Datasets:

Hantao

/

ChemImages

like 1

4081_chem202001513-fig-5016.jpg

Scheme 16Open in figure viewerPowerPointPrebiotic formation of diazomethane and a carbamoylation reagent from amino acids to achieve derivatization reagents for canonical RNA nucleosides.87

4081_chem202001513-fig-0003.jpg

Figure 3Open in figure viewerPowerPointg6A and t6A as examples for derivatives of RNA-nucleosides.92

4081_chem202001513-fig-5017.jpg

Scheme 17Open in figure viewerPowerPointPhosphorylation of thio-1 e enables the formation of the corresponding thioanhydronucleoside 46 and subsequent formation of deoxythiopyrimidine ribonucleosides.100

4081_chem202001513-fig-5018.jpg

Scheme 18Open in figure viewerPowerPointMechanism of a highly stereo- and furanoselective pathway towards deoxyribonucleosides 2.66

4081_chem202001513-fig-5019.jpg

Scheme 19Open in figure viewerPowerPointFormation of deoxyapiose nucleosides 50 as exemplary, selective pathways elucidating the precedence of deoxyribose over all further possible sugars.

21842_ansa202100067-fig-0001.jpg

FIGURE 1Open in figure viewerPercentage of papers covered in this annual review using each AIMS technique (above) and percentage of papers in specified fields of research (below)

The diversity of ambient ionization techniques extends far beyond those covered in this section, which solely represents techniques used in the studies explored in this review, as summarized in Table 1 and Figure #ansa202100067-fig-0001#1. The subsequent sections will discuss applications of AIMS in a number of key fields of research throughout 2021, alongside the highlighting of key technological developments within the field.

21842_ansa202100067-fig-0002.jpg

FIGURE 2Open in figure viewerPhotographs of the string sampling probe, a device developed for in situ sampling during endoscopy procedures. Reprinted with permission from Chen et al., 2021.41 Copyright 2021 American Chemical Society

Manoli et al. integrated REIMS with the Harmonic scalpel, a next-generation laparoscopic surgical tool that uses ultrasound to dissect tissues. This was applied to the direct analysis of various porcine tissues, with multiple points sampled across the surface of the tissues. Tissue-specific lipid profiles were produced, enabling the differentiation between tissues from the muscle, liver, colon and small intestine with an accuracy of 100%. The study also evaluated the use of REIMS with other electrosurgical dissection tools, showing distinct mass spectral differences from samples depending on the type of tool used, believed to be due to different mechanisms of aerosolization and droplet formation. Whereas the use of the Harmonic scalpel resulted in high intensities of diglycerides and triglycerides, other surgical tools produced high abundances of glycerophospholipids. Chen et al. developed a string sampling probe designed to be incorporated into endoscopes to perform in vivo chemical analysis during endoscopic procedures. The device consists of medical-grade silk moved by a stepping motor which places the probe in contact with a surface of interest during the procedure (Figure #ansa202100067-fig-0002#2). This contact allows analytes from the surface to adhere to the string and be transported to the ion inlet tube, where analytes are subsequently ionized by an electrospray ion source. In this study, the new technique was demonstrated with the in vivo analysis of gastric mucosa of a mouse using a gastrointestinal endoscope two meters in length. Finally, Fu et al. developed a new method referred to as coupling ultrasonic sputter desorption for the in vitro and in vivo analysis of biological tissues, demonstrating the use of the technique with liver, brain, kidney and lung tissues. The technique promises to be another potential tool for in situ analysis of biological tissues. Whereas the majority of AIMS techniques established for the operating room are intended for manual use by the surgeon, technologies have also been developed for use with surgical robots such as the Da Vinci robot. In a recent study, SpiderMass technology was coupled with a robotic arm for the topographical imaging of biological tissue. This enabled the production of 3D topographical images of various biological matrices including the whole body of a mouse during a post-mortem examination.

21842_ansa202100067-fig-0003.jpg

FIGURE 3Open in figure viewerRepresentative MasSpec Pen mass spectra from meat products, PCA plot differentiating samples, and image showing direct analysis of beef sample using the MasSpec Pen. Reprinted with permission from Gatmaitan et al., 2021.89 Copyright 2021 American Chemical Society

AIMS has furthermore proven to be a beneficial technique in the analysis of food products for the purpose of authentication and quality control. The MasSpec Pen was recently demonstrated in the authentication of meat and fish products. Meat fraud is an increasing concern in some parts of the world, typically seen in the form of the replacement of one type of meat with a cheaper alternative. In this study, direct analysis was performed on several animal products, including grain- and grass-fed beef, venison, salmon and trout. Using classification models, the discrimination of meat was achieved with an accuracy of 100% for the distinction of beef versus venison, 84% for classifying different fish types and 95% for the beef model. With an analysis time of 15 s per sample, an easy-to-use sampling mechanism and high accuracy models, the MasSpec Pen has been shown to be a promising tool for food authentication. REIMS has also been demonstrated in meat analysis. As the global transportation of food worldwide has become commonplace, it is more important than ever before to ensure the appropriate storage conditions of food products throughout their journey. This is particularly true of raw meat products, which must be maintained under specific storage conditions free from significant temperature fluctuations, which can cause problems ranging from poor quality due to structural damage of the meat to safety concerns resulting from degradation. In this study, REIMS was utilized for the lipidomic profiling of fresh and frozen-thawed beef muscle to evaluate the chemical variations caused by consecutive freeze-thaw cycles. Using multivariate analysis, fresh and frozen-thawed samples could be rapidly differentiated with an accuracy of 92–100%, with discrimination primarily based on changes in the content of fatty acids and phospholipids (Figure #ansa202100067-fig-0003#3). Various AIMS techniques have also been utilized to detect the presence of toxins in food products, such as LAESI to detect mycotoxins in mold-infected fruit, DART-MS coupled with a smartphone-based lateral flow to detect mycotoxin contamination and immuno-magnetic blade spray (a coated blade spray-based technique) to detect the marine toxin domoic acid.

21842_ansa202100067-fig-0004.jpg

FIGURE 4Open in figure viewerSample surface before and after analysis by LA-REIMS with the 3D MS scanner and 3D visualization of molecular distributions across the sample surface. Reprinted with permission from Nauta et al., 2021.125 Copyright 2021 American Chemical Society

More recently, imaging technology has developed to enable topological imaging of 3D objects. A laser-assisted REIMS technique was recently developed to profile uneven surfaces and produce 3D maps. With this method, the sample is automatically moved in relation to a surgical laser and the MS inlet, allowing rapid chemical profiling of multiple points across the surface of the sample. In this study, the technique was applied to the analysis of biological materials, such as a human femoral head, marrowbone, and apple, to demonstrate its applicability to large, uneven surfaces with complex matrices (Figure #ansa202100067-fig-0004#4). With the automated system, a 3D image composed of hundreds of datapoints could be produced in as little as 1 h. 3D MSI techniques, such as laser-assisted REIMS technique, could allow for the automated chemical profiling of large and uneven surfaces, providing new insight into molecular distributions across biological materials. This kind of information could be particularly beneficial in understanding the chemical basis of structural changes in biological materials such as osteoarthritis in bones.

21842_ansa202100067-fig-0005.jpg

FIGURE 5Open in figure viewerFigures depicting the design and appearance of the MasSpec Pointer (A-D), the waveform of the pulsed high-voltage of the device (E), and the use of use of the pointer (F). Reprinted with permission from Li et al., 2021.160 Copyright 2021 American Chemical Society

A primary advantage of AIMS is the potential to take the instrument into the field to perform rapid on-site testing, yet many techniques have not yet been validated in such a way that they are usable outside of the laboratory. As such, recent developments in AIMS have focused on the improvement of existing ambient technologies to improve performance in addition to creating user-friendly devices that could feasibly be used for remote analysis. Currently, many AIMS techniques require high-voltage power supplies, gases and carefully controlled setups to operate. In order for ambient ionization to translate to the field, various aspects of existing technologies need to be modified and refined. Li et al. recently aimed to resolve some of these requirements by developing a hand-powered ion source. The device is based on a piezo-electric element to achieve the transformation of hand mechanical energy into electrical energy and, in this study, was able to generate voltages of over 10 kV. The device was used to power several ion sources, including paper spray and nanoelectrospray, for the analysis of small molecules. Although just a proof-of-concept study, the research does demonstrate potential alternatives to ion sources requiring power supplies, though the mass detector coupled to the ion source would still require mains power. The same group also developed the MasSpec Pointer, a hand-held and wireless ion source based on arc discharge, a device that was produced from a commercially available electronic lighter (Figure #ansa202100067-fig-0005#5). Stelmack et al. evaluated the use of PSI coupled with a portable mass spectrometer under field conditions, specifically to assess the ruggedness of the technique under environmental conditions such as different wind speeds and directions in addition to a range of temperature and humidity levels. The study concluded that wind speeds greater than approximately 7 mph had substantial effects on signal intensity and duration, and certain wind directions (particularly a flow perpendicular to the ion source) were particularly problematic. The study explores some of the vulnerabilities of ambient ionization sources in the field and highlights points for consideration when moving from the laboratory into the field.

26069_celc202101135-fig-0001.jpg

Figure 1Open in figure viewerPowerPointPlatinum electrochemical active surface area (ESCA) as a function of Ar flow applied during electrodeposition. The ECSA of platinum was determined from the adsorbed hydrogen oxidation charge, in 0.1 m HClO4, applying several Ar flow rates (see colour code for FlowCV ranging from 0 to 40 mL⋅min−1), providing several data points for each flow condition during Pt electrodeposition (FlowPt-dep). The values are normalized to the geometrical surface area of the exposed hollow fibre electrode.

While qualitatively the Pt distribution observed in the SEM images can be understood on the basis of Figure #celc202101135-fig-0003#3, we will now discuss the flow rate dependent trend shown in Figure #celc202101135-fig-0001#1 for the obtained PtESCA, with the aid of Figure #celc202101135-fig-0004#4. Figure #celc202101135-fig-0004#4 shows a schematic representation of gas-liquid distributions present in various stages of stagnant, or exiting gas for a certain pore size. The pore size is determined by the distance between two Ti-particles. Stage A shows most of the surface of the Ti particles is in contact with liquid. This is likely the situation representing i) no exiting flow at all, or ii) the smaller pores at low gas velocities (when these do not participate in the exiting flow). Stages B to D show the transient in gas-liquid-solid contact occurring during the exiting of a bubble, implying that the average time-dependent extent of contact between the electrolyte and the particle is significantly smaller than in situation A. In stage D the velocity of the bubble induces liquid convection, enhancing mass transfer of dissolved species towards the surface of the electrode. A clear effect of flow is readily observed in Figure #celc202101135-fig-0005#5A, showing an increasingly larger limiting current as a function of increasing gas velocity. In Figure #celc202101135-fig-0005#5B the current found at 1.2 V vs RHE is plotted as a function of gas velocity. Initially, from 0 to 5 mL⋅min−1, the effect of flow on increasing the current is the largest, which can be explained by considering that any gas purge will induce convection near the HFE. As previously explained for the curve associated with the deposition of Pt (Figure #celc202101135-fig-0001#1), the degree of effective mixing is determined by two factors; i) the participation of a certain fraction of the pore sizes present near the surface of the fibre, and ii) the flux through a pore. Increasing the flow rate in the range from 5 to around 20 or 25 mL⋅min−1 allows participation in exiting gas of not only the larger, but also the predominantly present smaller pores. Contrary to the obtained PtESCA, above 25 mL⋅min−1 to at least 40 mL⋅min−1 the current linearly increases. It would therefore seem these gas fluxes still enhance the mass transfer by intensified mixing of the electrolyte, which is relevant for the kinetically favourable Fe(II) oxidation reaction as is summarized in Figure #celc202101135-fig-0006#6. The increasing gas flux, above values when the degree of pore participation approximates unity, likely affects the value of the so-called Nernst diffusion layer, for example used to describe the hydrodynamics involved in rotating disc electrodes (RDEs).

26069_celc202101135-fig-0002.jpg

Figure 2Open in figure viewerPowerPointPlatinum coverage of the titanium hollow fibre electrode. A) SEM image of deposit obtained without gas flow through the electrode wall. B) SEM image of deposit obtained when 30 mL⋅min−1 gas flow was applied through the electrode wall. C) HR-SEM ESB image of the cross section of the Pt/Ti HFE obtained using 30 mL⋅min−1 of flow through the electrode wall.

Representative SEM images of the platinized Ti-HFEs are presented in Figures #celc202101135-fig-0002#2A and 2B, revealing two types of deposition. The scattered and partial coverage of Pt on Ti as shown in image A is obtained by electrodeposition in the absence of gas flow. Image B shows the Pt coverage when 30 mL⋅min−1 was used. Generally, Pt is deposited on the outer wall of the hollow fibre, while a limited amount of small Pt particles deposits in the interior of the pore structure up to ≈12 μm inward, as can be deduced from the image shown in Figure #celc202101135-fig-0002#2C. Such distribution is likely the result of the limited depth of penetration (≈12 μm) of the PtCl62− solution into the pore structure, a depth similar to the dimensions of the two outer-most Ti particles. Cross sectional and surface images where taken for all of the samples and are available in the supporting information, Figures S8 to S25. For the samples from 5 to 20 mL⋅min−1 different surface regions could be distinguished, showing one or the other type of surface coverage (Figure #celc202101135-fig-0002#2A or Figure #celc202101135-fig-0002#2B). We will now discuss why an apparent bimodal distribution is obtained at low flow rates, gradually transforming to a homogeneous deposition alike the image shown in Figure #celc202101135-fig-0002#2B. Situation A leads to the scattered distribution of the Pt particles on a single Ti particle, observed in Figure #celc202101135-fig-0002#2A, and the accompanying relatively low PtESCA. Nucleation of Pt in this situation apparently occurs on preferred locations of the Ti particle(s). During electrodeposition the electrolyte close to the electrode is depleted of Pt ions, resulting in concentration polarization. In other words, deposition is limited by transport of the PtCl62− ions by diffusion through the stagnant liquid boundary layer towards the surface. When gas is effectively exiting a pore (the case at low gas velocities only for the larger pore sizes, and at e. g. 30 mL for most of the pores), a more homogeneous coverage of Pt is obtained (in agreement with the observed larger PtESCA). This is the result of the effective transport of the PtCl62− ions towards the surface by convection, overcompensating the lower, time averaged, extent of contact between the liquid and the solid in these conditions.

26069_celc202101135-fig-0003.jpg

Figure 3Open in figure viewerPowerPointPore size distribution of the Ti-HFE used in this work and the estimated degree (Flow %, left axis) to which each pore contributes to the exiting flow when applied in electrochemistry (see colour code). The red curve shows the cumulative of these percentages (right axis). Per example, the graph shows that at 5 mL⋅min−1 (see red colour code) pores in the range of 1.65 to 3.0 μm participate in the exiting flow. The smaller pores, constituting the largest fraction of the total in a narrow size window, start to participate at higher flow rates, demonstrating a steep cumulatively rising contribution (in flow %) in the flow range of 10 to 25 mL⋅min−1.

The values applied in Equation (5) hold for aqueous electrolyte and titanium oxide (the presumed surface composition on the basis of the XPS data). Please see the supporting information, including Figures S26–S30) for more detail on how the Laplace equation was applied, and which assumptions and data representations were used. Within Figure #celc202101135-fig-0003#3 two representations of the fraction of pores participating in exiting flow are provided, i. e. as a value in percentage (left axis) of the total flow corresponding to a certain pore diameter, and the cumulative contribution (right axis) from large (3.0 μm), to small pores (≈1.25 μm). Figure #celc202101135-fig-0003#3 shows that the dominantly present pore sizes are in the range of 1.5 to ≈1.7 μm). While qualitatively the Pt distribution observed in the SEM images can be understood on the basis of Figure #celc202101135-fig-0003#3, we will now discuss the flow rate dependent trend shown in Figure #celc202101135-fig-0001#1 for the obtained PtESCA, with the aid of Figure #celc202101135-fig-0004#4. Figure #celc202101135-fig-0004#4 shows a schematic representation of gas-liquid distributions present in various stages of stagnant, or exiting gas for a certain pore size. The pore size is determined by the distance between two Ti-particles. Stage A shows most of the surface of the Ti particles is in contact with liquid. This is likely the situation representing i) no exiting flow at all, or ii) the smaller pores at low gas velocities (when these do not participate in the exiting flow). Stages B to D show the transient in gas-liquid-solid contact occurring during the exiting of a bubble, implying that the average time-dependent extent of contact between the electrolyte and the particle is significantly smaller than in situation A. In stage D the velocity of the bubble induces liquid convection, enhancing mass transfer of dissolved species towards the surface of the electrode. Furthermore, the phenomena shown in Figure #celc202101135-fig-0003#3 B–D likely reduce the thickness of the boundary-layer (also referred to as the Nernst diffusion layer). While less influential on the overall Pt coverage across the HFE (likely due to deposition of Pt on already present Pt particles/structures), an increasing flux at the highest gas flow rates does explain the slightly increasing trend in the PtECSA above 20 mL⋅min−1 (Figure 1) which can be interpreted as increased surface roughness (SEM images in the supporting information, Figure S8–S25).

