proven by the clinical efficacy of its blocker cytotoxic T-cell-associated antigen 4 (CD152)-IgG in RA patients [ 23 ]. IL-10 plays a predominant role in limiting immune and inflammatory responses by regulating the function of both macrophages and Th1 cells [ 1 ]. IL-10 inhibits the tyrosine phosphorylation of the CD28 molecule and the subsequent phosphatidylinositol 3-kinase binding in T cells, and thereby directly acts on T cells [ 24 ]. In the present study, we found that PB CD4 + T cells from patients with active RA, in the presence of IL-10, are able to produce higher levels of IFN-γ after CD3 and CD28 costimulation than normal CD4 + T cells. Despite high-level IL-10R1 expression and constitutive STAT3 activation, IL-10-induced tyrosine phosphorylation of STAT3 is suppressed in RA CD4 + T cells, in contrast to normal CD4 + T cells, where STAT3 phosphorylation is dose-dependently inducible by IL-10. Serum IL-6 from RA patients induces STAT3 but not STAT1 phosphorylation in normal CD4 + T cells, and exogenous IL-6 induces the resistance to IL-10 inhibition of IFN-γ production. RA CD4 + T cells contain higher levels of SOCS1 but contain lower levels of SOCS3 transcripts in comparison with normal CD4 + T cells. These findings indicate that CD4 + T cells become resistant to the inhibitory effect of IL-10 before migration into the inflamed ST, and suggest that this resistance may be attributable to impaired IL-10-dependent STAT3 activation, in association with sustained STAT3 phosphorylation and SOCS1 induction. IL-10-mediated inhibition of CD4 + T-cell cytokine production is principally dependent on its inhibition of macrophage antigen-presenting cell function [ 1 ]. However, this indirect inhibitory effect is thought to be restricted at the site of T-cell activation in RA, because macrophages in the ST express high levels of cytokines, CD80 and CD86 molecules, and MHC class II antigens [ 10 , 18 - 20 ]. More recently, IL-10 has been shown to induce the antigen-specific T-cell unresponsiveness by inhibiting CD28 tyrosine phosphorylation [ 33 ]. This direct effect also may be limited in active RA patients, because their PB CD4 + T cells showed a defective IL-10 inhibition of CD28-costimulated production of both IFN-γ and IL-2. Numerous cytokines, both proinflammatory and anti-inflammatory, have been detected in the ST of RA, and the balance between these opposing cytokine activities regulates disease severity [ 10 ]. Endogenous IL-10, produced mainly by macrophages and T cells, inhibits proinflammatory cytokine production by ST cells [ 12 ]. However, this regulatory activity seems to be restricted during chronic inflammation. The activation of both the extracellular stimulus-regulated kinase and p38 kinase pathways, induced by TNF-α and IL-1, inhibits the Jak1–STAT3 signaling pathway shared by IL-10 and IL-6 in adhered macrophages [ 13 ]. More importantly, IL-10-mediated STAT3 activation is mostly undetectable in RA synovial macrophages. This impaired IL-10 signaling is probably induced by chronic exposure to immune complexes in vivo , because both cell surface IL-10R1 expression and IL-10-induced Jak1 activation are suppressed in IFN-γ-primed macrophages by a protein kinase C-dependent pathway following ligation of the IgG Fc gamma receptor [ 14 ]. Furthermore, dendritic cells from RA synovial fluids are resistant to the immunoregulatory effect of IL-10 due to decreased transport of intracellular IL-10R1 in the presence of proinflammatory cytokine stimuli such as TNF-α, IL-1, and granulocyte–macrophage colony-stimulating factor [ 15 ]. We have demonstrated that the resistance of RA CD4 + T cells to IL-10 may be associated with defective IL-10-dependent STAT3 activation, but not with IL-10R1 expression. Inhibitory effects of IL-10 on these inflammatory cell types are therefore differentially modulated at the signal transduction level under the inflammatory environment in RA. In association with impaired IL-10-mediated STAT3 activation, STAT3 was found to be tyrosine phosphorylated persistently (up to 6 hours) in freshly isolated PB and ST CD4 + T cells from RA patients. STAT3 is activated by a variety of cytokines, notably the IL-6 family of cytokines (e.g. IL-6, IL-11, leukemia inhibitory factor, and oncostatin M) and growth factors, in addition to IL-10 [ 4 ]. Of these cytokines, IL-6 plays a predominant role in eliciting a systemic reaction such as the acute phase response in active RA, due mainly to its abundance in the blood circulation [ 27 ]. Consistent with this notion, IL-6 was the major STAT3-activating factor contained in the serum of active RA patients, and the responsiveness to IL-10 was suppressed in normal CD4 + T cells after 36 hours of incubation with IL-6. These results suggest that both the sustained STAT3 activation and the resistance to IL-10 inhibition found in RA CD4 + T cells may be induced after chronic exposure in vivo to high concentrations of serum IL-6. However, it is also possible that STAT3 activity could be constitutively induced in CD4 + T cells by their own IL-10 secretion, leading to the loss of sensitivity to exogenous IL-10, because RA CD4 + T cells in the ST are capable of producing significant levels of IL-10 [ 34 ]. CD4 + T cells isolated from the ST of RA also showed a defect in the IL-10-induced STAT3 signaling pathway. It is most probable that the resistance of CD4 + T cells to IL-10 can be even augmented after migration into the inflamed ST, because IL-6 is highly concentrated compared with the blood level [ 27 ]. In addition, the involvement of other essential proinflammatory cytokines in this process was suggested by our preliminary experiments demonstrating that IL-10-mediated IFN-γ inhibition in CD4 + T cells was reduced by pretreatment with IL-1β and TNF-α, although less effectively than by IL-6 (data not shown). Furthermore, IFN-γ and IL-10 produced by CD4 + T cells themselves could be responsible for impaired IL-10 signaling in the ST, because T-cell infiltrates produce both cytokines [ 34 , 35 ]. In an autocrine fashion, IL-10 may persistently stimulate STAT3 activation and IFN-γ can induce SOCS1 protein as a crosstalk inhibitor of IL-10 signaling [ 32 ]. The T-cell-inhibitory effect of IL-10 may therefore be modulated complicatedly upon exposure to an inflammatory environment in RA joints, where many cytokines are present substantially [ 10 ]. STAT3 activation has been implicated in the pathogenesis of RA. Active STAT3 is constitutively expressed in synovial fluid mononuclear cells from RA patients [ 36 ]. IL-6 is the major STAT3-activating factor present in synovial fluid, which has a crucial role in the activation of monocyte functions such as gene expression of the Fc gamma receptor type I and type III and of HLA-DR [ 31 ]. More recently, high levels of activated STAT3, thought to be induced mainly by IL-6, have been detected in the ST, and STAT3 activation has been shown to be involved in synovial fibroblast proliferation and IL-6 production [ 37 ]. In this regard, STAT3 is critical in the survival and expansion of growth factor-dependent synovial fibroblasts [ 38 ]. Furthermore, the significance of persistent STAT3 signaling in Th1-cell-dominated autoimmune arthritis has been suggested by studies of the gp130 F 759/ F 759 mice, in which the Src homology phosphatase-2 binding site of gp130 (the transmembrane glycoprotein β subunit of the IL-6 family cytokine receptor), tyrosine 759, was mutated to phenylalanine [ 39 ]. In the gp130 F 759/ F 759 mice, T cells, particularly the CD4 + T-cell subset, are chronically activated and resistant to activation-induced cell death through gp130-mediated STAT3 activation. The longevity of cytokine signals transduced by the JAK–STAT pathway is regulated by the SOCS family proteins [ 7 ]. We found that CD4 + T cells from patients with active RA expressed higher levels of SOCS1, but lower levels of SOCS3, compared with normal CD4 + T cells. SOCS1 prevents activation of JAK by directly binding to JAK, and SOCS3 inhibits the action of JAK by binding to the Src homology phosphatase-2-binding domain of receptors such as gp130 [ 40 ]. SOCS1 and SOCS3 are induced by various cytokines, including IL-6 and IL-10, as mediators of negative feedback and crosstalk inhibition [ 7 ]. Recent studies with mice lacking SOCS3 or SOCS1 revealed that SOCS3 is a negative regulator of IL-6 signaling but not of IL-10 signaling. Studies of conditional SOCS3-deficient mice have shown that SOCS3 deficiency, but not SOCS1 deficiency, results in sustained activation of STAT3 in response to IL-6 [ 8 , 41 ]. The analysis of SOCS3-deficient macrophages has indicated that SOCS3 is a crucial inhibitor of the IL-6-induced transcriptional response [ 42 ]. However, SOCS3 is dispensable for both the negative feedback inhibition and the immunoregulatory action of IL-10 in macrophages [ 41 ]. On the contrary, SOCS1 was found to directly inhibit IL-10-mediated signaling [ 43 ]. Increased SOCS1 expression in RA CD4 + T cells may therefore be associated with both the impaired responsiveness to IL-10 and to IL-10-mediated STAT3 activation, and defective SOCS3 expression may be responsible for persistent STAT3 activation in response to serum IL-6. There is a possibility that SOCS1 induction may be associated with the ability of CD4 + T cells to produce IFN-γ, because CD4 + T cells from active RA could produce high levels of IFN-γ in the presence of IL-10, and because IFN-γ has been known as a potent inducer of SOCS1 [ 32 ]. It is of interest in this regard to indicate that polarized Th1 and Th2 cells express high levels of SOCS1 and SOCS3 mRNA, respectively [ 44 ]. IL-12-induced STAT4 activation is inhibited by SOCS3 induction in Th2 cells, whereas IL-4-induced STAT6 signaling is diminished by SOCS1 induction in Th1 cells. SOCS1 and SOCS3 may thus have important roles as Th1-specific and Th2-specific, mutually exclusive, cross-talk repressors of the IL-4–STAT6 and the IL-12–STAT4 signaling pathways, respectively. Consistent with this notion, PB T cells from patients with allergic diseases significantly express high levels of SOCS3 transcripts, and the SOCS3 expression correlates well with serum IgE levels and disease pathology [ 45 ]. Higher SOCS1 expression with lower SOCS3 expression in PB CD4 + T cells from RA patients, compared with healthy controls, is therefore probably consistent with their systemic bias towards a Th1 phenotype, as has previously been demonstrated [ 46 - 49 ]. Conclusion CD4 + T cells from active RA patients are characterized by their resistance to IL-10 inhibition of IFN-γ production, due to constitutive STAT3 phosphorylation and impaired IL-10-mediated STAT3 activation. The defective STAT3 signaling is possibly associated with SOCS1 predominance over SOCS3. These abnormalities in active RA are thought to be induced mainly after chronic exposure to high concentrations of IL-6. The limited efficacy of IL-10 treatment of RA patients [ 50 ] may be explained in part by the unresponsiveness to IL-10 of inflammatory cells, including T cells. On the contrary, the therapeutic efficacy of anti-IL-6 receptor antibody has been reported in RA patients [ 51 ], and one of the effects of this therapy may be to normalize T cells through the inhibition of IL-6-dependent STAT3 activation. More specific therapy targeting STAT3 activation will be awaited; for example, the induction of the SOCS3 gene, the efficacy of which has been demonstrated in animal models [ 37 ]. Abbreviations BSA = bovine serum albumin; CRP = C-reactive protein; ELISA = enzyme-linked immunosorbent assay; Fc = crystallazibe fragment; FCS = fetal calf serum; FITC = fluorescein isothiocyanate; IFN-γ = interferon gamma; IL = interleukin; IL-10R = interleukin-10 receptor; IL-10R1 = type 1 interleukin-10 receptor; JAK = Janus kinase; mAb = monoclonal antibody; MHC = major histocompatibility complex; PB = peripheral blood; PBMC = peripheral blood mononuclear cells; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; RA = rheumatoid arthritis; SOXS = suppressor of cytokine signaling; ST = synovial tissue; STAT = signal transducer and activator of transcription; Th = T helper cells; TNF-α = tumor necrosis factor alpha. Competing interests The author(s) declare that they have no competing interests. Authors' contributions Jiro Yamana was responsible for the experiments and data analysis and wrote the report. Masahiro Yamamura was responsible for the planning of the research and wrote up the manuscript. Akira Okamoto, Tetsushi Aita, Mitsuhiro Iwahashi, and Katsue Sunahori assisted the experiments. Hirofumi Makino critically read the manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064873.xml

1064878

p53 tumor suppressor gene mutations in fibroblast-like synoviocytes from erosion synovium and non-erosion synovium in rheumatoid arthritis

Abnormalities in the p53 tumor suppressor gene have been detected in rheumatoid arthritis (RA) and could contribute to the pathogenesis of chronic disease. To determine whether synoviocytes from invasive synovium in RA have an increased number of mutations compared with non-erosion synoviocytes, p53 cDNA subclones from fibroblast-like synoviocytes (FLS) derived from erosion and non-erosion sites of the same synovium were examined in patients requiring total joint replacement. Ten erosion FLS lines and nine non-erosion FLS lines were established from nine patients with RA. Exons 5–10 from 209 p53 subclones were sequenced (114 from erosion FLS, 95 from non-erosion FLS). Sixty percent of RA FLS cell lines and 8.6% of the p53 subclones isolated from FLS contained p53 mutations. No significant differences were observed between the erosion and non-erosion FLS with regard to the frequency or type of p53 mutation. The majority of the mutations were missense transition mutations, which are characteristic of oxidative damage. In addition, paired intact RA synovium and cultured FLS from the same joints were evaluated for p53 mutations. Matched synovium and cultured synoviocytes contained p53 mutations, although there was no overlap in the specific mutations identified in the paired samples. Clusters of p53 mutations in subclones were detected in some FLS, including one in codon 249, which is a well-recognized 'hot spot' associated with cancer. Our data are consistent with the hypothesis that p53 mutations are randomly induced by genotoxic exposure in small numbers of RA synoviocytes localized to erosion and non-erosion regions of RA synovium. The determining factor for invasiveness might be proximity to bone or cartilage rather than the presence of a p53 mutation.

Introduction Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by synovial tissue proliferation with progressive joint destruction. The etiology of RA remains unknown, but many factors, including autoimmunity, cytokines and genetic factors, participate in its pathogenesis [ 1 , 2 ]. Although inflammation and joint destruction can be intimately related, the two processes might also be independent in some circumstances [ 3 , 4 ]. This observation might be explained, at least in part, by autonomous behavior by fibroblast-like synoviocytes (FLS) [ 5 ]. These cells exhibit some features of transformation in RA, including loss of contact inhibition, anchorage-independent growth, oncogene activation, autonomous invasion into cartilage and somatic gene mutations [ 4 , 6 - 9 ]. One potential gene that might contribute to this phenotype is the p53 tumor suppressor gene, which plays a critical role in cell-cycle regulation, DNA repair, senescence, genomic stability and apoptosis [ 10 ]. p53 is expressed in many inflammatory and autoimmune diseases [ 11 - 13 ], and it may serve a protective function by suppressing cytokine production and matrix destruction [ 14 , 15 ]. For instance, mice lacking the p53 gene have significantly greater joint destruction compared with wild-type controls in the collagen-induced arthritis model [ 16 ]. The function of p53 can be altered through somatic mutations in both neoplasia and non-malignant conditions such as ulcerative colitis, sun-exposed skin and chronic RA [ 8 , 17 , 18 ]. The mutations in RA synovium reside primarily in the intimal lining and have also been identified in cultured FLS, which are thought to originate from this region [ 8 , 9 ]. p53 sequencing studies in RA have until now focused on synoviocytes derived from non-erosion regions of the synovium that are readily obtained at the time of joint replacement surgery. Because some reports suggest that monoclonal expansion and oligoclonal expansion of synoviocytes occur at sites of invasion [ 19 ], we evaluated the relative frequency of p53 mutations in FLS in paired erosion and non-erosion synoviocytes from the same patients. Our data suggest that mutations are present in both regions and that the tendency to invade may be related to proximity to the extracellar matrix. Materials and methods Synovial tissues and preparation of FLS Synovial tissue samples were collected with informed consent at the time of joint replacement from patients with RA. The diagnosis of RA conformed to the 1987 revised American College of Rheumatology criteria [ 20 ]. Separate samples were obtained, at the University of Toronto, from within erosions at the periphery of the articular surface and from non-erosion sites collected from the intracapsular non-articular surface. Erosion FLS lines and non-erosion FLS lines were then prepared from each patient. FLS cell lines were established as previously described [ 21 ]. Briefly, tissues were minced by sterilized scissors, and were incubated with 1 mg/ml collagenase in serum-free DMEM for 2 hours at 37°C, were filtered through a nylon mesh, were extensively washed, and were cultured in DMEM containing 10% fetal calf serum, 2 mmol/l glutamine, 50 μg/ml gentamicin, 100 U/ml penicillin, and 100 μg/ml streptomycin, in a humidified 5% CO 2 atmosphere. After overnight culture, cells were trypsinized, split in a 1:3 ratio, and were recultured in medium. FLS from passages 5–8 were used in these experiments, during which time they represented a homogeneous population of FLS (< 1% CD 11b, < 1% phagocytic, and < 1% Fc-gamma RII receptor-positive). A second set of FLS samples obtained from the University of California at San Diego were obtained with a matched sample of synovium, which was embedded in TissueTek OCT compound (Miles Diagnostics, Elkhart, IN, USA), snap frozen, and stored at -80°C until use. Frozen tissues with approximate area of 10 mm 2 were cut into 10 μl sections at the time of PCR analysis for p53 mutations. Production of immunoreactive MMP-1 FLS were cultured to near confluence in six-well tissue culture plates (Falcon, Bedford, MA, USA). IL-1β (3 ng/ml) or medium was added to the wells and was incubated for 72 hours at 37°C in a humidified 5% CO 2 atmosphere. Supernatants were collected and MMP-1 concentrations were determined by ELISA according to the manufacturer's instructions (Total MMP1Biotrak; Amersham Biosciences, Piscataway, NJ, USA) [ 22 ]. Statistical analysis Comparisons between two groups were analyzed by the Wilcoxon signed rank test. P < 0.05 was considered significant. Results p53 mutations in erosion FLS and non-erosion FLS To determine whether invasive synovium in RA has an increased number of mutations, p53 cDNA subclones from FLS derived from erosion sites and non-erosion sites were examined. Ten erosion FLS lines and nine non-erosion FLS lines were established from nine patients with RA (two erosion lines and one non-erosion line were examined in one patient). A total of 209 p53 subclones were subjected to sequence analysis (114 from erosion FLS, 95 from non-erosion FLS), and p53 exons 5–10 were examined. As shown in Table 1 , p53 mutations were identified in 11 out of 19 FLS lines (four of 10 erosion FLS lines, and seven of nine non-erosion FLS lines). Eighteen subclones out of the total 209 (8.6%) contained mutations. There were no significant differences in the frequency of p53 mutations between erosion FLS and non-erosion FLS (7.9% and 9.5%, respectively). Nested PCR was required for one of the FLS lines (RA4 non-erosion FLS line). The rate of mutation with this line was similar to the other erosion and non-erosion FLS. As in previous reports [ 8 , 9 , 23 ], most p53 mutations (eight of nine in erosion FLS and eight of nine in non-erosion FLS) were transition mutations (i.e. G>A or C>T), which are characteristic of mutations caused by oxidative damage [ 24 , 25 ], and no transversion mutations (i.e. G>T, A>T, C>A, C>G) were seen (see Fig. 1 for the pooled data). One single base deletion and one multinucleotide insertion were detected. The majority of p53 mutations (78%) were missense (see Fig. 1 for pooled data). Most of the mutations were identified in a single subclone, although multiple copies of one mutation in codon 321 AAA to GAA were observed in an erosion FLS line. These data suggest that the frequency and types of mutations are similar in FLS isolated from either sites of erosion or from regions that are not invading into bone or cartilage. MMP-1 expression in erosion and non-erosion FLS Two matched erosion and non-erosion FLS lines were also evaluated for medium-stimulated and IL-1β-stimulated collagenase gene expression (MMP-1). There were no differences between the erosion or non-erosion samples with regard to either basal or cytokine-stimulated MMP-1 protein concentrations in the culture supernatants (basal, 5.5 ± 1.2 ng/ml; IL-1 stimulated, 24.4 ± 3.2 ng/ml). p53 mutations in matched RA FLS and synovial tissue samples After evaluating the matched FLS lines, we then examined the mutations in whole RA synovium and FLS isolated from the same joint. The matched erosion and non-erosion synovia from the preceding analysis were not available, so subclones from four additional paired RA FLS and synovial tissues from non-erosion regions were sequenced for p53 mutations (see Table 2 ). Mutations were detected in 12 of the 43 p53 subclones from RA FLS (28%), and in five of 46 subclones from the paired RA synovial tissues (11%). The relatively higher percentage of mutations in this limited number of lines compared with those presented in Table 1 is within the range observed for RA FLS in other studies. A few of the subclones contained two mutations (two of the RA FLS subclones, and one of the RA tissue subclones). Distinct patterns were found in the RA FLS compared with the paired tissue p53 subclones (see Table 2 ), although the frequency of mutations was somewhat higher in these samples compared with those presented in Table 1 and previous reports [ 8 , 9 , 23 ]. Mutation analysis of FLS revealed eight transitions, three transversions, and three deletions among 14 mutations in RA FLS. Of the base changes, 11 were missense and three were silent (Table 2 ). The RA tissue samples had five transition mutations (four A>G, one C>T) and one transversion mutation (G>T), with four mutations identified as missense and two as silent. Nested PCR was required for two FLS lines, RA11 and RA12 FLS, and these results were similar to the cell lines that did not require nested PCR. RA13 synovial tissue had no mutations identified despite the use of nested PCR, indicating the fidelity of this process. Interestingly, FLS from one patient had multiple subclones containing a mutation at codon 249 (AGG>GGG [Arg>Gly]) (see Table 2 ). Codon 249 missense mutations have been detected frequently in malignant tissues [ 26 , 27 ]. In another patient, a silent codon 213 base change (CGA>CGG [Arg>Arg]) was detected in 50% of p53 subclones from both FLS and synovial tissues. This same base change was also present in peripheral blood mononuclear cells of the patient (data not shown) and probably represents a known p53 germline polymorphism [ 28 ]. Discussion The aggressive nature of RA synovium and the ability of cultured synoviocytes to invade autonomously in cartilage suggest that these cells might be permanently altered or imprinted by their sojourn through the rheumatoid joint [ 4 , 5 ]. Additional data evaluating expression of X-linked genes indicate that oligoclonal or monoclonal expansion of synoviocytes can occur in chronically inflamed rheumatoid synovial tissue, especially at sites of erosions [ 19 ]. More recently, we showed that islands of cells expressing mutant p53 genes are present in the rheumatoid synovial intimal lining and that these regions produce significantly higher amounts of IL-6 [ 9 ]. However, the relationship between mutations and the synovial invasion has not been systematically examined. In the present study, we first examined paired samples of synoviocytes isolated from the erosive front of synovium and from regions not directly adherent to bone or cartilage for p53 mutations. Sixty percent of the cell lines examined had mutations, and 8.6% of the subclones isolated from either site contained mutations. No significant differences were observed between erosion and non-erosion FLS with regard to the frequency or type of mutation. Previous reports describe p53 base changes in 15–50% of RA FLS lines, with a frequency of mutations within the p53 cDNA pool varying from 0% to 30% [ 8 , 23 , 29 ]. The broad range might relate to the methods used to identify mutations, some of which are less sensitive (e.g. single-strand conformation polymorphism), or may be due to the evaluation of different numbers of exons. Other investigators failed to find mutations, perhaps because the experiments focused on sequencing the unfractionated p53 cDNA pool rather than subclones, on evaluation of less severe disease, or on sequencing a limited number of subclones [ 30 , 31 ]. The mutations observed in this study are mainly transition missense base changes, as previously described [ 8 , 9 , 23 ]. These are characteristic of oxidative deamination by nitric oxide or oxygen radicals [ 24 , 25 ] and are consistent with the hypothesis that the p53 mutations are caused by oxidative stress in the inflammatory environment, although this still has not been proven [ 32 ]. Relaxation of the DNA mismatch repair mechanisms in synoviocytes also probably contributes to susceptibility to DNA damage. For instance, suppression of hMSH6 expression in RA synovium has been associated with synovial microsatellite instability [ 33 ]. The majority of mutations in the present study were only identified in individual subclones. However, multiple copies were detected in the sequenced subclones from three patients. One of these at codon 249 is a well-recognized 'hot spot' associated with lung cancer and hepatocellular carcinoma [ 26 , 27 ]. Inazuka and colleagues previously identified another 'hot spot' codon 245 transition mutation in RA FLS lines from two patients [ 23 ]. Table 3 summarizes p53 mutation clusters (i.e. detected in more than one subclone) in the present study and in the literature in cultured RA FLS or synovial tissue. More than 90% of the repeat p53 mutations are missense and have been frequently detected in malignant tissues. Nishioka and colleagues demonstrated oligoclonal expansion or monoclonal expansion of synoviocytes at the sites of erosion in RA [ 19 ]. Furthermore, p53 expression tends to be greatest at sites of invasion in the severe combined immunodeficiency mouse model where cultured FLS erode into cartilage explants [ 34 ]. We expected to see an increased number of mutations at these sites as well. However, there were no differences between matched erosion and non-erosion FLS with regard to the number or types of p53 mutations. The mutations in very late stage of disease are therefore equally abundant in all regions of the rheumatoid synovium. Although they are clearly present at the invasive front, which might contribute to local tissue destruction, they are not over-represented compared with other sites that have been exposed to the genotoxic environment for the same period of time. The lack of association between invasion into bone and p53 mutations is consistent with recent data suggesting that osteoclasts, rather than synoviocytes, are the primary mediators of bone erosions [ 35 ]. FLS might play a more important role in cartilage damage through the elaboration of proteolytic enzymes and cytokines, which are increased in cells lacking functional p53 protein. Mutations in erosion had unique sequences when compared with the paired non-erosion mutations from the same patient, which may not be surprising given the results of microdissection studies demonstrating multiple independent islands of mutant cells rather than diffuse monoclonal expansion [ 9 ]. In addition to studying paired erosion and non-erosion FLS, we analyzed a second set of samples where we had the opportunity to examine paired whole non-erosion synovium with FLS isolated from the same joint. Mutations were identified in the matched samples, with similar ratios of transitions and missense changes between cell lines and tissues. There was no overlap in the specific base changes. Because cells in the lining form islands with oligoclonal expansion of individual mutations, cell lines derived from a different fragment of synovium would be expected to have different mutations compared with another region. The percentage of cDNA subclones with mutations was higher in the FLS than in matched synovium. This is most probably because the cells bearing mutations in the intact tissue (synoviocytes in the intimal lining) represent only about 20% of the total compared with 100% of the cells in the homogeneous cell cultures. Our data are consistent with the proposed hypothesis that p53 mutations are randomly induced by genotoxic exposure in small numbers of RA synovial lining cells in both erosion and non-erosion regions. Based on the association of these same mutations with neoplasia and our previous studies showing that these mutations can be dominant negative [ 36 ], it is reasonable to suggest that some of the p53 mutant cells in RA have selective growth advantage and thus form clusters in RA synovial tissue. Subsequently, the islands can influence cells in the environment through the elaboration of cytokines and factors that are normally suppressed by wild-type p53 (such as IL-6). A careful evaluation of erosion and non-erosion sites suggests that cells in both regions are equally likely to contain mutant cells, and that the expression of proteins related to matrix invasion, like MMP-1, was similar in the two cell populations. The determining factor for invasiveness might be proximity to bone or cartilage rather than the presence of a p53 mutation. Hence, cells directly adjacent to the matrix can potentially adhere and invade, whereas those cells in non-erosion regions would only have an indirect influence on destruction. Conclusions Mutations of the p53 tumor suppressor gene were present in synoviocytes isolated from both erosion and non-erosion sites in longstanding RA. Clusters of mutations can occur in RA synovium, but the abnormal clones are not unique to sites of joint destruction. The determining factor for invasiveness might be proximity to bone or cartilage rather than the presence of a p53 mutation. Hence, cells directly adjacent to the matrix can potentially adhere and invade whereas those cells in non-erosion regions would only have an indirect influence on destruction. Abbreviations DMEM = Dulbecco's modified Eagle's medium; ELISA = enzyme-linked immunosorbent assay; FLS = fibroblast-like synoviocytes; IL = interleukin; MMP-1 = matrix metalloproteinase-1; PCR = polymerase chain reaction; RA = rheumatoid arthritis Competing interests The author(s) declare that they have no competing interests. Authors' contributions Yuji Yamanishi designed and executed the study and prepared the manuscript. Edward Keystone, Alison Connor, and Susan Zollman developed erosion and non-erosion FLS. David Boyle assisted with the design of experiments and obtained UCSD clinical samples. Douglas Green evaluated and interpreted data and assisted with preparation of the manuscript. Gary Firestein supervised the project, evaluated and interpreted data, and prepared the manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064878.xml

1064879

The critical role of arginine residues in the binding of human monoclonal antibodies to cardiolipin

Previously we reported that the variable heavy chain region (V H ) of a human beta 2 glycoprotein I-dependent monoclonal antiphospholipid antibody (IS4) was dominant in conferring the ability to bind cardiolipin (CL). In contrast, the identity of the paired variable light chain region (V L ) determined the strength of CL binding. In the present study, we examine the importance of specific arginine residues in IS4V H and paired V L in CL binding. The distribution of arginine residues in complementarity determining regions (CDRs) of V H and V L sequences was altered by site-directed mutagenesis or by CDR exchange. Ten different 2a2 germline gene-derived V L sequences were expressed with IS4V H and the V H of an anti-dsDNA antibody, B3. Six variants of IS4V H , containing different patterns of arginine residues in CDR3, were paired with B3V L and IS4V L . The ability of the 32 expressed heavy chain/light chain combinations to bind CL was determined by ELISA. Of four arginine residues in IS4V H CDR3 substituted to serines, two residues at positions 100 and 100 g had a major influence on the strength of CL binding while the two residues at positions 96 and 97 had no effect. In CDR exchange studies, V L containing B3V L CDR1 were associated with elevated CL binding, which was reduced significantly by substitution of a CDR1 arginine residue at position 27a with serine. In contrast, arginine residues in V L CDR2 or V L CDR3 did not enhance CL binding, and in one case may have contributed to inhibition of this binding. Subsets of arginine residues at specific locations in the CDRs of heavy chains and light chains of pathogenic antiphospholipid antibodies are important in determining their ability to bind CL.