26069_celc202101135-fig-0004.jpg

Figure 4Open in figure viewerPowerPointSchematic representation of the changing electrode-electrolyte contact area induced by gas bubbles forming in, and exiting from a pore of the HFE. The grey circles represent the Ti particles at the HFE surface, the blue surrounding is the electrolyte and the white area represents the gas phase. The black arrow indicates the direction of gas flow out of the pore. A) Initially the electrolyte occupies the pore. B) The electrolyte is displaced by the gas phase, when the pressure on the inside of the HFE is sufficiently large. C) A gas bubble occupies a significant part of the surface of the HFE. D) Desorption of the bubble results in convective mixing and re-establishment of the liquid-solid interface. The circular arrows represent this convective mixing induced by release of the gas bubble.

While qualitatively the Pt distribution observed in the SEM images can be understood on the basis of Figure #celc202101135-fig-0003#3, we will now discuss the flow rate dependent trend shown in Figure #celc202101135-fig-0001#1 for the obtained PtESCA, with the aid of Figure #celc202101135-fig-0004#4. Figure #celc202101135-fig-0004#4 shows a schematic representation of gas-liquid distributions present in various stages of stagnant, or exiting gas for a certain pore size. The pore size is determined by the distance between two Ti-particles. Stage A shows most of the surface of the Ti particles is in contact with liquid. This is likely the situation representing i) no exiting flow at all, or ii) the smaller pores at low gas velocities (when these do not participate in the exiting flow). Stages B to D show the transient in gas-liquid-solid contact occurring during the exiting of a bubble, implying that the average time-dependent extent of contact between the electrolyte and the particle is significantly smaller than in situation A. In stage D the velocity of the bubble induces liquid convection, enhancing mass transfer of dissolved species towards the surface of the electrode.

26069_celc202101135-fig-0005.jpg

Figure 5Open in figure viewerPowerPointA) Current as a function of potential in the FeII oxidation to FeIII, using 10 mm K4Fe(CN)6 in 0.5 m KCl. The dependency of the limiting current on Ar flow rate is clearly visible in the voltage range of 1.1 to 1.4 V. B) Current at 1.2 V vs RHE, from (A), as a function of Ar flow rate (FlowCV).

A clear effect of flow is readily observed in Figure #celc202101135-fig-0005#5A, showing an increasingly larger limiting current as a function of increasing gas velocity. In Figure #celc202101135-fig-0005#5B the current found at 1.2 V vs RHE is plotted as a function of gas velocity. Initially, from 0 to 5 mL⋅min−1, the effect of flow on increasing the current is the largest, which can be explained by considering that any gas purge will induce convection near the HFE. As previously explained for the curve associated with the deposition of Pt (Figure #celc202101135-fig-0001#1), the degree of effective mixing is determined by two factors; i) the participation of a certain fraction of the pore sizes present near the surface of the fibre, and ii) the flux through a pore. Increasing the flow rate in the range from 5 to around 20 or 25 mL⋅min−1 allows participation in exiting gas of not only the larger, but also the predominantly present smaller pores. Contrary to the obtained PtESCA, above 25 mL⋅min−1 to at least 40 mL⋅min−1 the current linearly increases. It would therefore seem these gas fluxes still enhance the mass transfer by intensified mixing of the electrolyte, which is relevant for the kinetically favourable Fe(II) oxidation reaction as is summarized in Figure #celc202101135-fig-0006#6. The increasing gas flux, above values when the degree of pore participation approximates unity, likely affects the value of the so-called Nernst diffusion layer, for example used to describe the hydrodynamics involved in rotating disc electrodes (RDEs). In Figure #celc202101135-fig-0007#7 the cyclic voltammetry of the HER is shown for a platinized Ti-HFE, varying the Ar flow rate in a range from 0 to 40 mL⋅min−1. In Figure #celc202101135-fig-0007#7A multiple changes with increasing flow rate can be seen in the H2 evolution region. In the experiment performed without any gas flow, H2 evolution starts at a negative potential of ≈ −0.02 V, and some of the (surface adsorbed) H2 is reversibly oxidized once the scan direction is reversed, explaining the peak at approximately ≈−0.03 V (compare Figure S2). However, when a flow is applied through the electrode wall, these two specific features are affected. First, the onset potential for reduction shifts to slightly positive potentials (≈0.01 V). Second the peak of oxidation of H2 has disappeared. In Figure #celc202101135-fig-0007#7B the reductive current is plotted against flow rate (at −30 mV), at first sight showing a less linear trend than observed for the oxidation of Fe(II). However, when the trend above approximately 15 mL⋅min−1 gas flow is considered, the increase in current (by approximately 37–33=4 mA) is actually similar to observed in Figure #celc202101135-fig-0005#5B (by approximately 9–5=4 mA). The initially very large increase in current is likely related to forced removal of H2 gas from the surface by the exiting inert gas, while the second phase of the curve is related to i) the increasing participation of smaller pores of the fibre, and ii) convective mixing, enhancing proton transport towards the surface. Both phenomena are illustrated in Figure #celc202101135-fig-0008#8.

26069_celc202101135-fig-0006.jpg

Figure 6Open in figure viewerPowerPointSchematic representation of the [Fe(CN)6]4− to [Fe(CN)63− electrochemical oxidation reaction on the platinized (Orange) Ti-HFE (Grey) and the relevant mass transfer (Indicated by the arrows) at an active pore.

A clear effect of flow is readily observed in Figure #celc202101135-fig-0005#5A, showing an increasingly larger limiting current as a function of increasing gas velocity. In Figure #celc202101135-fig-0005#5B the current found at 1.2 V vs RHE is plotted as a function of gas velocity. Initially, from 0 to 5 mL⋅min−1, the effect of flow on increasing the current is the largest, which can be explained by considering that any gas purge will induce convection near the HFE. As previously explained for the curve associated with the deposition of Pt (Figure #celc202101135-fig-0001#1), the degree of effective mixing is determined by two factors; i) the participation of a certain fraction of the pore sizes present near the surface of the fibre, and ii) the flux through a pore. Increasing the flow rate in the range from 5 to around 20 or 25 mL⋅min−1 allows participation in exiting gas of not only the larger, but also the predominantly present smaller pores. Contrary to the obtained PtESCA, above 25 mL⋅min−1 to at least 40 mL⋅min−1 the current linearly increases. It would therefore seem these gas fluxes still enhance the mass transfer by intensified mixing of the electrolyte, which is relevant for the kinetically favourable Fe(II) oxidation reaction as is summarized in Figure #celc202101135-fig-0006#6. The increasing gas flux, above values when the degree of pore participation approximates unity, likely affects the value of the so-called Nernst diffusion layer, for example used to describe the hydrodynamics involved in rotating disc electrodes (RDEs).

26069_celc202101135-fig-0007.jpg

Figure 7Open in figure viewerPowerPointA) The I−V curves of hydrogen evolution on a platinized titanium hollow fibre electrode (50 mV⋅s−1) recorded at several Ar flow rates, in 1.0 m HClO4. B) The current at −30 mV vs RHE as a function of Ar flow rate through the electrode wall.

In Figure #celc202101135-fig-0007#7 the cyclic voltammetry of the HER is shown for a platinized Ti-HFE, varying the Ar flow rate in a range from 0 to 40 mL⋅min−1. In Figure #celc202101135-fig-0007#7A multiple changes with increasing flow rate can be seen in the H2 evolution region. In the experiment performed without any gas flow, H2 evolution starts at a negative potential of ≈ −0.02 V, and some of the (surface adsorbed) H2 is reversibly oxidized once the scan direction is reversed, explaining the peak at approximately ≈−0.03 V (compare Figure S2). However, when a flow is applied through the electrode wall, these two specific features are affected. First, the onset potential for reduction shifts to slightly positive potentials (≈0.01 V). Second the peak of oxidation of H2 has disappeared. In Figure #celc202101135-fig-0007#7B the reductive current is plotted against flow rate (at −30 mV), at first sight showing a less linear trend than observed for the oxidation of Fe(II). However, when the trend above approximately 15 mL⋅min−1 gas flow is considered, the increase in current (by approximately 37–33=4 mA) is actually similar to observed in Figure #celc202101135-fig-0005#5B (by approximately 9–5=4 mA). The initially very large increase in current is likely related to forced removal of H2 gas from the surface by the exiting inert gas, while the second phase of the curve is related to i) the increasing participation of smaller pores of the fibre, and ii) convective mixing, enhancing proton transport towards the surface. Both phenomena are illustrated in Figure #celc202101135-fig-0008#8.

26069_celc202101135-fig-0008.jpg

Figure 8Open in figure viewerPowerPointSchematic representation of the hydrogen evolution reaction (HER) on the platinized (Orange) Ti-HFE (Grey) and the relevant mass transfer phenomena (Indicated by the arrows) at a pore, with the small blue circles representing H2 bubbles.

In Figure #celc202101135-fig-0007#7 the cyclic voltammetry of the HER is shown for a platinized Ti-HFE, varying the Ar flow rate in a range from 0 to 40 mL⋅min−1. In Figure #celc202101135-fig-0007#7A multiple changes with increasing flow rate can be seen in the H2 evolution region. In the experiment performed without any gas flow, H2 evolution starts at a negative potential of ≈ −0.02 V, and some of the (surface adsorbed) H2 is reversibly oxidized once the scan direction is reversed, explaining the peak at approximately ≈−0.03 V (compare Figure S2). However, when a flow is applied through the electrode wall, these two specific features are affected. First, the onset potential for reduction shifts to slightly positive potentials (≈0.01 V). Second the peak of oxidation of H2 has disappeared. In Figure #celc202101135-fig-0007#7B the reductive current is plotted against flow rate (at −30 mV), at first sight showing a less linear trend than observed for the oxidation of Fe(II). However, when the trend above approximately 15 mL⋅min−1 gas flow is considered, the increase in current (by approximately 37–33=4 mA) is actually similar to observed in Figure #celc202101135-fig-0005#5B (by approximately 9–5=4 mA). The initially very large increase in current is likely related to forced removal of H2 gas from the surface by the exiting inert gas, while the second phase of the curve is related to i) the increasing participation of smaller pores of the fibre, and ii) convective mixing, enhancing proton transport towards the surface. Both phenomena are illustrated in Figure #celc202101135-fig-0008#8.

26069_celc202101135-fig-0009.jpg

Figure 9Open in figure viewerPowerPointMass transfer coefficients (KM) in relation to the Ar flow rate through the Pt@Ti-HFE for the oxidation of Fe(II) and the hydrogen evolution reaction. The dashed red line indicates the range of flow rates in which a linear increase in current is observed.

A linear relation between the flow rate and KM is shown in Figure #celc202101135-fig-0009#9 for the oxidation of Fe(II), and only in the range of high flow rates for the formation of H2. The values of KM for the oxidation of Fe(II) are not unlike the order of magnitude reported for rotating disk electrodes at rotation velocities of a few thousands rpm. This suggests that the Nernst diffusion layer obtained by the gas exiting a pore in HFEs is similar to obtained by rotation in rotating disk electrodes, but certainly this requires further systematic investigation. The KM values for the hydrogen evolution reaction are two orders of magnitude smaller than for the oxidation of Fe(II), and thus significantly smaller than typically obtained for RDEs. We speculate this is related to the gas evolution, effecting the fractional surface area in contact with the electrolyte, but, again, this requires further investigation.

26069_celc202101135-fig-0010.jpg

Figure 10Open in figure viewerPowerPointSchematic representation of the HFE assembly in the electrochemical cell.

Electrochemical experiments were performed by positioning the HFE in the centre of a cylindrical glass vessel (42 mm high, 32 mm ID). A concentric platinized titanium mesh was used as a counter electrode and placed around the edge of the vessel. The reference electrode (REF: Ag/AgCl, 3 m NaCl) was attached to the HFE assembly, which was placed in the centre of the vessel. A schematic representation of the cell is provided in Figure #celc202101135-fig-0010#10. The Nernst equation was used to convert the potential vs REF to the RHE scale. A Biologic potentiostat was used to control the potentials at the electrodes in the cell.

12638_cmdc202100467-fig-0001.jpg

Figure 1Open in figure viewerPowerPointAssembly of GDB-17 fragment screening set and discovery of divalent metal transporter inhibitors.

We composed a fragment collection by selecting commercially available molecules with fragment-like properties appearing in GDB-17, a database of all possible organic molecules up to 17 atoms following simple rules of chemical stability and synthetic feasibility. This fragment-based discovery project allowed us to experimentally challenge GDB-17 as a source of diverse fragments for screening, thereby significantly extending our previous drug discovery project with GDB databases, which all had been guided by virtual screening on targets with known pharmacology such as NMDA receptors, nicotinic acetylcholine receptors, glutamate transporters and Janus kinases. As detailed below, design and screening of a GDB-17 derived fragment set resulted in the identification of the trifluoromethylsulfone 1 a and thiophene carboxylic acid 2 as weak inhibitors of DMT1, and of tetrahydrocarbazole (S)-3 as the first inhibitor of ZIP8 (Figure #cmdc202100467-fig-0001#1). While GDB molecules lie well within the size range required for fragments, most of them are highly functionalized and structurally too complex to be considered as realistic synthetic targets. Therefore, we defined subsets containing molecules of reduced complexity with fragment-like, drug-like or ChEMBL-like features. To assemble a diverse collection of GDB-17 fragments for experimental screening, we selected molecules as follows (Figure #cmdc202100467-fig-0001#1): 1) starting from commercially available molecules, retrieve all molecules with up to 17 atoms and following Congreve's rule of 3 (Ro3); 2) eliminate reactive functional groups (FG); 3) select molecules containing at least one saturated trivalent carbon or quaternary center, a structural feature present in most GDB-17 molecules but underrepresented in commercial compounds; 4) check that every molecule selected is indeed present in GDB-17 using a previously reported GDB-17 search tool. We applied this procedure to the catalog of a single commercial provider (Princeton BioMolecular Research, Inc.) and obtained 1,900 fragment-like molecules from GDB-17. A total of 1,676 of these molecules were purchased and actually delivered, providing a collection compatible with our screening capacity for DMT1. To adjust for our limited screening capacity concerning ZIP8 (see below), we performed an additional diversity selection of these 1,676 fragments by clustering and minimizing similarities between nearest neighbors to obtain a smaller subset of 511 fragments.

12638_cmdc202100467-fig-0002.jpg

Figure 2Open in figure viewerPowerPointComparison of purchased fragments with fragments from FDB-17 and with commercial fragments. (a) TMAP (tree-map) layout of the 1,676 purchased fragments (orange+red) including its diversity subset of 511 compounds (red) combined with 1,700 randomly selected fragments from FDB-17 (blue) and from commercial fragments (cyan). Each point corresponds to a molecule. The active compounds are shown in magenta. An interactive version of this TMAP is accessible at https://tm.gdb.tools/fragment_project/tmap/1.7k_pcc_fdb_mcf.html (b) Analysis of molecular shingles from each subset. The shingle counts are given as the average±standard error across 10 different selections of FDB-17 subsets and commercial fragments. (c) Histogram of fsp3C (fraction of sp3 carbon atoms). (d) Nearest neighbor similarity analysis in terms of Jaccard distance (dj) according to the molecular fingerprint MAP4. (e–h) Histograms of HAC (heavy atom count), HBD (hydrogen bond donor atom count), HBA (hydrogen bond acceptor atom count), and CLogP (calculated octanol:water partition coefficient). The standard error in histogram values across the ten FDB-17 subsets (blue lines) and the ten commercial fragment set (cyan lines) were below 1 % for all histograms and are not shown except for the nearest neighbor Jaccard distance histogram.

To assess the properties of our purchased fragments, we compared them with a random selection of 1,700 fragments from FDB-17, a database of 10 million fragments selected from GDB-17, and a second, equally sized random selection from a cumulated set of 18,151 commercially available fragments up to a size of 17 heavy atoms (Figure #cmdc202100467-fig-0002#2). The differences between purchased, commercial and FDB-17 fragments was visible in a TMAP (tree-map) layout computed using the MAP4 fingerprint combining the three compound sets (Figure #cmdc202100467-fig-0002#2a). In this TMAP, the FDB-17 subset was in large part separated from purchased and commercial fragments. This difference is caused by the abundance of 3D-shaped non-aromatic fragments in the FDB-17 subset, which are typical of GDB molecules but not well represented among commercially available molecules, including those in our purchased set. Not surprisingly therefore, the purchased fragments or its diverse subset shared a slightly larger fraction of their respective molecular shingles (circular substructures up to a radius of three bonds) with the commercial fragment subset (44 % or 51 %) than with the FDB-17 subsets (42 % or 46 %) (Figure #cmdc202100467-fig-0002#2b, shingle counts were averaged over 10 different random subsets of FDB-17 and commercial fragments). Furthermore, our selection rule requiring at least one tri- or tetrasubstituted tetrahedral atom in each fragment, to avoid the planarity of commercial fragments, did not shift significantly the histogram of fsp3C (fraction of sp3 carbon atoms) values towards the high values observed in the FDB-17 random subset (Figure #cmdc202100467-fig-0002#2c). Comparing the dissimilarity between nearest neighbors within each set, measured by the Jaccard distance (dj) using the molecular fingerprint MAP4, showed that our purchased subset had many pairs of similar compounds reflecting acquisition from a single commercial provider. However, our diversity subset of 511 compounds showed clearly dissimilar molecules as found in the FDB-17 and commercial fragment sets (Figure #cmdc202100467-fig-0002#2d). The purchased fragments had a similar size distribution to the commercial fragments, while the FDB-17 subset contained relatively smaller molecules due to its flat size distribution in the range 12≤HAC (heavy atom count)≤17 reflecting the sampling procedure used to assemble FDB-17 (Figure #cmdc202100467-fig-0002#2e). Purchased fragments matched commercial fragments in terms of HBD (hydrogen bond donor atom count, Figure #cmdc202100467-fig-0002#2f), but their HBA (hydrogen bond acceptor atoms count) distribution was closer to the FDB-17 subset (Figure #cmdc202100467-fig-0002#2g). This difference in HBA/HBD distribution probably explains the lower polarity of the purchased fragments compared to commercial fragments and the FDB-17 subset visible in the CLogP histogram (calculated octanol:water partition coefficient, Figure #cmdc202100467-fig-0002#2h).

12638_cmdc202100467-fig-0003.jpg

Figure 3Open in figure viewerPowerPointDiscovery of DMT1 and ZIP8 inhibitors by GDB fragment screening. IC50 curves determined for compounds 1 a (a) and 2 (c) using HEK293 cells stably overexpressing DMT1. Cells were pre-incubated for 5 min in the presence of the indicated concentrations of 1 a and 2. Next, radioactive 55Fe2+ was added (1 μM), and cells were incubated for 15 minutes. Each data point represents the Mean±SD (N=6–8) of the 55Fe2+-uptake determined in the presence of the indicated compound concentrations. (b) Structure of analogs 1 b–1 n found to be inactive against DMT1. (d) Representative trace of the Cd2+-flux recorded in HEK293T cells transiently transfected with ZIP8 or the empty vector. Cd2+-flux was monitored using the FLIPR Calcium 5 Assay Kit. A baseline was recorded for 30 sec, then, cells were incubated for 5 minutes in the presence of rac-3 (50 μM), and finally, Cd2+ (5 μM) was added, and the signal (AU) was recorded for 15 minutes. (e) IC50 curve determined for rac-3 measuring the Cd2+-uptake in the presence of the indicated compound concentrations. Cd2+-uptake was determined as the Area Under the Curve (AUC) of the change in florescence intensity observed upon substrate addition (459–750 s). Each data point represents the Mean±SD (N=6–8) of the AUC determined in the presence of each compound concentration. (f) Calcein quenching assay at pH 7.4. Ligand (20 μM), Fe2+ (4 μM) and ascorbic acid (400 μM) were preincubated for 5 min. Then, an equal volume of calcein (2 μM) in uptake solution was added to give the final concentrations of 1 μM calcein, 10 μM rac-3 or bipyridine, 2 μM Fe2+ and 200 μM ascorbic acid, which were incubated for 5 min before the fluorescence measurement. Data was obtained from an experiment performed in quintuplicate, and results are presented as the Mean±SD. See Supporting Information for details. (g) Structure of inhibitor rac-3 and analogs 4, 9 and 10. IC50 values in (a), (c) and (e) are the Mean±SD (N=4–6) of the IC50 values calculated from two independent experiments performed in triplicate.

To identify DMT1 inhibitors, we screened the entire library of 1,676 GDB-17 fragments, conditioned as 10 mM stock solutions in DMSO, by monitoring the uptake of radioactive 55Fe in HEK293 cells stably overexpressing the transporter. We used a concentration of 10 μM screening compound, which we judged sufficiently low to avoid non-specific effects on cells but high enough to indicate even a weak inhibition, as described earlier. The assay revealed 162 compounds with only weak activity (15–25 %), which were then retested at higher concentration (50 μM) to observe a stronger effect for confirmation. By repurchasing, purification and retesting, we were able to confirm the activity of trifluoromethylsulfone 1 a (IC50=64.5±1.1 μM) corresponding to a ligand efficiency of LE=0.42 kcal.mol−1, an acceptable value for an initial fragment hit (Figure #cmdc202100467-fig-0003#3a). Unfortunately, testing of 13 purchasable analogs of this hit did not indicate any other active compound, indicating that the relatively weak activity of 1 a was highly sensitive to structural changes (1 b–1 n, Figure #cmdc202100467-fig-0003#3b). We also characterized a second hit, thiophenecarboxylic acid 2, however this compound showed an even lower potency (IC50=525±65 μM, LE=0.35 kcal.mol−1, Figure #cmdc202100467-fig-0003#3c). These modest results were in line with previous efforts in our laboratory showing that DMT1 activity screens give a very low hit rate. To test inhibition of the human zinc transporter ZIP8 (SLC39A8), we established a cell-based assay in HEK293T cells transiently transfected with ZIP8, detecting uptake of divalent cadmium with the FLIPR Calcium 5 Assay Kit (Figure #cmdc202100467-fig-0003#3d). This assay was adapted from a previously published assay for the related ZIP2 using transient transfection because stably expressing the transporter is toxic to the cells. Due to the limited throughput of screening using transiently transfected cells, we performed activity assays with the smaller, diversity selected subset of 511 GDB-17 fragments rather than the entire set and tested directly at a higher concentration (50 μM) to not miss any activity. A first round of screening resulted in 95 compounds with an inhibition of the cadmium flux greater than 50 %. Secondary screening of these initial hits at 25 μM followed by re-screening of fresh solutions of the 10 best secondary hits resulted in tetrahydrocarbazole rac-3 as a confirmed inhibitor (IC50=15.5±1.2 μM, Figure #cmdc202100467-fig-0003#3e). The relative positioning of the two amines in rac-3 suggests that the observed iron uptake inhibition might reflect metal chelation as reported recently for pyrazolyl-pyrimidones. To exclude this possibility, we performed a calcein competition assay, where the quenching of calcein fluorescence indicates the presence of free Fe2+. In contrast to the positive control bipyridine, we could not detect any significant level of iron chelation by rac-3 (Figure #cmdc202100467-fig-0003#3f). Additionally, attempts to detect iron complexation by 1H-NMR were unsuccessful. A survey of twenty-one close analogs of rac-3 (compounds 4–24, Supporting Information Figure S1), which were either retrieved from the complete set of 1,676 fragments or purchased additionally, revealed three analogs with comparable although slightly lower potency than rac-3, showing that the chloro substituent could be replaced by bromo, and that the amino group was essential for inhibition but could be placed slightly differently in the molecule (4, 9 and 10, Figure #cmdc202100467-fig-0003#3g and Supporting Information Figure S2a). We performed an enantioselective synthesis of (R)-3 and (S)-3 by reductive amination from the parent ketone 25 with (R)- or (S)-α-methylbenzylamine to form the corresponding amine (R)-26 or (S)-26 stereoselectively followed by hydrogenation using a previously reported procedure. Testing the individual enantiomers showed that (S)-3 inhibited ZIP8 slightly stronger than (R)-3 (Figure #cmdc202100467-fig-0004#4b and Supporting Information Figure S2b). Considering that the chiral center in 3 is adjacent to the primary amino group, the low enantioselectivity was in line with the SAR study discussed above showing that analogs with a differently placed amino group (9 and 10, Figure #cmdc202100467-fig-0003#3g) retained activity against ZIP8. Both enantiomers of 3 also inhibited ZIP14 to a similar extent.

12638_cmdc202100467-fig-0004.jpg

Figure 4Open in figure viewerPowerPointCharacterization of ZIP inhibition by rac-3. (a) Average Cd2+-flux recorded in the absence or presence of rac-3 in HEK293T cells transiently transfected with DMT1, ZIP2, or ZIP14. Cd2+-uptake was determined as the Area Under the Curve (AUC) of the change in florescence intensity observed upon substrate addition. Results were expressed as % of the Cd2+-flux recorded in the absence of rac-3. Data from three independent experiments performed in triplicate are represented as Mean±SD (N=12–33). (b) Enantioselective synthesis and activity of (R)-3 and (S)-3. Conditions: a) i) (S)-α-methyl-4-methoxybenzylamine, conc. aq. HCl, toluene, reflux overnight, ii) NaBH4, EtOH, −30 °C – RT, overnight, iii) HCl, MeOH, toluene, RT, 60 min. (59 %); b) same as a) using (R)-α-methyl-4-methoxybenzylamine (66 %); c) BCl3, CH2Cl2, 10 °C, overnight (R: 18 %, S: 47 %). (c–d) detoxification gene expression assay with ZIP8 and ZIP14. MT2A gene expression in HEK293T cells transiently transfected with ZIP8 (blue, left panel), ZIP14 (green, right panel) or the empty vector (EV, red) treated for 2 hours with the indicated solutions. Expression of MT2A mRNA was determined by Real-Time PCR. Obtained Ct values for MT2A gene were normalized to the housekeeping gene GAPDH following the ΔCt method. Data was obtained from 5 independent experiments performed in triplicate, and results are presented as the Mean±SD (N=10–13). Statistical differences were determined by t-test or Mann-Whitney U test ((a) Cd2+ vs Cd2++rac-3; (c–d) Zn2+ (10 μM) vs. each other condition), p values are indicated on top of the corresponding graphs.

To evaluate the specificity of rac-3, we tested its inhibitory activity with other human divalent metal transporters available in our laboratory, including DMT1, ZIP2 and ZIP14 (Figure #cmdc202100467-fig-0004#4a). Using the Cd2+-uptake fluorescence-based assay, we observed that DMT1 was not inhibited by 50 μM rac-3. However, while ZIP2 was only weakly inhibited, rac-3 inhibited ZIP14 as strongly as ZIP8 (IC50=10.57±0.54 μM). This cross-inhibition probably reflects the fact that ZIP8 and ZIP14 are evolutionary closely related proteins and belong to the same subfamily within the ZIP family. We performed an enantioselective synthesis of (R)-3 and (S)-3 by reductive amination from the parent ketone 25 with (R)- or (S)-α-methylbenzylamine to form the corresponding amine (R)-26 or (S)-26 stereoselectively followed by hydrogenation using a previously reported procedure. Testing the individual enantiomers showed that (S)-3 inhibited ZIP8 slightly stronger than (R)-3 (Figure #cmdc202100467-fig-0004#4b and Supporting Information Figure S2b). Considering that the chiral center in 3 is adjacent to the primary amino group, the low enantioselectivity was in line with the SAR study discussed above showing that analogs with a differently placed amino group (9 and 10, Figure #cmdc202100467-fig-0003#3g) retained activity against ZIP8. Both enantiomers of 3 also inhibited ZIP14 to a similar extent. To test if the effect of rac-3 on ZIP8 indeed affected the transport of divalent metal ions into cells, we tested the expression of MT2A, a metal detoxification gene that is normally activated if cells are exposed to high levels of divalent metal ions, in the presence or absence of our inhibitor and 10 μM Zn2+. Indeed, expression of the MT2A gene upon exposure to zinc was strongly down-regulated in the presence of rac-3, confirming that our inhibitor strongly reduced divalent metal ion uptake into the cells (Figure #cmdc202100467-fig-0004#4c).

24168_cssc201601583-fig-0001.jpg

Figure 1Open in figure viewerPowerPointSchematic illustration of (a) photocatalytic water splitting, (b) photoelectrochemical water splitting, and (c–e) PV+electrolyzer configurations: (c) wired, (d) partially wired, and (e) wireless configurations.

Hydrogen can be produced from ubiquitous and abundant water by using the electricity generated from renewable energy sources. Electrochemical water splitting has been intensively studied for more than 200 years. In 1789, water electrolysis was first reported by van Troostwijik and Deiman. More than a century later, in the 1920s and 1930s, a large number of large-scale electrolysis plants were built in Canada, Norway, and elsewhere. More efficient electrolyzers were developed in the middle of the 20th century. The cost breakdown for electrolyzers is available in the literature. At the current stage of the process development for water electrolysis, three main objectives have been considered: (1) greater efficiency of large-scale water electrolyzers; (2) water electrolysis in conjunction with renewable energy sources; and (3) integrated photovoltaic (PV) electrolysis. PV electrolysis is further categorized into three: (a) photocatalytic water splitting, (b) photoelectrochemical water splitting, and (c) PV+electrolyzer configurations. A schematic illustration of these processes is presented in Figure #cssc201601583-fig-0001#1. Detailed techno-economic analyses of these PV electrolysis technologies have been carried out to validate their feasibility. A 10 % solar-to-hydrogen efficiency, which corresponds to approximately 8.2 mA cm−2 (geometric), has been considered as a benchmark (or an approximate value of 10 mA cm−2). Photocatalytic (Figure #cssc201601583-fig-0001#1 a) and photoelectrochemical (Figure #cssc201601583-fig-0001#1 b) water splitting directly utilize semiconductor (photon absorber) surfaces for surface redox reactions (hydrogen and oxygen production from water through reduction/oxidation reactions). In these cases, light irradiation on a photon absorber initiates the overall reaction, in which an exciton is generated. After separation of the electron–hole pair, the electron and hole can diffuse to the interface between solid and liquid. Finally, the electron and hole can be transferred into the reduction and oxidation reaction sites, where hydrogen and oxygen evolve, respectively. In contrast, in the PV+electrolyzer configuration, the surface potential, or the Fermi level, of the semiconductor does not equilibrate with the potential of the surface redox reactions. This PV+electrolyzer configuration is classified into three types: wired (Figure #cssc201601583-fig-0001#1 c), partially wired (Figure #cssc201601583-fig-0001#1 d), and wireless configurations (Figure #cssc201601583-fig-0001#1 e).

24168_cssc201601583-fig-0002.jpg

Figure 2Open in figure viewerPowerPointThe square scheme for proton-coupled electron transfer. ET=electron transfer, PT=proton transfer, and CPET=concerted proton–electron transfer. Reproduced from Reference 48 with permission from the Royal Society of Chemistry.

These reaction schemes are presented in Figure #cssc201601583-fig-0002#2.

24168_cssc201601583-fig-0003.jpg

Figure 3Open in figure viewerPowerPointa) The square scheme for proton-coupled electron transfer. ET= electron transfer, PT=proton transfer, and CPET=concerted proton–electron transfer. Reproduced from Reference 48 with permission from the Royal Society of Chemistry. b) Four concerted proton-coupled electron transfer mechanisms for the OER (in red) that occurs at a single surface site (S) and possible sequential proton–electron transfers along the path, which are constructed under the assumption that the charge of the reaction intermediates does not exceed one electron. The positive and negative intermediates are denoted in orange and blue, respectively. Oxygen is evolved in the final step (in gray). Adapted from Reference 46.

On the other hand, the decoupled pathway can be considered for molecular catalysts. Figure #cssc201601583-fig-0003#3 a summarizes the possible pathways and intermediates for the HER by both the decoupled and concerted pathways. The situation is complicated for the OER because ET or PT can be the sole rate-determining step on the inorganic oxide surface. An overall picture of possible pathways is presented in Figure #cssc201601583-fig-0003#3 b. For detailed descriptions of the rate expression using Tafel analysis, the reader is referred to another article.

24168_cssc201601583-fig-0004.jpg

Figure 4Open in figure viewerPowerPointIllustration of the concept of matching water affinity. Adapted from Reference 61.