Introduction The identification of antiphospholipid antibodies (aPL) is a key laboratory feature in the diagnosis of patients with antiphospholipid antibody syndrome (APS). The cardinal manifestations of this syndrome are vascular thrombosis, recurrent pregnancy loss, livedo reticularis and thrombocytopenia [ 1 , 2 ]. APS may affect any organ of the body, leading to a broad spectrum of manifestations [ 3 ]. It is the commonest cause of acquired hypercoagulability in the general population [ 4 ] and a major cause of pregnancy morbidity. APS may occur as a 'freestanding' syndrome (primary APS) [ 5 ] or in association with other autoimmune rheumatic diseases (secondary APS) [ 6 ]. In both primary APS and secondary APS, recurrence rates of up to 29% for thrombosis and a mortality of up to 10% over a 10-year follow-up period have been reported [ 7 ]. The only treatment that reduces the risk of thrombosis in APS is long-term anticoagulation [ 8 ]. This treatment may have severe side effects, notably bleeding. It is therefore important to develop a greater understanding of how aPL interact with their target antigens so that new treatments for APS, which are both more effective and more accurately targeted to the causes of the disease process, may be developed. aPL occur in 1.5–5% of healthy people and may also occur in various medical conditions without causing clinical features of APS [ 9 ]. The aPL that are found in patients with APS differ from those found in healthy people in that they target predominantly negatively charged phospholipid antibodies and are in fact directed against a variety of phospholipid binding serum proteins. These proteins include protein C, protein S, prothrombin and beta 2 glycoprotein I (β 2 GPI) [ 10 - 13 ]. β 2 GPI is the most extensively studied of these proteins and appears to be the most relevant clinically [ 14 - 16 ]. Furthermore, high levels of IgG aPL, rather than IgM aPL, are closely related to the occurrence of thrombosis in APS [ 17 , 18 ]. Sequence analysis of human monoclonal aPL has shown that IgG aPL, but not IgM aPL, often contain large numbers of somatic mutations in their variable heavy chain region (V H ) and variable light chain region (V L ) sequences [ 19 ]. The distribution of these somatic mutations suggests that they have accumulated under an antigen-driven influence [ 20 ]. These monoclonal aPL tend to have accumulations of arginine residues, asparagine residues and lysine residues in their complementarity determining region (CDRs). Arginine residues have also been noted to play an important role in the CDRs of some murine monoclonal aPL [ 21 , 22 ]. Arginine residues, lysine residues and asparagine residues also occur very commonly in the CDRs of human and murine antibodies to dsDNA (anti-dsDNA) [ 23 - 25 ], particularly arginine residues in V H CDR3 [ 25 - 27 ]. It has been suggested that the structure of these amino acids allows them to form charge interactions and hydrogen bonds with the negatively charged DNA phosphodiester backbone [ 25 , 28 ]. We hypothesise that the same types of interaction may occur between negatively charged epitopes upon phospholipid antibodies/β 2 GPI and arginine residues, asparagine residues and lysine residues at the binding sites of high-affinity pathogenic IgG aPL. We have previously described a system for the in vitro expression of whole IgG molecules from cloned V H and V L sequences of human monoclonal aPL antibodies [ 29 ]. This system was used to test the binding properties of combinations of heavy chains and light chains derived from a range of human antibodies. One of these antibodies, IS4, is an IgG antibody derived from a primary APS patient. IS4 binds to anionic phospholipid antibodies only in the presence of β 2 GPI, can bind to β 2 GPI alone and is pathogenic in a murine model [ 30 ]. It is therefore likely to be relevant in the pathogenesis of APS. We found that the sequence of IS4V H was dominant in conferring the ability to bind cardiolipin (CL) while the identity of the V L paired with this heavy chain was important in determining the strength of CL binding [ 29 ]. Modelling studies have shown that multiple surface-exposed arginine residues were prominent features of the heavy chains and light chains that conferred the highest ability to bind CL. The CDR3 region of IS4V H contains five arginine residues, of which four are predicted by the model to be surface-exposed, and therefore is potentially important in binding to CL [ 29 ]. The purpose of the study reported in this paper was to define the contribution of different CDRs, and of individual arginine residues within those CDRs, in binding to CL. Patterns of CDR arginine residues in the cloned V H and V L sequences were altered by site-directed mutagenesis or by CDR exchange. The altered heavy chains and light chains were expressed transiently in COS-7 cells. Binding of the different heavy chain/light chain combinations to CL was tested by direct ELISA. Materials and methods Human monoclonal antibodies IS4, B3 and UK4 are all human IgG monoclonal antibodies produced from lymphocytes of three different patients. IS4 was derived from a primary APS patient by the Epstein–Barr virus transformation of peripheral blood mononuclear cells and fusion with the human-mouse heterohybridoma K6H6/B5 cell line [ 31 ]. IS4 binds to CL in the presence of bovine and human β 2 GPI, and to human β 2 GPI alone [ 31 ]. B3 [ 32 ] and UK4 [ 33 ] were isolated by fusion of peripheral B lymphocytes from systemic lupus erythematosus patients with cells of the mouse human heteromyeloma line CB-F7. B3 binds single-stranded DNA, dsDNA, CL and histones [ 32 , 34 ]. UK4 binds negatively charged (but not neutral) phospholipid antibodies in the absence of β 2 GPI and does not bind DNA [ 33 ]. Assembly of constructs for expression Wild-type heavy chain and light chain constructs Constructs containing the wild-type heavy chain and light chain were prepared as detailed fully in previous articles [ 29 , 35 ]. UK4V H could not be cloned into the appropriate plasmid, hence only UK4V L was available for analysis. The expression vectors (pLN10, pLN100 and pG1D210) were all kind gifts from Dr Katy Kettleborough and Dr Tarran Jones (Aeres Biomedical, London, UK). Hybrid V L chain constructs Each hybrid V L chain construct contained the CDR1 of one of the human monoclonal IgG antibodies IS4, B3 or UK4 and the CDR2 and CDR3 of a different one of these antibodies. Two hybrid V L chains (BU and UB) had previously been made by Dr Haley and colleagues [ 36 ], and a further four chains (IB, IU, BI and UI) were made by a similar method, as follows. Two different wild-type V L expression vectors were digested with Acc 65 I and Pvu I (Promega, Southampton, UK). Acc 65 I cuts IS4, B3 or UK4 V L sequences at a position in FR2 that is 106 base pairs from the beginning of V L , but does not cut the expression vector outside the insert. Pvu I cuts the vectors at a single site approximately 1 kb downstream of the insert. Each vector was therefore digested into two linear bands; one of approximately 1.5 kb and the other of approximately 6 kb. The 1.5 kb fragment contained CDR2 and CDR3 of the IgG V L region and also part of the downstream expression vector containing the lambda constant region cDNA, while the 6 kb fragment contained CDR1 and the rest of the vector. The 6 kb fragment derived from one V L expression vector was ligated with the 1.5 kb fragment derived from the other. The resulting plasmid would contain CDR1 of one V L sequence and CDR2 and CRD3 of another V L sequence. Since IS4, B3 and UK4 V L sequences differ in their content of the restriction sites Aat II and Ava I, we checked that the desired parts of each sequence were present in the new hybrid sequences by carrying out Aat II, Hind III/ Ava I and Aat II/ Bam HI digests. Site-directed mutagenesis of IS4V H We generated six mutant forms of IS4V H in which particular arginine residues were mutated to serine, using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA) according to the manufacturer's protocol. Serine was chosen because it is nonpolar. Germline reversion could not be performed because the exact germline D H gene of IS4V H CDR3 is unknown. Four mutants, named IS4V H i, IS4V H ii, IS4V H iii and IS4V H iv, contained single mutations of arginine residues at positions 96, 97, 100 and 100 g, respectively. The remaining two forms contained two arginine to serine mutations, at positions 96 and 97 in the IS4V H i&ii mutant and at all four sites in mutant IS4V H x. Expression of whole IgG molecules The whole IgG molecules were expressed in COS-7 cells as described previously [ 29 , 37 ]. Detection and quantitation of whole IgG molecules in COS-7 supernatant by ELISA Whole IgG molecules were detected and quantitated in the COS-7 cell supernatants using a direct ELISA, as described in previous papers [ 29 , 35 , 37 ]. Detection of binding to CL by ELISA The binding of IgG molecules to CL was measured by direct ELISA as described previously [ 29 ]. Results Sequences of light chains expressed Amino acid sequences of IS4V L , UK4V L , B3V L and germline gene 2a2 are shown in Fig. 1a . All of these light chains contain numerous somatic mutations. Previous statistical analysis has shown that the observed pattern of replacement mutations in the CDRs of these sequences is consistent with antigen-driven selection [ 32 , 33 , 35 , 38 - 40 ]. The light chain B3aV L , shown in Fig. 1a , was derived from B3V L by site-directed mutagenesis of Arg27a to serine [ 37 ]. The V L sequences of IS4, B3 and UK4 are all encoded by the germline V λ gene 2a2, but differ in their patterns of somatic mutation. B3V λ contains two adjacent arginine residues in CDR1, both produced by somatic mutations. UK4V λ has a single somatic mutation to arginine in CDR3 at position 94. A serine residue in CDR3 of IS4V L is replaced by asparagine. Figure 1a also shows the amino acid sequences of the V λ CDR hybrids in which each newly formed chain construct contains CDR1 of one antibody with CDR2 and CDR3 of a different antibody. These hybrid sequences were named by combining the names of the two parent antibodies such that the first letter represented the antibody from which CDR1 was derived and the last letter represented the antibody from which both CDR2 and CDR3 were derived. Hybrid IB thus contains CDR1 from IS4, and CDR2 and CDR3 from B3, whereas hybrid BI contains the reverse combination (CDR1 from B3, and CDR2 and CDR3 from IS4). Sequences of heavy chains expressed The amino acid sequences of IS4V H and B3V H chain and the corresponding germline genes are displayed in Fig. 1b . B3V H has a single somatic mutation to arginine in CDR2. The CDR2 of IS4V H contains an asparagine residue created by somatic mutation and in CDR3 there are multiple arginine residues, which are highly likely to have arisen as a result of antigen-driven influence. The four surface-exposed arginine residues that were mutated to serine to create the six mutant forms of IS4V H are underlined in Fig. 1b . Expression of whole IgG Each of the 10 light chains shown in Fig. 1a was paired with B3V H and IS4V H . Each of the six mutant forms of IS4V H was paired with IS4V L and B3V L . A total of 32 heavy chain/light chain combinations were expressed in COS-7 cells. At least two expression experiments were carried out for each combination. IgG was obtained in the supernatant for all of the combinations. The range of concentrations of IgG obtained in COS-7 cell supernatants, determined by ELISA, from each of the 32 heavy chain/light chain combinations are presented in Table 1 . Identical concentrations were obtained for the combination IS4V H ii/B3V L from two different expression experiments. In each case the negative control sample, in which COS-7 cells were electroporated without any plasmid DNA, contained no detectable IgG. Consistently high yields were obtained with the B3V H /BIV L , B3V H /UIV L and IS4V H /UIV L combinations compared with the other antibody combinations. The phenomenon of variable expression with different V H and V L constructs is well documented both in this antibody expression system and in other systems [ 35 , 37 ], although the reason for the occurrence of variable expression is not clear. Results of anti-CL ELISA For each heavy chain/light chain combination that bound CL, the linear portion of the binding curve for absorbance against antibody concentration was determined empirically, by dilution of antibody over a wide range of concentrations. Similar patterns of binding were obtained for each combination from repeated expression experiments, hence representative results from a single experiment only are shown in Figs 2 , 3 , 4 . The importance of arginine residues in IS4V H As reported previously, the presence of the heavy chain of IS4 plays a dominant role in binding to CL [ 29 ]. IS4V H binds CL in combination with six of the 10 light chains tested (see Figs 2a and 3 ): B3V L , B3aV L , BIV L , IS4V L , IBV L and UIV L . Only one of these light chains (B3V L ) binds CL in combination with B3V H (Fig. 2b ). To identify the features of IS4V H that enhance binding to CL, we focused on the combination IS4V H /B3V L . This combination shows high binding to CL. This binding could be altered by the replacement of some or all of the four surface-exposed arginine residues in IS4V H CDR3 to serine, as shown in Fig. 4 . Substitution of all four arginine residues with serine residues (IS4V H x) abolished CL binding completely. This effect seems probably due entirely to the changes at positions 100 and 100 g. This is supported by the fact that heavy chain combinations containing arginine to serine mutations at these positions (IS4V H iii and IS4V H iv) displayed approximately 50% weaker binding to CL in combination with B3V L than did the wild-type IS4V H /B3V L combination. In contrast, there were no reductions in CL binding for the heavy chains containing arginine to serine mutations at position 96 (IS4V H i), at position 97 (IS4V H ii) or at both positions (IS4V H i&ii). The importance of arginine residues in the light chain CDRs Six light chains bound CL in conjunction with IS4V H (Figs 2a and 3 ). The strongest binding was seen with light chains containing B3V L CDR1, namely B3V L , B3aV L and BI V L , in combination with IS4V H . In contrast, light chains IB and UB, containing CDR2 and CDR3 from B3, showed weak binding and no binding to CL, respectively, in combination with IS4V H . To test the hypothesis that the arginine at position 27a in B3V L CDR1 is responsible for the favourable effect of this CDR on binding to CL, we expressed combinations of IS4V H and B3V H with B3aV L , in which Arg27a has been mutated to serine. As shown in Fig. 3 , there was a significant decrease in CL binding of B3V H /B3aV L compared with B3V H /B3V L . Although the combination IS4V H /B3aV L binds CL less strongly than does IS4V H /B3V L , reduction in binding is not as great as that seen when these light chains are combined with B3V H . This observation is consistent with the idea that IS4V H plays a dominant role in binding to CL. Despite being tested at a range of concentrations up to 75 times higher than those that gave maximal CL binding for the other combinations containing IS4V H , none of the light chains containing CDR2 and CDR3 derived from UK4V L , including UK4 wild-type, IU and BU, showed any binding to CL. Discussion Previously we have shown the important roles played in antigen binding by IS4V H and B3V L , which both contain multiple nongermline-encoded arginine residues in their CDRs, supporting the idea that this amino acid is important in creating a CL binding site [ 29 ]. The results described in the present study demonstrate that it is not just the presence of, but the precise location of arginine residues in the CDRs that is important in determining the ability to bind CL. The importance of arginine residues at specific positions in the V H and V L sequences of anti-DNA antibodies has been examined by many groups, by expressing the antibodies in vitro and then altering the sequence of the expressed immunoglobulins by chain swapping or mutagenesis [ 27 , 37 , 41 - 43 ]. In general, these studies have shown that altering the numbers of arginine residues in the CDRs of these antibodies can lead to significant alterations in binding to DNA. Arginines in V H CDR3 often play a particularly important role in binding to this antigen [ 27 , 37 , 41 - 43 ]. Behrendt and colleagues recently demonstrated that the affinity of human phage-derived anti-dsDNA Fabs from a lupus patient correlated with the presence of somatically mutated arginine residues in CDR1 and CDR2 of the heavy chain [ 44 ]. Previous studies of the contribution of aPL heavy chains or light chains to CL binding have yielded conflicting results. Different groups have reported important contributions from the heavy chain [ 21 , 45 ], from the light chain [ 46 ], or from both chains [ 43 , 47 ]. In one of these studies the role of arginine residues was examined in a murine antibody (3H9) with dual specificity for phospholipid antibodies and DNA [ 21 ]. The introduction of arginine residues into the V H at positions known to mediate DNA binding enhanced binding to phosphatidylserine–β 2 GPI complexes and to apoptotic cell debris, which may be an important physiological source of both these antigens [ 48 ]. Our data show that combinations of IS4V H with light chains containing CDR1 of B3 (B3V L , B3aV L and BIV L ) produced the strongest binding to CL. The CDR1 of B3V L and BIV L contains two surface-exposed arginine residues at positions 27 and 27a, while B3aV L contains only one arginine at position 27. Previous modelling studies have suggested that the binding of B3V H /B3V L to dsDNA is stabilised by the interaction of dsDNA with Arg27a in CDR1 and Arg54 in CDR2 of the light chain [ 34 ]. Expression and mutagenesis studies from our group confirmed that mutation of Arg27a to serine led to a reduction in binding to DNA [ 37 ]. In the present study the same change has been shown to reduce binding to CL, supporting the conclusion of Cocca and colleagues that arginines at particular positions can enhance binding to both DNA and CL [ 21 ]. It is important, however, not to overlook the possible contribution of other amino acids in B3V L to CL binding. For example, substitution of histidine at position 53 with lysine and substitution of serine at position 29 with glycine could significantly influence the stability of the antigen binding site. In fact, we have previously shown that introduction of the Ser29 to glycine mutation in addition to the Arg27a to serine mutation in the light chain of B3V L /B3V H leads to a further reduction in binding to dsDNA [ 37 ]. The presence of UK4V L CDR2 and CDR3 in any light chain blocked binding to CL, even when combined with B3V L CDR1 (light chain BU). UK4V L CDR1, however, does not block binding. We have previously shown that the presence of UK4V L CDR2 and CDR3 blocks binding to DNA and histones but not to the Ro antigen [ 36 , 37 ]. Modelling studies have shown that an arginine at position 94 in CDR3 of UK4V L hinders DNA binding sterically. A similar effect may be occurring with regards to the binding of UK4V L to CL. The effect of point mutations of specific arginine residues in CDR3 of IS4V H upon CL binding is shown in Fig. 4 . The low binding of IS4V H /IS4V L was abolished by inclusion of any one of these mutations. This is not the case, however, when these mutants are expressed with B3V L . In this case the arginine residues at 100 and 100 g confer a greater effect on CL binding compared with the arginine residues at positions 96 and 97. Substitutions of all four of these IS4V H CDR3 arginine residues were sufficient to completely abolish all binding to CL. An accumulation of arginine residues in V H CDR3 has been noted in most, but not in all, sequences of pathogenic monoclonal aPL. From our detailed analysis of all published sequences of monoclonal aPL we found that of 13 monoclonal aPL that had been examined in various biological assays, eight monoclonal aPL had been shown to be pathogenic [ 49 ]. Three aPL derived from patients with primary APS and a healthy subject induced a significantly higher rate of foetal resorptions and a significant reduction in foetal and placental weight following intravenous injection into mated BALB/c mice [ 50 , 51 ]. Five other aPL derived from patients with primary APS and systemic lupus erythematosus/APS were found to be thrombogenic in an in vivo model of thrombosis [ 30 ]. We compared the sequences of these eight pathogenic antibodies with those of the other five antibodies, observing no evidence of pathogenicity in these bioassays. There was no evidence of preferential gene usage in either antibody group and somatic mutations were common in both groups. The presence of arginine residues in V H CDR3, however, did differ between pathogenic aPL and nonpathogenic aPL. Six of the eight pathogenic aPL, but only one of five nonpathogenic aPL, contain at least two arginine residues in V H CDR3 [ 49 ]. Our data confirm that the effect of arginine residues on binding to CL is highly dependent on the positions that they occupy in the sequence. The precise location of arginine residues has been shown to be important in the binding of both murine and human anti-dsDNA to DNA in numerous studies [ 25 , 26 , 37 ]. Interestingly, Krishnan and colleagues have demonstrated a strong correlation between specificity for dsDNA and the relative position of arginine residues in V H CDR3 [ 52 , 53 ]. They reported that the frequency of arginine expression among murine anti-dsDNA antibodies was highest at position 100, and they postulate that the importance of this residue in binding to dsDNA lies in its position at the centre of the V H CDR3 loop in the structure of the antigen combining site [ 52 ]. Assuming that this loop would be projected outward from the antigen combining site, an arginine residue at position 100 would be located at the apex of the V H CDR3 loop. Conclusion We have demonstrated the relative importance of certain surface-exposed arginine residues at critical positions within the light chain CDR1 and heavy chain CDR3 of different human monoclonal antibodies in conferring the ability to bind CL in a direct ELISA. It is now important to test the effects of sequence changes involving these amino acids on pathogenic functions of these aPL, by expressing the altered antibodies in larger quantities from stably transfected cells, and then testing them in bioassays. Abbreviations aPL = antiphospholipid antibodies; APS = antiphospholipid syndrome; β 2 GPI = beta 2 glycoprotein I; CDR = complementarity determining region; CL = cardiolipin; dsDNA = double-stranded DNA; ELISA = enzyme-linked immunosorbent assay; Fab = antigen-binding fragment; V H = variable heavy chainregion; V L = variable light chainregion. Competing interests The author(s) declare that they have no competing interests. Authors' contributions IG produced four hybrid light chains, participated in the production of the mutant heavy chains, antibody expression and study design, and drafted the manuscript. NL participated in the production of the mutant heavy chains and antibody expression. PC and RC produced the human monoclonal aPL IS4. DL and DI participated in study design and coordination. AR conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064879.xml

1064880

Relaxin's induction of metalloproteinases is associated with the loss of collagen and glycosaminoglycans in synovial joint fibrocartilaginous explants

Diseases of specific fibrocartilaginous joints are especially common in women of reproductive age, suggesting that female hormones contribute to their etiopathogenesis. Previously, we showed that relaxin dose-dependently induces matrix metalloproteinase (MMP) expression in isolated joint fibrocartilaginous cells. Here we determined the effects of relaxin with or without β-estradiol on the modulation of MMPs in joint fibrocartilaginous explants, and assessed the contribution of these proteinases to the loss of collagen and glycosaminoglycan (GAG) in this tissue. Fibrocartilaginous discs from temporomandibular joints of female rabbits were cultured in medium alone or in medium containing relaxin (0.1 ng/ml) or β-estradiol (20 ng/ml) or relaxin plus β-estradiol. Additional experiments were done in the presence of the MMP inhibitor GM6001 or its control analog. After 48 hours of culture, the medium was assayed for MMPs and the discs were analyzed for collagen and GAG concentrations. Relaxin and β-estradiol plus relaxin induced the MMPs collagenase-1 and stromelysin-1 in fibrocartilaginous explants – a finding similar to that which we observed in pubic symphysis fibrocartilage, but not in articular cartilage explants. The induction of these proteinases by relaxin or β-estradiol plus relaxin was accompanied by a loss of GAGs and collagen in joint fibrocartilage. None of the hormone treatments altered the synthesis of GAGs, suggesting that the loss of this matrix molecule probably resulted from increased matrix degradation. Indeed, fibrocartilaginous explants cultured in the presence of GM6001 showed an inhibition of relaxin-induced and β-estradiol plus relaxin-induced collagenase and stromelysin activities to control baseline levels that were accompanied by the maintenance of collagen or GAG content at control levels. These findings show for the first time that relaxin has degradative effects on non-reproductive synovial joint fibrocartilaginous tissue and provide evidence for a link between relaxin, MMPs, and matrix degradation.