Although the Debye–Hückel theory and its extension have been identified as promising tools to quantitatively analyze the coefficient, the electrostatic interaction is overemphasized, and other parameters, such as water affinity and volumetric changes, are overlooked. Particularly, a quantitative theoretical discussion of the mean activity coefficient for different ions is under debate, and is thus not practical. Nevertheless, on the basis of the “like sees like” and “matching water affinity” concepts, a qualitative discussion can be attempted. Figure #cssc201601583-fig-0004#4 summarizes the “matching water affinity” concept. The concept correlates the ion size with the mean activity coefficient based on a consideration of the hydration sphere. The ions can be categorized into two types:

24168_cssc201601583-fig-0005.jpg

Figure 5Open in figure viewerPowerPointSchematic representation of water dissociation, formation of M−Had intermediates, the subsequent recombination of two Had atoms to form H2 (magenta arrow), and OH− desorption from the Ni(OH)2 domains (red arrows) followed by the adsorption of another water molecule on the same site (blue arrows). Water adsorption requires the concerted interaction of O atoms with Ni(OH)2 (broken orange spikes) and H atoms with Pt (broken magenta spikes) at the boundary between the Ni(OH)2 and Pt domains. The Ni(OH)2-induced stabilization of the hydrated cations (AC+) (broken dark blue spikes) likely occurs through noncovalent (van der Waals-type) interactions. The hydrated AC+ can interact further with water molecules (broken yellow spikes), which alters the orientation of water as well as the nature and strength of the interaction of the oxide with water. Reproduced from Reference 50 with permission from AAAS.

The HER proceeds through water molecule reduction at alkaline pH, where the dissociation of the O−H bond in the water molecule (or the Volmer step for water molecule reduction) is kinetically sluggish. When the Pt metal electrode was decorated with Ni(OH)2 islands, a drastic improvement in the HER performance was observed, plausibly because the water dissociation was facilitated at the interface around the Ni(OH)x islands. This finding suggests the significance of switching the reactant, which is associated with the reaction pH. Interestingly, when a small amount of LiOH (1 mmol L−1) was added to the 0.1 mol L−1 KOH electrolyte, the HER performance over the Ni(OH)2–Pt electrode was further improved by a factor of approximately two. The improvement was ascribed to the presence of a complex, Ni(OH)2–Li+–OH–H, which presumably enhanced the probability of the water dissociation (Figure #cssc201601583-fig-0005#5). A similar improvement in the HER performance was reported over Ir and Ru, in the order of supporting cations, K+<Li+<Ba2+ in 0.1 mol L−1 KOH electrolyte. For more details on this subject, the readers are referred to the recent review published by the group. These observations indicate that the local hydrolysis phenomenon (O−H bond dissociation) on the negatively charged surface is facilitated by a hydrated cation cluster, which is a function of the charge density of the ion cluster. Interestingly, under similar acidic conditions, cations were shown to influence the HER performance in 0.05 mol L−1 H2SO4+0.05 mol L−1 M2SO4, M=Li+≈Na+>K+<Rb+≈Cs+, although protons are believed to be the direct reactants. As such, local hydrolysis cannot be used to explain this observation. A scientific explanation for the observation has not been identified, and thus, further studies are required.

24168_cssc201601583-fig-0006.jpg

Figure 6Open in figure viewerPowerPointLinear sweep voltammograms over (a) Pt(111) and (b) Au(111) disk electrodes in 0.1 mol L−1 potassium perchlorate at various pH levels (1–13) at a disk-rotation speed of 1600 rpm, recorded at a scan rate of −50 mV s−1. Reproduced from Reference 118 with permission from Macmillan Publishers Ltd (copyright 2013).

A systematic study on the influence of pH on the HER was reported by Markovic and co-workers in 2013. In their study, 0.1 mol L−1 KClO4 (unbuffered) was used as a supporting electrolyte across the entire pH range (pH 1–13), and a single-crystal Pt (111) electrode was employed as a model electrode. The linear sweep voltammograms (LSVs) are depicted in Figure #cssc201601583-fig-0006#6 a on the SHE scale. In the figure, in acidic solutions of pH≤2 and at alkaline pH (pH>10), only sharply increasing cathodic currents are obtained. In contrast, a two-step reduction is observed at near-neutral pH levels. The first reduction is observed below approximately 0 V on the RHE scale, which reaches a constant value at approximately −300 mV on the SHE scale, and a further increase in the reduction current is obtained below approximately −700 mV on the SHE scale. On the other hand, Markovic and co-workers claimed that a microkinetic description, in conjunction with Fick's law for the HER (hydronium ion reduction and water molecule reduction), and the HOR (hydrogen oxidation with a water molecule and hydroxide ion) matches the experimental observation. Similarly to the previous case, the initial reduction events at a lower overpotential and near-neutral pH are attributed to the diffusion-limited hydronium ion reduction. However, in their scenario, a further increase in the reduction currents below approximately −700 mV versus SHE at near-neutral pH levels is attributed to water molecule reduction at near-neutral pH, which kinetically does not depend on the pH. In their study, Au(111) was also investigated under identical conditions, which showed a similar two-step reduction at near-neutral pH, as depicted in Figure #cssc201601583-fig-0006#6 b. As Au is not as active as Pt, the thermodynamics and the consideration of diffusion cannot solely explain the observation, and the contribution of kinetics cannot be ignored. Accordingly, the second reduction event at a larger overpotential in the near-neutral pH solutions should be explained by water molecule reduction, which kinetically does not depend on the pH. The proposed reduction of the water molecule is also supported by the observations for polycrystalline Ni and Ni-modified Pt. Furthermore, the difference between the performance for the HER observed in acidic and alkaline environments clearly suggests the significance of kinetics in the HER even when Pt is used, which implies the occurrence of water molecule reduction. When a typical supporting electrolyte, such as sodium sulfate solution, is used at pH 5, a two-step reduction is observed (Figure #cssc201601583-fig-0006#6 a), in which an overpotential of approximately 500 mV is required to achieve 10 mA cm−2. Once the significance of the reactant switching is realized, the buffered solution can be used as a “supporting electrolyte”. In this case, 10 mA cm−2 is achieved at an overpotential of approximately 120 mV in 0.1 mol L−1 NaH2PO4 when using a Pt disk electrode. The HER in the solutions is predominantly governed by the significant concentration overpotential, which can be minimized by fine tuning of the electrolyte properties. In the optimized electrolyte of 1.5 mol L−1 potassium phosphate (monobasic/dibasic=80:20), an overpotential of only 40 mV is sufficient to achieve 10 mA cm−2. Under these conditions, the performance is much better than that in alkaline solution (0.1 mol L−1 KOH), and it is almost comparable to that in acidic solution (0.1 mol L−1 HClO4). These findings demonstrate the promise of near-neutral pH conditions for the HER. It should be emphasized that the solution resistance originates from the migration of ions, which is highly correlated with the diffusion of ions, as discussed in Section 2. Thus, as expected, the electrolyte composition that maximizes ion diffusion (optimum conditions for the HER in buffered conditions) also minimizes the solution resistivity in buffered near-neutral pH conditions.

24168_cssc201601583-fig-0007.jpg

Figure 7Open in figure viewerPowerPointSteady-state positive-going sweeps of the HER polarization curves of Pt obtained in selected H2-saturated buffered electrolytes. The sweep rate is 10 mV s−1 and the rotating speed is 1600 rpm. The polarization curves have been corrected for the solution resistance. Adapted from Reference 80.

There have been a couple of studies reported regarding the HER at near-neutral pH in buffered solutions that function to maintain the local pH level. Experimentally, it was clearly demonstrated that the use of a buffered solution as a “supporting electrolyte” successfully prevented the diffusion limitation of hydronium ions at near-neutral pH levels. Figure #cssc201601583-fig-0007#7 shows the polarization curves obtained by using Pt in various electrolytes (including acetate, phosphate, and carbonate buffer) on the RHE scale.

24168_cssc201601583-fig-0008.jpg

Figure 8Open in figure viewerPowerPointSimulated kinetic, concentration, and solution resistance overpotentials at −10 mA cm−2 as a function of the NaH2PO4 concentration. Adapted from Reference 125 with permission from the American Chemical Society (copyright 2015).

Thus, the impact of the phosphate concentration on the HER over a wide concentration range was investigated. In Figure #cssc201601583-fig-0008#8, the overpotential distribution for the HER when using a Pt disk electrode in 0.5–3.0 mol L−1 sodium phosphate solution at pH 5 is depicted (at −10 mA cm−2). In the low concentration regime, the HER performance increased with increasing concentration, then reached a maximum at approximately 2.0 mol L−1, and decreased in more highly concentrated solutions, which is a clear volcano-shaped trend. In general, there are at least three critical parameters that affect the overall performance of an electrochemical reaction: The continuity equation can be solved by using the associated parameters and the kinetic description under the steady-state approximation, the solution of which provides the current–potential relationships. The details of the treatment are available in the literature. The calculation results are presented quantitatively in Figure #cssc201601583-fig-0008#8. The analysis predicts that the overall performance is strongly dependent on the concentration overpotential in buffered near-neutral pH conditions: the concentration overpotential exceeds 50 %, and the kinetic overpotential accounts for less than 10 % of the overall potential. These characteristics suggest that the HER in buffered near-neutral pH conditions is predominantly governed by the mass transport of weak acid species. The mass transport during the reaction is not related to the electrode identity but is solely determined by the electrolyte properties; therefore, it is inferred that fine-tuning of the electrolyte, or electrolyte engineering, to minimize the concentration overpotential leads to the improved apparent HER performance in the buffered near-neutral pH solutions. The associated electrolyte properties are as follows:

24168_cssc201601583-fig-0009.jpg

Figure 9Open in figure viewerPowerPointSchematic illustration of the reaction scheme for the HER in densely buffered conditions.

In a typical chemical engineering model, the chemical reaction can be represented by the mass transport and the surface reaction, with the mass balance considered. Figure #cssc201601583-fig-0009#9 shows a schematic illustration of the model that describes the HER in densely buffered conditions.

24168_cssc201601583-fig-0010.jpg

Figure 10Open in figure viewerPowerPointExperimentally observed iR-free overpotential (kinetic and concentration overpotentials) at 10 mA cm−2 over a polycrystalline Pt disk electrode in 0.5 mol L−1 sodium sulfate (pH 4.0), 0.1 mol L−1 NaH2PO4 (pH 4.4), and 1.5 mol L−1 potassium phosphate solution (K1.2H1.8PO4; KH2PO4/K2HPO4=80:20, pH 5.8), recorded at a scan rate of −50 mV s−1, at a disk rotation speed of 3600 rpm and at 298 K. Adapted from Reference 127.

The significance of electrolyte engineering at near-neutral pH for the HER is summarized in Figure #cssc201601583-fig-0010#10.

24168_cssc201601583-fig-0011.jpg

Figure 11Open in figure viewerPowerPointPotential–pH boundaries for the common transition metals (Pourbaix diagram). Reproduced from Reference 151 with permission from Macmillan Publishers Ltd (copyright 2013).

Improving not only the activity but also the stability is a challenge for OER electrodes, particularly in acidic and near-neutral pH environments. A trade-off between the activity and stability is reported for the OER in acidic environments, even for the most active oxides, IrO2 and RuO2. As expected from the Pourbaix diagram shown in Figure #cssc201601583-fig-0011#11, the cationic state of the metals is thermodynamically more favored at lower pH levels, which is consistent with the nature of their dissolution into solution. In addition to the influence of pH (the activity of free H+) on the OER, the supporting ions that coexist in the electrolyte have been reported to have a significant impact on the performance, which needs to be considered to rationalize and improve the OER performance, as will be discussed in the following sections.

24168_cssc201601583-fig-0012.jpg

Figure 12Open in figure viewerPowerPointa) pH dependence of the catalytic current determined at 1.35 V (versus NHE) in various buffer systems. The respective electrolytes consisted of the indicated buffer system at a concentration of 0.1 m. The lines were obtained by means of simulations, assuming that the catalytic activity was proportional to the relative concentration of the base. HEPES=2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid, MES=2-(N-morpholino)ethanesulfonic acid, and TRIS=2-amino-2-hydroxymethyl-propane-1,3-diol. b) The proposed situation that prevails in the CoCat material, that is, catalysis of water oxidation within the bulk of the amorphous oxide materials but proton transfer to an electrolyte buffer base (abbreviated as B−) at the bulk oxide surface. Adapted from reference 162.

A pioneering work was reported by Nocera and co-workers in 2008, which demonstrated a stable OER performance over a Co-based material in phosphate solution at near-neutral pH levels. In their study, a current density of approximately 1.0 mA cm−2 was achieved at 1.24 V versus SHE at pH 7 (ca. 1.65 V versus RHE). Their follow-up study proposed a self-healing mechanism, in which the cobalt cation released into the solution can be redeposited on the electrode during the OER that occurs when the phosphate is present. For more details about the Co–phosphate system, the readers are referred to the tutorial review by the group. Following this study, some groups attempted to elucidate and further improve the performance of Co-based materials at near-neutral pH conditions. Dau and co-workers investigated the OER performance of Co electrodes in various electrolytes (phosphate, carbonate, glycine, TRIS, acetate, HEPES, chloride, borate, and MES), which revealed a similarity between the OER–pH relationship and pH titration (Figure #cssc201601583-fig-0012#12 a). Based on this observation, it was proposed that the deprotonation of the surface proton that is generated during the OER is the limiting process (Figure #cssc201601583-fig-0012#12 b), which is assisted by the supporting ions.

24168_cssc201601583-fig-0013.jpg

Figure 13Open in figure viewerPowerPointa) pH dependence of the onset potential for the oxidation current (Uon,j, red squares) and optical absorption at 510 nm (Uon,A510, blue circles). The solid line represents the standard potential for oxygen evolution. Reproduced from 165 with permission from the American Chemical Society (copyright 2012). b) Plot of pH versus potential showing the pH dependence of the onset potential (Uon,j) defined at 130 μA cm−2 for water oxidation in the presence of the indicated bases at pH ranging from 5 to 9. Adapted from Reference 167.

A systematic investigation was also reported for Mn electrodes. When an unbuffered solution was used as a supporting electrolyte (0.5 mol L−1 Na2SO4), the onset potential for the OER was found to be independent of the solution pH on the SHE scale (Figure #cssc201601583-fig-0013#13 a), which interestingly coincides with the observation for the HER (see Section 3.2). The observed pH independence in their study was ascribed to the formation of Mn3+ from Mn2+. The OER–pH relationship was drastically altered when buffering action was introduced into the system. Particularly, Nocera and co-workers claimed that the self-healing mechanism was also present in the Mn–phosphate system, which results in the high stability of the Mn OER catalysts. Nakamura et al. reported a strong correlation between the OER performance and pKa of the buffering equilibrium: even at similar pH, the performance was significantly dependent on the pKa of the buffering action. Particularly, the larger pKa resulted in an improved OER onset potential (Figure #cssc201601583-fig-0013#13 b and c). This observation indicates the involvement of the deprotonation/protonation process in the rate-determining step, which was also observed for the Co OER catalysts.

24168_cssc201601583-fig-0014.jpg

Figure 14Open in figure viewerPowerPointParameters considered in the fine-tuning of the electrolyte properties (electrolyte engineering); these parameters affect the hydrogen production and dissolved gas conversion reactions.

These parameters are dependent on each other, and they define the ion activity, diffusion coefficient, and diffusion-layer thickness, which in turn determine the mass transport flux of the weak acid and dissolved gasses, as summarized in Figure #cssc201601583-fig-0014#14.

3895_ejoc202100950-fig-0001.jpg

Figure 1Open in figure viewerPowerPointCompounds tested previously as RNA cleaving catalysts.

Bis- and oligoguanidines related to 1 (Figure #ejoc202100950-fig-0001#1) have been shown not only to bind phosphates but also to accelerate nucleophilic displacement reactions quite effectively. Catalysts of this type have been applied already to manipulate ribonucleic acids. Hydrolytic cleavage of RNA in most cases results from nucleophilic attack of the 2’ hydroxy group forming a 2’,3’ cyclic phosphate and a free 5’ OH of the second fragment. The intramolecular nature of this attack accounts for the large rate increase when compared to analogous intermolecular reactions. The unmodified bisguanidine 1, however, failed as a catalyst for RNA cleavage. The role of guanidines in such reactions may be threefold: as a general base to deprotonate 2’ OH, as an electrophile/general acid to stabilize the pentavalent transition structure at the phosphorus atom and as a general acid for leaving group protonation (5’ OH). With pKa values around 14, normal guanidines will stay protonated at pH 7 and, in consequence, behave as poor general acids or bases in neutral aqueous solution. Better catalysts, therefore, can be expected from guanidine analogs with pKa values shifted towards neutrality. Consistent with that idea, acyl guanidines with structures similar to 1 have been reported to cleave RNA and RNA models more effectively. An alternative approach to lower the pKa values of amidines and guanidines is to incorporate them into heterocyclic structures (Figure #ejoc202100950-fig-0001#1). We previously investigated derivatives of 2-aminobenz-imidazoles (2, pKa≈7.0) and 2-aminopyridines (3, pKa≈6.5). Compound 2 is active and finally led us to the development of the tris(2-aminobenzimidazole) 4. Conjugates 5 of this molecule with DNA and PNA cleave complementary RNA strands with high sequence specificity and substrate half-lives in the range of 3.5–20 h. Surprisingly, compound 3 failed as RNA cleaving catalyst, in spite of similar pKa values. We discussed the lack of rotational symmetry around the bond connecting the heterocycle and the exocyclic nitrogen as a possible reason: Binding to phosphate ions requires two parallel NH groups of the protonated heterocycle to form hydrogen bonds. Unfavorable conformational equilibria thus may prevent substrate binding and catalysis (Figure #ejoc202100950-fig-0001#1).

3895_ejoc202100950-fig-0002.jpg

Figure 2Open in figure viewerPowerPointStructures of compounds 6–12.

To test the conformation hypothesis, we have investigated the RNA cleaving potential of some simple heterocycles (6–12, Figure #ejoc202100950-fig-0002#2), all containing the structural element of amidines or guanidines. Conformations unfavorable for catalysis as in the case of 3 do not exist in compounds 6–12. Cleavage experiments look simple, but should be interpreted with caution: The energy differences for each compound are calculated by subtracting the energy of the tautomer shown in the Figure #ejoc202100950-fig-0002#2, Figure #ejoc202100950-fig-0005#5 and Figure #ejoc202100950-fig-0006#6 of its corresponding ketimin tautomer. Compound 11, 18–23, 25 and 26 are secondary ketimines and for them just another secondary ketimin tautomer exists. As expected all energy differences are≥0.

3895_ejoc202100950-fig-0003.jpg

Figure 3Open in figure viewerPowerPointRNA substrates 13–15 and accessory oligonucleotides 16–17. The RNA part of 15 is built from enantiomeric nucleotides.

The dye labeled RNA 13 (Figure #ejoc202100950-fig-0003#3) was incubated for 20 h at 37 °C and pH 7 with test compounds in concentrations of 10 mM. Afterwards, fluorescently labeled RNA fragments were separated with an ALFexpress II sequencer and the corresponding signals integrated as reported before. Such reaction mixtures are far from being ideal solutions. Instead they can be heavily aggregated, depending on the nature of the heterocycle (see below). Standard loading buffers may precipitate the sample when it is transferred to the separation gel thus preventing analysis. In addition, many runs showed increased cleavage between pyrimidines and adenosines, a pattern typical for minor contaminations with natural RNases. Exact quantification of weak cleavage activities is hardly possible under such conditions. Therefore, all data shown in Table 1 were finally obtained from assays using enantiomeric L-RNA 14, known to be stable against natural RNases. Achiral compounds 6–12, in contrast, are unable to distinguish between both enantiomeric forms. A freshly purified sample of enantio RNA 14 still contained detectable traces of cleavage fragments in the 14mer ribo part, adding up to 0.9 % of the total peak area. When incubated at pH 7 for 20 h at 37 °C the combined area of fragments rose to 1.03 – 1.06 %. This value was unchanged in the presence of compounds 7–9 or 11 (10 mM) or by addition of DMSO (20 %). Increased cleavage was only seen with compounds 10, 12 and in particular with 2-aminoimidazole 6. 2-Aminopyridine 7, although the pKa (6.7) comes close to the ideal value of 7, did not show noticeable RNA cleavage. This observation clearly disproves our initial attempt to explain the catalytic incompetence of compound 3. The lacking activity of compounds 8 and 11 at pH 7 results from their unfavorable pKa. However, melamine 9 remained almost inactive even at pH 6.1 and 5.1. Matching the buffer pH and the pKa of the catalyst is important: When aminoimidazole 6 was tested at pH 6.1, the amount of RNA cleavage dropped from 3.1 % to 0.8 %.

3895_ejoc202100950-fig-0004.jpg

Figure 4Open in figure viewerPowerPointHeterocyclic guanidine analogs, when acting as proton shuttles, are converted into tautomeric forms that may be unfavorable as shown in the case of compound 7.

An alternative view on catalytic efficiencies starts to ask for the protonation state of the reacting phosphate esters during RNA hydrolysis. Computational and experimental methods have determined values of 8–9 as first pKa and approximately 14 as second pKa of equatorial phosphorane hydroxy groups. It is thus reasonable to consider protonation of dianionic phosphoranes in the pH region of 7–8: Breslow suggested a proton transfer from 2’-OH to the phosphorane to occur in the mechanism of RNase A. Detailed QM/MM simulations have supported this idea and assigned the role of the proton shuttle to His12. A mechanism proposed by Cleland for the pH-independent nonenzymatic cleavage of RNA assumes a water molecule acting simultaneously as a general acid and base to transfer a proton from 2’-OH to the pentavalent phosphorane. While computational studies have emphasized the importance of this proton transfer step, the role of water as a proton shuttle has been supported by some but not all authors. However, as Lönnberg pointed out, it may well be that heterocyclic guanidine analogs function as proton shuttles. Proton inventory studies have shown that even weakly acidic guanidinium ions can contribute to catalysis by protonation of the phosphorane. Heterocyclic guanidines acting as proton shuttles would be tautomerized to form an exocyclic imine as shown in Figure #ejoc202100950-fig-0004#4. This mechanism is feasible in such cases only when the energy difference of both tautomers is not too large. Fast semiempirical calculations are sufficient to get a first estimate of the energy (heat of formation) differences of tautomers (Table 1; see below for a comparison with more advanced methods) and to find a weak correlation with the catalytic potential of the heterocycles: Smaller values (15.7–43.7 kJ mol−1) are obtained for the active compounds 6, 10 and 12, larger energy differences (63.4–71.0 kJ mol−1) for compounds 7–9. Compound 11, however, has a smaller energy difference than 6 and 12 but is nevertheless inactive due to its unfavorable pKa. Furthermore, the sterically less hindered aminoimidazole 6 is a much better catalyst when compared to the benzimidazole analog 10, in spite of its less favorable ΔrH°298 value. This shows that the energy difference alone cannot predict the catalytic potential but may be one criterion in addition to pKa values, steric effects and the tendency of compounds to aggregate. 2-Methylaminoperimidine 25 was active in the same concentration range as 24 but, possibly due to steric hindrance, RNA degradation was much slower. Almost no reaction was seen with 2-dimethylaminoperimidine 26. To form stable ion pairs with phosphates, protonated guanidine analogs must have two parallel NH groups not available in compound 26. The structure also does not allow proton shuttling as shown in Figure #ejoc202100950-fig-0004#4. To synthesize conjugates of 2-aminoperimidines with oligonucleotides, attachment as 2-alkylamino derivatives in analogy to compound 4 would be attractive. The diminished activity of such derivatives, however, motivated us to investigate other linking modes. As expected, for all perimidines 27–31 tautomeric forms with similar energy differences were found (Table 3). The aim of the present study is to identify heterocyclic amidines and guanidines as building blocks of improved artificial ribonucleases related to conjugate 5 – and to understand why some are better catalysts than others. Parameters we consider relevant are the pKa values, the steric demand and the hydrophobicity of the compounds: The weakly basic candidates such as 8, 9, and 11 all fail in the cleavage assay. In addition, we have found a weak correlation between catalytic activity and the energy difference between amino and imino tautomers (Figure #ejoc202100950-fig-0012#12). This parameter prompted us to investigate 2-aminoperimidine 24. At first glance, 24 is the most effective catalyst of this study, a powerful RNA cleaver even in the micromolar concentration range. In contrast, much less cleavage occurs in the presence of 2-aminobenzimidazole 10, the active component of our previous artificial ribonucleases. A direct comparison, however, is complicated by the fact that RNA cleavage is not catalyzed by a single guanidine. Instead, a concerted interaction of two or more subunits is required. When monomeric building blocks are tested, this cooperativity can result from high initial concentrations or by local enrichment caused by aggregation phenomena. The latter is obviously the case with compound 24. We assume that in general RNA hydrolysis by monomeric guanidine analogs occurs in non-ideal solutions. Compounds with large, flat and hydrophobic ring systems, therefore, may become active at lower concentrations than small and hydrophilic molecules. On the other hand, when linked together by appropriate frameworks, derivatives of 2-aminobenzimidazole such as 4 and 5 turn into powerful catalysts. Against this background, the cleavage activity of aminoimidazole 6 is quite remarkable - a compound with minimal aggregation potential. It is superior to compound 10, in spite of a less favorable value of ΔrH°298. The difference may be attributed to the steric hindrance caused by the benzene ring of 10. When ΔrH°298 is reduced to zero by the symmetry of imidazoimidazole 18, a further increase in reactivity is seen. Although this does not prove the relevance of proton shuttling as depicted in Figure #ejoc202100950-fig-0004#4, it may be wise to take tautomeric equilibria into account. The effort to predict them by AM1 calculations is insignificant.

3895_ejoc202100950-fig-0005.jpg

Figure 5Open in figure viewerPowerPointGuanidine analogs 18–23.

To further test the tautomer hypothesis, the catalytic potential of guanidine analogs which are expected to have a low or even zero energy difference was investigated (Figure #ejoc202100950-fig-0005#5, Table 2). Imidazoimidazole 18 exists in structurally identical tautomeric forms. This property and the pKa value of 7.4 made 18 an interesting candidate that turned out to be even twice as active as aminoimidazole 6. Unfortunately, the compound slowly degraded in aqueous buffer. All our attempts to attach side chains for conjugation with oligonucleotides failed due to the low stability of the intermediates. In contrast, no degradation was observed with compounds 19 and 20–23 (Figure #ejoc202100950-fig-0005#5). The energy difference between tautomeric forms of compound 20, 21 and 23 is also zero due to symmetry. The benzene rings, however, shifted the pKa to unfavorable values, increased steric hindrance and also reduced solubilities in water. As a consequence, even at pH 5 and in the presence of 40 % DMSO less than 2 mM of compounds 20–22 were soluble. No RNA cleavage was seen at pH 7 or 6.1 and only minor effects at pH 5.0. In contrast, compound 19 is characterized by a pKa value of 6.5 and tautomers of similar energy. Although hardly active at pH 6.9, it is a good RNA cleaver at pH 5.9 and outperforms compound 10 by a factor of 3. We could not determine the pKa of compound 23 by our UV-spectroscopic method but the corresponding hydrochloride is a strongly acidic compound outside of the useful pKa range. The energy differences for each compound are calculated by subtracting the energy of the tautomer shown in the Figure #ejoc202100950-fig-0002#2, Figure #ejoc202100950-fig-0005#5 and Figure #ejoc202100950-fig-0006#6 of its corresponding ketimin tautomer. Compound 11, 18–23, 25 and 26 are secondary ketimines and for them just another secondary ketimin tautomer exists. As expected all energy differences are≥0.

3895_ejoc202100950-fig-0006.jpg

Figure 6Open in figure viewerPowerPoint2-Aminoperimidine 24 and derivatives 25–31.

2-Aminoperimidine 24 turned out in our systematic search as a promising candidate with a ΔrH°298 value of only 20.5 kJ mol−1 (Figure #ejoc202100950-fig-0006#6 and Table 3). In fact, compound 24 appeared to be the most effective catalyst among all monomeric guanidine analogs we have tested so far. At the optimal concentration of 250 μM it almost attained the activity of trisbenzimidazole 4. Increasing the concentration of 24 into the mM range caused reduced cleavage, presumably by massive aggregation (Figure #ejoc202100950-fig-0007#7). This result prompted us to take a closer look on 24 and its derivatives 25–31. The energy differences for each compound are calculated by subtracting the energy of the tautomer shown in the Figure #ejoc202100950-fig-0002#2, Figure #ejoc202100950-fig-0005#5 and Figure #ejoc202100950-fig-0006#6 of its corresponding ketimin tautomer. Compound 11, 18–23, 25 and 26 are secondary ketimines and for them just another secondary ketimin tautomer exists. As expected all energy differences are≥0.

3895_ejoc202100950-fig-0007.jpg

Figure 7Open in figure viewerPowerPointCleavage of RNA substrate 13 by tris(2-aminobenzimidazole) 4 and 2-aminoperimidines 24, 25, and 26 as a function of catalyst concentration. Conditions: 150 nM RNA 13, 3.13–500 μM cleaver, 50 mM TRIS-HCl, pH 8, 37 °C, 20 h. Data points, determined at least in duplicate, are connected by lines for the sake of clarity.

2-Aminoperimidine 24 turned out in our systematic search as a promising candidate with a ΔrH°298 value of only 20.5 kJ mol−1 (Figure #ejoc202100950-fig-0006#6 and Table 3). In fact, compound 24 appeared to be the most effective catalyst among all monomeric guanidine analogs we have tested so far. At the optimal concentration of 250 μM it almost attained the activity of trisbenzimidazole 4. Increasing the concentration of 24 into the mM range caused reduced cleavage, presumably by massive aggregation (Figure #ejoc202100950-fig-0007#7). This result prompted us to take a closer look on 24 and its derivatives 25–31.

3895_ejoc202100950-fig-0008.jpg

Figure 8Open in figure viewerPowerPointCleavage of RNA substrate 13 by 2-aminoperimidine derivatives 24 and 27–31 as a function of catalyst concentration. Conditions: 150 nM RNA 13, 3.13–500 μM cleaver, 50 mM TRIS-HCl, pH 7 or 8, 37 °C, 20 h. Data points, determined at least in duplicate, are connected by lines for the sake of clarity.

RNA cleavage experiments were conducted near the pH optimum for each compound: pH 7 for the less basic derivatives 27 and 29 and at pH 8 for compounds 28, 30, and 31. Most perimidine derivatives have activities comparable to those of 24 (Figure #ejoc202100950-fig-0008#8). Ester 30 and in particular the bromo derivative 27 are distinctly weaker catalysts. The importance of the reaction pH coming close to the pKa of the catalyst is demonstrated in Figure #ejoc202100950-fig-0009#9. Methylperimidine 28 (pKa=8.1) cleaves faster at pH 8 whereas ester 29 (pKa=6.7) works better at pH 7. Bromoperimidine 27 (pKa=7.2), a weak catalyst already at pH 7, does not work at all under more basic conditions (Figure #ejoc202100950-fig-0009#9).

3895_ejoc202100950-fig-0009.jpg

Figure 9Open in figure viewerPowerPointpH-dependent cleavage of RNA substrate 13 by aminoperimidine derivatives 27, 28, and 29. Conditions: 150 nM RNA 13, 6.25–500 μM catalyst, 50 mM TRIS-HCl pH 7 or 8, 37 °C, 20 h. Data points are connected by lines for the sake of clarity.

RNA cleavage experiments were conducted near the pH optimum for each compound: pH 7 for the less basic derivatives 27 and 29 and at pH 8 for compounds 28, 30, and 31. Most perimidine derivatives have activities comparable to those of 24 (Figure #ejoc202100950-fig-0008#8). Ester 30 and in particular the bromo derivative 27 are distinctly weaker catalysts. The importance of the reaction pH coming close to the pKa of the catalyst is demonstrated in Figure #ejoc202100950-fig-0009#9. Methylperimidine 28 (pKa=8.1) cleaves faster at pH 8 whereas ester 29 (pKa=6.7) works better at pH 7. Bromoperimidine 27 (pKa=7.2), a weak catalyst already at pH 7, does not work at all under more basic conditions (Figure #ejoc202100950-fig-0009#9).

3895_ejoc202100950-fig-0010.jpg

Figure 10Open in figure viewerPowerPointCleavage pattern of TAR RNA 15 induced by different agents and analyzed with an ALFexpress II sequencer detecting the fluorescence of Cy5-labeled fragments. a) 150 nM RNA 15, 100 μM 2-aminoperimidine 24, 50 mM TRIS-HCl pH 8, 37 °C, 20 h. b) 150 nM RNA 15, 2 M imidazole pH 7, 40 mM NaCl, 500 μM EDTA, 37 °C, 20 h. The non-denaturing imidazole buffer cleaves RNA 15 preferentially in the single-stranded bulge and loop regions (indicated in grey, see Figure 3).

The cleavage pattern induced in the enantiomeric TAR RNA 15 by compound 24 (Figure #ejoc202100950-fig-0010#10a) and by imidazole buffer (Figure #ejoc202100950-fig-0010#10b) reveals a striking difference. Probing of the stable stem-loop structure with imidazole shows, as expected, cleavage restricted to the bulge and the loop regions. In contrast, 24 induces hydrolysis in all possible positions. This requires full denaturation of the RNA stem-loop and suggests binding of 15 to polycationic aggregates formed by 24. At 20 μM, however, the amount of cleavage induced by 24 drops drastically and a pattern similar to Figure #ejoc202100950-fig-0010#10b occurs. In contrast, the cleavage pattern of compounds 6 and 18 represents the bulge and loop structure of RNA 15 even at the highest concentration of 10 mM. Direct evidence for aggregates formed from aminoperimidine 24 and oligonucleotides is provided by fluorescence correlation spectroscopy (FCS). Dye labeled DNA 16 (19 nM) when diluted with oligo 17 (131 nM; 150 nM total oligonucleotide concentration as in the cleavage experiments) in buffer free of heterocyclic guanidines shows high mobility consistent with a monomeric state. At concentrations of 250 μM and more, 24 massively raises the diffusion time of 16 indicating the presence of large aggregates. According to Figure #ejoc202100950-fig-0010#10a, however, aggregation is not absent even at lower concentrations. Aminoimidazole 6, in contrast, has no impact on substrate mobility up to 10 mM. Some effects are found for compounds 10 and 18, but only in the absence of cosolvent whereas compound 19 due to its more lipophilic structure causes considerable aggregation above 2 mM.