Introduction In certain sites in and around joints, ligaments and tendons subjected to complex tensile and compressive loading specialize into fibrocartilaginous tissues [ 1 - 3 ] containing types I and II collagens and cartilage-specific proteoglycans. These tissues include specific regions of the metacarpophalangeal ligament and the deep flexor tendon, the temporomandibular joint (TMJ) disc, and the pubic symphysis. Within the pubic symphysis of several species, the reproductive hormone relaxin induces matrix remodeling activity during pregnancy and parturition, causing a marked decrease in collagen content through partly characterized mechanisms that transform this tissue into a ligamentous structure [ 4 - 9 ]. The relaxin-mediated loss of matrix macromolecules in the pubic symphysis and other tissues is exacerbated by estrogen [ 4 , 7 , 8 , 10 ]. The relative contribution of matrix synthesis and degradation to these relaxin-mediated changes is not clear, although collagen loss through increased proteolysis has been suggested [ 4 ], and studies in relaxin-knockout mice have implicated increased collagenase activity [ 11 ]. To understand the potential basis for relaxin and estrogen's modulation of the composition of fibrocartilaginous tissues, we previously studied cells isolated from rabbit TMJ discs. Relaxin induced the expression of the matrix metalloproteinases (MMPs) collagenase-1 (MMP-1) and stromelysin-1 (MMP-3) in a dose-dependent fashion but had little effect on the expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) or TIMP-2 [ 12 ]. In cells primed with β-estradiol, however, the relaxin concentration required for maximal induction of collagenase-1 and stromelysin-1 was 90–99% lower than in unprimed cells. Notably, the MMP response to relaxin was specific to fibrocartilaginous cells and was not observed in TMJ synoviocytes. These findings suggest that relaxin, by targeting fibrocartilage, might predispose women to musculoskeletal diseases of fibrocartilaginous joints. One such disease is TMJ disorders, which affect some 11 million adults in the USA [ 13 , 14 ], predominantly women, with a female : male ratio of 2:1 to 6:1 [ 14 ]. Unlike similar diseases of other joints, TMJ disorders occur primarily in women of reproductive age [ 14 ]. Given the gender and age distribution of these disorders and the relaxin-induced loss of matrix macromolecules in the pubic symphysis fibrocartilage [ 4 , 6 , 7 , 9 ] and isolated TMJ fibrocartilaginous cells [ 12 ], we have proposed that relaxin compromises the integrity of fibrocartilaginous tissues by enhancing the degradation of their matrices directly through the induction of specific MMPs. However, although relaxin causes a loss of collagens and proteoglycans in reproductive organs [ 6 , 7 ] and also increases MMP expression in specific tissues [ 6 , 12 , 15 - 21 ], the induction of MMPs by relaxin has not been demonstrated in joint fibrocartilaginous tissues or its induction of MMPs has not been linked to the loss of matrix macromolecules in any tissue. In this study we determined the effects of relaxin with or without β-estradiol on the modulation of MMPs, and assessed the contribution of these proteinases to the changes in collagen and glycosaminoglycan (GAG) content in fibrocartilaginous disc explants. Our findings are consistent with the hypothesis that relaxin-mediated induction of MMPs is associated with the loss of matrix macromolecules that could compromise tissue function and biomechanics and might lead to joint disease. Materials and methods Materials Twenty-week-old female New Zealand white rabbits were obtained from Nita Bell Laboratories (Hayward, California, USA). Ketamine hydrochloride was from Parke Davis (Morris Plains, New Jersey, USA), and xylazine was from Rugby Lab (Rockville Center, New York, USA). Lactalbumin hydrolysate, α-casein, β-estradiol-17-valerate, pepsin, papain, chondroitin sulfate A sodium from bovine trachea, Safranin-O, Fast Green, cetylpyridinium chloride, and other reagents were from Sigma (St Louis, Missouri, USA). 1,9-Dimethylmethylene blue (DMMB) was from Molecular Probes (Eugene, Oregon, USA), and 35 S was from Amersham (Arlington Heights, IL, USA). Protein assay kits, gelatin (EIA grade), and nitrocellulose membrane were from Bio-Rad (Hercules, California). α-Minimal essential medium, trypsin, penicillin–streptomycin, and Fungizone ® were from Gibco (Grand Island, New York, USA). All other standard chemicals were from Sigma or Fisher Scientific (Pittsburg, Pennsylvania, USA). Rabbit anti-human collagenase-1 polyclonal antibody and rabbit anti-mouse stromelysin-1 monoclonal antibody, horseradish peroxidase-conjugated secondary antibodies, and the MMP inhibitor GM6001 and its control analog were from Chemicon International (Temecula, California, USA). Rabbit anti-human-TIMP-1 antibody that cross-reacts with the rabbit inhibitor [ 12 ] was from Triple Point Biologics (Forest Grove, Oregon, USA). Enhanced chemiluminescence reagent for western blotting was from Amersham International (Little Chalfont, Bucks., UK). Sircol collagen assay kit was from Accurate Chemical and Scientific Corporation (Westbury, New York, USA), and fluorescein isothiocyanate (FITC)-labelled collagen was from Chondrex (Seattle, Washington, USA). Recombinant human relaxin was kindly provided by Connetics Corporation (Palo Alto, California, USA). Retrieval and culturing of TMJ discs, pubic symphysis, and articular cartilage All procedures on rabbits were approved by the Committee on Animal Research of the University of California, San Francisco, and conducted in accord with accepted standards of humane animal care. Rabbits were anesthetized with ketamine hydrochloride (40 mg/kg) and xylazine (3–5 mg/kg), and the TMJ discs were harvested bilaterally under sterile conditions and immediately placed in calcium-free and magnesium-free phosphate-buffered saline (PBS) containing antibiotics (100 U/ml penicillin, 100 mg/ml streptomycin, and 100 U/ml Fungizone). After removal of the synovium under a dissecting microscope, each disc was washed three times in PBS and bisected longitudinally such that four samples from each rabbit were available (three for hormone treatments and one for control). The hemisections were weighed, placed in wells of a 24-well culture plate, covered with 1 ml of serum-free medium (phenol-free α-minimal essential medium with 0.2% lactalbumin hydrolysate, glutamine, nonessential amino acids, 100 U/ml penicillin, and 100 mg/ml streptomycin) with or without hormones, and cultured at 37°C in air containing 5% CO 2 . For determination of MMPs and GAG staining, 32 hemisections from eight rabbits were exposed to medium alone, β-estradiol (20 ng/ml), relaxin (0.1 ng/ml), or both hormones at the same doses for 48 hours. The conditioned medium was collected and stored for MMP assays, and the discs were processed for GAG staining. To assess the contribution of relaxin-induced MMPs to the loss of collagen and GAGs, 24 hemisections from six rabbits were cultured with the MMP inhibitor GM6001 or its control analog 2 hours before and during the hormone treatments. The inhibitor was used at 10 μM, because this concentration was shown to inhibit collagenase activity induced by 0.1 ng/ml relaxin in dose–response experiments to baseline levels. The conditioned medium was collected and stored at -70°C for total protein and MMP assays. The discs were dried in a SpeedVac ® , weighed, digested, and used for the determination of GAG and collagen content. To determine whether the observed induction of collagenase by relaxin is specific to fibrocartilage, experiments were performed with pubic symphysis fibrocartilage, which is a known target site for β-estradiol and relaxin as a positive control, and with articular cartilage from the knee. For retrieval of articular cartilage, the joint was shaved, the articular surfaces were exposed, and the cartilage was scraped from the articular surfaces of the femur and tibia and incubated in PBS with antibiotic as described above. Similarly, the pubic bones and symphyseal areas were exposed under sterile conditions and the pubic symphysis (fibrocartilaginous tissues between the pubic bones) was dissected, removed, and incubated in PBS with antibiotics. The tissues were weighed, placed in wells of a 24-well culture plate, and studied as described above. Western blotting Hormone-induced changes in collagenase-1, stromelysin-1, and TIMP-1 were determined by western blotting. Disc-conditioned medium was mixed with 4 × sample buffer and subjected to SDS–polyacrylamide-gel electrophoresis with 10% or 18% gels. Equal amounts of protein (determined with a bicinchoninic acid protein assay kit) were loaded in each lane. The proteins were transferred to nitrocellulose membranes, which were blocked, washed, and incubated for 1 hour with antibodies against TIMP-1 (1:250 dilution), collagenase-1 (1:250 dilution in Tris-buffered saline), or stromelysin-1 (1:500 dilution). The membranes were then washed, incubated with horseradish peroxidase-conjugated goat anti-rabbit antibody (1:1000 dilution), and washed again. Bands were revealed by incubation with enhanced chemiluminescence reagent and exposure to radiographic film. The bands for TIMP-1 western blots were quantified by videodensitometry as described [ 22 ]. Conditioned medium from pubic symphysis and articular cartilage explants was similarly subjected to western blot analysis for collagenase-1 and stromelysin-1. Substrate zymography Enzyme activities were quantified by substrate zymography of conditioned media from 32 hemisections (mean wet weight 13 ± 9 mg). The samples were standardized by total protein and subjected to SDS–polyacrylamide-gel electrophoresis with 10% gels containing 2 mg/ml gelatin or casein at 15°C as described [ 22 ]. The gels were washed in 2.5% Triton X-100 for 30 min with one change of wash buffer, incubated at 37°C for 60–72 hours in incubation buffer (50 mM Tris-HCl buffer pH 8, 5 mM CaCl 2 , 0.02% NaN 3 ), stained with 5% Coomassie blue, and destained in 10% acetic acid and 40% methanol until proteinase bands were clearly visible. Images of the gels were captured with a charge-coupled device camera and NIH image software. The levels of 53/58 kDa gelatinolytic and 51/54 kDa caseinolytic enzymes and their low-molecular-mass activated forms were quantified by videodensitometry [ 22 ]. The substrate zymograms rather than western blots were used to quantify hormone-mediated increases in proteinase levels because zymograms are more sensitive, often display both pro-forms and active forms of proteinases, show a greater linear range of densitometric values and have good reproducibility that together enable a reliable quantification of the enzymes from these gels [ 23 - 25 ]. In addition, gelatin zymograms selectively detect proteinase activity at 53/58 kDa and at 43 kDa attributable primarily to collagenase rather than stromelysin because gelatin is a poor substrate for stromelysin [ 25 , 26 ]. Histochemical staining and quantification of GAGs To assess changes in GAG levels, the discs were washed three times in PBS, frozen in OCT compound, and sectioned with a cryostat. The section were defrosted for 30 min, fixed for 10 min in methanol, air-dried for 15 min, stained with 1% Fast Green solution for 3 min, placed in 1% acetic acid for 1 min, stained with 2% Safranin-O for 2 min, dehydrated through successive ethanol and xylene washes, and mounted with coverslips. Ten sections of each hemisection were analyzed by an examiner blinded to the hormone treatment. The stained discs were videodigitized and analyzed with a software program that automatically outlined the total and Safranin-O-stained areas with threshold settings (Photoshop 4.0; Adobe, San Jose, California, USA). These areas were then quantified with NIH Image 1.62, and the percentage of disc staining positive for GAGs was calculated from the ratio of the stained area to the total area in each section. The average of the 10 values for each half disc was used for analysis. Determination of GAG synthesis by 35 S radiolabeling To quantify GAG biosynthesis, 32 disc hemisections (mean weight 14 ± 4 mg) were incubated at 37°C for 6 hours in 1 ml of phenol-free and serum-free medium with or without hormones and 165 kBq (0.0044 mCi) of 35 S as described [ 27 ]. The discs were washed three times with medium containing 1 mg/ml sodium sulfate and digested for 24 hours with 20 U/ml papain. The digest (500 μl) was incubated for 30 min with 100 μl of 5% cetyl pyridiuium chloride in 0.3 M potassium chloride at room temperature (20–22°C) to precipitate GAGs. After centrifugation (3000 g for 20 min), the supernatant was removed and the precipitate was dissolved in 600 μl of concentrated formic acid by heating to 70°C for 10 min. Aliquots (20 μl) of this solution were added to 3 ml of scintillation fluid and subjected to liquid scintillation counting. The radioactivity (counts/min) was standardized to the total dry disc weight. Quantification of GAGs and collagen Each disc hemisection was digested in 600 μl of 3 mg/ml pepsin in 0.05 M acetic acid and incubated at 37°C for 18–20 hours in a dry bath. DMMB binding assays for GAGs, and Sircol assays for collagen content, were performed in triplicate on 24 disc hemisections. The DMMB reagent was prepared as described [ 28 ]. Pepsin digests (200 μl) from each treatment group (GM6001 or analog control) were mixed with 1 ml of DMMB reagent, and absorbance at 525 nm was determined with a spectrophotometer. The GAG concentration (μg/ml) was determined by comparing the absorbance of the sample against a standard curve prepared from bovine chondroitin sulfate A, and the disc GAG content was standardized to the total dry tissue weight. For the collagen assay, 200 μl of pepsin digest was mixed with 1 ml of Sircol dye reagent, incubated for 30 min at room temperature, and centrifuged at 10,000 g to separate the unbound dye from the collagen-bound dye. After removal of the unbound dye, 1 ml of the alkali reagent was added to the collagen–dye complex and vortex-mixed to dissolve the collagen-bound dye completely. Aliquots (200 μl) were transferred to the 96-well plates, and absorbance at 550 nm was determined with a microtiter plate reader (Molecular Devices, Sunnyvale, California, USA). The collagen concentration (μg/ml) was determined against a collagen standard curve, and the disc collagen content was standardized to the total disc dry weight. Quantification of collagenase activity Collagenase activity in conditioned medium from discs cultured with GM6001 or control analog was assessed by FITC–collagen assay. A 96-well plate was coated with FITC–collagen (10 μg per well) overnight at 4°C and washed twice with PBS. Disc-conditioned medium (100 μl) was added to the wells, and the plate was incubated at 35°C for 1 hour. As a reference, 100 μl of blank medium containing 3000 ng of bacterial collagenase was added to one set of wells for complete digestion of FITC–collagen. After incubation, 90 μl from each well was transferred to another 96-well plate, and the fluorescence intensity of degraded FITC–collagen products was determined with a microplate spectrofluorometer (Spectramax Gemini XS; Molecular Devices) with excitation at 494 nm and emission at 518 nm. The data were converted to relative fluorescence units of collagenase activity as described by the manufacturer and standardized to the dry weight of each half disc. The fold differences in collagenase activity in medium from control and hormone-treated discs were determined for each experiment. All assays were performed in duplicate. Statistical analysis Because of inherent variability in matrix content and proteinase activity in discs from different rabbits, three disc hemisections from each rabbit were treated with hormones and one served as control. MMP levels and the GAG and collagen content in each hormone-treated disc hemisection were standardized to the values of the control hemisection within each animal and the fold changes were plotted as histograms. The statistical significance of differences was determined by single-factorial analysis of variance (ANOVA). Intergroup differences were analyzed by Fisher's multiple comparisons test; P < 0.05 was considered statistically significant. Values are expressed as means ± SD. Results Relaxin and β-estradiol induce collagenase-1 and stromelysin-1 in TMJ disc explants Explanted discs constitutively expressed collagenase-1 (MMP-1) (Fig. 1a , lane 1), and the expression of this proteinase was increased by exposure to relaxin alone or to β-estradiol plus relaxin (Fig. 1a , lanes 3 and 4). Gelatin substrate zymograms confirmed the induction of 53/58 kDa proteinase by these hormones and, because this assay is more sensitive than western blots, showed an additional 43 kDa gelatinolytic enzyme (Fig. 1b ). Because the gelatinolytic enzymes were inhibited by 1,10-phenanthroline (Fig. 1b , lane 5), these proteinases were characterized as MMPs, most probably procollagenase-1 and active collagenase-1. Western blots with conditioned medium from a disc explant exposed to relaxin showed that the 53/58 kDa and 43 kDa activities corresponded to procollagenase-1 and collagenase-1, respectively (Fig. 1b , lane 6). Proteinase expression was about 1.7-fold higher in relaxin-treated and β-estradiol plus relaxin-treated discs than in control cultures ( P < 0.05) and was not potentiated by β-estradiol (Fig. 1c ). Because the expression of stromelysin and collagenase is often coordinately regulated [ 29 ], we assessed stromelysin expression. Western blots showed that all three hormone treatments induced stromelysin-1 (MMP-3) (Fig. 1d ). Casein substrate zymograms demonstrated a 51/54 kDa caseinolytic proteinase (Fig. 1e , lanes 1–4) that was inhibited by 1,10-phenanthroline (Fig. 1e , lane 5), indicating a metalloprotease. This characterization was confirmed by western blotting (Fig. 1e , lane 6). Proteinase expression in relaxin-treated cultures was double that in control cultures ( P < 0.05) and was not potentiated by β-estradiol. Relaxin induces collagenase-1 and stromelysin-1 in fibrocartilage but not in articular cartilage In pubic symphysis fibrocartilage, which is a known target site for β-estradiol and relaxin, β-estradiol caused slight increases in collagenase-1, while relaxin alone or in combination with β-estradiol induced a substantially greater expression of collagenase-1 relative to untreated discs (Fig. 2a ). Relaxin also increased stromelysin-1 levels in pubic symphysis fibrocartilage. However, β-estradiol alone or in conjunction with relaxin produced substantially greater increases in stromelysin-1 levels than relaxin alone. In knee articular cartilage, although β-estradiol induced collagenase-1 and stromelysin-1, neither relaxin nor β-estradiol plus relaxin increased the expression of these proteinases over control levels (Fig. 2b,2d ). Indeed, relaxin alone seemed to inhibit stromelysin-1 expression in articular cartilage. Loss of GAGs parallels the induction of MMPs by relaxin but not by β-estradiol Because all hormone treatments induced stromelysin-1 expression in explanted discs, we assessed the level of a known substrate, proteoglycans, in Safranin-O-stained sections. The GAG-positive area was larger in control discs (30.1 ± 2.8% of total disc area) and discs treated with β-estradiol (29.7 ± 4.7%) than in those treated with relaxin (19.2 ± 3.3%) or β-estradiol plus relaxin (16.9 ± 2.7%) (Fig. 3a ). These findings reflect statistically significant differences ( P < 0.01, ANOVA) in GAG staining between control discs and those treated with relaxin ( P < 0.05, Fisher's test) or β-estradiol plus relaxin ( P < 0.01). Similarly, the GAG-positive area was significantly smaller ( P < 0.04, ANOVA) in discs treated with relaxin ( P < 0.05, Fisher's test) or β-estradiol plus relaxin ( P < 0.05) than in those treated with β-estradiol alone. β-Estradiol induces TIMP-1 To determine why GAG loss did not increase in parallel with stromelysin expression in explants treated with β-estradiol alone, we assessed GAG synthesis and TIMP-1 expression. Except for a significantly lower GAG synthesis in discs exposed to β-estradiol plus relaxin than in those exposed to β-estradiol alone ( P < 0.05), differences between the other groups were not significant (Fig. 3b ). β-Estradiol caused a significant ( P < 0.01) twofold induction in TIMP-1 expression over controls (Fig. 3c,3d ). However, neither relaxin alone nor β-estradiol plus relaxin modulated any changes in TIMP-1 expression in the disc explants. Inhibition of MMP activity prevents relaxin-mediated loss of GAGs To establish an association between the increased MMP activity and the loss of GAGs in explants treated with relaxin or β-estradiol plus relaxin, we cultured the explants with the MMP inhibitor GM6001 or its control analog. Western blot analysis showed a higher expression of stromelysin-1 in hormone-treated than untreated disc explants in the presence of GM6001 or its control analog (data not shown). However, zymography showed increased 51/54 kDa caseinolytic activity (stromelysin-1) only in hormone-treated explants incubated with the control analog (Fig. 4a ), and not in those incubated with GM6001 (Fig. 4b ). DMMB assays showed that hormone treatments in the presence of control analog decreased the GAG content ( P < 0.0001, ANOVA), which was 30% lower in relaxin-treated explants ( P < 0.001, Fisher's test) and 40% lower in those treated with β-estradiol plus relaxin ( P < 0.001) than in untreated explants (Fig. 4c ). Similarly, the GAG content was lower ( P < 0.0001, ANOVA) in discs treated with relaxin ( P < 0.05, Fisher's test) or β-estradiol plus relaxin ( P < 0.001) than in those treated with β-estradiol alone. In the presence of GM6001, however, hormone treatments did not affect the GAG content (Fig. 4d ). Relaxin-induced collagenase activity contributes to loss of disc collagen All three hormone treatments increased the expression of procollagenase-1 in the presence of GM6001 or its control analog similarly to that shown in Fig. 1a,1b,1c . However, as shown by FITC-collagen degradation assays, collagenase activity was significantly increased only by relaxin or β-estradiol plus relaxin in the presence of the control analog (Fig. 5a ). In discs incubated with GM6001, hormone-induced collagenase activity was inhibited to control levels (Fig. 5b ). Conversely, Sircol assays showed the collagen content was significantly decreased ( P < 0.0001, ANOVA) only in the presence of the control analog and only by relaxin (40% of control and β-estradiol alone; P < 0.0001, Fisher's test) or β-estradiol plus relaxin (60% versus control and β-estradiol alone; P < 0.0001) (Fig. 5c ). In the presence of GM6001, hormone treatments did not affect collagen content (Fig. 5d ). Discussion This study shows that relaxin induced the expression of collagenase-1 and stromelysin-1 in rabbit TMJ disc explants, accompanied by a loss of GAGs and collagen, but did not affect GAG synthesis. In explants cultured with the MMP inhibitor GM6001, collagenase-1 and stromelysin-1 activities in hormone-treated discs were inhibited to baseline levels, and collagen and GAG content were maintained at control levels. These findings show that relaxin has degradative effects on nonreproductive synovial joint fibrocartilaginous tissue and provide evidence that increases in MMP activity mediated by relaxin and β-estradiol plus relaxin contribute directly to the loss of disc collagen and GAGs. The lack of effect on GAG synthesis further validates the importance of the degradative component of the remodeling cycle in relaxin's modulation of matrix loss in fibrocartilage. Because the MMP inhibitor used in our studies is not specific for collagenase-1 and stromelysin-1, the hormone-induced loss of collagen and GAGs cannot be specifically linked to those two proteinases. Rather, our findings implicate MMPs in general in this response. However, because GM6001 has a low dissociation constant for both collagenase-1 and stromelysin-1 [ 30 ], and their induction by relaxin was accompanied by a loss of their matrix substrates, collagenase-1 and stromelysin-1 are probably involved in the relaxin-mediated loss of collagen and GAGs, respectively. In contrast to the results obtained with relaxin and β-estradiol plus relaxin, the induction of collagenase-1 and stromelysin-1 by β-estradiol alone was not accompanied by changes in GAG or collagen content within the disc. How can we explain this apparent discrepancy? β-Estradiol had little effect on GAG synthesis, as measured by 35 S incorporation, but it produced a statistically significant increase in TIMP-1 expression that could have counteracted any increases in degradative activity due to increased expression of collagenase-1 and stromelysin-1. Indeed, the results of the collagen degradation assay lend credence to this hypothesis. These findings imply that relaxin and β-estradiol selectively contribute to the degeneration of fibrocartilaginous tissue by differentially modulating MMP expression, matrix synthesis, and net matrix content. The potential similarities in the responsiveness of TMJ fibrocartilaginous explants and the pubic symphysis fibrocartilage to relaxin are reflected not only by the relaxin's induction of collagenase but also by the comparable loss of collagen on the exposure of these tissues to the hormone [ 4 , 9 ]. Thus, the extent of collagen loss in fibrocartilaginous disc explants exposed to relaxin (40%) or β-estradiol plus relaxin (60%) was similar to that in the pubic symphysis of unprimed and β-estradiol-primed ovariectomized nonpregnant rats (64 ± 4% and 68 ± 6%, respectively) [ 4 ]. Similarly, in pregnant ovariectomized rats, relaxin decreased collagen to 39% of the levels in nonpregnant animals [ 9 ]. Additionally, β-estradiol alone had minimal effects on the collagen content of the fibrocartilaginous TMJ disc, which is also similar to observations on the pubic symphysis [ 4 , 9 ]. Thus, relaxin with or without β-estradiol, but not β-estradiol alone, has a potent effect on the amount of collagen in fibrocartilaginous tissues from different sites, including the pubic symphysis and synovial joints. These findings also suggest that in fibrocartilaginous tissues, including the TMJ disc and possibly the pubic symphysis, relaxin decreases collagen and GAG content primarily by inducing MMP expression. The response of articular cartilage to relaxin or β-estradiol plus relaxin was substantially different from that of the TMJ disc and pubic symphysis fibrocartilages. Although the reasons for these differences remain to be determined, it is well accepted that articular cartilage is a cartilaginous tissue containing chondrocytic cells, whereas fibrocartilage is a heterogenous tissue composed of cartilage and fibrous tissue that contains cells of fibroblastic, chondrocytic, and fibrochondocytic phenotypes. It is plausible that of these cells, the fibroblastic and/or fibrochondrocytic cells found in fibrocartilage, rather than the chondrocytic cells, are those that produce the observed responses to relaxin and β-estradiol plus relaxin. Indeed, previous findings on both dermal fibroblasts showing a potent induction of MMP-1 [ 18 ] and on articular chondrocytes that show minimal modulation of total collagen synthesis by relaxin [ 31 ] lend credence to this hypothesis. Additional studies are indicated to address the mechanistic basis for the differences in responsiveness of fibrocartilaginous versus cartilaginous cells to relaxin. Our findings are consistent with emerging data suggesting that the mechanisms for the loss of matrix macromolecules caused by relaxin are tissue-specific [ 32 ]. Thus, for example whereas relaxin increases collagenase-1 expression in TMJ disc and pubic symphyseal [ 6 ] fibrocartilages, it had minimal effects on its expression in articular cartilage explants. In monolayer articular or multilayer growth plate rabbit chondrocytes, relaxin produces no net change in collagen synthesis and no alterations in type II collagen mRNA levels, but increases the expression of types I and III collagen mRNA, thereby amplifying the dedifferentiation process [ 31 ]. In contrast, relaxin downregulates collagen expression by up to 40% and induces collagenase expression in cultured dermal fibroblasts [ 18 ]. As in our study, relaxin increases collagenase activity in human cervical stromal cells; however, in contrast to our findings, it also increases GAG synthesis [ 15 , 16 ]. MMPs contribute substantially to tissue degeneration in inflammatory joint diseases, including rheumatoid arthritis and osteoarthritis [ 33 - 35 ]. Our findings show that relaxin directly modulates MMP expression and probably causes matrix loss in fibrocartilaginous tissues from a synovial joint. Although the effects of relaxin on loss of matrix macromolecules, particularly collagen, have been demonstrated in the fibrocartilaginous pubic symphysis [ 4 - 7 ], this is the first study to demonstrate a similar targeting of fibrocartilaginous tissues from the synovial TMJ, and may implicate this hormone in the pathogenesis of TMJ disease in a subset of women with these disorders. Because even subtle alterations in collagen and GAG composition can affect the structural properties and the ability of joint tissues to function normally, this modulation of MMPs and resulting matrix loss in the fibrocartilaginous TMJ disc by relaxin might explain the distinct age and gender distribution of TMJ diseases. Furthermore, these findings have potential physiologic relevance because the induction of collagenase-1 and stromelysin-1 and the loss of collagen and GAGs occurred at concentrations of relaxin found systemically in cycling women [ 36 - 38 ]. Although the ability of systemic relaxin to access the TMJ and reach the avascular disc remains to be determined, our recent findings in vivo showing relaxin-mediated decreases in GAG concentration in the TMJ discs of ovariectomized rabbits suggest that this systemic hormone can indeed access the TMJ disc and contribute to its degradation [ 39 ]. Conclusions Relaxin causes the targeted induction of collagenase-1 and stromelysin-1 in synovial joint and pubic symphysis fibrocartilages but not in articular cartilage. This induction of MMPs in joint fibrocartilage is accompanied by a loss of collagen and GAGs that is prevented by an MMP inhibitor, suggesting a link between relaxin, MMPs, and matrix degradation. These studies provide the first evidence that relaxin contributes to the degradative remodeling of joint fibrocartilage and that there is an association between relaxin-induced MMPs and matrix loss; they also suggest a potential mechanism of action of relaxin in contributing to TMJ diseases in a subset of women with these disorders. Competing interests The author(s) declare that they have no competing interests. Authors' contributions TN performed all experiments, assays and analysis in which MMP inhibitors were used. TTD performed all experiments to characterize the changes in MMPs and GAGs in joint fibrocartilage in response to relaxin and β-estradiol. GH and QZ characterized the responses of the pubic symphysis fibrocartilage and articular cartilage to the hormones. MS retrieved tissues from animals and assisted in several MMP assays. SK conceived the study, participated in its design and coordination, supervised the statistical analysis, and wrote the manuscript. All authors read and approved the final manuscript. Abbreviations ANOVA = analysis of variance; DMMB = 1,9-dimethylmethylene blue; GAG = glycosaminoglycan; FITC = fluorescin isothycyanate; MMP = matrix metalloproteinase; PBS = phosphate-buffered saline; TIMP = tissue inhibitor of metalloproteinase; TMJ = temporomandibular joint.

D:\keerthana\PMC001xxxxxx\PMC1064880.xml

1064881

Interleukin-7 deficiency in rheumatoid arthritis: consequences for therapy-induced lymphopenia

We previously demonstrated prolonged, profound CD4 + T-lymphopenia in rheumatoid arthritis (RA) patients following lymphocyte-depleting therapy. Poor reconstitution could result either from reduced de novo T-cell production through the thymus or from poor peripheral expansion of residual T-cells. Interleukin-7 (IL-7) is known to stimulate the thymus to produce new T-cells and to allow circulating mature T-cells to expand, thereby playing a critical role in T-cell homeostasis. In the present study we demonstrated reduced levels of circulating IL-7 in a cross-section of RA patients. IL-7 production by bone marrow stromal cell cultures was also compromised in RA. To investigate whether such an IL-7 deficiency could account for the prolonged lymphopenia observed in RA following therapeutic lymphodepletion, we compared RA patients and patients with solid cancers treated with high-dose chemotherapy and autologous progenitor cell rescue. Chemotherapy rendered all patients similarly lymphopenic, but this was sustained in RA patients at 12 months, as compared with the reconstitution that occurred in cancer patients by 3–4 months. Both cohorts produced naïve T-cells containing T-cell receptor excision circles. The main distinguishing feature between the groups was a failure to expand peripheral T-cells in RA, particularly memory cells during the first 3 months after treatment. Most importantly, there was no increase in serum IL-7 levels in RA, as compared with a fourfold rise in non-RA control individuals at the time of lymphopenia. Our data therefore suggest that RA patients are relatively IL-7 deficient and that this deficiency is likely to be an important contributing factor to poor early T-cell reconstitution in RA following therapeutic lymphodepletion. Furthermore, in RA patients with stable, well controlled disease, IL-7 levels were positively correlated with the T-cell receptor excision circle content of CD4 + T-cells, demonstrating a direct effect of IL-7 on thymic activity in this cohort.

Introduction Peripheral blood T-cell lymphopenia is long-lasting in patients with rheumatoid arthritis (RA) receiving lymphodepleting therapies, such as monoclonal antibodies [ 1 - 3 ] or high-dose cyclophosphamide with autologous stem cell rescue (autologous stem cell transplantation) [ 4 , 5 ]. It has now been extensively documented in a number of systems that IL-7 drives the survival and proliferation of human T-cells after lymphodepletion (for review [ 6 ]). In particular, high circulating levels of this cytokine have been documented in patients rendered lymphopenic either by lymphocytotoxic treatment [ 7 ] or by HIV infection [ 8 - 10 ]. IL-7 produced in response to lymphopenia stimulates proliferation of both naïve and memory human T-cells [ 7 ], but also has a direct stimulating effect on thymic activity [ 11 ]. IL-7 plays many other roles such as the induction/enhancement of a T-helper-1 immune response [ 12 , 13 ], maturation of monocytes into dendritic cells, recruitment and expansion of T-cell clones [ 14 - 16 ], and induction of natural killer cell lytic activity [ 17 - 19 ]. These make IL-7 a master modulator of T-cell-mediated immune responses, particularly in tumour surveillance and eradication, in addition to its role as master regulator of peripheral T-cell homeostasis [ 8 ] Specific abnormalities within the naïve T-cell compartment in RA, such as repertoire contraction and shortened telomeres, have suggested a possible defect in generating and/or maintaining naive T-cells [ 20 - 23 ]. Furthermore, we recently showed [ 24 ] that RA patients possessed fewer naïve CD4 + T-cells than did healthy control individuals and that a smaller proportion of these cells contained a T-cell receptor excision circle (TREC). Circulating C-reactive protein (CRP) levels correlated inversely with the TREC content of naïve CD4 + T-cells, suggesting that inflammation was driving naïve CD4 + T-cell proliferation and differentiation, leading to dilution of TREC-containing cells. We could not, however, exclude an additional intrinsic defect in thymic T-cell production in RA patients [ 24 ]. In recent studies we reported persistent and profound CD4 + T-cell lymphopenia in RA patients as long as 7 years after a single course of CAMPATH-1H monoclonal antibody treatment [ 25 ] and up to 36 months after autologous stem cell transplantation [ 26 ]. RA patients usually reconstitute their B and natural killer cells rapidly, whereas CD8 + T-cell reconstitution takes longer and full recovery of CD4 + T cells may never occur. This is in contrast to patients undergoing bone marrow or stem cell transplantation for haematological malignancy or solid tumours, in whom both T-cell compartments reconstitute within 1 year of follow up [ 27 - 29 ]. Poor reconstitution after lymphodepleting therapy is likely to result either from reduced de novo T-cell production from the thymus or from poor peripheral expansion of naïve and memory cells, both of which processes are driven by IL-7. Here we report on a deficit in circulating levels of IL-7 in a cross-section of RA patients. This is associated with a reduced production of IL-7 in bone marrow derived stromal cell cultures, and may contribute to the defective CD4 + T-cell reconstitution that occurs following therapeutic lymphodepletion, primarily at the level of mature T-cell expansion in the periphery. Furthermore, we show that TREC levels correlate with circulating levels of IL-7 in patients in whom inflammation is controlled. Methods Patient cohorts Ethical approval for the project was obtained from the Leeds Teaching Hospitals National Health Service Trust Ethics Committee, and informed consent was obtained from each participant. Healthy control individuals were recruited from among local blood donors ( n = 34). RA ( n = 28) and osteoarthritis (OA; n = 12) patients were recruited through routine clinics at the Leeds General Infirmary (Table 1 ). They included patients with early, drug naïve ( n = 7) and long-lasting, refractory ( n = 21) RA (CRP range 5–155 mg/l) and patients with established, long-lasting OA ( n = 12; CRP below detection range). For the reconstitution studies we analyzed three RA patient cohorts ( n = 31) and a cohort of non-RA patients with solid tumours ( n = 7; Table 2 ). Each RA patient received high-dose cytotoxic therapy followed by autologous haematological transplants [ 26 , 30 , 31 ]. Each had disease that had proved resistant to multiple conventional antirheumatic drugs. Cohort 1 received an unmanipulated graft; cohort 2 received a graft that had undergone selection for CD34 + cells; and cohort 3 received a graft that had been CD34 + cell selected and T-cell depleted. The clinical progress of these patients was previously described elsewhere [ 26 , 30 , 31 ]. Control patients (Table 2 ) included five individuals with lung carcinoma, one with breast carcinoma and one with melanoma. They received unmanipulated autologous grafts following high-dose chemotherapy, as previously documented [ 32 ]. For the IL-7 longitudinal studies, we analyzed four lymphoma and three sarcoma patients. All received intensive chemotherapy followed by reinfusion of unmanipulated autologous stem cells (Table 2 ). In addition, we studied three patients with systemic vasculitis who received the lymphocytotoxic monoclonal antibody CAMPATH 1H [ 33 ]. For our work on RA patients in clinical remission (Table 1 ), we recruited consecutive patients ( n = 36) attending the rheumatology outpatient clinics with stable RA. They possessed no clinically significant synovitis and were deemed to be in 'remission' by the assessing consultant rheumatologist. Patients satisfied all of the following inclusion criteria: previous certified diagnosis of RA; over 18 years of age; disease duration of at least 12 months before remission; no disease flare within preceding 6 months; stable treatment within preceding 6 months; nil or minimal clinical evidence of active inflammatory disease and CRP below 15 mg/l within preceding 6 months; and no clinical indication to change treatment. We further refined this cohort by separating patients into those who satisfied the American College of Rheumatology (ACR) remission criteria and those who did not (Table 3 ). Cytokine measurements IL-7, transforming growth factor-β 1 , IL-6, tumour necrosis factor (TNF)-α and oncostatin M levels in sera and in tissue culture supernatants were measured using enzyme-linked immunosorbent assay (ELISA; R&D, Abingdon, UK), in accordance with the manufacturer's instructions. The sensitivities of the assay were <0.1 pg/ml for IL-7, 0.2 pg/ml for IL-6, 0.5 pg/ml for TNF-α, and 20 pg/ml for oncostatin M. T-cell subset separation Peripheral blood mononuclear cells (PBMCs) were recovered as described previously [ 24 ], and CD4 + and CD8 + T cells were separated by negative selection (Metachem, Meylan, France). Purified CD4 + and CD8 + T cells (>92% pure for CD4 + and 89% pure for CD8 + T cells) were stained for CD45RB (FITC; Dako, Ely, UK), CD45RA (PE; Serotec, Oxford, UK), CD45RO (PE-CY5; Serotec) and CD62L (ECD Coulter, High Wycombe, UK) using conventional methods. Naïve T-cells were further sorted according to their CD45RB bright , CD45RA + and CD62L + phenotype, using a FACS-Vantage cell sorter (Becton Dickinson, Oxford, UK). Memory cells and other subsets were identified based on their expression of CD45RB bright/dull , CD45RA ± , CD45RO bright/dull , and CD62L ± [ 24 ]. Real-time polymerase chain reaction quantification of T-cell receptor excision circles DNA was extracted from the different lymphocyte populations using standard proteinase K digestion followed by a phenol/chloroform extraction, either from total CD4 + and CD8 + populations after magnetic separation or from naïve cells after further cell sorting. TRECs were quantified using a real-time polymerase chain reaction based assay, as described previously [ 24 ]. Briefly, TREC primers were F (d-CAC CTC TGG GCT ACG TGC TAG) and R (d-GAA CAC ATG CTG AGG TTT AAA GAG AAT); and glyceraldehyde-3-phosphate dehydrogenase primers were F (D-AAC AGC GAC ACC CAT CCT C) and R (d-CAT ACC AGG AAA TGA GCT TGA CAA). This analysis provided a final value that represented TREC DNA as a proportion of glyceraldehyde-3-phosphate dehydrogenase DNA, which is equivalent to the percentage of cells containing a TREC. Following the release of the entire T-cell receptor locus sequence late in 2002, we validated our assay utilizing an alternative set of TREC primers designed to minimize background signal when using PBMC DNA. Proliferation assays PBMCs were separated as above from 5 ml blood from RA patients and healthy control individuals. An aliquot of PBMCs was stained with a combination of CD127 (FITC; Serotec), CD19 (PE; Serotec) and CD4 or CD8 (PE-CY5; Serotec) to quantify IL-7 receptor expression on different cell types by flow cytometry. Cells were resuspended in RPMI 1640 supplemented with penicillin and streptomycin, glutamine and 10% human AB + serum (Sigma, Aldwich, UK) and proliferation was assessed in response to PHA (10 μg/ml, Sigma), IL-2 (20 units/ml; Sigma), IL-7 (1–100 ng/ml; Sigma) or anti-CD3 antibody (OKT3; 1 μg/ml) with or without anti-CD28 antibody (YTH913.12; 5 μg/ml) co-coated on plastic Proliferation was quantified by incorporation of 3 H-thymidine (1 μCi/well) after 5 days of culture. Long-term bone marrow cultures Bone marrow mononuclear cells were obtained from posterior iliac crest aspirates from RA patients and healthy control individuals after informed consent had been obtained (with local research ethics committee approval), following centrifugation on Lymphoprep (Nycomed Pharma AS, Oslo, Norway), as previously described [ 34 , 35 ]. Aspirates from RA patients were repeated after 6–8 months of therapy with infliximab (Remicade; Schering Plough, Kenilworth, NJ, USA). Long-term bone marrow cultures from 10 7 bone marrow mononuclear cells were grown, in accordance with standard techniques [ 34 , 35 ]. By allowing the formation of an adherent layer consisting mainly of macrophages and cells of mesenchymal origin, this culture system has been considered appropriate for evaluating the regulatory role played by the bone marrow microenvironment in haematopoiesis [ 36 ]. At weekly intervals, cultures were fed by demi-depopulation. The adherent layer was usually confluent after 3–4 weeks, and at that time point cell-free supernatants were harvested and stored at -70°C for cytokine quantification. Statistical methods Nonparametric tests were used throughout. The Mann–Whitney U-test for two independent samples was used to compare healthy control individuals with RA patients. The Spearman rank correlation coefficient was used to determine correlations between two variables. A Wilcoxon sign rank test was used to compare pretherapy and post-therapy outcomes. Results Basal interleukin-7 production is reduced in rheumatoid arthritis We measured serum levels of IL-7 in a cross-section of active RA patients ( n = 28), healthy control individuals ( n = 34) and OA patients ( n = 12). There was no correlation between serum levels of IL-7 and age in healthy control individuals [ 37 , 38 ], and sex did not make any difference. Circulating IL-7 levels (Fig. 1a ) were significantly lower in RA patients than in healthy control individuals ( P < 0.00001). In RA there was no association between levels of circulating IL-7 and disease duration, inflammation as measured by CRP (Fig. 1b ; nonsignificant correlation [R = 0.201, P = 0.161]), presence of a shared epitope ( n = 17), or antirheumatic therapy (nonsteroidal anti-inflammatory drugs, methotrexate, or steroids). OA patients exhibited slightly lower IL-7 levels than did control individuals ( P = 0.035) but they had significantly higher IL-7 levels than did RA patients ( P < 0.00001). After Bonferroni correction there was no longer a significant difference between control individuals and OA patients, but other results remained unaffected. Regression analysis did not reveal any further trends. There are several sources of IL-7 production, including stromal cells in the bone marrow, dendritic cells and epithelial cells in the thymus, skin and gut [ 6 ]. We compared the ability of bone marrow stromal cells, derived from RA patients ( n = 9) and healthy control individuals ( n = 15), to produce IL-7 spontaneously in long-term cultures (Fig. 1c ). The production of IL-7 was significantly lower in RA patients than in control individuals ( P = 0.001). Furthermore, production did not consistently change after clinical remission induced by therapeutic TNF-α blockade ( n = 8; P = 0.725). We also examined the PBMC response to IL-7 in RA patients and healthy control individuals. Whereas RA PBMCs responded suboptimally to IL-2, mitogen (PHA) or antigen (anti-CD3/CD28), as previously documented [ 39 ], their response to IL-7 was similar to that in control individuals (Fig. 1d ). Importantly, in a cross-sectional comparison of 10 RA patients and 10 healthy control individuals, we could not find a significant difference in the number of cells expressing the IL-7 receptor (CD127) or in its level of expression (data not shown). Altogether, these findings suggest a deficit in circulating levels of IL-7 in RA, possibly due to an inability to produce IL-7, at least in stromal cells of bone marrow origin. Defective T-cell expansion in rheumatoid arthritis Patients receiving lymphocytotoxic therapy for conditions other than RA reconstitute more rapidly and completely than do RA patients. We previously studied three cohorts of RA patients who had received high-dose chemotherapy followed by stem cell reinfusion (Table 2 ). As previously reported [ 26 , 30 , 31 ], CD4 + counts fell after treatment and subsequently remained low in all cohorts, with no significant differences due to graft manipulation (data not shown). In contrast, CD8 + T-cell counts initially rose before rapidly returning to basal levels. In the present study we compared T-cell reconstitution in 12 RA patients (six from each of cohorts 2 and 3) and seven patients with solid tumours (Fig. 2 and Table 2 ). To avoid potential confounding effects of immunosuppressive drugs, RA patients were removed from the analysis if it subsequently became necessary to reinstitute antirheumatic therapies at times when disease activity resumed. The figure therefore represents 12 RA patients pretreatment and seven at 9 months. The chemotherapy regimens differed between RA and non-RA patients, but the nadir lymphocyte counts were similar. Figure 2 illustrates the composition of the peripheral T-cell pool at baseline and at various times after treatment. The individual subsets were defined according to the lymphocyte differentiation pathway suggested by our previous work [ 24 ]. The most naïve cells are represented in grey at the top of each bar chart. These cells progress to conventional memory cells and their precursors (striped bars) via post-naïve cells (white bars). Presumed 'central' memory cells are presented in black. Notably, at baseline RA patients possessed no CD4 + and CD8 + central memory cells in peripheral blood, as reported previously [ 24 ]. After chemotherapy there was simultaneous accumulation of all subsets in cancer patients, resulting in rapid restitution of CD4 + T-cell counts within 3 months. The same was true of the CD8 + subsets except that there was an 'overshoot' of memory CD8 + T-cells. In contrast, there was no early expansion of any T-cell subset in RA, although some long-term restoration of naïve CD4 + subsets was observed. Naïve CD8 + T-cells also accumulated slowly, and there was a brief expansion of CD8 + memory cells. These marked differences between RA and cancer patients demonstrate that a limited early peripheral expansion after treatment may account, in part, for the lack of T-cell reconstitution in RA. Although graft manipulation differed between cancer patients (un-manipulated) and RA (CD34 selected ± T-cell depletion) as mentioned above, we found that graft manipulation did not affect the rate of reconstitution in RA patients (data not shown). Other factors that differ between the RA and control group reflect the underlying disease. For example, RA patients may have been exposed to low-dose corticosteroid therapy as part of their prior treatment. It is not possible to exclude an effect of such a factor on our data. Delayed thymic activity in rheumatoid arthritis In order to compare thymic activity after lymphodepletion, we measured TRECs longitudinally in CD4 + T-cells in the same RA and cancer patient cohorts. As a molecular marker of T-cell receptor rearrangement, TRECs provide a surrogate measure of recent thymic activity [ 40 , 41 ]. We measured TREC content in total CD4 + -T cells as well as in naïve CD4 + -T cells in isolation. Patterns of TREC variation were consistent between the seven cancer patients. TRECs rapidly accumulated after treatment but returned to baseline by 3 months (Fig. 3 ; open diamonds). Our data in cancer are therefore consistent with an early surge in thymic activity, followed by a slow return to baseline at a time when the T-cell counts have returned to baseline levels. Variation in the TREC content of naïve cells also followed that pattern. The reduction in TREC content of an individual subset such as naïve cells is better explained by proliferation within that subset [ 24 , 42 , 43 ], therefore suggesting that naïve T-cells underwent peripheral expansion, resulting in TREC dilution in CD4 + cells (open diamonds). Similar results were observed for CD8 + T cells (data not shown). The early thymic response to lymphopenia did not occur in the 12 RA patients. In contrast, the TREC content of total CD4 + T-cells climbed gradually for several months after treatment (Fig. 3 ; closed diamonds). The TREC measurements in naïve cells also did not return to baseline, however, suggesting a lack of proliferation of CD4 + naïve cells. Therefore, a delay in achieving good release of newly developed T-cells also appeared to contribute to slow T-cell reconstitution in RA after high-dose chemotherapy. Similar results were observed for CD8 + T-cells (data not shown). Lymphopenia-induced interleukin-7 production is defective in rheumatoid arthritis Figures 2 and 3 suggest that the development and expansion of CD4 + T-cells were compromised in lymphopenic RA patients. Both the development and expansion of T cells have been extensively documented in relation to IL-7 (for review [ 6 ]). The relative deficiency in circulating IL-7 levels in RA patients identified in Fig. 1 therefore suggests a significant role for IL-7 in impaired T-cell reconstitution following high-dose chemotherapy. We measured serum levels of IL-7 longitudinally in four RA patients after lymphodepleting therapy (cohort 3, without relapse within 12 months) and seven non-RA patients (Table 2 ). Figure 4 clearly demonstrates a four- to fivefold rise and subsequent decrease in IL-7 levels, coincident with short-term lymphopenia in non-RA patients (triangles). In marked contrast, IL-7 levels did not change significantly in four RA patients over 12 months of follow up (squares). Interleukin-7 levels correlated with thymic activity in patients with well controlled rheumatoid arthritis In RA patients whose disease was controlled by in vivo TNF blockade, spontaneous release of IL-7 from bone marrow derived stromal cell cultures was variable, remaining reduced in some patients but returning to normal in others (Fig. 1 ). We therefore decided to investigate IL-7 levels in patients with well controlled disease and minimal levels of disease activity for at least 6 months before recruitment ( n = 36; Table 1 ). Levels of IL-7 were heterogeneous and ranged from 2.47 to 16.25 pg/ml. No clinical parameter was significantly correlated with IL-7 levels (disease duration, remission duration, previous or current therapy, rheumatoid factor). We measured TREC in total CD4 + T-cells in these patients in clinical remission in relation to age. The results were also heterogeneous (Fig. 5a ; all triangles). Comparing these values with our previous results in healthy control individuals (small circles [ 24 ]), there appeared to be two distinct patient groups. One of these groups had a CD4 + T-cell TREC content similar to or higher than that in age-matched healthy control individuals, and the other group exhibited lower TREC content. We used the median TREC content to distinguish two groups. Open and closed triangles relate to group 1 (above the median TREC value) and group 2 (below the median TREC value), respectively. The relationship between TREC content and age was present in group 1 (thick line; R = -0.738, P = 0.001) but not in group 2. No clinical parameter was able to predict TREC content (disease duration, remission duration, previous or current therapy, rheumatoid factor). We reanalyzed the IL-7 data with respect to this dichotomy in TREC levels, and there was a significant difference in circulating levels of IL-7 between these two groups (group 1, n = 17: IL-7 12.71 ± 2.76 pg/ml, range 9.57–16.25 pg/ml; group 2, n = 19: IL-7 6.50 ± 1.88 pg/ml, range 2.47–9.30 pg/ml; P < 0.00001). Furthermore, there was a positive correlation between the levels of circulating IL-7 and the TREC content of total CD4 + T cells (Fig. 5b ; n = 36, all diamonds; R = 0.777, P < 0.00001). No similar relationship was observed in healthy control individuals ( n = 12; R = 0.219, P = 0.595). We subsequently reanalyzed the data according to the ACR criteria for clinical remission [ 44 , 45 ]. Patients fulfilling or not fulfilling the ACR criteria (Table 3 ) are shown as open and black diamonds, respectively, in Fig. 5b . The two populations were undistinguishable in terms of TREC content ( P = 0.807). There was no difference in their circulating levels of IL-7 (ACR positive: 9.07 ± 3.33 pg/ml, range 4.9–15.23 pg/ml; ACR negative: 9.31 ± 4.01 pg/ml, range 2.47–16.25 pg/ml; P = 0.838). Furthermore, the correlation between IL-7 and TREC content was maintained in both groups (ACR positive, n = 17: R = 0.680, P = 0.005; ACR negative, n = 19: R = 0.779, P = 0.001). These data suggest that, removing any influence of systemic inflammation, RA patients form two groups that are characterized by normal or low levels of thymic activity and IL-7. It is not possible to predict from these data whether these abnormalities are primary or, indeed, whether they have pathogenic significance. However, they may be important in the context of reconstitution capacity after lymphodepleting therapies. Discussion We previously demonstrated that RA patients failed to reconstitute their peripheral T-cell pool even several years after lymphodepleting therapy [ 25 , 26 , 30 , 31 ]. The aim of the present work was to identify possible factors underlying this observation. IL-7 drives the expansion of human T-cells [ 8 , 46 , 47 ], and moreover it is an important thymic stimulant [ 11 ]. We identified a deficit in circulating levels of IL-7 in a cross-section of patients with active RA (Fig. 1 ). It was therefore possible that a similar deficit in IL-7 was a critical factor in the suboptimal response to lymphopenia in RA patients. Our data suggest that the RA thymus has a similar reserve to the thymus of disease control individuals (Fig. 3 ; similar peak levels at 9 months in RA as at 1 month in cancer), although it exhibits a more sluggish response to lymphopenia. However, both naïve and memory RA T-cells expand poorly in response to lymphopenia (Fig. 2 ), and this appears to be the major factor limiting reconstitution. We have also demonstrated low levels of lymphopenia-induced circulating IL-7 in RA patients (Fig. 4 ), and low basal IL-7 production from stromal cells originating from the bone marrow (Fig. 1 ). Finally, we showed a direct correlation between circulating levels of IL-7 and thymic capacity to produce new T-cells in RA patients with clinically undetectable disease activity (Fig. 5 ). To date, IL-7 is not a cytokine that has been associated with RA. However, there are conflicting results regarding its expression in RA patients. In one study [ 48 ] IL-7 was present at high levels in the serum of adult RA patients, and it correlated with CRP. In contrast, in children with systemic juvenile RA, plasma levels of IL-7 were unrelated to disease activity (joint counts and circulating IL-6) and undetectable in synovial fluids [ 49 ]. In another study, IL-7 was elevated in RA synovial fluid but not in OA [ 50 ] and its production was associated with stromal cells in the synovium [ 51 ]. Circulating levels of IL-7 in healthy control individuals are also very heterogeneous between publications (ranging from 0.1 to 30 pg/ml), possibly because of the use of different ELISA systems (commercial IL-7 ELISA kits using monoclonal or polyclonal antibodies, in-house sandwich ELISA using polyclonal rabbit antisera). In our study we found that IL-7 levels were highly dependent on the condition of serum collection (in particular, the type of Vacutainer [Greiner Bio-one, Knemsmuster, Austria; standard NHS supply]) and we standardized our collection protocol (blood taken into plain glass tubes, clotting time of 2 hours at room temperature, centrifugation at 1000 g for 10 min, storage at -20°C). In addition, in a recent report from Fry and Mackall [ 8 ], circulating levels of IL-7 in CD4 + T-cell depleted and repleted HIV patients were in keeping with our findings (<30 pg/ml and 10–20 pg/ml, respectively). Peripheral T-cell expansion differed greatly between our patient groups, as shown in Fig. 2 . This was particularly obvious for memory cells and their precursors, and appeared sufficient to account for the reconstitution defect in RA. However, lack of TREC dilution in naïve T-cells (Fig. 3 ) also suggested an absence of expansion within that subset in RA. IL-7 deficiency may again be relevant. IL-7 is produced in response to lymphopenia [ 7 ] and stimulates proliferation of both naïve and memory human T-cells. Although serum was not available from our cohort of solid tumour patients, we found high circulating levels of IL-7 in lymphodepleted patients with other tumours and with systemic vasculitis (Fig. 4 ), which is in keeping with the literature. In contrast, we found that basal serum IL-7 levels were reduced in a range of RA patients, irrespective of inflammation or medication (Fig. 1b ). Furthermore, there was no IL-7 rise following lymphodepletion (Fig. 4 ). RA and control PBMCs responded equivalently to IL-7 stimulation in vitro , suggesting no defect in IL-7 receptor expression or signalling (Fig. 1d ). Circulating IL-7 levels may also reflect the availability of specific binding sites on T-cells [ 6 ], but our two patient groups were similarly lymphopenic, making this explanation unlikely. Lymph node-resident dendritic-like cells may also produce IL-7 [ 52 ]. Although we were unable to examine these cells directly, our data do not suggest compensatory production from that source. Therefore, although we cannot definitively exclude alternative explanations for reduced IL-7 levels, low levels in lymphopenic RA patients (Fig. 4 ) and the variable ability to recover IL-7 in remission (Fig. 5 ) strongly implicate an underlying defect in IL-7 regulation, also highlighted by the bone marrow derived stromal culture (Fig. 1 ). IL-7 expression is upregulated or downregulated by different cytokines in different tissues (transforming growth factor-β, interferon-γ, TNF-α, IL-1 and IL-2, among others) and further work is necessary to uncover the mechanisms that control circulating levels of IL-7. CD8 + lymphopenia is also associated with raised circulating IL-7 levels [ 53 ] but this correlation is less strong. This suggests that factors other than IL-7 can effectively drive CD8 + T-cell expansion, and it is notable that transient expansion of CD8 + memory T-cells did occur in RA patients. Our experience and that of others suggests that such expansions may be driven by intercurrent infections (Isaacs JD, unpublished observations) [ 54 ]. This may also underlie the CD8 + T-cell over-compensation observed in cancer patients (Fig. 2 ). The RA thymus was clearly capable of producing new T-cells. This was evident not only when comparing naïve T-cell reconstitution in RA and cancer cohorts (Fig. 2 ) but also when TREC-containing cells were examined (Fig. 3 ). There is a complex relationship between thymic activity, T-cell proliferation and death, and TREC measurements [ 24 , 42 , 43 , 55 ]. Just after lymphocytotoxic therapy, however, TREC levels and T-cell counts are low and their subsequent accumulation must therefore reflect thymic output. TRECs achieved similar peak levels in both RA and cancer patients, suggesting an equivalent thymic capacity for T-cell production in these two groups. In cancer patients, however, TREC levels peaked early, as compared with a slow rise in RA patients. An association between higher levels of IL-7 and thymic capacity to produce new T-cells was predictable, based on the direct stimulatory effect of IL-7 on thymic activity at many stages in T-cell progenitor development [ 6 , 11 , 56 - 60 ] High levels of IL-7, as detected in lymphopenic control patients, could therefore result in a burst of thymic activity. In contrast, it is not immediately obvious what other factor(s) could determine the delayed rise in thymic activity in RA patients. Other growth factors are also able to stimulate the thymus [ 61 ], but another plausible mechanism is the removal of inhibition. Several of the cytokines that are abundant in RA, such as IL-6, oncostatin M and leukaemia inhibitory factor, suppress thymic function [ 37 ]. Levels of TNF-α, IL-6 and oncostatin M fell after high-dose chemotherapy in RA patients (data not shown) as the disease entered remission, and this may have resulted in a corresponding slow increase in thymic activity. Our data have pathogenic and therapeutic implications. First, they provide further support for a stromal cell function defect in RA. Previous studies of bone marrow progenitor cell reserve and stromal function in RA patients were more consistent with a defect secondary to TNF-α associated toxicity [ 34 ]. In those studies, progenitor cell reserve was reduced, and RA stroma was unable to support haematopoiesis from healthy CD34 + progenitors. Both abnormalities correlated with TNF-α levels in bone marrow culture supernatants and significantly improved after in vivo TNF-α blockade. Those data therefore support a scenario in which the RA marrow was suppressed by chronic exposure to TNF-α and potentially other proinflammatory cytokines. However, our data relating both to circulating IL-7 and to bone marrow production demonstrate independence from the inflammatory process (Fig. 1 ) at least in some patients, and are consistent with a primary abnormality. Therefore, supplementation with recombinant IL-7 may be necessary to improve lymphocyte reconstitution in lymphopenic RA patients, with the caveat that this cytokine is also a co-stimulatory factor for T-cells. It may therefore encourage the expansion of autoreactive T-cells with a worsening of disease. For example, IL-7 has been associated with preferential expansion [ 62 ] and activation [ 63 ] of autoreactive T-cells in multiple sclerosis. Additionally, IL-7 has been associated with lymphoproliferative disorders [ 64 - 66 ] to which RA patients are already predisposed. Furthermore, our data do not exclude additional contributions to limited T-cell expansion, and proliferative exhaustion is a factor that may not be amenable to therapeutic intervention. It is therefore possible that terminally differentiated memory T-cells, resulting from chronic immune activation in RA, cannot proliferate in response to lymphopenia. This does not explain the lack of proliferation of naïve T-cells from RA patients, however (Figs 2 and 3 ). Conclusion In conclusion, although our data are necessarily an averaged view of events that occur after lymphodepletion, we have made a number of observations relevant to poor T-cell reconstitution in lymphopenic RA patients. Importantly, the RA thymus is capable of producing naïve T-cells but its function is compromised by an IL-7 deficiency. The latter also severely limits the peripheral expansion of both naïve and memory T-cells. Our data suggest potential approaches to correct lymphocyte reconstitution defects in RA patients receiving lymphocytotoxic therapies and provide further insights into the disease process itself. Abbreviations ACR = American College of Rheumatology; CRP = C-reactive protein; ELISA = enzyme-linked immunosorbent assay; IL = interleukin; OA = osteoarthritis; PBMC = peripheral blood mononuclear cell; RA = rheumatoid arthritis; TNF = tumour necrosis factor; TREC = T-cell receptor excision circle. Competing interests The author(s) delcare that they have no competing interests. Authors' contributions Frederique Ponchel designed, optimized and performed the work on TREC quantification, T-cell differentiation, ELISA and statistics. Robert J. Verburg provided clinical sample (RA, Leiden). Sarah J Bingham provided clinical sample (RA, Leeds). Andrew K Brown provided clinical sample (RA in remission, Leeds). John Moore provided clinical sample (RA, Sydney). Andrew Protheroe provided clinical sample (cancer, Leeds). Kath Short collected clinical samples (cancer, Leeds). Catherine A Lawson processed clinical samples. Ann W Morgan provided clinical sample (RA, Leeds). Mark Quinn provided clinical sample (RA, Leeds). Maya Buch provided clinical sample (RA, Leeds). Sarah L Field provided technical support. Sarah L Maltby provided technical support. Aurelie Masurel provided technical support. Susan H Douglas provided technical support. Liz Straszynski provided technical support. Ursula Fearon provided technical support. Douglas J Veale provided provided support. Poulam Patel is Head of Department (cancer, Leeds). Dennis McGonagle provided support (Leeds). John Snowden provided support (Sheffield). Alexander F Markham is Head of Department (Leeds). David Ma is Head of Department (Sydney). Jacob M van Laar provided support and clinical material (Leiden). Helen A Papadaki is Head of Department and provided clinical samples (Heraklion). Paul Emery is Head of Department (Leeds). John D Isaacs is Head of Department (Leeds).

D:\keerthana\PMC001xxxxxx\PMC1064881.xml

1064882

Natural killer cell dysfunction is a distinguishing feature of systemic onset juvenile rheumatoid arthritis and macrophage activation syndrome

Macrophage activation syndrome (MAS) has been reported in association with many rheumatic diseases, most commonly in systemic juvenile rheumatoid arthritis (sJRA). Clinically, MAS is similar to hemophagocytic lymphohistiocytosis (HLH), a genetic disorder with absent or depressed natural killer (NK) function. We have previously reported that, as in HLH, patients with MAS have profoundly decreased NK activity, suggesting that this abnormality might be relevant to the pathogenesis of the syndrome. Here we examined the extent of NK dysfunction across the spectrum of diseases that comprise juvenile rheumatoid arthritis (JRA). Peripheral blood mononuclear cells (PBMC) were collected from patients with pauciarticular ( n = 4), polyarticular ( n = 16), and systemic ( n = 20) forms of JRA. NK cytolytic activity was measured after co-incubation of PBMC with the NK-sensitive K562 cell line. NK cells (CD56 + /T cell receptor [TCR]-αβ - ), NK T cells (CD56 + /TCR-αβ + ), and CD8 + T cells were also assessed for perforin and granzyme B expression by flow cytometry. Overall, NK cytolytic activity was significantly lower in patients with sJRA than in other JRA patients and controls. In a subgroup of patients with predominantly sJRA, NK cell activity was profoundly decreased: in 10 of 20 patients with sJRA and in only 1 of 20 patients with other JRA, levels of NK activity were below two standard deviations of pediatric controls ( P = 0.002). Some decrease in perforin expression in NK cells and cytotoxic T lymphocytes was seen in patients within each of the JRA groups with no statistically significant differences. There was a profound decrease in the proportion of circulating CD56 bright NK cells in three sJRA patients, a pattern similar to that previously observed in MAS and HLH. In conclusion, a subgroup of patients with JRA who have not yet had an episode of MAS showed decreased NK function and an absence of circulating CD56 bright population, similar to the abnormalities observed in patients with MAS and HLH. This phenomenon was particularly common in the systemic form of JRA, a clinical entity strongly associated with MAS.