3895_ejoc202100950-fig-0011.jpg

Figure 11Open in figure viewerPowerPointEnergy difference between the relevant tautomers of the various compounds calculated by different theoretical methods (RI-CCSD(T) only for 6–12, 23–27 and those where it is zero due to symmetry). Data points are connected by lines for the sake of clarity.

To test how reliable the semi empirical description of the energy difference by AM1 is, more advanced quantum chemical calculations were performed. The comparison is shown in Figure #ejoc202100950-fig-0011#11. The values for ΔrH°298 obtained by AM1 have a similar order of magnitude as ΔrG°298 calculated by the other methods and show overall similar qualitative trends. Due to the semi-empirical nature of AM1, zero point vibrational energies and thermal corrections are already included in the definition of self-consistent field energies, so that we did not attempt to disentangle different contributions and omitted furthermore entropic corrections in directly comparing here ΔrH° values from semi-empirical AM1 with ΔrG° values from ab initio methods. For the relevant compounds 6–12, 19, 22, 24, 25 and 27–31 the mean absolute error (MAE) between AM1 and the hybrid density functional B3LYP is 7.7 kJ mol−1 which is similar to the MAE between AM1 and second-order Møller-Plesset perturbation theory with the resolution of identity approximation RI-MP2 (6.9 kJ mol−1) and not much larger than the MAE between B3LYP and RI-MP2 (5.9 kJ mol−1). For the relevant compounds ΔrG°298 by RI-MP2 is always larger than ΔrG°298 by B3LYP. ΔrG°298 based on high level coupled cluster theory with the resolution of identity approximation RI-CCSD(T) is between the values of B3LYP and RI-MP2. As very accurate energy differences seem not to be important in the classification whether the compound has cleaving abilities or not, AM1 is a fast method to get a first estimate.

3895_ejoc202100950-fig-0012.jpg

Figure 12Open in figure viewerPowerPointClassification of the cleavage behavior (“-“ marks compounds that do not cleave, “+” marks compounds that are active cleavers and “(+)” marks compounds that show only little cleavage at pH 7) of various compounds in dependence of their experimental pKa and computed ΔrH°298 values as obtained on the AM1 level (for a variant of this figure with ΔrG°298 from RI-MP2 see Supporting Information).

The aim of the present study is to identify heterocyclic amidines and guanidines as building blocks of improved artificial ribonucleases related to conjugate 5 – and to understand why some are better catalysts than others. Parameters we consider relevant are the pKa values, the steric demand and the hydrophobicity of the compounds: The weakly basic candidates such as 8, 9, and 11 all fail in the cleavage assay. In addition, we have found a weak correlation between catalytic activity and the energy difference between amino and imino tautomers (Figure #ejoc202100950-fig-0012#12). This parameter prompted us to investigate 2-aminoperimidine 24. At first glance, 24 is the most effective catalyst of this study, a powerful RNA cleaver even in the micromolar concentration range. In contrast, much less cleavage occurs in the presence of 2-aminobenzimidazole 10, the active component of our previous artificial ribonucleases. A direct comparison, however, is complicated by the fact that RNA cleavage is not catalyzed by a single guanidine. Instead, a concerted interaction of two or more subunits is required. When monomeric building blocks are tested, this cooperativity can result from high initial concentrations or by local enrichment caused by aggregation phenomena. The latter is obviously the case with compound 24. We assume that in general RNA hydrolysis by monomeric guanidine analogs occurs in non-ideal solutions. Compounds with large, flat and hydrophobic ring systems, therefore, may become active at lower concentrations than small and hydrophilic molecules. On the other hand, when linked together by appropriate frameworks, derivatives of 2-aminobenzimidazole such as 4 and 5 turn into powerful catalysts. Against this background, the cleavage activity of aminoimidazole 6 is quite remarkable - a compound with minimal aggregation potential. It is superior to compound 10, in spite of a less favorable value of ΔrH°298. The difference may be attributed to the steric hindrance caused by the benzene ring of 10. When ΔrH°298 is reduced to zero by the symmetry of imidazoimidazole 18, a further increase in reactivity is seen. Although this does not prove the relevance of proton shuttling as depicted in Figure #ejoc202100950-fig-0004#4, it may be wise to take tautomeric equilibria into account. The effort to predict them by AM1 calculations is insignificant.

464_cbic201700103-fig-0001.jpg

Figure 1Open in figure viewerPowerPointMass spectra of the tetraphosphorylated β-casein peptide in the presence of uranyl, A) without UV and B) after UV irradiation. Asterisked peaks are MALDI-induced neutral losses of H3PO4 from the intact peptide. C) Detail of mass spectra displaying the isotope distributions of the three major photocleavage products. D) Sequence of the tetraphosphorylated β-casein peptide. The observed cleavages are indicated with solid lines, with residue number (above) and mass (below). Int. Std.=internal standard (neurotensin).

Briefly, β-casein peptide solution was mixed with a fivefold molar excess of uranyl, as this ratio has previously been shown to be optimal for the photocleavage reaction, and incubated for 1 h at room temperature. The coordination of uranyl to the β-casein peptide was confirmed by MS (Figure S1 in the Supporting Information). The samples were placed on ice and irradiated with UV light (λ=365 nm) for 15 min in solution. The peptides were purified and desalted on reversed-phase micro columns (this step also removes uranyl) followed by MS or MS/MS analysis. Figure #cbic201700103-fig-0001#1 A shows the control MALDI TOF mass spectrum of the intact β-casein peptide before UV irradiation. The peak at m/z 3122.2 corresponds to the singly protonated β-casein peptide (theoretical mass [M+H]+ 3122.2 Da), with four peaks arising from sequential neutral losses of 98 Da (H3PO4) from the phosphorylated serine residues. This neutral loss is a well-known gas phase reaction that occurs in the MALDI process. The peak at m/z 1672.9 (“Int. Std.”) corresponds to singly protonated neurotensin, which was added as an internal mass calibrant. Figure #cbic201700103-fig-0001#1 B shows the mass spectrum of the peptide products formed by UV irradiation of uranyl-bound β-casein peptide. Uranyl-induced photocleavage gives rise to three products: m/z 1097.6 (peptide (1–9)), 1624.8 (1–14) and 1904.9, (1–16). Two of the three cleavage sites are close to the phosphorylation motifs at amino acid residues 15 and 17, as expected. The last cleavage product might be a result of intramolecular folding of the peptide in solution, but this is speculative as the specificity of this photocleavage reaction is not fully understood. Interestingly, the observed masses do not correspond to the theoretical masses of products formed by peptide bond hydrolysis. The expected fragments generated by hydrolysis would be m/z 1098.6, 1625.8 and 1905.9. These theoretical masses are 1 Da higher, thus suggesting that the photocleavage reaction does not results in peptide products with a free C-terminal -COOH. Rather, the observed masses strongly suggest that the peptides formed by uranyl photocleavage are C-terminally amidated (-CONH2). In order to obtain their masses with high accuracy, they were analysed with an ESI hybrid ion trap Orbitrap mass spectrometer (Table 1). The masses of the truncated peptides are in excellent agreement with the theoretical masses of C-terminally amidated sequences (mass deviations <1.5 ppm). At first glance, the MALDI mass spectra suggest that uranyl photocleavage proceeds solely by an α-amidation-like pathway to yield exclusively C-terminally amidated peptides. However, closer inspection of the MALDI spectrum (Figure #cbic201700103-fig-0001#1) reveals a minor peak at m/z 1125.6. We attribute this to the diamide product formed by a similar α-carbon-centred radical, as that leading to the α-amidation product (1–9) at m/z 1097.6. However, there are no other peaks in the MALDI spectrum corresponding to diamide products (m/z 1762.8 and m/z 2042.8) formed by a similar α-carbon-centred radical as that leading to the α-amidation products (1–14) and (1–16). Thus, the reaction studied here, as measured by MALDI MS, seems to favour the α-amidation pathway over the diamide pathway.

464_cbic201700103-fig-0002.jpg

Figure 2Open in figure viewerPowerPointCID spectra of peptides (1–9), (1–14) and (1–16) formed by uranyl photocleavage of the β-casein peptide. Mass spectra of photocleavage products with precursor masses: A) m/z 549.30, 2+ (1097.6 Da), B) m/z 812.93, 2+ (1624.9 Da), and C) m/z 953.47, 2+ (1904.9 Da). Sequences of the respective photocleavage products are shown below each spectrum, indicating the identified b- and y-ions.

To verify that their C termini are indeed amidated, the peptides were subjected to collision induced dissociation (CID) experiments. Figure #cbic201700103-fig-0002#2 A shows the CID spectrum for peptide (1–9). The b8-fragment ion comprises the first eight residues (1–8), and its mass agrees well with the unmodified sequence (mass deviation 0.5 ppm). The mass of the C-terminal residue was readily obtained by subtracting the mass of the b8-fragment ion from the mass of the peptide ion (1–9) (i.e. precursor ion mass). The experimental mass of the C-terminal proline residue clearly shows that it was amidated, as proved by the excellent agreement with the theoretical mass for an amidated proline amino acid (mass deviation 1.9 ppm). Consequently, the CID spectrum of peptide (1–9) confirms unambiguously that its C terminus was amidated. Similarly, the CID spectrum of peptide (1–14) (Figure #cbic201700103-fig-0002#2 B) demonstrates that residues 1–13 are unmodified, as the mass of the b13-fragment ion fits well with the theoretical mass of the unmodified sequence (mass deviation 1.3 ppm). The C-terminal residue is contained within the y6-fragment ion of residues 9–14 (Figure #cbic201700103-fig-0002#2 B), and the mass of this fragment ion is in excellent agreement with the theoretical mass of the amidated sequence (mass deviation 1.2 ppm). The masses of the two fragment ions (b132+ and y6+) thus provide conclusive evidence for the amidation of C terminus in peptide (1–14). The CID product spectrum of peptide (1–16) contains a rich set of b- and y-fragment ions (Figure #cbic201700103-fig-0002#2 C). Again, all b-fragment ions corresponded to unmodified sequences, whereas the masses of the y-fragment ions corresponded to sequences with an amidated C-terminal Leu residue (y4+, mass deviation 0.6 ppm). It should be noted that the absence of b14- and b15-fragment ions means that the amidation is localised to the last three residues of peptide (1–16), that is, -Glu-pSer-Leu. However, as the amidation was located at the C terminus in peptide (1–9) and peptide (1–14), we consider it is most likely that this is also the case for peptide (1–16).

464_cbic201700103-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointProposed mechanism for uranyl-dependent photocleavage. The two suggested pathways are A) the α-amidation pathway and B) the diamide pathway. R groups=amino acid side chains. Adapted from refs. 17a and 17b.

In order to investigate whether the uranyl photocleavage reaction exhibits a similar dependence on oxygen, we conducted the experiment under essentially oxygen-free conditions. Oxygen was efficiently removed by bubbling xenon gas through the solution in a glovebox containing an inert nitrogen atmosphere (Figure S5). Figure #cbic201700103-fig-0003#3 shows the mass spectrum obtained from UV irradiation of uranyl-bound β-casein peptide in an oxygen-free environment. Interestingly, in the absence of oxygen there was no formation of photocleavage products (Figure #cbic201700103-fig-0003#3 B). In contrast, uranyl photocleavage of DNA has been reported to occur also under oxygen-free conditions. The strong oxygen dependence of uranyl-induced photocleavage thus appears to be unique for polypeptide backbone cleavage.

464_cbic201700103-fig-0003.jpg

Figure 3Open in figure viewerPowerPointMass spectra obtained from the β-casein peptide incubated with uranyl in an essentially oxygen-free environment, A) without UV and B) after UV irradiation. Asterisked peaks are MALDI-induced neutral losses of H3PO4 from the intact peptide. The experiment was carried out in a glovebox (oxygen <1 ppm). Prior to UV irradiation the samples were degassed by bubbling xenon through the samples for 5 min. Finally the samples were desalted and the mass was analysed by MALDI-MS. Int. Std.=internal standard (neurotensin).

30105_cptc202100204-fig-5001.jpg

Scheme 1Open in figure viewerPowerPoint2,5-diaryl tetrazoles (T1–T9) synthesized with a combination of electron donating (D), electron withdrawing (A), and weak electron withdrawing (wA) or weak electron donating (wD) substituents on the aryl rings (X and Y), according to Hammet constants (Table 1). Photogenerated nitrile imines (NI) yielding: a) pyrazolines (P1, P2, P5–P9) by 1,3-dipolar cycloaddition in the presence of acrylamide and b) water soluble acyl hydrazides (AH1, AH5–AH9).

The UV-Vis absorption spectra of T1–T9 were measured in 1 : 1 acetonitrile : phosphate buffer (50 mM, pH 7.4) solution, due to the low solubility of T1–T5 in pure buffer. In addition, for the water soluble tetrazoles T6–T9, we recorded the absorption spectra in pure phosphate buffer (PBS) (Figure S1). Spectra of tetrazoles are reported in Figure #cptc202100204-fig-0001#1, according to the D or A electron properties of the substituents. As expected, substituents affect the position of the maximum absorption, in particular, most bathochromically shifted tetrazoles are T3 and T4, containing the nitro group (bold lines in D and B spectra in Figure #cptc202100204-fig-0001#1). In general, almost all compounds presented maximum absorbance among 290–310 nm, with tails extended up to 350–400 nm, suggesting that photoactivation could be efficiently achieved even in this spectral region. Absorption spectra of T6–T9 recorded in PBS (50 mM, pH 7.4) exhibit similar profiles to those obtained in ACN : PBS mixture (Figure S1), with slightly lower molar extinction coefficients (Table S3). Tetrazoles disclosed a great variety of efficiency in the NI photogeneration, as quantum yield values (Φ, in Table 1) spanned from 0.002 to 0.65. All the tetrazoles were photoreactive with Φ>0.11, except for the nitro derivatives T3 and T4, (Table 1), even if they were characterized by the highest molar absorptivity and longer absorption wavelength. This outcome suggests that the presence of nitrogroup has a remarkable effect, almost erasing the tetrazole photoreactivity (Φ≥0.2 %), favoring photophysical deactivation mechanisms. The correlation quantum yield vs Hammett constants (σp/m) for the substituents on the C-5-phenyl ring (Figure #cptc202100204-fig-0001#1, for numbering), suggests that electron donor substituents on the C-5-phenyl ring display comparable detrimental effects on NI photogeneration (Φ≤0.14). On the contrary, electron withdrawing groups on C-5-phenyl ring, increase the photogeneration efficiency with Φ paralleling the σ values, in ACN : phosphate buffer (Figure #cptc202100204-fig-0002#2).

30105_cptc202100204-fig-5002.jpg

Scheme 2Open in figure viewerPowerPointa) Benzensulfonyl hydrazide (1 eq.), EtOH, r.t.; b) NaNO2 (1 eq.), HCl 37 %, EtOH 50 %, 0 °C, 30 min; c) DIPEA (2.5 eq.), Pyridine 0 °C, 4 hours.

Tetrazoles disclosed a great variety of efficiency in the NI photogeneration, as quantum yield values (Φ, in Table 1) spanned from 0.002 to 0.65. All the tetrazoles were photoreactive with Φ>0.11, except for the nitro derivatives T3 and T4, (Table 1), even if they were characterized by the highest molar absorptivity and longer absorption wavelength. This outcome suggests that the presence of nitrogroup has a remarkable effect, almost erasing the tetrazole photoreactivity (Φ≥0.2 %), favoring photophysical deactivation mechanisms. The correlation quantum yield vs Hammett constants (σp/m) for the substituents on the C-5-phenyl ring (Figure #cptc202100204-fig-0001#1, for numbering), suggests that electron donor substituents on the C-5-phenyl ring display comparable detrimental effects on NI photogeneration (Φ≤0.14). On the contrary, electron withdrawing groups on C-5-phenyl ring, increase the photogeneration efficiency with Φ paralleling the σ values, in ACN : phosphate buffer (Figure #cptc202100204-fig-0002#2).

30105_cptc202100204-fig-0001.jpg

Figure 1Open in figure viewerPowerPointUV-VIS spectra of 2,5-diaryl tetrazoles (2×10−5 M) recorded in 1 : 1 ACN : PBS (50 mM, pH 7.4) solution.