Introduction The term 'macrophage activation syndrome' (MAS) in pediatric rheumatology refers to a set of symptoms caused by the excessive activation and proliferation of T cells and well-differentiated macrophages [ 1 - 4 ]. This activation leads to an overwhelming inflammatory reaction that can be fatal. The pathognomonic features of this syndrome are found in bone marrow aspirates: numerous, well-differentiated macrophages (or histiocytes) actively phagocytosing hematopoietic elements. Although MAS has been increasingly recognized in association with almost any rheumatic disease, it is by far most common in the systemic form of juvenile rheumatoid arthritis (JRA) [ 1 , 5 - 11 ]. Clinically, MAS has strong similarities to familial hemophagocytic lymphohistiocytosis (FHLH) and virus-associated or reactive hemophagocytic lymphohistiocytosis (HLH) [ 2 - 4 ]. The immune abnormalities in the familial form of HLH have been studied extensively, and the most consistent finding has been global impairment of cytotoxic lymphocyte and natural killer (NK) cell function [ 12 - 14 ]. In about 50% of patients with FHLH in North America, these immunologic abnormalities are secondary to mutations in the gene encoding perforin, a protein that mediates the cytotoxic activity of NK and T cells [ 15 ]. Although it has been proposed that abnormal cytotoxic cells might fail to provide appropriate apoptotic signals for the removal of activated macrophages and T-cells after infection is cleared [ 16 ], the exact pathways that would link the decreased NK and cytotoxic T cell function with macrophage expansion have not been confirmed. We have previously reported that, as in HLH, NK function is profoundly depressed in the vast majority of patients with MAS [ 17 ] suggesting that this immunologic abnormality might be relevant to the pathogenesis of the syndrome. In the present study we sought to assess the extent of NK dysfunction in the most common rheumatic disease of childhood, JRA. JRA is a chronic, idiopathic, inflammatory disorder with diverse clinical symptoms both at onset and during the course of the disease. Classification of this heterogeneous disease has been based primarily on the type of onset, namely the clinical manifestations during the first six months [ 18 , 19 ]. There are at least three major onset types: pauciarticular (four or fewer joints involved), polyarticular (five or more joints), and systemic. The systemic onset form, with its markedly febrile presentation, is certainly the most distinct clinical subtype of the disease. In contrast to patients with pauciarticular and polyarticular JRA, in whom the joint disease usually overshadows the more general symptomatology, in systemic onset JRA extra-articular features such as spiking fevers, evanescent macular rash, hepatosplenomegaly, lymphadenopathy, and, occasionally, polyserositis are most prominent [ 20 ]. The reason for the increased incidence of MAS in patients with systemic forms of JRA in comparison with other clinical forms of this disease is not clear, but NK cell abnormalities might have a role [ 3 ]. The main purpose of this cross-sectional study was to characterize numbers of circulating NK cells, their cytolytic activity, CD56 bright : CD56 dim subset ratio, and perforin/granzyme B expression in the major cytotoxic cell populations in patients with different clinical forms of JRA. Materials and methods Patients In all patients included in the study, the diagnosis of JRA was established on the basis of the American College of Rheumatology (ACR) diagnostic criteria [ 18 ]. The main clinical characteristics of the patients are summarized in Table 1 . Peripheral blood samples were collected from the patients after obtaining informed consent under an institutional review board-approved study of JRA. Flow cytometric analysis Relative and absolute numbers of NK, CD8 + , and NK T cells, as well as perforin and granzyme B expression in these cell populations, were determined as described previously [ 14 , 21 ]. In brief, whole blood samples were first surface stained with the following antibodies: fluorescein isothiocyanate-labelled T cell receptor (TCR)-αβ, CD8-peridinin chlorophyll protein (CD8-PerCP) (BD Immunocytometry Systems, San Jose, CA), and CD56-allophycocyanin (CD56-APC) (Immunotech, Brea, CA). Red cells were then lysed with FACSlyse (BD Immunocytometry Systems) and washed. The resultant white cell pellets were then fixed and permeabilized with Cytofix/Cytoperm (BD Pharmingen, San Diego, CA) and stained with either phycoerythrin-conjugated mouse IgG2b anti-perforin or phycoerythrin-conjugated mouse IgG1 anti-granzyme B antibodies (BD Pharmingen). After being washed, cells were resuspended in 1% paraformaldehyde and stored at 4°C until analyzed with a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA). The following gates were used to distinguish the three populations of interest: CD8 + T cells were defined as being TCR-αβ + , CD8 + , and CD56 - ; NK cells as TCR-αβ - and CD56 + ; and NK T cells as TCR-αβ + and CD56 + . All populations were also restricted to a live cell gate based on forward versus side scatter. The perforin-positive or granzyme B-positive regions were set by using isotype-matched negative control samples, and the percentage positive for each gate was reported. NK cell cytotoxicity analysis NK activity was assessed after co-incubation of peripheral blood mononuclear cell preparations (effector cells) with 51 Cr-labeled target cells at various effector : target cell ratios as described previously [ 21 ]. The NK-sensitive K562 line was used as a source of target cells. The levels of radioactivity released from target cells into supernatants were assessed by gamma scintillation after 4 hours of incubation. All experiments were performed in triplicate in a 96-well microtiter plate. Spontaneous and maximum release wells were included on each plate as controls. Spontaneous release was assessed in the wells containing 51 Cr-labeled target cells in medium without effector cells. Maximum release was determined in the wells containing labeled target cells in the presence of detergent to promote total lysis. The percentage lysis was calculated as described previously [ 21 ]: percentage lysis = 100 × (mean radioactivity of sample minus mean radioactivity of background)/(mean maximum radioactivity minus mean radioactivity of background). Lytic units were calculated from the curve of the percentage lysis. One lytic unit was defined as the number of effector cells needed to produce 10% lysis of 10 3 target cells during the 4 hours of incubation. Controls The results were compared with the normal ranges for age-matched controls that have been developed in our clinical laboratory by studying 41 pediatric samples obtained from the out-patient clinic during routine 'well-child' visits from children considered 'healthy' [ 14 ]. Statistical analysis The unpaired t -test and Wilcoxon two-sample test were used to compare NK cytolytic activity and perforin/granzyme B expression between the patient and control groups. The rank correlation test was used to characterize the relationship between NK cell activity and perforin expression. The unpaired t -test and logistic regression analysis were used to assess the possible contribution of treatment regimens to the development of NK cell dysfunction. Results NK cell cytolytic activity and NK cell numbers As shown in Fig. 1 , some decrease in NK cell cytolytic activity was noted in both clinical groups of JRA patients. This trend was particularly strong in patients with systemic JRA (sJRA). The mean cytolytic activity in the sJRA group was 4.0 (SEM = 1.2) compared with 8.2 (SEM = 1.6) in patients with pauciarticular/polyarticular JRA (unpaired t -test, P = 0.042; Wilcoxon two-sample test, P = 0.0062). Furthermore, in a subgroup of patients with sJRA, NK cell function was profoundly depressed. Thus, in 10 of 20 patients with sJRA and in only 1 of 20 patients with pauciarticular/polyarticular JRA, the NK cell cytolytic activity was below two standard deviations (SD) of the control group (χ 2 , P = 0.002). The same degree of NK cell dysfunction was observed in our previous studies of patients with MAS and HLH [ 17 ]. As shown in Table 1 , a significant proportion of the patients studied were on immunosuppressive medications, including prednisone, methotrexate, and tumor necrosis factor (TNF)-blocking agents. Because NK function can be affected by such medications [ 22 ], we sought to assess whether the differences in treatment regimens between patients with sJRA and pauciarticular/polyarticular JRA might have contributed to the observed differences in NK function. Overall, patients receiving immunosuppressive drugs had somewhat lower levels of NK cytolytic activity. However, logistic regression analysis showed no statistically significant differences in NK function between patients with or without immunosuppressive treatment (independent variables: prednisone, methotrexate, and TNF-blocking agents; dependent variable NK cytolytic activity below 2SD of control samples, χ 2 = 0.877, P = 0.831). Because low NK cell numbers in patients with sJRA have been previously noted in one study [ 23 ], we assessed whether the decrease in NK cell cytolytic activity might have been related to low NK cell counts. A moderate correlation between NK function and the proportion of peripheral blood mononuclear cells that were NK cells was found for both groups of patients with JRA ( r = 0.52, 95% confidence interval 0.08–0.8 in the sJRA group; r = 0.47, 95% confidence interval 0.7–0.75 in the other JRA group). Correlation coefficients between function and number of NK cells were not significantly different between the two JRA groups (Fisher's Z transformation; P = 0.7). The mean number of NK cells (expressed as a proportion of TCR - αβ - /CD56 + cells in a population of peripheral blood mononuclear cells) among the sJRA group was 0.077 (SD 0.04) in comparison with 0.081 (SD 0.034) among the other JRA group was not significantly different ( P = 0.72 on the basis of the two-tailed independent t -test). Thus, it seems that suppressed NK function is not simply a result of reduced numbers of NK cells among patients with sJRA. CD56 dim and CD56 bright NK cells On the basis of on the intensity of CD56 staining, human NK cells have been recently subdivided into two distinct subsets with distinct functional characteristics. CD56 bright NK cells have the ability to produce high levels of immunoregulatory cytokines, in particular interferon (IFN)-γ, but are in general poorly cytotoxic (reviewed in [ 24 ]). By contrast, CD56 dim NK cells produce relatively low levels of cytokines and are potent cytotoxic effector cells expressing high levels of peforin. We have previously described a lack of the circulating CD56 bright NK cells in patients with MAS and HLH [ 25 ]. The analysis of the fluorescence-activated cell sorting data in the current study revealed that three patients with sJRA had a similar abnormality, namely an almost complete absence of circulating CD56 bright cells. Figure 2 shows examples of CD56 staining in such patients. Perforin expression Because reduced perforin expression in cytotoxic effector cells has previously been reported in sJRA [ 26 , 27 ], we assessed perforin content and relative proportions of perforin-positive NK cells, CD8 + lymphocytes, and NK T cells. High variability was noted in both groups of patients with JRA. The comparison of the mean values between patients with sJRA versus other JRA groups did not reveal statistically significant differences. However, the examination of individual patterns of perforin staining in NK cells revealed mean channel fluorescence (MCF) values below 2SD of the control group in seven patients, five of whom had sJRA. In addition to low MCF, one of these patients with sJRA had profoundly decreased proportions of perforin-positive cells in all three cytotoxic cell populations, a pattern that we have previously described in patients with sJRA who have had multiple episodes of MAS [ 17 ]. Examples of such abnormal patterns of perforin expression are shown in Fig. 3 . In contrast, the patterns of staining for granzyme B were similar between patients and controls in all three cytotoxic cell populations (data not shown). Because perforin is a protein that mediates the cytotoxic activity of NK cells, we assessed whether decreased perforin expression might have contributed to the development of NK dysfunction in JRA. In the group of seven patients with low perforin levels in NK cells, only three had NK cytolytic activity below 2SD of the control group. Furthermore, there was no significant correlation between MCF in NK cells and their cytolytic activity ( r = 0.1596 for patients with sJRA, and r = 0.1991 for patients with other JRA subtypes). Discussion In this study, profoundly depressed NK cell activity was observed in a large subgroup of patients with sJRA and in only 1 of 20 patients with the polyarticular form of the disease. The extent of NK dysfunction in this group of patients was similar to that seen in patients with MAS [ 17 ] or HLH [ 12 , 13 ]. The two study groups (sJRA versus other JRA subtypes) were well matched in terms of age, duration of the disease, and treatment regimens with the exception of a slightly higher proportion of patients with sJRA receiving steroids. Steroids have been reported to suppress the cytolytic activity of NK cells [ 22 ], and this might potentially have contributed to the observed differences in NK function. However, the logistic regression analysis did not show significant differences between groups defined on the basis of treatment regimens. In addition, several patients with sJRA who demonstrated profoundly depressed NK cell cytolytic activity were receiving only non-steroidal anti-inflammatory drugs. Owing to the limitations of the statistical power with the numbers of study subjects enrolled, it is possible that some effects of the immunosuppressive medications might have been underestimated. Nevertheless, NK dysfunction seems to be a distinguishing feature of sJRA that is intrinsic to the disease itself. Further analysis of the flow cytometry data revealed that some of the patients with sJRA had a rather selective disappearance of the circulating immunoregulatory CD56 bright subset of NK cells, a pattern previously seen in patients with MAS or HLH [ 25 ]. These cells express low levels of perforin and are, in general, poorly cytotoxic [ 24 ]. The disappearance of CD56 bright NK cells from peripheral circulation is therefore unlikely to account for the defects in cytolytic activity of NK cells. In contrast, CD56 bright NK cells might have a function in regulating the CD56 dim perforin bright cells, and in this case their disappearance might have an effect on cytolytic activity in some sJRA patients. Alternatively, the apparent absence of immunoregulatory NK cells in peripheral circulation might reflect their active recruitment to sites of inflammation. Although the observed NK dysfunction in a subgroup of patients with sJRA might not be of primary etiological significance for JRA itself, the similarities to the immunologic abnormalities seen in MAS and HLH suggest that depressed NK cell activity is likely to be relevant to the pathogenesis of MAS in sJRA. It is important to mention that low NK cell activity has been noted in many rheumatic diseases [ 28 ], most notably in systemic lupus erythematosus [ 29 ]. In our study, however, in a subgroup of patients with sJRA, the extent of NK dysfunction was profound, with an almost complete absence of cytolytic activity. This parallels the fact that, although MAS has been described in association with almost any rheumatic disease and it is not uncommon in systemic lupus erythematosus [ 30 ], it is by far most common in sJRA [ 4 - 11 ]. Other groups have noted low levels of perforin expression in cytotoxic cells from patients with sJRA in comparison with other clinical forms of the disease, suggesting that this feature might be responsible for the increased incidence of MAS [ 26 , 27 ]. In our study, when patients with sJRA were analyzed as a group, perforin levels were not significantly different from those in patients with other JRA types. However, the examination of the individual patterns of perforin staining revealed a small subgroup of JRA patients with a very low perforin content in NK cells. Most of these patients had sJRA. Furthermore, one of them had profoundly decreased proportions of perforin-positive cells in all three major cytotoxic cell populations, a pattern that has been previously reported in patients with MAS [ 17 ] and in the carriers of perforin-deficient FHLH [ 14 ]. Although no overall correlation between perforin expression and NK cell cytolytic activty was noted in our study, we still cannot exclude the possibility that, at least in some patients, reduced perforin expression might have functional significance. In other words, there might be some heterogeneity in the mechanisms underlying NK cell dysfunction in sJRA. The existence of such heterogeneity was also noted in our previous study of MAS patients [ 17 ] that included ethnically diverse Caucasian, African American, and Latin American patients. The ethnic heterogeneity of the patients with JRA included in this study might also underlie the discrepancy between our results and the study by Wulffraat and colleagues [ 26 ], which showed that patients with sJRA as a group had lower perforin expression in cytotoxic effector lymphocytes. That study included a much more ethnically homogeneous population of Dutch children. Granzyme B is another important component of the perforin-mediated cytotoxicity pathway. In our study both patient groups had granzyme B expression patterns indistinguishable from those seen in healthy controls, suggesting that the observed NK dysfunction is not likely to be related to abnormal granzyme B expression. The cytolytic activity of NK cells in our study was measured by using NK-sensitive K562 cells, which are lymphoblasts derived from a patient with chronic myelogenous leukemia. The exact receptors involved in the NK-mediated lysis of K562 cells have not yet been identified. However, the lysis of some similar cell lines has been recently shown to be mediated through the natural cytotoxicity receptors (NKp46, NKp30, and NKp44) [ 31 ]. These receptors have important biologic functions in the innate immune system, and their abnormal expression might have a function in the development of NK dysfunction in sJRA. On the basis of our data, the feature that distinguishes systemic onset JRA from other forms of JRA, and is common to the major hemophagocytic syndromes, is NK cell dysfunction. The exact mechanisms that would link deficient NK cell function and, in some cases, depressed perforin expression with the expansion of activated macrophages are not clear. One possible explanation is that decreased NK function might be responsible for a diminished ability to clear the infecting pathogen and remove the source of antigenic stimulation at early stages of infection [ 32 ]. This would lead to persistent antigen-driven T cell activation associated with an increased production of cytokines, such as IFN-γ and granulocyte/macrophage colony-stimulating factor, that stimulate macrophages. Subsequently, the sustained macrophage activation would result in tissue infiltration and in the production of high levels of TNF-α, interleukin-1, and interleukin-6, which have a major role in the various clinical symptoms and tissue damage. Several recent studies using perforin-deficient and NKcell-depleted mice indicate that NK cells and perforin-based systems are also involved in the downregulation of immune responses through a direct effect of NK cells and/or perforin-based systems on the survival of activated lymphocytes [ 33 - 36 ]. NK dysfunction might therefore lead to a failure to provide homeostatic signals for the removal of activated T cells. For instance, Su and colleagues [ 33 ] demonstrated that the infection of NK-depleted mice with murine CMV results in an exaggerated immune response associated with more persistent expansion of cytotoxic CD8 + T cells that secrete IFN-γ, an important macrophage activator. Another possible explanation is related to the recently discovered ability of NK cells to lyse autologous antigen-presenting cells such as dendritic cells, thus limiting the magnitude of an immune response [ 37 ]. Interestingly, this interaction might involve the above-mentioned natural cytotoxicity receptors [ 38 ]. Conclusions NK cell dysfunction is the feature that distinguishes systemic onset JRA from other forms of JRA, and is common to the major hemophagocytic syndromes. This suggests that impaired cytotoxic functions and/or deficiency of immunoregulatory NK cells are relevant to the development of MAS. Patients with sJRA who have these immunologic abnormalities may therefore be a high-risk group that might benefit from closer observation. Abbreviations IFN = interferon; HLH = hemophagocytic lymphohistiocytosis; JRA = juvenile rheumatoid arthritis; MAS = macrophage activation syndrome; NK = natural killer; MCF = mean channel fluorescence; sJRA = systemic juvenile rheumatoid arthritis; TCR = T cell receptor; TNF = tumor necrosis factor. Competing interests The author(s) declare that they have no competing interests. Authors' contributions JV carried out sample collection, flow cytometry, data analysis and manuscript preparation. SL carried out NK cytotoxicity assays and manuscript preparation. EHG carried out statistical analysis and manuscript preparation. TBG carried out patient referral, clinical data analysis and manuscript preparation. MHP carried out patient referral, clinical data analysis and manuscript preparation. AF carried out study design, NK studies oversight and manuscript preparation. AAG carried out study design, project oversight, patient referral, data analysis and manuscript preparation. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064882.xml

1064883

Elevated matrix metalloproteinase-9 in patients with systemic sclerosis

Matrix metalloproteinase-9 (MMP-9) has been implicated in the pathogenesis of cancer, autoimmune disease, and various pathologic conditions characterized by excessive fibrosis. In this study, we investigated the expression of MMP-9 and its clinical significance in systemic sclerosis (SSc). The patients ( n = 42) with SSc had higher concentrations of MMP-9 and of tissue inhibitor of metalloproteinase-1 (TIMP-1) and a higher ratio of MMP-9 to TIMP-1 in sera than healthy controls ( n = 32). Serum MMP-9 concentrations were significantly higher in the diffuse type ( n = 23) than the limited type of SSc ( n = 19). Serum concentrations of MMP-9 correlated well with the degree of skin involvement, as determined by the Rodnan score and with serum concentrations of transforming growth factor β. Moreover, dermal fibroblasts from patients with SSc produced more MMP-9 than those from healthy controls when they were stimulated with IL-1β, tumor necrosis factor α, or transforming growth factor β. Such an increase in MMP-9 production was partially blocked by treatment with cyclosporin A. In summary, the serum MMP-9 concentrations were elevated in SSc patients and correlated well with skin scores. The increased MMP-9 concentrations may be attributable to overproduction by dermal fibroblasts in SSc. These findings suggest that the enhanced production of MMP-9 may contribute to fibrogenic remodeling during the progression of skin sclerosis in SSc.

Introduction Systemic sclerosis (SSc) is a generalized disorder of connective tissue characterized by microvacular damage and excessive fibrosis in the skin and internal organs, including the heart, lungs, and gastrointestinal tract. One of the major hallmarks of the disease is an increased amount of collagen deposits in the affected tissue. The relative proportion of two major types of skin procollagen, types I and III, is higher in SSc lesions than in healthy controls [ 1 , 2 ]. This increase in collagen deposits may be associated with changes in the dermal microvasculature in SSc. In particular, alterations in the structure of the basement membrane, a critical component of the vessel, may lead to changes in the surrounding tissue and to subsequent development of fibrosis in SSc [ 3 ]. The finding that the synthesis of type IV collagen, a major collagen type in basement membrane, is disproportionately increased in the dermal fibroblasts and sera of patients with SSc supports this notion [ 4 , 5 ]. The enhanced expression of matrix collagen is presumably associated with abnormal immune responses to collagen in SSc [ 6 - 10 ]. For example, autoantibodies to type IV collagen have been observed in some SSc patients and may be involved in endothelial injury [ 7 , 8 ]. Immunization of mice with autologous type IV collagen leads to the activation of fibroblasts and to fibrosis [ 9 ]. Furthermore, type IV collagen activates T cells from patients with SSc [ 10 ], suggesting that the selective immunity to type IV collagen may influence the clinical expression of SSc. The excessive production of type IV collagen and subsequent autoimmune T-cell responses to type IV collagen may set off a self-perpetuating cycle in SSc through the interaction between lymphocytes and fibroblasts. The matrix metalloproteinases (MMPs) are a family of extracellular endopeptidases that selectively degrade the components of various extracellular matrixes. Of these, MMP-9 (92–96 kD gelatinase B), whose substrates include type IV collagen in basement membrane [ 11 ], has been thought to be involved in the cellular invasion of the basement membrane by cells involved in arthritis and cancer (e.g. T cells, mononuclear phagocytes, synovial fibroblasts, and metastatic tumor cells) [ 12 - 15 ]. MMP-9 has been associated with chronic inflammatory autoimmune diseases, including rheumatoid arthritis, Sjögren's syndrome, idiopathic uveitis, and systemic lupus erythematosus [ 16 - 19 ]. Moreover, the overexpression of MMP-9 has been reported in various pathologic conditions characterized by excessive fibrosis, including idiopathic pulmonary fibrosis, bronchial asthma, experimental biliary cirrhosis, and chronic pancreatitis [ 20 - 23 ], suggesting that elevated MMP-9 is closely linked to fibrogenic remodeling in target organs. In the present study, we measured the expression of MMP-9 and tissue inhibitor of metalloproteinase-1 (TIMP-1), an inhibitor of MMP-9, in the sera and culture supernatants of dermal fibroblasts from SSc patients and compared them with serum concentrations of transforming growth factor β (TGFβ) and with clinical and laboratory parameters of SSc. Materials and methods Patients This study was conducted in accordance with the principles embodied in the Declaration of Helsinki and was approved by the Ethical Committees in the Catholic Research Institutes of Medical Sciences. Before the study, informed consent was obtained from all patients and healthy controls. Forty-two patients (1 man and 41 women), all of whom fulfilled the criteria of the American Rheumatism Association for SSc [ 24 ], were studied; their mean age was 43.7 years (range 24–69 years). The mean duration of disease was 80.8 months (range 5–276 months). The comparisons were made with 32 healthy controls (all women) who had no rheumatic disease; their mean age was 44.2 years (range 21–62 years). The ages and sexes of the patient and control groups did not differ significantly. Clinical and laboratory evaluation Clinical and laboratory assessments were done at the time of sampling. The clinical variables were age, sex, disease duration, type of SSc [ 25 ], modified Rodan score [ 26 ], presence or absence of esophageal involvement on endoscopy and esophageal manometry, interstitial lung disease on chest radioagrapy and/or high-resolution computerized tomography, diffusion capacity (DLCO; diffusion of carbon monoxide in the lung) on the pulmonary function test, arthritis, sicca syndrome, and antibodies to Scl-70 or centromere using ELISA kits (MBL, Nagoya, Japan). Interstitial lung disease was defined as bibasilar interstitial fibrosis on chest radiographs, or, in patients with no abnormalities on chest radiographs, as the presence of alveolitis on high-resolution computerized tomography. ELISA for serum MMP-9, TIMP-1, and TGFβ The total MMP-9 and TIMP-1 concentrations were determined in the serum and the culture supernatant using a commercial ELISA kit (R&D Systems Inc, Minneapolis, MN, USA). In accordance with the manufacturer's recommendations, the aliquots of serum were diluted to a ratio of 1:100 in the assay buffer. The detection limits of the MMP-9 and TIMP-1 kits were 0.15 ng/ml and 0.08 ng/ml, respectively. The MMP-9 assay kit detected pro-MMP-9 and complexes of pro-MMP-9 with TIMP-1 and had no significant cross-reactivity with MMP-1, MMP-2, MMP-3, TIMP-1, or TIMP-2. Again, the TIMP-1 detection kit detected TIMP-1 either free or in complex with some MMPs and showed no cross-reactivity or interference with TIMP-2. Circulating TGFβ was measured in the same samples using ELISA, as described previously [ 27 ]. Briefly, 2 μg/ml of monoclonal antibodies to TGFβ1, β2, and β3 (R&D Systems) were added to 96-well plates (Nunc Inc, Roskilde, Denmark). They were incubated overnight at 4°C and blocked with PBS containing 1% bovine serum albumin and 0.05% Tween 20 for 2 hours at room temperature. A sample (50 μl) of each patient's serum was diluted 1:2 with PBS, acidified with 50 μl of 2.5 M acetic acid and 10 M urea for 10 minutes at room temperature and then was neutralized with 50 μl of 2.7 M NaOH and 1 M HEPES. The patient's sera and the standard recombinant TGFβ (R&D Systems) were then put into 96-well plates and incubated at room temperature for 2 hours. Biotinylated polyclonal antibodies (50 ng/ml) to human TGFβ (R&D Systems) were added and the reaction was allowed to proceed for 2 hours at room temperature. Streptavidin–alkaline phosphatase (Sigma Bioscience, St Louis, MO, USA) diluted 1:2000 with PBS was added, and the reaction was again allowed to proceed for 2 hours. p -Nitrophenylphosphate (1 mg/ml) (Sigma Bioscience) dissolved in diethanolamine (Sigma Bioscience) was added to induce a color reaction, and 1 N NaOH (Fisher Scientific, Pittsburgh, PA, USA) was used to stop the reaction. An automated microplate reader (Vmax, Molecular Devices, Palo Alto, CA, USA) was used to measure the optical density at 405 nm. Between each of these steps, the plates were washed four times with PBS containing 0.05% Tween 20. A standard curve was drawn by plotting the optical density versus the log of the recombinant TGFβ concentration. The detection limit for TGFβ was 30 pg/ml. Detection of MMP-9 activities by gel zymography MMP-9 and MMP-2 activities were also tested by gelatin zymography. A 0.5-μl sample of serum diluted in 30 μl of SDS buffer was separated in 10% SDS–PAGE gel polymerized with 1 mg/ml gelatin (Invitrogen Life Technologies, Carlsbad, CA, USA). Culture supernatants of HT1080 cell lines (malignant human fibroblasts) stimulated with 10 μg/ml of concanavalin A were used as a positive control. Gels were washed once for 3 hours in 2.5% Triton X-100 to remove the SDS and once for 30 minutes in the reaction buffer containing 50 mM Tris/HCl, 200 mM NaCl, 10 mM CaCl 2 , and 0.02% (w/v) Brij 35 (pH 7.5). The reaction buffer was changed to a fresh one, and the gels were incubated at 37°C for 24 hours. Gelatinolytic activity was visualized by staining the gels with 0.5% Coomassie brilliant blue and was quantified by densitometry. Isolation and culture of dermal fibroblasts Dermal fibroblasts were obtained from affected skin of two SSc patients and from two healthy controls, as described previously [ 28 ]. Fibroblasts were grown from explants in Dulbecco's modified Eagle's medium (DMEM) at 37°C in 5% CO 2 . The cells were then centrifuged at 500 g , resuspended in DMEM supplemented with 10% fetal calf serum (Gibco-BRL, Grand Island, NY, USA), 2 mM glutamine, penicillin (100 U/ml), and streptomycin (100 μg/ml), and plated in 25-cm 2 flasks. The cultures were kept at 37°C in 5% CO 2 and the culture medium was replaced every 3 days. When cells approached confluence, they were detached with trypsin, passed after dilution 1:3 with fresh medium, and recultured until use. Cells were housed in a 37°C humidified incubator with 5% CO 2 . Second- or third-passage cells were used for all experiments. Fibroblasts were seeded in 24-well plates at 5 × 10 4 cells per well in serum-free DMEM supplemented with insulin–transferrin–selenium A (ITSA; Gibco BRL). After the cells had been grown in selected medium alone for 12 hours, we added cytokines – IL-1β (10 ng/ml), tumor necrosis factor α (TNF-α) (10 ng/ml), and TGFβ (10 ng/ml) – to stimulate the fibroblasts. After 24 hours of incubation, cell-free media were collected and stored at -20°C until assay. All cultures were set up in triplicate or quadruplicate. Statistics Data are expressed as means ± standard error of the mean (SEM). Numerical data for groups were compared using the Mann–Whitney rank sum test, and data for categories were compared using a chi-square test. Correlation between two variables was tested using Spearman's rank correlation coefficient. P values less than 0.05 were considered statistically significant. Results Elevated serum MMP-9 and TIMP-1 concentrations in SSc patients The serum concentrations of MMP-9 were significantly higher in patients with SSc ( n = 42) than in healthy controls ( n = 32) (317.6 ± 33.5 ng/ml versus 81.2 ± 6.8 ng/ml, P < 0.001) (Fig. 1 ). The serum concentration of TIMP-1, an inhibitor of MMP-9, was also higher in SSc patients than in healthy controls (157.1 ± 13.2 ng/ml versus 77.7 ± 12.5 ng/ml, P < 0.001), but SSc patients had higher MMP-9/TIMP-1 ratios than healthy controls (233.0 ± 27.1 versus 69.5 ± 24.3, P < 0.001). There was no correlation between MMP-9 and TIMP-1 concentrations in SSc patients or in healthy controls. SSc patients with the diffuse type ( n = 23) and had higher concentrations of MMP-9 than those with the limited type ( n = 19) (364.6 ± 32.4 ng/ml versus 260.0 ± 34.6 ng/ml, P = 0.034) (Fig. 2 ). No significant differences were found between the two groups of patients with regard to age, sex, disease duration, and prednisolone usage or the kinds of immunosuppressive agents being used (e.g. D-penicillamine and cyclosphosphamide) (Table 1 ). MMP-9 activities measured by gel zymography We used gel zymography to study sera of 20 SSc patients and 10 healthy controls, all selected unsystematically, to ascertain the serum gelatinase activity of MMP-9. As can be seen in Fig. 3 , the 92 kDa band, consistent with the latent form of MMP-9, was detected in the sera of all subjects. The bands in Fig. 3 represent the latent form of MMP-9 (92 kDa, upper band) and the latent form of MMP-2 (72 kDa, lower band). The serum MMP-9 activities of SSc patients were higher than those of healthy controls. Densitometric analysis in sera of 20 SSc patients and 10 healthy controls indicated that the mean MMP-9 activity for SSc patients was 137.2 ± 21.7 densitometry units and for healthy controls, 38.5 ± 4.2 densitometry units ( P < 0.001). Furthermore, a good linear correlation was found between the densitometry units measured by zymogram and the respective concentrations of MMP-9 measured by immunoassay in the sera of SSc patients ( r = 0.875 and P < 0.001; data not shown). However, the intensity of the 86 kDa band (active MMP-9) was generally weak and was often not measurable. Correlation of serum MMP-9 concentrations with skin scores To determine the association of MMP-9 concentrations with a definite clinical manifestation of SSc, we compared the serum MMP-9 concentrations with clinical and laboratory characteristics in patients ( n = 35) with SSc. The patients with severe skin involvement ( n = 18), defined by a modified Rodnan score ≥20, had significantly higher concentrations of circulating MMP-9 than those with mild to moderate skin involvement ( n = 17) (modified Rodnan score <20) (Table 2 ). Moreover, the serum MMP-9 concentrations correlated well with the Rodnan scores ( n = 35, r = 0.425, P = 0.011) and with the serum TGFβ concentrations ( n = 41, r = 0.736, P < 0.001) (Fig. 4a,4b ). However, a correlation between MMP-9 and TGFβ was not found in the sera from healthy controls (data not shown). There were no differences in the MMP-9 concentrations with respect to age, the presence of esophageal involvement, interstitial lung disease, decrease of diffusion capacity (DLCO < 70%), digital ulcer, arthritis, sicca syndrome, and antibodies to Scl-70 or centromere-B (Table 2 ). MMP-9 production by dermal fibroblasts The finding that MMP-9 concentrations correlated with skin scores prompted us to investigate the in vitro MMP-9 production by dermal fibroblasts from SSc patients. The spontaneous MMP-9 concentrations in the culture supernatants of dermal fibroblasts were not greatly different between SSc patients and healthy controls (Fig. 5 ). However, stimulation of SSc fibroblasts with IL-1β, TNF-α, or TGFβ strongly increased MMP-9 production relative to the unstimulated concentration, by factors of 3.5, 3.2, and 2.3, respectively, whereas fibroblasts of healthy controls responded weakly to these cytokines (by factors of 1.6, 1.5, and 1.2, respectively). The increase in MMP-9 production by IL-1β and TNF-α appears to be triggered at least in part by a cyclosporin A (CsA)-sensitive pathway, since 500 ng/ml CsA limited MMP-9 production in SSc fibroblasts stimulated with IL-1β or TNF-α to 63% and 57% of original responses, respectively. Discussion We have shown that circulating MMP-9 is higher in patients with SSc than in healthy controls, particularly in the diffuse type of SSc, and correlates well with the extent of skin fibrosis. This finding supports earlier reports that overexpression of MMP-9 is closely linked with various diseases characterized by excessive fibrosis [ 20 - 23 ]. Recent studies support the evidence for a crucial role of MMP-9 in fibrotic diseases. For example, MMP-9-deficient mice exhibit significantly less pulmonary fibrosis in response to bleomycin than their with MMP-9 +/+ littermates [ 29 ]. In the hepatic fibrosis model infected by Schistosoma mansoni , the severity of fibrosis was most closely associated with the increased MMP-9 activity [ 30 ]. Similarly, in response to bleomycin, mice deficient in γ-glutamyl transpeptidase showed a reduction in pulmonary fibrosis, in part associated with lower MMP-9 activity in lung tissues [ 31 ]. What, then, are the plausible mechanisms by which MMP-9 participates in fibrotic response? One possible explanation comes from the role of MMP-9 in chronic inflammation, resulting in fibrosis. MMP-9 can trigger inflammation directly, by tissue destruction, or indirectly, by generation of an inflammatory signal or recruitment of inflammatory cells [ 32 ]. Infiltration of inflammatory cells is closely associated with an abnormal fibrotic response [ 33 ]. Moreover, in mice, targeted deletion of MMP-9 attenuated collagen accumulation, which was correlated with decreased infiltration of neutrophils and macrophages in resolving experimental myocardial infarction [ 34 ]. In SSc, several proinflammatory cytokines activate fibroblasts to increase MMP-9 production, as depicted in Fig. 5 . The overproduced MMP-9 may induce microvascular damage and leakage of substances that further augment endothelial cell damage or fibroblast activation in SSc. This damage may facilitate the movement of inflammatory cells across the basement membrane [ 11 , 35 ], ultimately leading to excessive fibrosis. In this context, type IV collagen autoimmunity, as mentioned in the Introduction, would play an additional role in fibroblast activation through the interaction between T lymphocytes and fibroblasts [ 9 , 10 ]. Such a hypothesis is supported by the findings in SSc patients that microvascular injury precedes fibrosis and that the degree of hypoxia is correlated with skin fibrosis [ 36 , 37 ]. Although the role of TGFβ in SSc remains elusive, several reports have suggested that it may be an ideal candidate as a mediator of skin fibrosis in SSc [ 38 , 39 ]. In the present study, the circulating TGFβ strongly correlated with the MMP-9 concentrations, a finding consistent with the observation that MMP-9 concentrations correlated best with skin scores of SSc. It is known that TGFβ increases the production of MMP-9 in several cell types, possibly through a process requiring protein synthesis that leads to increased statility of MMP-9 mRNA [ 40 , 41 ]. On the other hand, the increased MMP, in turn, is able to cleave latent TGFβ, leading to activation of TGFβ [ 42 ], in a process that may constitute a self-perpetuating cycle. If this is the case in SSc patients, MMP-9 may indirectly participate in the fibrotic reaction through the activation of TGFβ, a potent fibrogenic growth factor. The expression of MMP-9 has been reported to be elevated in the culture medium of alveolar macrophages from patients with idiopathic pulmonary fibrosis or bronchial asthma [ 20 , 21 , 43 ]. Serum MMP-9 and the MMP-9/TIMP-1 ratio also correlate with the severity of the airway inflammation [ 44 ]. In the present study, we did not find any association between serum MMP-9 and the presence or severity of interstitial lung disease, even in a subgroup of SSc patients with diffuse or limited disease (data not shown). The contribution of interstitial lung disease to MMP-9 elevation may be obscured by the stronger effect of skin fibrosis. The sources of MMP-9 are keratinocytes, monocytes, leukocytes, macrophages, and fibroblasts [ 12 - 15 ]. Fibroblasts from patients with early SSc exhibited higher concentrations of other MMPs (MMP-1 and MMP-3) than fibroblasts from normal individuals [ 45 ]. In addition, the finding that MMP-9 correlated best with skin scores prompted us to explore the production of MMP-9 by dermal fibroblasts in SSc patients. This study has shown that SSc fibroblasts produced more MMP-9 after stimulation with IL-1β, TNF-α, and TGFβ than fibroblasts of healthy controls. These findings show that one of the sources for MMP-9 production in SSc is dermal fibroblasts. Moreover, CsA, a calcineurin inhibitor, partially blocked IL-1β-induced or TNF-α-induced MMP-9 production by SSc fibroblasts. This finding suggests that activation of calcineurin and further downstream dephosphorylation of nuclear factor of activated T cells plays a role in the induction of MMP-9 [ 46 ] and that CsA may exert its therapeutic effect against SSc [ 47 ] by modulating MMP-9 activity. The findings we report here are in sharp contrast to those in a recently published paper by Kikuchi and colleagues [ 48 ], who found decreased concentrations of the active form of MMP-9 in the sera of patients with diffuse SSc. It seems unlikely that this discrepancy is attributable to a difference in the ELISA method (e.g. assay for total MMP-9 in this study versus active MMP-9 in the earlier report), because our patients showed a strong correlation between total MMP-9 and active MMP-9 in the additional test using the ELISA kit (R&D Systems; r = 0.745, P < 0.001; data not shown). In our study, 33 patients (79%) required corticosteroid plus penicillamine or cyclosphosphamide to control the disease, whereas in the study by Kikuchi and colleagues, only 13 (21%) of 62 patients had been treated with these drugs, suggesting that our patients were in a more active and inflammatory stage of the disease. Given that MMP-9 is abundant in highly inflammatory lesions [ 32 ], differences in the stage of disease and clinical features of the patients assessed could account for the opposite results. Accumulating evidence indicates the importance of TIMP activities in the progression of fibrosis in various pathologic conditions, including asthmatic bronchitis, cirrhosis of the liver, and SSc [ 49 - 51 ]. Moreover, both TIMP1- and TIMP-2 can promote the proliferation of fibroblasts in vitro [ 52 , 53 ]. Therefore, it remains to be defined whether the elevated expression of MMP-9 relative to that of TIMP-1 in SSc is directly involved in skin fibrosis or merely reflects biological compensation for excessive fibrosis. Studies of the effect of active MMP-9 or its inhibitor on fibrogenic remodeling in animal models of SSc are needed to clarify this issue. Conclusion Circulating MMP-9 concentrations were elevated in the patients with SSc and correlated best with the skin scores and serum TGFβ concentrations. The production of MMP-9 by dermal fibroblasts of SSc patients was strongly upregulated by stimulation with IL-1β, TNF-α, and TGFβ and such an increase was suppressed by a CsA-sensitive mechanism. Our findings suggest that MMP-9 may play a role in the progression of skin fibrosis in SSc. Abbreviations CsA = cyclosporin A; DMEM = Dulbecco's modified Eagle's medium; ELISA = enzyme-linked immunosorbent assay; IL-1β = interleukin-1β; MMP = matrix metalloproteinase; PBS = phosphate-buffered saline; SSc = systemic sclerosis; TGFβ = transforming growth factor β; TIMP = tissue inhibitor of metalloproteinase; TNF-α = tumor necrosis factor α. Competing interests This work was supported by grants from the Korea Research Foundation Grant (KRF-2002-041-E00107) and the Catholic Research Institutes of Medical Science, Republic of Korea. Authors' contributions W-UK collected the clinical data and analyzed it. S-YM and Y-JS cultured dermal fibroblasts and measured the MMP-9 concentration in the culture supernatant. M-LC performed the gel zymography. K-HH determined the concentrations of MMP-9 and TIMP-1 in the sera. M-LC drafted the manuscript. C-SC designed the study. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064883.xml

1064884

Identification of a SmD3 epitope with a single symmetrical dimethylation of an arginine residue as a specific target of a subpopulation of anti-Sm antibodies

Anti-Sm antibodies, identified in 1966 by Tan and Kunkel, are highly specific serological markers for systemic lupus erythrematosus (SLE). Anti-Sm reactivity is found in 5–30% of SLE patients, depending on the autoantibody detection system and the racial background of the SLE population. The Sm autoantigen complex comprises at least nine different polypeptides. All of these core proteins can serve as targets of the anti-Sm B-cell response, but most frequently the B and D polypeptides are involved. Because the BB'Sm proteins share cross-reactive epitopes (PPPGMRPP) with U1 specific ribonucleoproteins, which are more frequently targeted by antibodies that are present in patients with mixed connective tissue disease, the SmD polypeptides are regarded as the Sm autoantigens that are most specific to SLE. It was recently shown that the polypeptides D1, D3 and BB' contain symmetrical dimethylarginine, which is a component of a major autoepitope within the carboxyl-terminus of SmD1. In one of those studies, a synthetic dimethylated peptide of SmD1 (amino acids 95–119) exhibited significantly increased immunoreactivity as compared with unmodified SmD1 peptide. Using immobilized peptides, we confirmed that the dimethylated arginine residues play an essential role in the formation of major SmD1 and SmD3 autoepitopes. Moreover, we demonstrated that one particular peptide of SmD3 represents a more sensitive and more reliable substrate for the detection of a subclass of anti-Sm antibodies. Twenty-eight out of 176 (15.9%) SLE patients but only one out of 449 (0.2%) control individuals tested positive for the anti-SmD3 peptide (SMP) antibodies in a new ELISA system. These data indicate that anti-SMP antibodies are exclusively present in sera from SLE patients. Thus, anti-SMP detection using ELISA represents a new serological marker with which to diagnose and discriminate between systemic autoimmune disorders.

Introduction Systemic rheumatic diseases are characterized by circulating autoantibodies to defined intracellular targets (for review [ 1 ]). Historically, among the earliest of these autoantibodies to be identified was anti-Sm, which was subsequently considered a serological hallmark of systemic lupus erythematosus (SLE) [ 2 ]. Thus, anti-Sm antibodies have been included among the American College of Rheumatology (ACR) criteria for classification of this disease [ 3 ]. In addition to autoantibodies that target the Sm complex, anti-double-stranded DNA (dsDNA), anti-proliferating cell nuclear antigen, anti-U1-RNP, anti-nucleosome, anti-histone, anti-Ro/SS-A, anti-La/SS-B, anti-ribosomal RNP, and anti-phospholipid antibodies are also frequently found in sera from SLE patients [ 1 ]. Of interest, certain SLE-associated autoantibodies have been shown to be present before the clinical onset of the disease and thus have high prognostic value [ 4 ]. On average, anti-Sm reactivity is found in 5–30% of patients with SLE, although the specific frequency depends on the detection system used and the racial and genetic makeup of the SLE population [ 5 , 6 ]. The Sm autoantigen is part of the spliceosomal complex that participates in the splicing of nuclear pre-mRNA [ 7 ]. The complex itself is comprised of at least nine different core polypeptides with molecular weights that range from 9 to 29.5 kDa [ 8 ]: B (B1; 28 kDa), B' (B2; 29 kDa), N (B3; 29.5 kDa), D1 (16 kDa), D2 (16.5 kDa), D3 (18 kDa), E (12 kDa), F (11 kDa) and G (9 kDa). All of these core proteins can be targets of the anti-Sm immune response, but the most prevalent response is to the B and D polypeptides, which are therefore considered the major antigens [ 8 - 10 ]. Because SmBB' share cross-reactive epitopes with U1-specific RNPs, which are more frequently targeted by antibodies that are present in patients with mixed connective tissue disease (MCTD), SmD is regarded as the Sm autoantigen that is most specific to SLE [ 11 ]. Within the SmD family, the SmD1/D3 reactivity pattern is at least four times more common than SmD1/D2/D3 recognition, with immunoreactivity to SmD1 being the most dominant [ 11 ]. Several linear and conformational epitopes have been mapped on the SmB and SmD proteins [ 12 - 14 ]. On SmD1 and SmBB' the major reactivity was found in the carboxyl-terminal regions [ 13 - 15 ]. The epitope PPPGMRPP, which occurs three times within the carboxyl-terminus of SmBB', was shown to crossreact with other proline-rich structures of spliceosomal autoantigens, including the U1-specific RNPs, and of retroviral proteins such as HIV-1 p24gag [ 16 ]. Follow-up studies and immunization experiments revealed this motif to be consistently the earliest detectable SmBB' epitope, indicating that it acts as a potential starting point for epitope-spreading events associated with the SmBB' molecule and SmD polypeptides [ 17 , 18 ]. A recent study [ 19 ] identified five linear epitopes on SmD2 and four on SmD3 that were distributed along the full length of the molecules. All of these epitopes share basic properties and are exposed on the surface of the protein, rendering them antigenic [ 19 ]. One of the B-cell epitopes on SmD3 (epitope 4; amino acids 104–126) exhibited sequence similarity with an antigenic region from the SmD1 protein, and this may account for some cross-reactivity [ 19 ]. For diagnostic purposes, a synthetic peptide corresponding to the carboxyl-terminal domain of SmD1 was used to develop an ELISA system with diagnostic sensitivities and specificities ranging from 36% to 70% and from 91.7% to 97.2%, respectively [ 6 , 20 ]. It was recently shown that the polypeptides D1, D3 and BB' contain symmetrical dimethylarginine (sDMA), which constitutes a major autoepitope within the carboxyl-terminus of SmD1 [ 21 , 22 ]. The aims of the present study were to develop a peptide-based ELISA system for the detection of anti-Sm antibodies and to evaluate the diagnostic properties of the peptide assay. Moreover, epitope-mapping experiments were performed to shed more light on the controversial findings of the importance of sDMA residues within the SmD1 and SmD3 sequences and their relationship to SLE. Methods Patients and sera evaluated Sera ( n = 628) were collected from patients suffering from SLE ( n = 176), rheumatoid arthritis ( n = 86), Sjögren's syndrome ( n = 24), MCTD ( n = 26), systemic sclerosis ( n = 26) and polymyositis/dermatomyositis ( n = 13), as well as from patients with overlap syndromes ( n = 8). All patients were classified according to the ACR criteria for each disease [ 3 , 23 - 27 ]. Clinical, serological and demographic data were available from 101 SLE patients (SLE panel 1). This cohort contained 34 samples from white patients, 51 from black patients, five from hispanic patients, one from an east Indian patient, and one from an oriental patient. The racial background of four patients was not known. Of these patients, 13 were male and 86 female (in two the sex was unknown), and the mean age was 40 years (range 16–80 years). The sera were kindly provided by Drs R Mierau and E Genth (Rheumaklinik Aachen, Aachen, Germany), Prof. Dr MJ Fritzler (University of Calgary, Calgary, Canada) and by Labor Limbach (Heidelberg, Germany). To assess further the assay specificity, we analyzed a group of sera from patients with infectious diseases ( n = 77), including hepatitis C virus ( n = 30), cytomegalovirus ( n = 22) and Epstein–Barr virus (EBV; n = 25) infections, as well as from 192 healthy blood donors. All sera were stored at -80°C until use. For the epitope-mapping study, a panel of five sera (all from SLE patients) containing anti-Sm antibodies that were available in greater quantities preselected by ELISA (Varelisa ® Sm; Pharmacia Diagnostics, Freiburg, Germany) was used. Autoimmune sera with antibody specificities other than anti-Sm were selected as negative controls, including anti-RNP, anti-SS-A (Ro), anti-SS-B (La), anti-PM/Scl, anti-centromere protein (CENP) and anti-Scl-70. Serological characterization of randomly selected systemic lupus erythematosus sera The sera from SLE patients represented in panel 1 with clinical, serological and demographic data, as well as the autoimmune controls, were tested for autoantibodies to histones (cutoff 30 U/ml), dsDNA (cutoff 55 U/ml) and the Sm complex (cutoff 15 U/ml, or ratio 1) using quantitative ELISA tests (Varelisa ® ; Pharmacia Diagnostics; catalog nos 14196, 16296 and 16496). SLE sera and samples that exhibited unexpected results were also measured using the semiquantitative Split anti-nuclear antibody (ANA) profile (Pharmacia Diagnostics; cutoff ratio 1). The latter assay contains the autoantigens U1-68 kDa, U1-A, U1-C, SmBB', SmD, Ro-52, Ro-60 and La. All ELISAs were performed in accordance with the manufacturer's instructions. Reference serum panels The US Centers for Disease Control and Prevention (CDC) ANA serum panel [ 28 ] and the Association of Medical Laboratory Immunologists (AMLI) samples [ 29 ] were used to characterize the new Sm peptide-based immunoassay. Epitope-mapping with immobilized oligopeptides The published sequences of SmD1 (P13641) and SmD3 (P43331) were used to synthesize overlapping 15 mer peptides with a pipetting robot (ASP222; Abimed, Langenfield, Germany), in accordance with the protocol described by Gausepohl and Behn [ 30 - 33 ]. The carboxyl-terminal extensions of both polypeptides were synthesized with an offset of two amino acids (13 amino acid overlap). Each arginine-containing peptide was synthesized as three variants, one with natural arginine, one with sDMA and one with asymmetrical dimethylarginine at the respective positions. In addition, a highly reactive SmD3 peptide was synthesized with certain combinations of natural arginine and sDMA. Following completion of the peptide synthesis, nonspecific binding sites were blocked by overnight incubation of the membranes in blocking buffer (2% milk in Tris-buffered saline) at room temperature. After one washing step for 5 min, the membranes were incubated with serum samples diluted 1:100 in blocking buffer for 2 hours at room temperature. Unbound antibodies were removed by three washing steps in Tris-buffered saline–0.2% Tween. The membranes were incubated for 75 min at room temperature in a peroxidase-conjugated goat-anti-human IgG antibody (Dianova, Hamburg, Germany) that was diluted 1:5000 in blocking buffer. Unbound secondary antibodies were removed by three changes of Tris-buffered saline. Finally, bound antibodies were visualized using the ECL detection system (Amersham Bioscience, Freiburg, Germany). Synthesis of the SmD3 peptide The candidate SmD3 peptide (SMP) identified in the epitope-mapping study ( 108 AARG sDMA GRGMGRGNIF 122 ) was synthesized with an additional cysteine residue at the carboxyl-terminus, in accordance with Fmoc-chemistry at the Peptide Specialty Laboratories GmbH (Heidelberg, Germany). sDMA (Ref. B-3345.0001) was purchased from Bachem AG (Bubendorf, Switzerland) and used for synthesis. Crude fraction was purified using high-performance liquid chromatography. Quality and purity of the peptide was assessed by mass spectrometry and analytical high-performance liquid chromatography. The molecular mass was found to be 1708.1 Da (average; monoisotopic mass 1706.9), and a purity in excess of 95% was identified. SmD3 peptide ELISA Microtiterplates (Maxisorb, Nunc, Denmark) were coated with the uncoupled 16 mer peptide at a concentration of 2.5 μg/ml in phosphate-buffered saline (pH = 7.6). After an incubation time of 15 hours at 15°C, the plates were blocked with 1% bovine serum albumin in phosphate-buffered saline for 30 min at room temperature. The assay was performed in accordance with the general protocol for the Varelisa ® system (Pharmacia Diagnostics). In brief, following a prewashing (300 μl/well) step, the serum samples were diluted 1:101 in sample buffer (phosphate-buffered saline, containing bovine serum albumin and Tween), added to the wells and then incubated for 30 min (100 μl/well). After three washing steps (300 μl/well), horseradish peroxidase conjugated anti-human IgG was added and incubated for 30 min (100 μl/well). Visualization was done by incubation in 3,3',5,5'-tetra-methyl benzidine substrate for 10 min (100 μl/well), and the reaction was terminated by adding 50 μl stop solution (0.5 mol/l H 2 SO 4 ) to each well. All steps were carried out at room temperature. A standard curve was developed using a highly reactive index serum that bound the SmD3 peptide and had a defined reactivity of 40,000 U/ml. The curve was plotted at six standard points (0, 3, 7, 16, 40 and 100 U/ml). Each serum sample was tested in duplicate and serum samples that had reactivity above the assay range were serially diluted to 1:500, 1:2500, 1:12,500 and 1:62,500. To define further the assay characteristics, 192 normal human sera were assayed in accordance with the instructions for use. Blood donors exhibited a reactivity range of 0.4–11.5 U/ml, a mean value of 2.2 U/ml and a standard deviation of 1.2 U/ml. The cutoff was set at 13 U/ml following receiver operating characteristic (ROC) analysis. Positive predictive values (PPVs) and negative predictive values (NPVs) were calculated at different cutoff values using Analyse-it software (Version 1.62; Analyse-it Software Ltd, Leeds, UK). Precision and reproducibility Measurements of imprecision (interassay and intra-assay variability) were taken over four and six replicates, respectively. To assess the precision of the anti-SMP ELISA, suitable anti-Sm sera – a low value sample (L), a medium value sample (M) and a high value sample (H) – were assayed in five independent tests on one day (interassay) or in a single run (intra-assay). For within-run precision, the L, M and H samples were measured in six replicates on one solid phase. The precision data were calculated using analysis of variance. Linearity Linearity was analyzed by testing dilutions (1:1, 2:3, 1:2, 1:4, 1:8, 1:16, 1:32) of the highest standard point (S6) and of the high value sample from the precision analysis (H). For each dilution point, a ratio of the measured reactivity to the expected value was calculated, and 1 was subtracted from this quotient. Results Epitope fine mapping of the carboxyl-terminal extensions of SmD1 and SmD3 To evaluate the effect of arginine dimethylation on the antigenicity of SmD1 and SmD3 and to localize relevant epitopes on both polypeptides, a panel of anti-Sm sera was tested for reactivity with peptide arrays (15 mer, two offset) covering the carboxyl-terminal region of SmD1 (P13641) and SmD3 (P43331). The results show that dimethylation of arginine residues significantly affects the binding of anti-Sm antibodies to carboxyl-terminal SmD1 and SmD3 peptides (Fig. 1 ). All anti-Sm sera exhibited increased binding to SmD1 peptides containing sDMA as compared with those containing unmethylated arginine (Fig. 1a ). In particular, peptides that exclusively consist of glycine and sDMA repeats exhibited strong reactivity with the antibodies (peptide nos 9, 10 and 11). Nevertheless, SmD1 peptides containing sDMA represent a rather unspecific substrate for anti-Sm antibodies because they were also bound by sera that contained anti-centromere antibodies. Interestingly, those anti-centromere antibodies also bound to peptides containing the asymmetrical form of DMA but to a lesser extent. Binding experiments with peptides derived from SmD3 yielded similar results. Only SmD3 peptides containing sDMA reacted with anti-Sm antibodies, confirming the importance of the symmetrical dimethylation of arginine residues (Fig. 1b ). In contrast to SmD1, none of the control sera reacted with SmD3 derived peptides, which reflects high binding specificity. One particular peptide (no. 77; 108 AA sDMA G sDMA G sDMA GMG sDMA GNIF 122 ) was strongly recognized by three out of five anti-Sm sera. Using a mutational analysis in which arginine residues of peptide no. 77 ( 108 AARGRGRGMGRGNIF 122 ) were successively replaced by sDMA, we were able to show that a particular peptide with a single dimethylated arginine residue at position 112 exhibited immunoreactivity with all of the five anti-Sm sera but not with the control sera (Fig. 1c ). Thus, by introducing only one sDMA and at a defined position (amino acid 112) in SmD3, the sensitivity of binding to this peptide ( 108 AARG sDMA GRGMGRGNIF 122 ; SMP) was remarkably increased without loss in specificity. This candidate peptide was subsequently synthesized as a soluble antigen and used as a substrate in ELISA. Immunoserological characterization of the systemic lupus erythematosus patient cohort In order to characterize the SLE cohort with regard to autoantibody profiles, 101 SLE patient sera were tested for U1-68 kD, U1-A, U1-C, SmBB', SmD, Ro-52/SS-A, Ro-60/SS-A, La/SS-B, histone, dsDNA and β 2 -glycoprotein I reactivity. The prevalences of the different autoantibodies were as follows: 15.8% for U1-68, 24.8% for U1-A, 25.7% for U1-C, 21.8% for SmBB', 15.8% for SmD, 21.8% for Ro-52, 47.5% for Ro-60, 21.8% for La, 37.6% for histone, 51% for dsDNA and 17% for β 2 -glycoprotein I. The prevalences were therefore consistent with those in previous studies [ 1 , 6 ]. Thus, with regard to their autoantibody profiles, our SLE cohort appears to be representative of SLE patients in general. Anti-SmD3 peptide ELISA A 15 amino acid soluble peptide exhibited highest sensitivity and specificity in the SPOT assay ( 108 AARG sDMA GRGMGRGNIF 122 ) was, for coupling purposes, synthesized with an additional cysteine at the carboxyl-terminus. Nevertheless, the uncoupled 16 mer peptide was subsequently used to develop an ELISA system based on the general protocol of the Varelisa ® tests (Pharmacia Diagnostics). Assay performance characteristics To evaluate the performance of the assay, the precision, reproducibility and linearity were analyzed. The intra-assay and interassay variabilities (coefficient of variation in %) for three samples ranged from 1.82% to 6.52% and from 2.27% to 7.42%, respectively. Even after five serial dilutions, two samples exhibited a linear range of reactivity (<20% deviation). The cutoff was defined by ROC analysis, performed with SLE and control sera. The assay performance characteristics of the new anti-SMP test are summarized in Fig. 2 , including intra-assay and interassay variability (Fig. 2a ), linearity (Fig. 2b ), and ROC analysis, PPV, NPV and efficiency (Fig. 2c ). To determine the cutoff, the focus was set to yield a high specificity and a technical cutoff was defined at 13 U/ml. Sera from 176 SLE patients, 181 other autoimmune patients, 77 patients with infectious diseases, and from 192 normal donors were analyzed in the new ELISA system (Table 1 ). Twenty-eight SLE patients (15.9%) tested positive for anti-SMP antibodies, exhibiting a significantly increased reactivity of up to 1190 U/ml with a mean value of 43.0 U/ml (standard deviation 160.2 U/ml). Sera from patients with related disorders had significantly reduced reactivity (mean 3.36 U/ml). Only one patient in the rheumatoid arthritis group had a positive test result (24.6 U/ml). None of the remaining control individuals, including patients suffering from systemic sclerosis ( n = 26), polymyositis/dermatomyositis ( n = 13), MCTD ( n = 26), Sjögren's syndrome ( n = 24), or infectious diseases ( n = 77), exhibited reactivity to the SmD3 peptide. The serum samples from patients with infectious diseases demonstrated reduced reactivity (mean 0.67 U/ml; top value 3.3 U/ml), even when compared with sera from healthy donors (mean 2.21 U/ml; top value 11.5 U/ml). The highest assay value in the infectious disease sera was found in patients with EBV infection (3.3 U/ml). In summary, 28 samples in the SLE cohort ( n = 176) and only one serum sample from the control group ( n = 449; 0.2%) tested positive. This resulted in a diagnostic specificity of 99.8% and a sensitivity of 15.9%. PPV, NPV and diagnostic efficiency were calculated at 96.6%, 75.3% and 76.3%, respectively (Fig. 2c ). These data indicate that, within the assay parameters used here, anti-SMP antibodies appear to be exclusively present in sera from SLE patients. Apart from anti-SMP reactivity, it is interesting to note that the positive rheumatoid arthritis serum contained high titres of antibodies to U1-RNPs 68 kDa (ratio 4.5), U1-C (ratio 9.4) and histone (133.8 U/ml). Anti-SmBB'and anti-SmD titres, as determined by ELISA, were elevated when compared with controls, but they were still below the cutoff values. No reactivity could be found to U1-A, Ro-52, Ro-60, La, dsDNA, or β 2 -glycoprotein. Correlation with other autoantibodies A statistical evaluation was performed using sera from a cohort of 101 patients with clinically defined SLE to evaluate correlations between anti-SmD3 peptide antibodies and other autoantibodies. Significant correlations were found with dsDNA ( P = 0.0058, χ 2 = 7.6), U1-68 ( P < 0.0001, χ 2 = 15.42), U1-A ( P < 0.0001, χ 2 = 25.49), U1-C ( P < 0.0001, χ 2 = 18.05), SmBB' ( P < 0.0001, χ 2 = 24.04) and SmD ( P < 0.0001, χ 2 = 38.76), but not to histone ( P = 0.0259, χ 2 = 4.96), La ( P = 0.8747, χ 2 = 0.02), Ro-52 ( P = 0.4034, χ 2 = 0.7), Ro-60 ( P = 0.0143, χ 2 = 6.0) and β 2 -glycoprotein antibodies ( P = 0.3819, χ 2 = 0.74; Table 2 ). When reactivity with components of the Sm complex was evaluated, five samples of the clinically defined SLE patients ( n = 101) reacted with the purified SmD antigen, but not with SMP. The remaining 11 SmD positive sera (68.8%) also tested positive in the new anti-SMP ELISA. Interestingly, among the patients studied we found four (nos 89, 92, 20627 and 9811) who fulfilled SLE criteria and were all anti-SmD negative, but exhibited anti-SMP reactivities of 15.4, 21.3, 41.3 and 13.9 units, respectively. Correlation with racial and clinical parameters When correlating autoantibody specificities with race, there was a statistically significant association of autoantibodies to U1-68 kDa ( P = 0.002), U1-A ( P < 0.0001), U1-C ( P = 0.0002), SmBB' ( P = 0.0004), dsDNA ( P = 0.0128) and SmD ( P = 0.0002) with black race among SLE patients. There was no statistically significant association of other autoantibody specificities, including SmD3 peptide ( P = 0.0253), Ro-52 ( P = 0.8023), Ro-60 ( P = 0.0399), La ( P = 0.7137) and histones ( P = 0.9831), with the race of the patients under investigation (data not shown). In addition, there was no significant correlation of SMP reactivity with renal ( P = 0.2810) or central nerve system involvement ( P = 0.5066). Reference panels The CDC and the AMLI reference panels for ANA were evaluated using the new SMP ELISA. Increased titres were found in ANA 1 (10.3 U/ml) and ANA 5 (910 U/ml) from the CDC panel and in samples I (1420 U/ml) and J (13.9 U/ml) from the AMLI panel [ 28 , 29 ] (Table 3 ). Discussion In the present study we analyzed the anti-Sm immune response directed toward the Sm antigens D1 and D3, which are considered SLE-specific autoantigens [ 11 ]. Using immobilized peptides prepared by the SPOT technology, it was shown that symmetric dimethylation of arginine residues plays an important role in the B-cell epitope recognition of both autoantigens. This observation is in accordance with the findings of Brahms and coworkers [ 21 ] but it is not in keeping with those of Riemekasten and colleagues [ 20 ]. In addition, we found the specificity of antibody binding to SmD3 peptides to be higher than that to SmD1 peptides, both of which were prepared using the SPOT method. McClain and colleagues [ 19 ] described four antigenic regions on SmD3, of which antigenic region 4 encompasses amino acids 104–126. In their study, peptides synthesized on pins were subjected to analysis without using the modified form of arginine. In our study, we found reactivity within this region only when natural arginine was replaced by sDMA. These apparently contradictory results might be explained by the use of different sera or methology, and/or by the different peptide length used in the two studies. Three out of five of our sera specifically recognized the peptide 108 AA sDMA G sDMA G sDMA GMG sDMA GNIF 122 . Interestingly, the dimethylation of only one arginine at a defined position (amino acid 112) was able to increase further the sensitivity of this particular peptide without loss of specificity. Although it has been shown that all arginine residues within the peptide 108 AA sDMA G sDMA G sDMA GMG sDMA GNIF 122 become dimethylated in vivo , it is unclear whether the identified peptide with single dimethylation occurs in vivo and thus serves as the triggering epitope, or rather whether it represents an artificial structure that is more suitable for in vitro assays [ 21 ]. Fewer than 20 proteins have been identified during the past 40 years as containing dimethylated arginines [ 34 ]. The two major catalyzing enzymes of this reaction are the type I and type II protein arginine methyltransferases, which preferentially methylate arginines located in RG clusters. Recently, however, using arginine methyl-specific antibodies and HeLa cell extracts, it was shown that more than 200 proteins contained symmetrically dimethylated arginines; among these were a remarkable number of known autoantigens [ 34 ]. Further studies are necessary to screen known autoantigens containing dimethylated arginine residues for epitopes. Based on data from epitope analysis, we used a candidate peptide ( 108 AARG sDMA GRGMGRGNIF 122 ) to develop an ELISA system. The new anti-Sm assay (anti-SMP) had a sensitivity of 15.9% and a specificity of 99.8% for SLE, resulting in a high PPV (96.6%) and NPV (75.3%), and a remarkable diagnostic efficiency of 76.3%. Therefore, this test appears to offer a new approach to serological evaluation and diagnosis of SLE. Because no international 'gold standard' is available for detection of anti-Sm antibodies, we compared results with the new peptide-based ELISA with the results of an anti-Sm ELISA using purified Sm antigen (Varelisa ® Sm; Pharmacia Diagnostics). Using this approach, we found comparable sensitivities but significant differences in the specificity. The specificity of the conventional ELISA (88%) was significantly lower than the specificity of the new peptide-based assay (99.8%; data not shown). Further studies are in progress to compare the assay performance of the anti-SMP assay with that of other commercially available anti-Sm immunoassays. For epitope-mapping, anti-Sm sera from SLE patients were preselected, based on ELISA results (Varelisa ® Sm). That this selection method might have affected the results of epitope-mapping cannot be excluded. Nevertheless, the high sensitivity and specificity of the SmD3 peptide with a single dimethylated arginine could be confirmed using the soluble peptide in ELISA. Evaluation of the biochemical properties of the identified Sm epitopes suggested that the isoelectric point (pI) of the peptide can be regarded as a predictor of antigenicity on the Sm complex. On U1-RNP-A, SmB' and SmD1, the average pI of antigenic regions was 10.4 (nonantigenic 6.0) and on SmD2 and SmD3 the pIs were 9.0 or higher [ 19 ]. These findings fit well with the observed pI (>12.88) of the SMP. Further investigation is warranted to determine whether the basic character of the epitope simply increases the probability of surface exposure of these regions and thus accessibility for immune recognition. Epstein–Barr virus, Epstein–Barr virus nuclear antigen 1 and anti-SmD antibodies Epitope-mapping studies on SmD1 have identified an epitope motif (amino acids 95–119) that crossreacts with a homologous sequence (amino acids 35-58) of the EBV nuclear antigen 1 [ 35 , 36 ]. A more recent study showed that this epitope also crossreacts with a homologous region of SmD3 containing glycine and arginine repeats (RGRGRGMGR) [ 19 ]. It is also evident that GPRR (amino acids 114–119 on SmD1) represents a common crossreactive autoepitope motif, which is present not only on EBV nuclear antigen 1, but also on a variety of autoantigens including CENP-A, CENP-B, CENP-C, SmBB', SmD1 and Ro-52, to name but a few [ 30 ]. Thus, patients suffering from infectious mononucleosis or SLE-related disorders may have a positive carboxyl-terminal SmD1 or SmD3 ELISA that might be regarded as a false-positive result. Of interest, several studies have suggested an influence of EBV on the development of SLE-like conditions [ 37 , 38 ]. Among the 25 EBV disease controls we evaluated, we found no false-positive samples, confirming the suggested high specificity of the anti-SMP assay. Therefore, we consider the use of EBV-positive sera with high titres of EBV-associated antibodies to be important reagents for developing highly specific and reliable anti-SmD immunoassays. Unfortunately, other investigators did not include an EBV patient group in the evaluation of anti-Sm antibody assays [ 20 ]. Correlations with other autoantibodies Coincident reactivity with dsDNA and Sm antigens has been reported by several authors [ 39 - 41 ]. Although in those studies full-length SmD was used, in our investigation there was also a correlation of anti-dsDNA and anti-SMP reactivity ( P = 0.0058, χ 2 = 7.6). Apart from DNA we found also a positive correlation of anti-SMP antibodies with U1-68 ( P < 0.0001, χ 2 = 15.42), U1-A ( P < 0.0001, χ 2 = 25.49), U1-C ( P < 0.0001, χ 2 = 18.05), SmBB' ( P < 0.0001, χ 2 = 24.04) and SmD ( P < 0.0001, χ 2 = 38.76), but not to histone ( P = 0.0259, χ 2 = 4.96), La ( P = 0.8747, χ 2 = 0.02), Ro-52 ( P = 0.4034, χ 2 = 0.7), Ro-60 ( P = 0.0143, χ 2 = 6.0) and β 2 -glycoprotein ( P = 0.3819, χ 2 = 0.74) antibodies. Whether the observed associations are caused by cross-reactivity or by different autoantibody species that often occur simultaneously remains unclear. Preliminary results of inhibition experiments have shown no inhibiting effect of the SmD3 peptide on the binding of anti-Sm antibodies to the native Sm antigen in ELISA (Varelisa ® Sm; data not shown). The absence of inhibition can be explained by the variety of different anti-Sm antibody subpopulations and by the variety of corresponding epitopes. Recently, two polyclonal antibodies (SYM10, SYM11) were generated that specifically bind to the symmetrical form of dimethylarginine and react with a variety of other known autoantigens [ 42 ]. Those antibodies may shed more light on the correlation of anti-SMP antibodies with other known autoantibody specificities. Reference sera The reference sera for ANA obtained from the CDC and the AMLI were tested using the anti-SMP assay [ 28 , 29 ]. Although sample J (AMLI) was defined as a RNP-positive serum, we found reactivity to the SMP (13.9 U/ml). None of five investigators found precipitating anti-Sm antibodies, and only two out of 21 reported Sm reactivity in their enzyme immunoassay in this serum, which was derived from a patient with SLE. In the immunoblot of the AMLI study, both serum I and serum J exhibited reactivity to the SmD proteins. Surprisingly, the reactivity to SmD3 was significantly greater in sample J than in serum I. Thus, the anti-SMP antibody test had a higher sensitivity and clinical accuracy than the anti-Sm tests used in most of the participating laboratories [ 29 ]. The apparent disparity between the results of the present study and those of Riemekasten [ 20 ] and Brahms [ 21 ] and their groups might be explained by the existence of different epitopes on the carboxyl-terminal extensions of SmD1 and SmD3. The peptide (amino acids 83–119) [ 20 ] may form a conformational epitope, whereas the shorter peptides used in the second study contain primarily linear, sDMA-dependent binding sites [ 21 ]. Furthermore, the reduced reactivity against the full-length SmD1 [ 20 ], as compared with SmD1 83–119 peptide, suggests that this epitope represents a cryptic structure. This observation raises the issue of which epitopes are 'seen' in vivo and which ones play a central role in the pathogenesis of SLE. In a recent study [ 43 ] it was observed that the injection of SmD1 83–119 fused to a carrier protein is able to accelerate the pathogenic process in SLE-prone mice. Summary In the present study we showed that dimethylation of arginine residues of the major SmD1 and SmD3 autoepitopes results in remarkably increased binding by SLE autoantibodies. Moreover, it could be shown that one particular SmD3 peptide represents a highly specific substrate for detecting a subclass of anti-Sm antibodies by ELISA. At a defined cutoff value of 13 U/ml, the sensitivity was 15.9% and the specificity was 99.8%, yielding a diagnostic efficiency of 76.3%. Conclusion Based on the findings of the present study, we conclude that anti-SMP antibodies are exclusively present in sera from SLE patients and that the new anti-SMP ELISA test appears to offer a new serological reagent that will improve our ability to diagnosis SLE and to discriminate SLE from other autoimmune and infectious diseases. Abbreviations ACR = American College of Rheumatology; AMLI = Association of Medical Laboratory Immunologists; ANA = anti-nuclear antibody; CDC = Centers for Disease Control and Prevention; CENP = centromere protein; dsDNA = double-stranded DNA; EBV = Epstein–Barr virus; ELISA = enzyme-linked immunosorbent assay; MCTD = mixed connective tissue disease; NPV = negative predictive value; pI = isoelectric point; PPV = positive predictive value; ROC = receiver operating characteristic; sDMA = symmetrical dimethylarginine; SLE = systemic lupus erythematosus; SMP = SmD3 peptide. Competing interests MM receives royalties for the commercial ELISA system from Pharmacia Diagnostics (Freiburg, Germany). Authors' contributions MM planned and initiated the present study. He carried out the epitope-mapping of SmD1 and SmD3, and developed and evaluated the ELISA system. Based on the results he filed a draft verison of the manuscript. MB advised MM regarding the planning of the epitope -mapping experiments and contributed to the preparation of the manuscript. MF delivered clinically defined sera, advised MM on evaluating the clincal part of the study, and contributed to the preparation of the manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064884.xml