Subsequently, to define the pyrazolines with higher emission and analyze the correlation between fluorescence intensity and the electronic properties of the substituents, we have measured the fluorescence intensity of the synthesized compounds, in both water (Figure #cptc202100204-fig-0003#3) and DMSO (Figure S8).

30105_cptc202100204-fig-0002.jpg

Figure 2Open in figure viewerPowerPointQuantum yield Φ vs Hammett constants (σp/m) correlation.

30105_cptc202100204-fig-0003.jpg

Figure 3Open in figure viewerPowerPointA) Fluorescence emission of pyrazoline P1, P2 and P5-P9 in water solution, at 5×10−6 M concentration, exciting at λexc=350 nm. B) Magnification of the fluorescence spectra for the low emitting P1, P2, P6 and P9 in water solution, at 5×10−6 M concentration.

21077_cbic202000571-fig-0001.jpg

Figure 1Open in figure viewerPowerPointExpansion microscopy of intracellular organelles. HEK293 cells were grown on glass coverslips and transfected to achieve fluorescent protein (FP)-expression (optional). Cells were then fixed, immunostained, treated with the crosslinking reagent AcX and embedded into the gel. Subsequently, the sample was treated with proteinase K, which digests all cellular proteins to peptides, that are crosslinked to the gel. This allows them to expand together with the gel when incubated in water. The expanded cells were imaged by conventional confocal or STED microscopy. Unexpanded controls were imaged after immunostaining. The area or volume of the structure of interest was then measured and used to calculate the expansion factor.

We chose the cellular nucleus as a first organelle to measure the EF. We thus immunostained the nuclear pore complex in fixed HEK293 cells with a primary antibody against Nucleoporin 153 (NUP153) and a dye conjugated secondary antibody. Afterwards, the sample was gelled and expanded (see Figure #cbic202000571-fig-0001#1 for an overview of the procedure). Using confocal microscopy, the cells were optically sectioned and the maximal extent of the nucleus determined and imaged. In these images, the extent of the nucleus was then manually traced and its area recorded (Figure #cbic202000571-fig-0002#2). By comparing the area from nuclei of expanded (Figure #cbic202000571-fig-0002#2a) and unexpanded (Figure #cbic202000571-fig-0002#2b) cells a microscopic EF was calculated, shown in Figure #cbic202000571-fig-0002#2c and d. The linear expansion factor is calculated by taking the square root of the expansion in area or the third root of the expansion in volume. The volumes can be calculated from a series of images (z-stack), since the distance between the slices is known. Figure #cbic202000571-fig-0002#2d shows significant standard deviations for the measured areas, and the resulting error in the EF is also large. This is due to the inherent variance of biological systems. In the case of the nucleus, cells are morphologically different and are at different points in the cell cycle at the time of fixation. However, the narrow 95 % confidence interval shows that the EF can be reliably determined in this fashion if the sample size is sufficiently large.

21077_cbic202000571-fig-0002.jpg

Figure 2Open in figure viewerPowerPointExpansion of the nucleus. a) HEK293 cell immunolabelled with an antibody against the nuclear pore complex protein NUP153 in an expanded gel. For imaging, the confocal plane with the maximal extent of the nucleus was chosen. The size of the nucleus was measured by manually tracing the NUP153 signal, and the resulting area was recorded for analysis. b) For unexpanded cells treated in the same manner, the analysis was performed analogously. c) Box-whisker plot comparing the measured areas for maximal extent of the nucleus between expanded and unexpanded cells. The median value is denoted by the bar; the box shows the quartile ranges. Whiskers extend from the 5th to the 95th percentile. d) Expansion factor calculated from pooled median areas of unexpanded (n=80) and expanded (n=54) nuclei across two independent replicates. The EF was calculated using median values.

We chose the cellular nucleus as a first organelle to measure the EF. We thus immunostained the nuclear pore complex in fixed HEK293 cells with a primary antibody against Nucleoporin 153 (NUP153) and a dye conjugated secondary antibody. Afterwards, the sample was gelled and expanded (see Figure #cbic202000571-fig-0001#1 for an overview of the procedure). Using confocal microscopy, the cells were optically sectioned and the maximal extent of the nucleus determined and imaged. In these images, the extent of the nucleus was then manually traced and its area recorded (Figure #cbic202000571-fig-0002#2). By comparing the area from nuclei of expanded (Figure #cbic202000571-fig-0002#2a) and unexpanded (Figure #cbic202000571-fig-0002#2b) cells a microscopic EF was calculated, shown in Figure #cbic202000571-fig-0002#2c and d. The linear expansion factor is calculated by taking the square root of the expansion in area or the third root of the expansion in volume. The volumes can be calculated from a series of images (z-stack), since the distance between the slices is known. Figure #cbic202000571-fig-0002#2d shows significant standard deviations for the measured areas, and the resulting error in the EF is also large. This is due to the inherent variance of biological systems. In the case of the nucleus, cells are morphologically different and are at different points in the cell cycle at the time of fixation. However, the narrow 95 % confidence interval shows that the EF can be reliably determined in this fashion if the sample size is sufficiently large.

21077_cbic202000571-fig-0003.jpg

Figure 3Open in figure viewerPowerPointExpansion of the cell area and mitochondria within the same cells. To stain the plasma membrane, HEK293 cells expressing GPI-GFP were additionally immunolabelled with antibodies against the mitochondrial outer-membrane protein TOM20. The cells were expanded and imaged as z-stacks on a spinning disc microscope setup. a) Example images of 1) expanded and 2) unexpanded cells expressing GPI-GFP on the cell membrane. The z-slice with the maximum extent of the cell was chosen, and the cell area was measured manually. b) Example surface renderings of immunostained mitochondria derived from the z-stacks of 1) expanded and 2) unexpanded cells are shown. The red-to-yellow shading of the surface renderings illustrates the depth, where yellow objects are further away from the viewer. The z-stacks were thresholded, and the volume of all voxels was summed to obtain the volume of the whole mitochondrial network. c) Box-whisker plot showing measured volumes for TOM20 and areas for GPI-GFP. The median value is denoted by the bar; the box shows the quartile ranges. Whiskers extend from the 5th to the 95th percentile. d) Expansion factors calculated from median volumes of the mitochondrial network of unexpanded (n=76) and expanded (n=80) cells. The expansion factor for cell areas was calculated from median areas of unexpanded (n=206) and expanded (n=72) cells. The pooled data shown were obtained from three independent replicates.

We next attempted to use the volume of the whole cell and the volume of the mitochondrial network to measure the microscopic EF. To this end, the plasma membrane was labelled by expression of glycosylphosphatidylinositol (GPI)-anchored GFP in HEK293 cells and an ATTO488-conjugated single-domain antibody directed against GFP was used to boost this signal. In the same cells, mitochondria were immunostained for TOM20, a marker for the outer mitochondrial membrane. Expanded and unexpanded cells where then imaged as z-stacks on a spinning disc confocal microscope setup. The signal intensity of the GPI-GFP plasma membrane staining after expansion was too low to measure the volume of the cell accurately. This was most likely due to the aggressive permeabilization required for the ExM protocol, which destabilized the plasma membrane. Hence a lot of the GPI-GFP was washed away and after expansion the remaining signal was too weak to enable the reconstruction of a 3D image of the cell that could be used to estimate its volume. Instead, the z-slice with the maximum extent of the cell was chosen and its area recorded (Figure #cbic202000571-fig-0003#3a). The antibody-based labelling of the mitochondrial outer membrane using the TOM20 antibody gave a better signal-to-noise ratio. The TOM20 signal was segmented via thresholding in every stack and then all the voxels (dimensions per voxel: 0.135×0.135×0.370 μm3) were counted to obtain the volume of the organelle (Figure #cbic202000571-fig-0003#3b). While comparison of the cell areas obtained a linear EF of 2.7, comparison of mitochondrial volumes between expanded and unexpanded cells yielded a linear EF of 1.9 (Figure #cbic202000571-fig-0003#3c and d).

21077_cbic202000571-fig-0004.jpg

Figure 4Open in figure viewerPowerPointExpansion of peroxisomes. HEK293 cells expressing GFP-PTS1 were immunolabelled with an antibody against PEX14 (cyan), expanded and imaged in two-colour STED. The GFP signal was boosted with an ATTO488-labelled nanobody against GFP (magenta). a) One expanded HEK cell; insets are highlighted on the right with a visualization of the data analysis (right). a1) As clear assignment of the PEX14 signal in clustered peroxisomes was not possible, these were excluded from the analysis. a2) In more isolated peroxisomes, the peroxisomal membrane was manually traced according to the PEX14 signal, and the area was determined for analysis. a3) For the peroxisomal matrix, the GFP-PTS1 signal was thresholded, and its area was determined automatically. b) An unexpanded cell treated and stained by the same method was used for the gels. Areas were measured analogously to the analysis of expanded cells. c) Box-whisker plots showing areas of peroxisomal matrix and membrane before and after expansion. The median value is denoted by the bar; the box shows the quartile ranges. Whiskers extend from the 5th to the 95th percentile. d) Median areas of unexpanded (n=744) and expanded (n=654) peroxisomal membranes were used to calculate the expansion factor. Similarly, for the peroxisomal matrix, median areas of unexpanded (n=3657) and expanded (n=3322) matrices were compared. The pooled data shown were obtained from three independent replicates.

We thus labelled the peroxisomal matrix with GFP-PTS1. The peroxisomal targeting signal directed the GFP across the peroxisomal membrane into the lumen of the peroxisomes. Then, an ATTO488-conjugated-anti-GFP single domain antibody was applied to enhance (or boost) this signal. To label the membrane, a polyclonal antibody against PEX14 was used. This antibody has a very high affinity to its target and gives a good signal to noise ratio. When used in combination with stable dyes compatible with STED microscopy, such as Aberrior STAR RED, the labelling density is sufficient for ExM-STED. A representative image of the obtained two-colour STED images of peroxisomal matrix and membrane is shown in Figure #cbic202000571-fig-0004#4a) shows expanded cells in the gel, while b) shows unexpanded cells. We have to note that HEK cells, when mildly permeabilized, conventionally embedded, and imaged using STED microscopy allow resolution of greater detail than that achieved in the unexpanded cells shown here. This effect is most likely caused by the harsh permeabilization needed for the gelation. The size of the peroxisomal membrane could not be measured automatically, as the signal for PEX14 appeared to be dotted along the membrane. Assigning the PEX14 signal was not possible for clustered peroxisomes (Figure #cbic202000571-fig-0004#4a-1), as it could not be distinguished to which peroxisome the signal belongs. These peroxisomes were thus excluded from further analysis. For peroxisomes whose borders could be defined by the membrane staining (Figure #cbic202000571-fig-0004#4a-2), the PEX14 signal was manually traced to obtain the outline of the peroxisomal membrane and then its extent was recorded. For the peroxisomal matrix, we thresholded the boosted GFP-PTS1 signal and then measured the area of the matrix using the Fiji-Plugin “Particle Analyzer” (Inset 3 of Figure 4a). As before, we thus obtained two sets of areas for expanded and unexpanded cells that were employed to calculate the linear EF. With a median expansion factor of 4.1 for the gel, the linear microscopic EF for the peroxisomal membrane was 2.1, while the expansion of the peroxisome matrix was considerably smaller. The linear EF for the peroxisomal matrix was 1.3. This result can however not be attributed to a poor signal-to-noise-ratio, since the chosen labelling approach also gave strong fluorescence in the expanded gel. Instead, this result is most likely caused by the dense environment of the peroxisomal matrix, which in some cases even contains a crystalloid protein core. It has also been shown that peroxisomes do not expand like other organelles, including ER, endosomes, lysosomes or mitochondria, upon incubation of cells in hypotonic solution. This is probably also due to the dense protein matrix in the lumen of these organelles. During the digestion step, proteinase K probably cannot cleave proteins in this environment efficiently, resulting in incomplete expansion of these protein structures afterwards. To overcome this issue, we tried to prolong the digestion with proteinase K, but this led to significant loss of signal (data not shown. We also tried to use a monoclonal antibody against PEX5, the peroxisomal import receptor, which is shuttling between the cytosol and the peroxisomal membrane to import PTS1-containing cargo proteins. At the membrane, PEX5 is interacting with PEX14, triggering the translocation of the cargo protein. The monoclonal PEX5 antibody used here, only binds to PEX5 when it is located at the peroxisomal membrane and not to the cytosolic pool of the shuttling receptor. Here, we obtained good STED images using PEX14 and PEX5 in a dual colour STED colocalization study but the signal with the PEX5 antibody was too weak for ExM-STED (data not shown). The weak signal might be due to the fact that monoclonal antibodies bind to only one epitope on their target protein, therefore the use of polyclonals is recommended for ExM-STED, as these can bind multiple epitopes on their target protein.

10192_cmdc202100210-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointA) difluoroboron dipyrromethene (BODIPY) scaffold. B) BODIPY dyes 1 with a carboxylic acid functionality for bioconjugation and its corresponding NHS ester 2 used in this study.

A graphical representation of the crystal structure of 2 is given in Figure #cmdc202100210-fig-0001#1 and corresponding crystallographic data is provided in the Supporting Information. Structural features are comparable to literature reports with BODIPY moiety formed by three conjugated heterocyclic rings (the central six-membered ring with two adjacent five-membered rings) being almost planar. Conjugation of 1 to the Glu-CO-Lys binding motifs with different spacer entities was performed in solution using two different coupling methods. Either compound 1 together with HATU or the NHS derivative 2 were reacted with corresponding PSMA conjugates in the presence of DIPEA. This way, BODIPY-PSMA conjugates 3–6 were obtained in overall yields of 37–44 % after purification by semi-preparative HPLC, respectively. No significant differences in terms of coupling efficiency between compound 1 and HATU vs. NHS ester 2 were noted. Final bioconjugates 3–6 were fully characterized by one- and two-dimensional 1H, 13C, 19F-NMR spectroscopy, mass spectrometry, fluorescence spectroscopy, and analytical HPLC. Corresponding data are provided in the Supporting Information.

10192_cmdc202100210-fig-5002.jpg

Scheme 2Open in figure viewerPowerPointSynthetic procedure for the preparation of BODIPY-PSMA bioconjugates 3–6 using a combination of solid-phase and solution-based chemistry.

During the establishment of the synthetic routes of the BODIPY-PSMA conjugates 3–6 and the radiolabeling experiments, we observed the decomposition of the BODIPY fluorophore under acidic conditions. For example, when performing the syntheses entirely on solid support and cleaving corresponding BODIPY-PSMA conjugates from the resin using a TFA/H2O/triisopropylsilane (95/2.5/2.5, % v/v) cocktail, we isolated fractions by HPLC that eluted prior to the product peaks and that differed by a mass difference of m/z=-48 in the mass spectrometry compared to the intact compounds confirming loss of the BF2-core (Figure S25). To investigate the stability of the BODIPY dye under these conditions in more detail, compound 1 was incubated in 95 % TFA and samples were measured by fluorescence spectroscopy for up to 24 h (Figure #cmdc202100210-fig-0002#2A+B). While the absorption spectra of 1 in TFA only showed a small blue-shift of maxima from λabs=496 nm to λabs=488 nm (Figure #cmdc202100210-fig-0002#2A), a significant change of corresponding emission spectra was noted. After a few seconds of the incubation process a slight red shift from λem=509 nm to λem=515 nm is observed (Figure #cmdc202100210-fig-0002#2B), probably due to the initial coordination of a proton to a nitrogen atom of the BODIPY core. Within 2 h a steady decrease of the emission intensity was noted, presumably as a result of the proton-mediated loss of the BF2 moiety. At 24 h, the emission intensity was quenched completely, while the absorption intensity was still observable. The plot of the emission intensity vs. time gave the decay curve in Figure S34. From a tentative biexponential fit, two kinetic constants for decomposition/fading with k=0.168±0.011 min and k=0.018±0.006 min were obtained. Corresponding half-lives of 1 in TFA were calculated to τ1/2=4.1±0.3 min and τ1/2=37.9±9.4 min. Our results are in agreement with results from Rumyantsev, who postulated a two-step dissociation process of BODIPY complexes with a rapid protonation of the BODIPY core followed by a slow release of BF2 unit. We therefore confirm that proton coordination finally results in the release of the BF2 unit, while the chromophore remained intact. This finding is corroborated by the observation that fluorescence can widely be restored after 24 h upon addition of BF3-diethyletherate (data not shown). The TFA-mediated loss of the BF2 entity of compound 1 was also confirmed by analytical HPLC and MS (Figures S33B–S34). A similar finding was observed in the HPLC chromatograms of the 18F-labeling reactions (vide infra). To gain more insight, the impact of the Lewis acid SnCl4, again on the decomposition of the precursor 1, was investigated in acetonitrile by UV/vis and fluorescence spectroscopy (Figure #cmdc202100210-fig-0002#2C+D). The corresponding emission spectra showed a bathochromic shift of the maximum from λem=509 nm to λem=520 nm compared to intact 1 immediately after dissolving in the SnCl4 solution (Figure #cmdc202100210-fig-0002#2D). Moreover, a second emission band with λem=545 nm was noted, which can unambiguously be traced back to an excitation maximum at λexc=530 nm (Figure S29). A similar observation was made for corresponding absorption spectra with an initial red shift and formation of a second absorption band varying slightly over the course of the measurements (Figure #cmdc202100210-fig-0002#2C). These changes might be attributed to the formation of an intermediate between the Lewis acid and the fluorine atoms of 1. Another possible explanation could be the substitution of fluoride with chloride or even an insertion of SnClx. At least, both the pronounced spectral shift and the maintained fluorescence hint to distinctly disturbed, but still intact chromophore with a bridging heteroatom between the two pyrrole moieties. In summary, the acidic conditions for radiolabeling, mainly due to the Lewis acid SnCl4 and subsequent aqueous work-up, also resulted in the partial loss of the BF2 core, which occurred to 5 % to 10 % depending on the reaction conditions. When higher amounts of SnCl4 were used, formation of by-products increased. These impurities, however, could easily be separated from the 18F-labeled bioconjugates by HPLC. And even more important, once formed, no degradation or formation of (radioactive) by-products of HPLC-purified [18F]F-4 was observed in phosphate buffered saline (PBS) at pH 7.4 for 2 h, and the RCP was still >97 % after this incubation time, indicating high stability of the radiotracer.