1064885

Nifedipine protects against overproduction of superoxide anion by monocytes from patients with systemic sclerosis

We have reported previously that dihydropyridine-type calcium-channel antagonists (DTCCA) such as nifedipine decrease plasma markers of oxidative stress damage in systemic sclerosis (SSc). To clarify the cellular basis of these beneficial effects, we investigated the effects in vivo and in vitro of nifedipine on superoxide anion (O 2 •- ) production by peripheral blood monocytes. We compared 10 healthy controls with 12 patients with SSc, first after interruption of treatment with DTCCA and second after 2 weeks of treatment with nifedipine (60 mg/day). O 2 •- production by monocytes stimulated with phorbol myristate acetate (PMA) was quantified by the cytochrome c reduction method. We also investigated the effects in vitro of DTCCA on O 2 •- production and protein phosphorylation in healthy monocytes and on protein kinase C (PKC) activity using recombinant PKC. After DTCCA had been washed out, monocytes from patients with SSc produced more O 2 •- than those from controls. Nifedipine treatment considerably decreased O 2 •- production by PMA-stimulated monocytes. Treatment of healthy monocytes with nifedipine in vitro inhibited PMA-induced O 2 •- production and protein phosphorylation in a dose-dependent manner. Finally, nifedipine strongly inhibited the activity of recombinant PKC in vitro . Thus, the oxidative stress damage observed in SSc is consistent with O 2 •- overproduction by primed monocytes. This was decreased by nifedipine treatment both in vivo and in vitro . This beneficial property of nifedipine seems to be mediated by its cellular action and by the inhibition of PKC activity. This supports the hypothesis that this drug could be useful for the treatment of diseases associated with oxidative stress.

Introduction Systemic sclerosis (SSc) is a connective tissue disease, characterised by vascular involvement with generalised microangiopathy culminating in systemic fibrosis. Several lines of evidence suggest that the generation of oxygen free radicals is of major importance in the pathogenesis of SSc [ 1 ]. Frequent episodes of reperfusion injury generate oxygen free radicals locally, but increased lipid peroxidation is not related to Raynaud's phenomenon only [ 2 ] and the inflammatory process might also generate oxidative stress [ 1 ]. Histological studies of cutaneous SSc lesions have revealed early mononuclear cell infiltration of perivascular spaces around small vessels [ 3 , 4 ], and mononuclear cells might affect vascular and tissue lesions by producing various molecules [ 5 ]. Monocytes seem to have a key function in several disorders (for example atherosclerosis) associated with free radical generation [ 6 ]. Monocytes from patients with SSc produce greater amounts of superoxide anion (O 2 •- ) than those from healthy subjects and patients with primary Raynaud's phenomenon [ 7 ]. Dihydropyridine-type calcium-channel antagonists (DTCCA) are an essential treatment in SSc because they decrease vasospastic propensity. These drugs are suspected to have anti-oxidant properties in other diseases [ 8 , 9 ]. We reported previously that nifedipine and nicardipine decrease circulating markers of oxidative stress damage in patients with SSc [ 10 , 11 ]. The aim of the present study was to investigate O 2 •- production ex vivo by monocytes from patients with SSc after a wash-out and during nifedipine treatment. We also investigated the effects of nifedipine in vitro on O 2 •- release from human monocytes, on protein phosphorylation with phorbol myristate acetate (PMA) as stimulator and on protein kinase C (PKC) activity. Methods Reagents, except when specified, were provided by Sigma (St Louis, MO, USA). Patients Twelve non-smoking patients with SSc were included (three men and nine women); their mean age was 56 (±10 SD) years and the mean disease duration was 8 ± 5 years. Each of the patients had been hospitalised for systematic follow-up of the disease. SSc was classified as limited cutaneous or diffuse cutaneous according to the criteria of LeRoy and colleagues [ 12 ]. The clinical features of their disease were assessed as recommended [ 13 ]; results are detailed in Table 1 . The initial evaluation included biological tests performed on the morning of admission, after 1 hour of rest at room temperature; patients stopped taking calcium-channel blockers 3 days before admission [ 10 ]. This wash-out period was long enough for DTCCAs to have ceased to have an effect because the half-lives of nifedipine and nicardipine lie between 6 and 11 h and these drugs are converted into inactive metabolites. Patients were also evaluated 2 weeks later, when they were receiving stable treatment (including 20 mg of nifedipine three times a day). As before, this was done in the morning, after 1 h of rest at room temperature. All patients gave written informed consent. The control subjects were 10 healthy non-smokers from the laboratory staff (8 women and 2 men, mean age 48 ± 9 years). Monocyte preparation Blood samples (20 ml) were collected on EDTA. Whole blood was diluted 1:2 in RPMI 1640 solution (Eurobio, Les Ulis, France), layered on 15 ml of Ficoll-Hypaque (relative density 1.077; Eurobio) and centrifuged at 150 g for 25 min at room temperature to separate mononuclear cells from red blood cells and polymorphonuclear neutrophils. Mononuclear cells collected at the interface were washed twice in RPMI 1640 solution and separated into lymphocytes and monocytes by gradient centrifugation with 47% Percoll (Pharmacia, Uppsala, Sweden) at 150 g for 25 min. The purity of the monocyte population (more than 90%) was assessed by May–Grunwald–Giemsa staining. Cells were diluted in RPMI and used immediately. O 2 •- (respiratory burst) assay O 2 •- was quantified by the cytochrome c reduction assay at 550 nm [ 14 ] with a Uvikon ® spectrophotometer equipped with a thermostat-controlled cuvette holder and a magnetic stirrer. In brief, 840 µl of monocytes (2 × 10 6 cells) were incubated at 37°C with 80 µM cytochome c in 1 ml of PBS with stirring before stimulation with 10 µl of PMA (100 nM). Absorbance was read against a reference blank curve containing PBS and cytochrome c at 5, 10 and 15 min. Results are expressed in nmoles of O 2 •- per 10 6 cells using 21.2 mM/cm as the extinction coefficient. For in vitro evaluation of the effects of DTCCAs, control monocytes were preincubated for 30 min with the desired concentration of drugs or vehicle (dimethyl sulphoxide, 100 µM) before stimulation with PMA. The respiratory burst of control and nifedipine-treated monocytes induced by formyl-Met-Leu-Phe (fMLP) was measured by a sensitive fluorimetric assay for hydrogen peroxide under standard conditions [ 15 ] using the Amplex Red fluorescent probe (Molecular Probes, Eugene, OR, USA). The increase in the fluorescent signal (excitation and emission wavelengths of 530 and 590 nm respectively) was monitored continuously after the stimulation of monocytes with 1 µM fMLP. Results are expressed as the percentage of control values, representing the total hydrogen peroxide production. The scavenging effect of nifedipine was studied in an assay system containing nifedipine at the desired concentration with 0.15 mM hypoxanthine, 1.5 mM EDTA and 0.025 mM cytochrome c . The reaction was started by adding 0.1 U/ml xanthine oxidase. Absorbance was read at 550 nm. Protein phosphorylation in monocytes and in a cell-free system Phosphorylated proteins were detected by using the fluorescent Pro-Q-Diamond dye (Molecular Probes), which can directly detect phosphate groups attached to tyrosine, serine or threonine residues in gels. In brief, monocytes from healthy donors were treated with or without nifedipine and stimulated for 10 min with 100 nM PMA (which induces the activation of PKC). Reactions were stopped with ice-cold buffer and cell lysates were prepared by incubating monocytes for 30 min in 0.5 ml of ice-cold extraction buffer (350 nM NaCl, 20 mM Hepes-KOH, pH 7.9, 1 mM EDTA, 0.1 mM EGTA, 20% glycerol, 1% Nonidet P40, 1 mM dithiothreitol, 0.5 mM phenylmethylsulphonyl fluoride, 20 mM ß-glycerophosphate, 0.1 mM Na 3 VO 4 and 1 µg/ml each of aprotinin, pepstatin and leupeptin). Cells were then disrupted by sonication (three times 5 s) and proteins were delipidated and desalted before being subjected to SDS gel electrophoresis. Protein phosphorylation was also studied after treatment of a cytosolic fraction of resting monocytes with PMA for 10 min. Protein samples (150 µg of a 1 mg/ml protein sample) were treated in vitro with nifedipine for 30 min before stimulation of PKC with a mixture of diacylglycerol and calcium. This was done with the PKC Biotrak ® enzyme assay (Amersham Biosciences, Little Chalfont, Buckinghamshire, UK) in accordance with the manufacturer's instructions. Proteins (50 µg) were separated by SDS gel electrophoresis. Proteins were fluorescently stained by fixing the gels overnight in 50% methanol and 10% trichloracetic acid. The gels were washed with deionised water for 10–20 min, stained with Pro-Q-Diamond for 180 min and destained by three washes in 4% acetonitrile in 50 mM sodium acetate, pH 4, for 2 hours. Gels were scanned with a fluorimager, a Typhoon 9400 laser scanner (Amersham Biosciences), with excitation at 532 nm and a 580 nm band pass emission filter for Pro-Q-diamond dye detection. Phosphorylated proteins were quantified densitometrically with the Image Quant software. Results represent the net phosphorylation induced by PMA, expressed as percentages of the control PMA lane (taken as 100%). PKC assay PKC assays were performed in vitro with a biologically active full-length recombinant isoform (Calbiochem, EMB Biosciences, San Diego, CA, USA) and a non-radioactive PKC kinase assay kit (Stressgen, Victoria, BC, Canada). The principle is based on an enzyme-linked immunosorbent assay method that uses a specific synthetic peptide as a substrate for PKC and a polyclonal antibody that recognised the phosphorylated form of the substrate. In brief, 1 µl of recombinant PKC (5 µg/ml) in 30 µl of kinase assay dilution buffer was incubated for 20 min at 37°C with the drug vehicle (control) or with nifedipine or two other PKC inhibitors (calphostin and GFI09203). PKC was then activated by 10 µl of PMA (100 nM) for 10 min at 37°C and the reaction was blocked by incubating tubes in ice-cold water. PKC activity was then determined as recommended by the manufacturer. Results are expressed as percentages of the net activity induced by PMA (taken as 100%). Approval and consent The study was approved by local ethics committee, and all patients gave written informed consent. Statistical analysis Data were compared with the nonparametric Mann–Whitney (unpaired data) and Wilcoxon (paired data) tests. P < 0.05 was considered significant. All quantitative data are expressed as means ± SD. Results Monocyte O 2 •- production and respiratory burst Monocytes from patients with SSc ( n = 12) not treated by nifedipine produced a significantly greater amount of O 2 •- than controls ( n = 10) after stimulation ex vivo with the PKC activator PMA for 15 min (9.1 ± 1.7 versus 5.4 ± 0.7 nmol per 10 6 cells; P < 0.01; Fig. 1a ). This difference was also observed when monocytes were stimulated with PMA for 10 min (7.6 ± 2.3 versus 4.3 ± 0.8 nmol per 10 6 cells; P < 0.01) but not when they were stimulated for 5 min (2.7 ± 0.8 versus 2.1 ± 0.4 nmol per 10 6 cells; not significant). These observations suggest that the biochemical modifications affecting monocytes from patients with SSc might not affect the kinetics of NADPH oxidase but rather its late regulatory mechanisms. Monocytes from patients treated with nifedipine (60 mg/day) for 14 days produced a significantly smaller amount of O 2 •- than monocytes from untreated patients with SSc after stimulation with PMA for 5 min (1.0 ± 0.7 versus 2.7 ± 0.8 nmol per 10 6 cells; P < 0.05), 10 min (1.6 ± 1.4 versus 7.6 ± 2.3 nmol per 10 6 cells; P < 0.001) and 15 min (1.8 ± 1.0 versus 9.1 ± 1.7 nmol per 10 6 cells; P < 0.001; Fig. 1a ). Disease subsets, in particular the cutaneous subtype, did not influence baseline O 2 •- production by monocytes or the decrease in O 2 •- production with nifedipine treatment. To determine whether monocytes were directly altered by nifedipine, monocytes from healthy donors were treated in vitro with nifedipine for 30 min before stimulation with PMA. Nifedipine induced a concentration-dependent inhibition of O 2 •- production at concentrations of 10 and 50 µM ( n = 5) (Fig. 1b ). PMA induction was inhibited by 50% (IC 50 ) by approximately 3–5 µM nifedipine. This inhibitory effect of nifedipine was not due to any cytotoxic effect, as determined by the trypan blue exclusion test (cell death less than 5%). The effects of nifedipine were next studied on the monocyte respiratory burst induced by an inflammatory agonist such as fMLP. However, fMLP induced very weak monocyte O 2 •- production in standard conditions, so respiratory burst was measured with a fluorimetric assay. Treatment of monocytes for 30 min with 5 and 10 µM nifedipine significantly inhibited hydrogen peroxide production by 55 ± 7% and 90 ± 9%, respectively, relative to controls ( n = 3). Next we studied whether other calcium-channel blockers alter the O 2 •- production of healthy monocytes ( n = 5) (Fig. 1c ). A strong and similar inhibitory effect was observed with nifedipine or nicardipine, but a less marked inhibitory effect was observed with diltiazem. To investigate whether the inhibitory effect of nifedipine was due to the scavenging of free radicals, O 2 •- was generated by the xanthine/xanthine oxidase system in the presence of various concentrations of nifedipine. In these conditions, nifedipine did not affect O 2 •- generation, ruling out a scavenger effect (data not shown). Protein phosphorylation The production of O 2 •- by phagocytes is dependent on the activation of PKC, a family of isoenzymes that phosphorylate various proteins including some components of the NADPH oxidase. Because PMA directly activates various PKC isoforms, we studied the effect of nifedipine on PKC-dependent protein phosphorylation in healthy monocytes. Phosphorylated proteins were directly detected in polyacrylamide gels by using Pro-Q-Diamond, a fluorescent dye that binds to phosphate groups on proteins. In the absence of PMA, nifedipine inhibited the basal phosphorylation of various proteins in monocytes (Fig. 2a ). In monocytes not treated with nifedipine, stimulation with PMA markedly increased the phosphorylation state of various proteins and nifedipine strongly decreased the PMA-induced phosphorylation of proteins. Densitometric analysis of 10 major phosphorylated protein bands with molecular masses of between 60 and 15 kDa showed that nifedipine inhibited the net protein phosphorylation induced by PMA in a concentration-dependent manner. The inhibition of phosphorylation was also investigated in this cell system under the same conditions with the classical PKC inhibitor calphostin; the inhibition of protein phosphorylation had a quite similar profile (Fig. 2b ) with closed results on densitometric analysis. The inhibitory effects of nifedipine were also demonstrated in a cell-free system in which the activity of PKCs present in the cytosolic fraction of resting monocytes was directly stimulated in the presence of a mixture of calcium and diacylglycerol. In these conditions, treatment of the cytosolic fraction with nifedipine abolished PKC-dependent protein phosphorylation (Fig. 2c ). PKC activity in vitro To determine whether PKC is a direct target of nifedipine, we investigated kinase activity in vitro with a biologically active PKC recombinant that was pretreated with or without various concentrations of nifedipine or with other classical PKC inhibitors such as calphostin or GF109203X [ 16 ]. Nifedipine (1, 10 and 50 µM) dose-dependently inhibited PKC activity with an inhibition reaching about 80%. Similar inhibition was observed with calphostin and GF109203X ( n = 3; Fig. 3 ). These results strongly suggest that nifedipine might directly inhibit PKC. Discussion In this report we show that, first, ex vivo monocytes from patients with SSc not treated with calcium-channel blockers produce more O 2 •- than control monocytes; second, nifedipine (60 mg/day) strongly inhibits the ability of monocytes to produce O 2 •- ex vivo ; and third, nifedipine inhibits O 2 •- production by healthy monocytes in vitro , a property associated with the inhibition of PKC-dependent protein phosphorylation and PKC activity. Since the proposal that free radicals have a function in SSc [ 17 ], several reports have provided evidence that free radicals are involved in the pathogenesis of this complex disease [ 1 , 2 , 7 , 18 ]. Vasospasm and ischaemia–reperfusion are known to generate free radicals, but other factors such as the inflammatory process might also be involved [ 1 ]. It has been suggested that resting and PMA-stimulated monocytes produce more O 2 •- than controls [ 7 ]. This is consistent with our observation that circulating monocytes from patients with SSc stimulated ex vivo by PMA release about twice as much O 2 •- as controls, which is similar to previous findings [ 7 ]. This PMA-dependent increase in O 2 •- suggests that circulating monocytes are in a primed state, which is consistent with their prior exposure to a substimulatory dose of agonists such as cytokines or immune complexes [ 19 ]. These data show that free radicals are overproduced in SSc and that this is due at least partly to monocytes. The mechanism of monocyte activation remains unknown. However, ischaemic insult does not seem sufficient to explain this phenomenon, which is associated with NADPH oxidase activation and might involve cytokines [ 7 ]. The main finding of this study is that the treatment of patients with SSc with nifedipine (60 mg/day) strongly decreases the production of O 2 •- by monocytes and leads to the disappearance of the monocyte priming state. PMA was chosen because it mimics the activation of NADPH oxidase by PKC, a mechanism also induced by most of the physiological activators of monocytes. Mononuclear cells are thought to be important in SSc [ 3 - 5 ] and in oxidative stress because they release various mediators leading to tissue injury. Unlike monocytes, polymorphonuclear neutrophils do not seem to be activated in SSc [ 7 ]. The results reported here are consistent with our previous data showing that dihydropyridine-type calcium-channel blockers can reduce plasma markers of oxidative stress [ 10 ]. Together with data showing a concomitant improvement of endothelial injury [ 11 ], these results suggest that calcium-channel blockers, especially the dihydropyridine type, are not only of major importance for the treatment of Raynaud's phenomenon and vasospastic propensity [ 20 , 21 ] but should also be regarded as essential drugs that limit monocyte activation and oxidative stress. Several reports have demonstrated the anti-oxidant properties of calcium-channel blockers. In hypertensive rats, dihydropyridine calcium-channel blockers decrease lipoprotein oxidation and prolong survival independently of modifications of blood pressure [ 22 ]. In hypertensive patients, treatment with nifedipine decreases lipoperoxide and isoprostane concentrations and increases the plasma antioxidant capacity while increasing endothelium-dependent vasodilation by restoring nitric oxide bioavailability [ 23 ]. At the cellular level, nifedipine can prevent ischaemia-induced endothelial permeability mediated primarily by the inhibition of PKC-a [ 24 ]. A study with activated human and rabbit neutrophils [ 25 ] and another study in vascular smooth muscle cells [ 26 ] suggested that PKC can be inhibited by nicardipine. Our results are consistent with these findings and further emphasise the ability of nifedipine to inhibit PKC activity in vitro , as shown by the strong decrease in PMA-induced protein phosphorylation in monocytes and in a cell-free system and the inhibition of a recombinant isoform of PKC. Analysis of the protein phosphorylation pattern showed that nifedipine also inhibits basal phosphorylation of proteins, indicating that other protein kinases might be altered. The molecular mechanism by which nifedipine inhibits PKC cannot be derived from our results and will require further experiments. However, nifedipine inhibits both basal and PMA-induced protein phosphorylation as well as PKC activity, which suggests that nifedipine might interact with both the regulatory and catalytic domains of PKC. The inhibition of PKC-dependent protein phosphorylation might be part of the molecular mechanism responsible for the loss of the primed state of monocytes from patients with SSc treated with nifedipine. This is emphasised by the fact that IC 50 values were above 3–5 µM nifedipine for PMA-induced O 2 •- production in vitro (Fig. 1b ) and also for PKC-dependent phosphorylation (Fig. 2a ). However, this hypothesis must be tested with monocytes from patients with SSc. Nifedipine (10 and 50 µM) significantly inhibited O 2 •- release by activated monocytes in vitro ; this is consistent with results previously obtained with human neutrophils [ 25 ]. The IC 50 for calcium channels antagonist activity of nifedipine is close to 100 pM [ 27 ] and the clinical plasma concentration of nifedipine is close to 0.2 µM [ 28 ]. We used higher nifedipine concentrations for experiments in vitro ; however, nifedipine can accumulate in membrane-bounded structures, resulting in higher localised concentrations [ 29 ]. Thus, our data with nifedipine and other calcium-channel blockers in vitro could explain the results obtained with monocytes from patients with SSc treated with nifedipine. We found that diltiazem had a weaker effect on O 2 •- production by activated monocytes than other calcium-channel blockers evaluated; this was previously reported using human neutrophils [ 30 ] and suggests that dihydropyridine-type calcium-channel blockers are the strongest inhibitors of O 2 •- production. Most of the classical properties of nifedipine are exerted through the modulation of L-type voltage-gated calcium channels on smooth muscle cells. Such channels are also present on monocytes; we suspect that the molecular mechanism highlighted is independent of these channels, but this will require further investigations. Calcium channel-independent properties have previously been reported and because these are not represented in endothelial cells they cannot account for these cellular effects [ 31 ]. It was suggested that these drugs act as scavengers; however, our data, together with those on human polymorphonuclear neutrophils [ 24 ], do not support this hypothesis. Conclusion Monocytes from patients with SSc adopt a priming state resulting in increased ex vivo PKC-dependent production of O 2 •- relative to healthy monocytes. Treatment of patients with SSc or of healthy monocytes in vitro markedly decreased the ability of monocytes to generate O 2 •- . Biochemical analysis of protein phosphorylation in monocytes and cell-free systems suggested that nifedipine directly inhibits PKC activity. These results demonstrate the potential anti-oxidant effects of this drug, which might have important clinical implications for SSc and other oxidative stress-associated diseases. Abbreviations DTCCA = dihydropyridine-type calcium-channel antagonists; fMLP = formyl-Met-Leu-Phe; IC 50 = concentration for 50% inhibition; PKC = protein kinase C; PMA = phorbol myristate acetate; SSc = systemic sclerosis. Competing interests The author(s) declare that they have no competing interests. Authors' contributions YA, DB and AP devised the study. YA, HL and AP performed all the experiments. OGE and AK assisted in the writing of the report. All authors read and approved the final manuscript.

D:\keerthana\PMC001xxxxxx\PMC1064885.xml

1064886

Gene expression profiles in the rat streptococcal cell wall-induced arthritis model identified using microarray analysis

Experimental arthritis models are considered valuable tools for delineating mechanisms of inflammation and autoimmune phenomena. Use of microarray-based methods represents a new and challenging approach that allows molecular dissection of complex autoimmune diseases such as arthritis. In order to characterize the temporal gene expression profile in joints from the reactivation model of streptococcal cell wall (SCW)-induced arthritis in Lewis (LEW/N) rats, total RNA was extracted from ankle joints from naïve, SCW injected, or phosphate buffered saline injected animals (time course study) and gene expression was analyzed using Affymetrix oligonucleotide microarray technology (RAE230A). After normalization and statistical analysis of data, 631 differentially expressed genes were sorted into clusters based on their levels and kinetics of expression using Spotfire ® profile search and K-mean cluster analysis. Microarray-based data for a subset of genes were validated using real-time PCR TaqMan ® analysis. Analysis of the microarray data identified 631 genes (441 upregulated and 190 downregulated) that were differentially expressed (Delta > 1.8, P < 0.01), showing specific levels and patterns of gene expression. The genes exhibiting the highest fold increase in expression on days -13.8, -13, or 3 were involved in chemotaxis, inflammatory response, cell adhesion and extracellular matrix remodelling. Transcriptome analysis identified 10 upregulated genes (Delta > 5), which have not previously been associated with arthritis pathology and are located in genomic regions associated with autoimmune disease. The majority of the downregulated genes were associated with metabolism, transport and regulation of muscle development. In conclusion, the present study describes the temporal expression of multiple disease-associated genes with potential pathophysiological roles in the reactivation model of SCW-induced arthritis in Lewis (LEW/N) rat. These findings improve our understanding of the molecular events that underlie the pathology in this animal model, which is potentially a valuable comparator to human rheumatoid arthritis (RA).

Introduction Rheumatoid arthritis (RA) is an autoimmune chronic inflammatory disease of unknown aetiology that is characterized by infiltration of monocytes, T cells and polymorphonuclear cells into the synovial joints. The pathogenesis of this disease is still poorly understood, and fundamental questions regarding the precise molecular nature and biological significance of the inflammatory changes remain to be answered [ 1 , 2 ]. A powerful way to gain insight into the molecular complexity and pathogenesis of arthritis has arisen from oligonucleotide-based microarray technology [ 3 ], because this platform provides an opportunity to analyze simultaneously the expression of a large number of genes in disease tissues. The earliest preclinical stages of human RA are not easily accessible to investigation, but a diverse range of experimental arthritis models are considered valuable tools for delineating mechanisms of inflammation and autoimmune phenomena. An animal model that shares some of the hallmarks of human RA is the reactivation model of streptococcal cell wall (SCW)-induced arthritis in rats. In this model, a synovitis with maximal swelling at 24 hours is induced by local injection of SCW antigen directly into an ankle joint. The initial response is reactivated by systemic (intravenous) challenge with SCW, which produces a more prolonged and severe inflammation confined to the joint previously injected with SCW. In contrast to some other animal models, in which the arthritic response develops gradually and unpredictably, in this model the flare response develops synchronously, allowing precise analysis of pathophysiological mechanisms [ 4 , 5 ]. Some pathological changes observed in SCW-induced arthritis that are of relevance to human RA include infiltration of polymorphonuclear cells, CD4 + T cells and macrophages, hyperplasia of the synovial lining layer, pannus formation and moderate erosion of cartilage and bone [ 4 ]. Previous reports have shown the dependency of this model on tumour necrosis factor (TNF)-α, IL-1α, IL-4, P-selectin, vascular cell adhesion molecule-1, macrophage inflammatory protein (MIP)-2, MIP-1α and monocyte chemoattractant protein (MCP)-1 [ 6 , 7 ]. Although the involvement of nitric oxide synthase (NOS) [ 8 ] and cyclo-oxygenase [ 9 ] in the development of SCW-induced arthritis has also been noted, a global analysis of coordinated gene expression during the time course of disease in this experimental arthritis model has not been investigated. Arthritis involves many cell types from tissues adjacent to the synovium. Therefore, as shown in previous studies [ 10 , 11 ], analysis of gene expression profiles by processing whole homogenized joints can provide useful information about dysregulated genes, not only in synoviocytes but also in other, neighbouring cells (myocytes, osteocytes and chondrocytes) that may also contribute to disease pathology. In the present study, whole homogenized rat ankle joints from naïve, SCW-injected and phosphate-buffered saline (PBS; vehicle)-injected animals, included in a time-course study, were analyzed for differential gene expression using the RAE230A Affymetrix GeneChip ® microarray (Affymetrix Inc., Santa Clara, CA, USA). In order to identify different patterns of gene expression during the course of SCW-induced arthritis, a selected set of genes whose expression was statistically significantly different between arthritic and control animals on days -13.8, -13 and 3 were analyzed using agglomerative hierarchical clustering, Spotfire ® (Spotfire Inc., Cambridge, MA, USA) profile search and K-means cluster analysis. Validation of microarray data for a subset of genes was performed by real-time RT-PCR TaqMan ® (Applied Biosystems, Foster City, CA, USA) analysis, which provides a highly accurate method for quantifying mRNA expression levels for any particular differentially expressed gene. To further investigate the possible association of 20 selected upregulated genes with arthritis pathogenesis, their chromosomal locations and the chromosomal locations of their corresponding human orthologue were compared with the locations of previously reported quantitative trait loci (QTLs) for inflammation, arthritis and other autoimmune diseases. Our findings show, for the first time, the gene expression profiles and kinetics of expression of hundreds of genes that are differentially expressed in arthritic joints from the reactivation model of SCW-induced arthritis in Lewis (LEW/N) rat, thereby improving our understanding of the biological pathways that contribute to the pathogenesis of arthritis in this animal model and providing a valuable comparator to human RA. Methods Reagents The peptidoglycan–polysaccharide (PG-PS) 100p fraction of SCW was purchased from Lee Laboratories (Grayson, GA, USA). RAE230A Affymetrix GeneChip ® were purchased from Affymetrix Inc. All reagents required for RT-PCR were from PE Applied Biosystems (Warrington, UK). Forward and reverse primers were purchased from Invitrogen™ Life Technologies (Invitrogen Ltd, Paisley, UK). TaqMan ® probes were synthesized by PE Applied Biosystems. RiboGreen, used to quantify RNA, was obtained from Molecular Probes Inc. (Leiden, The Netherlands) and RNA 6000 Nano LabChip Kit ® , used to assess RNA integrity, was from Agilent Technologies Inc. (Stockport, UK). Animals All in vivo studies were undertaken in certified, dedicated in vivo experimental laboratories at the GlaxoSmithKline Medicines Research Centre (Stevenage, UK). The studies complied with national legislation and with local policies on the care and use of animals, and with related codes of practice. Male Lewis (LEW/N) rats obtained from Harlan UK Ltd (Oxon, UK), at age 6–7 weeks, were housed under standard conditions and received food and water ad libitum . Animals were habituated to the holding room for a minimum of 1 week before the experimental procedures. Induction and assessment of SCW-induced arthritis SCW arthritis was induced in 6- to 8-week-old male Lewis (LEW/N) rats (weight 125–150 g) following a method similar to that previously described by Esser and coworkers [ 4 ]. A SCW preparation (PG-PS, 100p fraction) was suspended in PBS and 10 μl of the suspension containing 5 μg PG-PS from Streptococcus pyogenes was injected into the right ankle joint (day -14). Animals from control groups were injected similarly with 10 μl PBS. A group of noninjected rats was also included in our study to assess gene expression profiles in joints from naïve animals. Reactivation of the arthritic inflammation was induced 14 days after intra-articular injection (designated day 0) by intravenous injection of 200 μg PG-PS. This resulted in monoarticular arthritis involving the joint originally injected with PG-PS [ 7 ]. Ankle swelling at different time points was measured using a caliper. The inflammatory response is expressed as the change in ankle diameter relative to the starting diameter. Five animals injected with PG-PS or PBS were killed at different time points (4 hours after intra-articular injection [day -13.8], day -13, day -10, day 0, 6 hours after intravenous challenge [day 0.4], day 1, day 3 and day 7) and ankle joints were dissected, snap frozen in liquid nitrogen and stored at -80°C for subsequent analysis. Total RNA isolation from rat joints Frozen ankle joints were pulverized in liquid nitrogen using a mortar and pestle, and total RNA was isolated from individual homogenized joints (four or five animals/group) using RNeasy ® Mini-kits (Qiagen Ltd, Crawley, UK), following the manufacturer's instructions. In our experimental design, a nonpooling strategy for total RNA samples was used (a total of 75 samples from different animals were analyzed). In order to ensure that no contamination with genomic DNA occurred, samples were treated for 15 min with 10 units of RNase-free DNase (Qiagen Ltd) at room temperature. RiboGreen ® RNA Quantitation Kit (Molecular Probes Inc.) with optical densities at 260 nm and 280 nm was used to determine the total RNA concentration of the samples. The quality of the RNA was assessed based on demonstration of distinct intact 28S and 18S ribosomal RNA bands using RNA 6000 Nano LabChip Kit ® (Agilent 2100 Bioanalyser; Agilent Technologies UK Ltd, Stockport, UK). Five of the 75 total RNA samples exhibited evidence of RNA degradation and were excluded from subsequent analyses. Oligonucleotide microarray analysis The rat RAE230A GeneChip ® oligonucleotide microarray (Affymetrix Inc.), containing about 16,000 probe sets, representing 4699 well annotated full-length genes, 10,467 expressed sequence tags (ESTs) and 700 non-ESTs (excluding full lengths), was used to analyze gene expression profiles in joints from SCW-injected or PBS-injected animals during the course of disease. Isolated total RNA (10 μg/chip) was used to generate biotin-labelled cRNA. Aliquots of each sample ( n = 70) were then hybridized to RAE230A Affymetrix GeneChip ® arrays at 45°C for 16 hours, followed by washing and staining, in accordance with the standard protocol described in the Affymetrix GeneChip ® Expression Analysis Technical Manual [ 12 ]. The GeneChip ® s were scanned using the Affymetrix 3000 Scanner™ and the expression levels were calculated for all 16,000 probe sets (about 12,000 genes) with Affymetrix ® MicroArraySuite software (MAS 5.0). Statistical analysis of microarray data The Affymetrix GeneChip ® data were processed, normalized and statistically analyzed (analysis of variance [ANOVA]) using Rosetta Resolver ® v3.2 software (Rosetta Biosoftware, Kirkland, WA, USA). Genes with P < 0.01 (ANOVA) were considered to be differentially expressed. Fold changes in gene expression were calculated by dividing the mean intensity signal from all the individual SCW-injected rats included in each group by the mean intensity signal from the corresponding PBS control group. The level of statistical significance was determined by ANOVA. Subsequent data analysis involved two-dimensional data visualization, principal component analysis (PCA) using SIMCA-P v10.2 Statistical Analysis Software (Umetrics, Windsor, UK) [ 13 ] and agglomerative hierarchical clustering analysis [ 14 ]. For identification of different temporal patterns and levels of gene expression, Spotfire ® profile search analysis and K-means clustering analysis [ 15 ] were performed using the Spotfire ® DecisionSite for Functional Genomics programme. In this analysis the mean signal intensity of gene expression in each group included in the study (four to five samples/group) was used. Identification of the ontology, accession number and chromosomal location of the genes of interest was performed combining information from GlaxoSmithKline Bioinformatics Databases and other existing public databases . The mapping of the differentially expressed genes to QTLs for arthritis was investigated using Rat and Human Genome browsers from Ensembl , Rat Genome Database and the ARB Rat Genetic Database . Quantitative real-time PCR (TaqMan ® ) Expression levels of selected genes found to be upregulated by gene array analysis were validated by real-time RT-PCR TaqMan ® analysis using the ABI Prism 7900 Sequence Detector System ® (PE Applied Biosystems, Foster City, CA, USA), as previously described [ 16 ]. For cDNA synthesis 600 ng total RNA (from the same samples analysed by RAE230A GeneChip ® microarray) were reverse transcribed using TaqMan ® RT reagents (PE Applied Biosystems) in a MJ Research PTC-200 PCR Peltier Thermal Cycler (MJ Research, Rayne Brauntree, Essex, UK). TaqMan ® probes and primers for the genes of interest were designed using primer design software Primer Express™ (PE Applied Biosystems) and optimized for use. The forward primers, reverse primers and probes used are summarized in Table 1 . The final optimized concentrations of forward primer, reverse primer and probe for all of the target genes were 900 nmol/l, 900 nmol/l and 100 nmol/l, respectively, except for CD14, for which the concentrations were 300 nmol/l, 300 nmol/l and 100 nmol/l. Standard curves for each individual target amplicon were constructed using sheared rat genomic DNA (BD Biosciences, Cowley, Oxford, UK). All PCR assays were performed in duplicate, and results are represented by the mean values of copy no./50 ng cDNA. Ubiquitin [ 17 ] was used as a housekeeping gene against which all samples were normalized. Data presentation The data included in Table 2 show the mean fold change (Delta) increase or decrease in gene expression in joints from SWC-injected rats compared with the expression in the corresponding PBS control group, along with the P value. As selection criteria to present the most relevant genes, a cutoff of 1.8-fold increased/decreased expression and P < 0.01 were arbitrarily chosen. Gene expression profile plots (Fig. 6 ) represent data from Affymetrix Rat Genome RAE230A GeneChip ® and real-time RT-PCR TaqMan ® as the mean of signal intensity or the mean of normalized copy no./50 ng cDNA for all the samples from the same group (four to five), respectively. Results Time course of inflammation in the SCW-induced arthritis model Intra-articular injection of SCW resulted in increased ankle swelling that peaked 24 hours after injection (day -13), followed by a gradual reduction by day 0 (Fig. 1 ). At this time point intravenous challenge with SCW led to reactivation of the inflammatory response, which peaked 72 hours thereafter (day 3). Animals injected intra-articularly with PBS (vehicle in which the SCW was suspended) were used as control groups at each specific time point. Another group of naïve animals (noninjected rats) was used to assess a possible inflammatory response due to the intra-articular injection alone. Gene expression profiling in SCW-induced arthritis Analysis of RAE230A GeneChip ® microarray data identified about 9000 probes (5479 upregulated and 3898 downregulated) that were differentially expressed to a highly significant degree ( P < 0.01) in arthritic rat joints from the time course study. After applying selection criteria (Delta > 1.8 and P < 0.01), 631 of the dysregulated probes had well characterized full-length sequences in databases (441 upregulated and 190 downregulated) and 697 were unknown (ESTs; 444 upregulated and 253 downregulated). These genes are too numerous to describe in detail, and therefore we present a selected list of upregulated genes in Table 2 and Fig. 2 , and a selection of downregulated genes based on the ontologies that reflect the major changes occurring in arthritic animals (Fig. 3 ). ESTs were excluded from Table 2 and from subsequent clustering analysis. See Additional file 1 , which contains all genes that were upregulated and downregulated. Principal component analysis and hierarchical clustering An overview of the experimental RAE230A GeneChip ® data was obtained using PCA (graphs not shown) [ 13 ] and agglomerative hierarchical clustering [ 14 ]. Both two-dimensional analyses identified day -13.8 (4 hours after intra-articular injection of SCW), day -13 and day 3 as the time points at which the greatest changes in gene expression in arthritic joints occurred in comparison with corresponding PBS control groups. The results from the hierarchical clustering are shown for visual inspection as a coloured heat map in Fig. 4 . As shown on the x-axis (panel at the top of Fig. 4 ), the majority of the PBS samples clustered together, except the PBS samples from day -13.8, which clustered close to the SCW-injected animals from day 3. This observation indicated the presence of a mild inflammatory response in joints from rats killed 4 hours after the initial intra-articular injection of PBS, when compared with expression levels in joints from naïve animals or the PBS samples from later time points. PCA and hierarchical clustering analysis allowed us to identify two outliers corresponding to arthritic animals from day 3, which did not show any sign of measurable inflammation after intravenous challenge. Both samples were excluded from subsequent mean or Delta calculations. Identification of different patterns of gene expression The selected 631 dysregulated genes ( P < 0.01 and Delta > 1.8) were analyzed using Spotfire ® profile search analysis and K-means clustering [ 15 ], allowing the identification of different patterns and levels of gene expression throughout the time course of disease. As shown in Fig. 5 , the upregulated genes were grouped into seven clusters (C-1 to C-7) according to their kinetics of expression. Thus, all genes exhibiting similar patterns of expression at the analyzed time points were grouped into the same cluster (e.g. C-1 for those genes whose expression reached a peak on day -13.8). These genes were also sorted into three K-means clusters according to their level of expression (low, medium and high). The cluster number to which each gene belongs is summarized in Table 2 . Interestingly, the expressions of different markers for cell types associated with the pathogenesis of RA were found to be upregulated throughout the time course of SCW-induced arthritis. These markers were grouped into different clusters as follows: C-2 = CD44 (leucocytes, erythrocytes); C-3 = CD2 (T cell, natural killer [NK] cells), E-selectin (SELE; activated endothelial cells); C-4 = L-selectin (SELL; lymphocytes, monocytes and NK cells); C-5 = CD14 (monocytes), ICAM1 (endothelial cells), α M integrin (ITGAM or CD11b; granulocytes, monocytes, NK cells), P-selectin (SELP; endothelial cells, activated platelets), lipocalin 2 (LCN2; neutrophils); C-6 = CD74 (B cells, monocytes), CD38 (activated T cells, plasma cells), CD8a (cytotoxic/suppressor T cells, NK cells); and C-7 = CD3d (T cells), CD4 (helper–inducer T cells). The different temporal expression of these markers highlights that expression levels for CD3d and CD4 were significantly upregulated only at day 3 after challenge, in contrast to CD2 and E-selectin, whose expression was found to be upregulated only at day -13. The rest of the markers exhibited significant fold changes in gene expression at both phases of disease (4 hours after intra-articular injection of SCW, day -13 and day 3 after challenge), except CD8a, CD74 and CD38, which were found to be upregulated at a later time point in the pre-reactivation phase (day -13). Only CD44 was not found to be upregulated on day 3 after challenge. Lipocalin 2, αM integrin and CD8a exhibited the greatest fold changes in gene expression. Functional grouping of dysregulated genes In order to establish functional annotations for the selected dysregulated genes, the biological processes and molecular functions of the genes were investigated using different databases. This search identified 19 ontologies for the upregulated genes, allowing us to organize them according to their major functions (Table 2 and Fig. 2 ). Because of space limitations in the manuscript, we could not include all of the upregulated genes in Table 2 and Fig. 2 . The genes not included were involved in blood coagulation, catabolism, defence response, G-protein-coupled receptor protein signalling pathways, metabolism and protein modification, or were genes with unknown functions (for more information, please see Additional file 1 ). A hallmark of RA is infiltration of leucocytes into synovial tissue mediated by a complex network of cytokines, adhesion molecules and chemoattractants [ 18 ]. Interestingly, most of the genes exhibiting the greatest fold increase in gene expression (Delta > 5) on days -13.8, -13 or 3 were involved in chemotaxis. These included several CC chemokine ligands (CCLs; CCL20, CCL2 [also called SCYA2 or MCP-1]), CXC chemokine ligands (CXCLs; CXCL2, CXCL6 and GRO1), CC chemokine receptors (CCRs; CCR1, CCR2, CCR5), CXC chemokine receptors (CXCRs; CXCR2) and a recently characterized cytokine called chemokine-like factor 1 [ 19 ]. Our results also showed marked upregulation (Delta > 5) for numerous genes that are involved in the immune and/or inflammatory response, such as IL-1β, IL-6, TNF-α, TNFRSF1b, IL-1Rn, NOS2, CD8a, VAV1, LST1 (leukocyte specific transcript 1), LCP2 (lymphocyte cytosolic protein 2), FCGR2 (Fc receptor, IgG, low affinity Iib), PTGES (microsomal prostaglandin E synthase-1) and the major histocompatibility complex (MHC) class Ib gene (RTAW2). Other components of the MHC such as MHC class II (HLA-DMA and HLA-DMB) and MHC class Ib RT1.S3 genes were also found to be upregulated in this model. Genes participating in cell adhesion such as TNFIP6, FCNB (ficolin B), CSPG2 (versican), ICAM1 and αM integrin (ITGAM) also exhibited a significant fold increase in gene expression (Delta > 5). Among other genes, some mediators controlling extracellular matrix (ECM) turnover and breakdown under normal and disease conditions, including five matrix metalloproteinases (MMPs; MMP-3, -12, -13, -14 and -23a), the aggrecanase ADAMTS-1, tissue inhibitor of metalloproteinases (TIMP)1, and the secretory leucocyte protease inhibitor (SLPI) were also found to be significantly upregulated in arthritic joints. The majority of the downregulated genes were associated with regulation of metabolism, myogenesis, or regulation of muscle development and transport (Fig. 3 ). Differentially expressed genes: QTL association From the 441 selected genes that were upregulated during SCW-induced arthritis, we selected a list of 20 genes that exhibited a greater than fivefold change in gene expression and that had not previously been linked to autoimmune arthritis. To further investigate the possibility that these genes play a role in arthritis pathogenesis, their rat chromosomal locations and the locations of their human orthologues were identified and compared with those of rat and human QTLs for autoimmune diseases. Interestingly, 10 of these genes were found to be located in chromosomal regions that mapped to rat and/or human QTLs previously reported to be associated with inflammation, arthritis, or autoimmune diseases, such as systemic lupus erythematosus, multiple sclerosis, allergic rhinitis and asthma (Table 3 ). Analysis of expression profiles of specific transcripts In order to validate microarray data, mRNA expression levels for a subset of genes were quantified by real-time RT-PCR TaqMan ® analysis. As shown in Fig. 6 , there was a significant correlation (Pearson product moment correlation coefficient r > 0.9 and P < 0.01) between the gene expression profiles for the proinflammatory cytokines IL-1β, TNF-α and IL-6, the chemokine GRO1 and the cell markers CD14 and CD3, when microarray data were compared with RT-PCR TaqMan ® data. Although the fold changes in gene expression calculated using data from both methods were not exactly the same (probably due to differences in the sensitivities of the assays), the quantitative real-time RT-PCR TaqMan ® method verified the results of the gene array analysis. Discussion The temporal expression of multiple disease-associated genes with potential pathophysiological roles in the reactivation model of SCW-induced arthritis in Lewis (LEW/N) rat has not previously been fully addressed. The present study analyzed gene expression profiles in rat joints with SCW-induced arthritis using RAE230A GeneChip ® oligonucleotide microarray (Affymetrix Inc.). We chose to profile gene expression in whole ankle joint tissues, which comprises heterogeneous cell types, with the aim of gaining a global insight into the molecular changes associated with arthritis pathology in this model. Analysis of the time course data generated by microarray identified 631 genes (441 upregulated and 190 downregulated) with full-length sequences in databases that were significantly differentially expressed (Delta > 1.8 and P < 0.01). Our experimental design (time course study) and use of K-means cluster analysis allowed us to identify specific patterns of gene expression for the different dysregulated genes, highlighting the importance of performing kinetic studies to identify the time point at which a particular gene is maximally expressed. Thus, these gene expression data indicate optimal times for measuring potential disease biomarkers in rat joints, and our approach offers a useful tool with which to investigate the clinical efficacy and mechanism of action of novel therapeutic agents in rat SCW-induced arthritis. Changes in gene expression may reflect regulation at the mRNA level or changes in the number of cells (proliferation or infiltration) that synthesize these mRNAs. Thus, optimally, microarray analysis should be conducted in isolated populations of cells so that differential gene expression may be directly correlated with transcription of the genes. However, complex diseases such as RA involve extensive tissue injury, and not all of the cell types that contribute to RA pathogenesis have been identified. Hence, analysis of the damaged tissue, rather than analysis of an isolated cell type, increases the probability that differential gene expression will be examined in those cells that are important in RA pathogenesis. In the present study we conducted a global analysis of coordinated gene expression in injured tissue. Further bioinformatic analysis of the data to examine cell markers, and genes whose expression may correlate with them, in combination with analysis of the cell populations present in the arthritic joint using immunohistochemistry or fluorescence activated cell sorting techniques, would be required to corroborate the differential gene expression of a particular gene of interest. Previous studies have already shown that cell-specific gene expression patterns can indicate the presence of immune cells [ 20 ]. RAE230A GeneChip ® oligonucleotide microarray analysis identified the expression of different markers for cell types associated with the pathogenesis of RA. Based on the level of gene expression and Delta values detected for the different markers, our results suggest that the main cell types present in arthritic joints in this model are T cells, neutrophils, monocytes/macrophages and B cells, confirming previous descriptions of the joint cell composition in this model [ 6 , 21 ]. Gene expression profiling of arthritic rat joints revealed a spectrum of genes exhibiting extensive inflammatory activity, infiltration of activated cells, angiogenesis, regulation of apoptosis and ECM remodelling activities. Most of the genes found to be upregulated in SCW-induced arthritic joints have also been reported to be highly expressed in human RA synovial tissue [ 22 , 23 ] or in joints from other rodent experimental arthritis models [ 10 , 11 , 24 , 25 ]. The upregulated expression of TNF-α, IL-1α, IL-1β, IL-4R, P-selectin, MIP-1α (CCL3), MCP-1 (CCL2), NOS2 and NOS3 [ 6 - 8 ] demonstrated in the present study is in agreement with previous observations of the dependency of the rat SCW-induced arthritis model on these mediators. The SLPI has previously been reported to be upregulated in arthritic joints and to mediate tissue destruction and inflammation in a rat model of arthritis induced by intraperitoneal injection of SCW [ 26 ]. Similar results were found in our study, because significant upregulation of SLPI gene expression was observed during both phases of the disease. Additionally, previous studies have shown that nuclear factor-κB (NF-κB) is activated in the synovium of rats with SCW-induced arthritis and that inhibition of the activity of this transcription factor enhances synovial apoptosis, which is consistent with the potential involvement of NF-κB in synovial hyperplasia [ 27 ]. In accord with these observations, the microarray data showed early upregulation of genes involved in the NF-κB signalling pathway, such as NF-κB1 (p50 or p105), NFKBIA (IκBα), TNF-α, TNFRSF1a and TNFRSF1b, suggesting a possible regulatory role of NF-κB in the transcription of genes that mediate disease progression in SCW-induced arthritis. Histopathological studies in arthritic rat joints from the reactivation SCW-induced arthritis model have shown that only moderate histological changes in articular cartilage, with few erosive effects on bone, occur at early stages in the flare reaction (day 3), whereas evident cartilage degradation is observed at later time points (20 days after intravenous challenge with SCW) [ 4 ]. The microarray data suggest that tissue remodelling is an active process in this model because abundant expression of collagen-related genes (Col5A2, Col5A3, Col12A1 and Col18A1), enzymes that degrade matrix molecules such as MMPs and the aggrecanase ADAMTS-1 (a disintegrin-like and metalloproteinase with thrombospondin type 1 motif, which is capable of cleaving versican), together with other genes that control ECM turnover and breakdown (TIMP1, PLAU [plasminogen activator, urokinase], PLAU receptor [PLAUR]), were found to be upregulated in arthritic joints. MMP-3 (stromelysin) appears to be pivotal in the activation of collagenases, whereas MMP-13 is crucial in collagen breakdown [ 28 ]. The PLAU/PLAUR system plays a critical role in cartilage degradation during osteoarthris by regulating pericellular proteolysis mediated by serine proteases [ 22 , 29 ]. The complement system has also been reported to participate in tissue injury during inflammatory and autoimmune diseases [ 30 ], and ficolins can initiate the lectin pathway of complement activation through attached serine proteases (Mannan-binding lectin serine proteases [MASPs]) [ 31 ]. Interestingly, the microarray data revealed significant upregulation of the first complement component C1, which exerts collagenolytic activity in addition to the role it plays in the classic cascade [ 29 ]. In addition, upregulation of the expression of C2, C3, ficolin B (FCNB) and MASP1 was also noted, supporting the concept that activation of the complement system, together with the imbalance between MMPs, TIMPs and other related molecules, could mediate cartilage destruction in this experimental model of RA. In our analysis we also identified 10 genes that are differentially expressed in arthritic joints and that that map to genomic regions previously reported to be QTLs for autoimmune diseases. Although it is premature to suggest that the 10 genes are candidates for these QTLs, our observations suggest that expression of these genes may influence the onset, severity and/or susceptibility to arthritis in this animal model. Of particular interest is KDAP (napsin) because of the high fold increase in gene expression observed in arthritic joints from SCW-injected animals (D = 48.2 on day 3). This aspartic protease was shown to be expressed in kidney, lung and lymphoid organs of mice [ 32 ], and it has been suggested that it functions as a lysosomal protease involved in protein catabolism in renal proximal tubules [ 33 ]. However, little is known about the role of KDAP in other organs and tissues. Interestingly, human KDAP resides on chromosome 19q13.3–19q13.4, a region previously identified to be involved in susceptibility to autoimmune diseases, including systemic lupus erythematosus, multiple sclerosis and insulin-dependent diabetes mellitus [ 34 , 35 ]. Our results show, for the first time, that KDAP gene expression is upregulated in experimental arthritis tissue, and suggest that further characterization is required to unravel the biological/pathological activities of this gene in RA. The microarray data also revealed high upregulation in runt-related transcription factor 1 (RUNX1) and a group of transporter genes (SLC11A1, SLC13A3, SLC1A3, SLC21A2 [MATR1], SLC28A2, SLC29A3, SLC5A2 and SLC7A7), from which the prostaglandin transporter gene MATR1 exhibited the greatest upregulation on day 3 after intravenous challenge with SCW. The rat MATR1 gene maps to the type II collagen induced arthritis severity QTL6 (Cia6) [ 36 ], and its human orthologue is located within autoimmune disease QTLs for asthma, psoriasis and atopic dermatitis [ 37 - 39 ]. Several authors reported linkage of SLC11A1 (also named NRAMP1) to human RA [ 40 - 42 ]. The Z-DNA forming polymorphic repeat in the RUNX1-containing promoter region of human SLC11A1 may contribute to the differing allelic associations observed with infectious versus autoimmune disease susceptibility [ 43 ]. Recent studies reported that regulation of expression of organic cation transporter gene SLC22A4 by RUNX1 is associated with susceptibility to RA [ 44 ]. Other transporter genes (SLC12A8 and SLC9A3R1) have also been linked to susceptibility to other autoimmune diseases such as psoriasis [ 45 ]. These observations together suggest that RUNX1 and the transporter genes found to be differentially expressed in arthritic joints may contribute to arthritis susceptibility and to the inflammatory processes that mediate the pathology of this model. Conclusion The present study identified the temporal gene expression profiles of hundreds of genes, including cytokines, chemokines, adhesion molecules, transcription factors, apoptotic and angiogenesis mediators, whose expression is associated with onset and progression of arthritis pathology in rat joints from the reactivation model of SCW-induced arthritis in Lewis (LEW/N) rat. This transcript profiling offers not only the optimal kinetics of expression for different potential disease biomarkers, but it also improves our understanding of the molecular events that underlie the pathology in this animal model of RA. In addition, although the majority of genes found to be differentially expressed in this model were previously associated with human RA, further genes not previously linked to autoimmune diseases were identified, providing a resource for future research and for the development of new therapeutic targets. Abbreviations ANOVA = analysis of variance; CCL = CC chemokine ligand; CCR = CC chemokine receptor; CXCL = CXC chemokine ligand; CXCR = CXC chemokine receptor; ECM = extracellular matrix; EST = expressed sequence tag; IL = interleukin; MCP = monocyte chemoattractant protein; MHC = major histocompatibility complex; MIP = macrophage inflammatory protein; MMP = matrix metalloproteinase; NF-κB = nuclear factor-κB; NK = natural killer; NOS = nitric oxide synthase; PBS = phosphate-buffered saline; PCA = principal component analysis; PCR = polymerase chain reaction; PG-PS = peptidoglycan–polysaccharide; QTL = quantitative trait locus; RA = rheumatoid arthritis; RT = reverse transcription; SCW = streptococcal cell wall; SLPI = secretory leucocyte protease inhibitor; TIMP = tissue inhibitor of matrix metalloproteinase; TNF = tumour necrosis factor. Competing interests The author(s) declare that they have no competing interests. Authors' contributions RI carried out the study design, in vivo experiments, total RNA extractions, RT-PCR analysis of data and manuscript preparation. CC and SG performed the microarray experiments and statistical analysis of the array data. MD and PL carried out the study design and collaborated in the preparation of the manuscript. Supplementary Material Additional File 1 Excel spreadsheets summarizing all of the genes upregulated (Delta > 1.8 and P < 0.01) and downregulated (Delta < 1.8 and P < 0.01) in ankle joints from SCW-induced arthritis in Lewis (LEW/N) rats on days -13.8 (4 hours after intra-articular injection of SCW), -13 and 3. Data are expressed as the mean fold increase in gene expression (D = Delta) in SCW-injected animals compared with the expression in the corresponding PBS control group, along with P values. Click here for file

D:\keerthana\PMC001xxxxxx\PMC1064886.xml