10192_cmdc202100210-fig-0001.jpg

Figure 1Open in figure viewerPowerPointORTEP representation of 2 with thermal ellipsoids drawn at 50 % probability level.

Next, the bioconjugates 3–6 were labeled at room temperature with 18F via Lewis-acid assisted 19F/18F isotopic exchange in dry acetonitrile using tetraethylammonium-[18F]fluoride ([18F]F-TEA) and SnCl4 within 15 min incubation. A representative scheme of the radiolabeling process is given in Scheme #cmdc202100210-fig-5003#3 for compound 4. Optimization of the labeling conditions included variation of the reaction temperature and the amount of SnCl4. The highest radiochemical yields (RCYs) were achieved when a 100-fold molar excess of SnCl4 compared to the labeling precursor was used. An increase of the labeling temperature to 37 °C or 95 °C or higher amounts of SnCl4 (e. g. 500 eq.) did not result in an increase of RCYs. Under optimized labeling conditions, [18F]F-3, [18F]F-4, [18F]F-5 and [18F]F-6 were obtained in radiochemical yields of 20–30 % and radiochemical purities (RCP) of >99 %. The mean molar activities Am were ∼0.7 MBq ⋅ nmol−1. Following the labeling reactions, the radiolabeled products [18F]F-3, [18F]F-4, [18F]F-5 and [18F]F-6 were purified by C18 SepPak cartridges and reconstituted in saline (5 % EtOH) for further experiments. While the RCPs were >99 %, an additional peak in the UV/vis trace of the HPLC chromatograms of [18F]F-3, [18F]F-4, [18F]F-5 and [18F]F-6 was observed, which matched the HLPC trace of the BF2-free conjugates. This finding is exemplarily shown for [18F]F-4 in Figure #cmdc202100210-fig-0003#3.

10192_cmdc202100210-fig-0002.jpg

Figure 2Open in figure viewerPowerPointNormalized A) absorption and B) emission spectra (λexc=460 nm) of 1 in 95 % TFA at different time points. Normalized C) absorption and D) emission spectra (λexc=460 nm) of 1 in 1 M SnCl4 in acetonitrile at different time points.

In addition to the competitive binding studies, we also investigated the internalization of [18F]F-4 into LNCaP cells for up to 2 h. To determine PSMA-specific cell internalization, additional blocking experiments were performed using 2-(phosphonomethyl)-pentanedioic acid (2-PMPA), a highly specific PSMA inhibitor. A graphical representation of the PSMA-mediated internalization and PSMA-specific cell-surface binding of [18F]F-4 is given in Figure #cmdc202100210-fig-0004#4. Given P values apply to differences between blocked and unblocked experiments. As expected, the PSMA-mediated internalized fraction of [18F]F-4 increased gradually over time. At 15 min and 30 min, [18F]F-4 was internalized to 2.24 % and 2.17 % (P<0.0057), respectively, which further increased to 3.3 % at 60 min (P<0.0015). The maximum was reached at 120 min with 3.83 % of [18F]F-4 being internalized (P<0.0002). In contrast, the specific surface bound fraction of [18F]F-4 remained essentially constant over the course of the experiment with ∼0.5 %.

10192_cmdc202100210-fig-5003.jpg

Scheme 3Open in figure viewerPowerPointRepresentative radiolabeling of BODIPY-PSMA conjugate 4 by Lewis acid mediated isotopic 19F/18F exchange.

The specific cell surface binding and internalization of 4 in LNCaP cells at different time points was also visualized by fluorescence microscopy using a structured illumination microscope (SIM) (Figure #cmdc202100210-fig-0005#5). PSMA-mediated uptake of compound 4 into LNCaP cells was confirmed by fluorescence microscopy demonstrating gradual internalization over the period of 1 h. At 15 min, a significant accumulation of 4 was found in the cell membrane of the LNCaP cells, in which the large extracellular domain of the expressed PSMA is located (Figure #cmdc202100210-fig-0005#5A). At later time points (30 min and 1 h), an increasing number of intracellular fractions of high intensity was noted (Figure #cmdc202100210-fig-0005#5B+C), being in concordance with the results of the cell internalization experiment using [18F]F-4. Internalization of compound 4 occurs presumably via clathrin-coated pits and subsequent endocytosis after interaction with PSMA. Moreover, our results are in line with a recent study by Eder et al., which investigated the subcellular fate of bimodal PSMA inhibitors in more detail by STED nanoscopy. At 4 °C, accumulation of 4 in the cell membrane was observed. To further demonstrate the specificity of the surface binding and internalization of 4, we also used 2-PMPA for blocking studies (Figure #cmdc202100210-fig-0005#5E). After incubation with excess 2-PMPA, neither specific accumulation in the membrane nor internalization were observed, indicating PSMA-specific membrane accumulation and cellular internalization. The visible intracellular signals in the blocking study are caused by the autofluorescence of the cell compartments as can be seen from the negative control of fixed cells without addition of compound 4 (Figure #cmdc202100210-fig-0005#5F).

10192_cmdc202100210-fig-0003.jpg

Figure 3Open in figure viewerPowerPointA) Representative UV/vis HPLC chromatogram of 4 (insert: MS spectrum of 4). B) Radio-HPLC chromatogram of [18F]F-4. C) Representative UV/vis HPLC chromatogram of 4 without BF2 (insert: MS spectrum of 4 -BF2). D) UV/vis HPLC chromatogram of [18F]F-4 showing loss of the BF2 entity during radiolabeling by Lewis acid mediated isotopic 19F/18F isotopic exchange.

10192_cmdc202100210-fig-0004.jpg

Figure 4Open in figure viewerPowerPointPSMA-specific surface bound and internalized fractions of [18F]F-4 in LNCaP cells. Values are given as mean±SD. Results are from two independent experiments in triplicate.

10192_cmdc202100210-fig-0005.jpg

Figure 5Open in figure viewerPowerPointPSMA-mediated uptake of 4 into LNCaP cells, confirmed by SIM-fluorescence microscopy over time at 25 °C at A) 15 min, B) 30 min, C) 1 h, D) 1 h at 4 °C, E) 1 h+2-PMPA, F) DAPI control.

16601_syst202000062-fig-0001.jpg

Figure 1Open in figure viewerPowerPointMorphologies of calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogels ([Ca2+]=1 M) and orthophosphate-pyrophosphate solutions ([Pi]+[PPi]=0.7 M). These images show structures grown for 1 h.

The primary variable in these experiments was the concentration ratio of orthophosphate to pyrophosphate totalling 0.7 M in the solution phase (Table 1). Figure #syst202000062-fig-0001#1 shows a macroscopic view of tube formation generated by the orthophosphate-pyrophosphate solutions ([Pi]+[PPi]=0.7 M) being placed onto the hydrogel surface containing calcium ions ([Ca2+]=1 M) after 1 h. Structures form via the same mechanism as a chemical garden developing from a salt seed, with the hydrogel-solution interface facilitating the necessary chemical gradients (osmotic, pH, concentration) to initiate growth. Firstly, an initial semi-permeable calcium orthophosphate-pyrophosphate membrane formed on the surface of the hydrogels, which promotes the advective transport of ions and water from the solution toward the hydrogel phase. This causes a build-up of osmotic pressure at the hydrogel-solution interface. The pressure is subsequently released through streams of calcium ions, which pierce the semipermeable membrane. Upon entering the orthophosphate-pyrophosphate solution reservoirs, precipitation nucleates around the calcium streams, leading to development of tubular architectures. Tubular elongation is maintained through continuing cycles of osmotic pressure that drive the release of calcium ions through the established structures, which contribute to furthered precipitation at the tip of the tube.

16601_syst202000062-fig-0002.jpg

Figure 2Open in figure viewerPowerPointTube height as a function of time for calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogels ([Ca2+]=1 M) and orthophosphate-pyrophosphate solutions ([Pi]+[PPi]=0.7 M) (a) at native pH (n=4) and (b) with solutions adjusted to pH 9.5 (n=4).

As shown in Figure #syst202000062-fig-0002#2a, tube height in the Z direction was monitored at 5 min intervals over a period of 1 h. Measurements of tube height were taken between the gel/solution interface and the representative height of the majority of tube structures. Measurable changes in growth rate (mm/min) can be determined from the slope of tube height (mm) over time (min). In all cases, there was a linear increase in tube height with time over the first 15 min of formation. The respective initial growth rates are provided in Table 3. Tubes prepared from [PPi]=0 % displayed an approximate growth rate of 1 mm/min, whilst growth rates of tubes grown from solutions containing pyrophosphate were in the range of 1.1–1.5 mm/min. Interestingly, growth rates associated with tubes generated from pyrophosphate containing solutions were observed to reduce at around 20 min to≤1 mm/min, whilst tubes grown in the absence of pyrophosphate grew at a consistent rate of 1 mm/min until reaching the growth vessel ceiling at approximately 35 min. Growth velocities in all cases levelled off by 40 min (approximately 0 mm/min). This resulted in notable differences in the final tube height as a consequence of pyrophosphate content, with [PPi]=12 % being most detrimental to tube development overall, despite these structures being associated with the greatest initial growth rate of 1.44 mm/min. Tubes grown from [PPi]=0 % and [PPi]=21 % demonstrated comparable final growth heights. As shown in Figure #syst202000062-fig-0002#2b, tube height in the Z direction was monitored at 5 min intervals over a period of 1 h for pH adjusted solutions. The respective initial growth rates are provided in Table 4. Adjusting the pH of [PPi]=0 % solution from 9.35 to 9.5 increased the growth rate from approximately 1 mm/min to 1.34 mm/min, which was comparable to the initial growth velocities displayed by tubes grown from solutions containing pyrophosphate prior to pH adjustment. As the hydrogel phase is slightly acidic, increasing the alkalinity of the [PPi]=0–6 % solutions effectively steepens the pH gradient between hydrogel and solution phases. This may promote the elevated advective transport of ions at the hydrogel-solution interface, which in turn augments the tubular growth rate. Interestingly, the growth rate of tubes grown from [PPi]=18–21 % solutions also increased, despite the pH of the solutions being reduced relative to native pH of these solutions. Additionally, the apparent growth ceiling of tubes grown from [PPi]=18–21 % solutions was notably reduced by adjustment of the pH to 9.5. Compared to tubes grown from [PPi]=0 % solutions adjusted to pH 9.5, adjusting the pH of [PPi]=21 % solutions from 9.64 to 9.5 resulted in a 5 mm deficit in final tube height between these tubes. Although [PPi]=12 % solutions required the smallest adjustment in pH from 9.54 to 9.5, this condition remained the most detrimental to tube development,

16601_syst202000062-fig-0003.jpg

Figure 3Open in figure viewerPowerPointTube height as a function of time for calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogel ([Ca2+]=1 M) and [PPi]=12 % solutions (a) adjusted to pH 9 (n=4) and (b) adjusted to pH 9.5 (n=4). Tube morphologies for the respective conditions are also shown. These images show structures grown for 1 h.

In order to explore the effects of pH alteration and PPi content further, tubes were grown from [PPi]=0 % and [PPi]=12 % solutions adjusted to pH 9. Tube growth was compared to structures grown from solutions adjusted to pH 9.5. As shown in Figure #syst202000062-fig-0003#3a, the initial growth rate of tubes prepared with [PPi]=0 % solutions exceeded those of tubes prepared from [PPi]=12 % solutions by approximately 0.5 mm/min when the solutions were adjusted to pH 9. Comparably, Figure #syst202000062-fig-0003#3b reveals that when the solutions were adjusted to pH 9.5, the distinction between initial growth rates was less extensive, differing by 0.1 mm/min. Whilst tubes form more rapidly in the absence of pyrophosphate, the differences in growth are more apparent when the pH was decreased. There were also apparent differences in the final growth height, with tubes grown with [PPi]=12 % solutions adjusted to pH 9 failing to grow beyond 25 mm, whereas adjustment of solutions to pH 9.5 resulted in tubes reaching up to 30 mm. The deficit in final tube heights was therefore approximately 15 mm and 10 mm, respectively, between these tubes and those grown from [PPi]=0 % solutions. Finally, [PPi]=12 % solutions generally resulted in fewer tubes compared to [PPi]=0 % solutions, however, the population of forming tubes was further diminished by adjustment of [PPi]=12 % solutions to pH 9.

16601_syst202000062-fig-0004.jpg

Figure 4Open in figure viewerPowerPoint(a) Powder X-ray diffraction patterns and (b) Raman spectra for calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogels ([Ca2+]=1 M) and orthophosphate-pyrophosphate solutions ([Pi]+[PPi]=0.7 M).

X-ray diffraction (XRD) patterns are shown in Figure #syst202000062-fig-0004#4a. For tubes grown from 0 %=[PPi] solutions, the diffraction pattern is consistent with that of apatitic calcium orthophosphate and was matched to ICDD pattern 01-074-9761 for hydroxyapatite. For tubes grown from [PPi]=6 % solutions, the corresponding hydroxyapatite peaks appear less defined whilst the background intensity increases. For tubes grown from [PPi]=12–21 % solutions, the patterns consist broad signal between 26–32° 2θ degrees, indicating that the mineral precipitated is X-ray amorphous. It has been determined that pyrophosphate inhibits crystallisation by adsorbing onto sites of hydroxyapatite crystal growth, which prevents further binding of ionic building blocks (i. e. calcium and orthophosphate). In effect, this mechanism increases the temporal stability of amorphous calcium orthophosphate, accounting for the poorer crystallinity attributed to precipitates grown from solutions of increasing pyrophosphate content. Raman spectra are shown in Figure #syst202000062-fig-0004#4b. The Raman spectrum of tubes grown from [PPi]=0 % solutions shows an intense peak located at 960 cm−1, which correlates specifically to the symmetric stretching vibration mode of the PO4 group as found in hydroxyapatite. Additional peaks located at 430 cm−1, 590 cm−1 and 1045 cm−1 are consistent with hydroxyapatite PO4 group symmetric bending, asymmetric bending and asymmetric stretching, respectively. For tubes grown from [PPi]=6–21 % solutions, the apatite symmetric stretching peak becomes broader and is downshifted to approximately 950–590 cm−1, indicative of an amorphous apatitic phase. Further, peaks indicative of apatite symmetric bending, asymmetric bending and asymmetric stretching are substantially diminished. Peaks associated with PPi content appear at approximately 740 cm−1 and 1040 cm−1, indicative of P−O−P and P−O stretching vibrations respectively. As the pyrophosphate content increases, the intensity of the 1040 cm−1 peak is enhanced whilst the corresponding intensity of the apatite symmetric stretching peak between 950–590 cm−1 decreases.

16601_syst202000062-fig-0005.jpg

Figure 5Open in figure viewerPowerPointScanning electron microscopy (SEM) images of calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogels ([Ca2+]=1 M) and orthophosphate-pyrophosphate solutions ([Pi]+[PPi]=0.7 M).

Microstructural features and surface topology of precipitated structures was analysed using scanning electron microscopy (SEM) as shown in Figure #syst202000062-fig-0005#5. The typical radius of the tubes analysed ranged between 50–100 μm. Tube wall thickness was also observed to range between 5–10 μm. Both the outside and inside facing walls of tubes grown from [PPi]=0 % solutions showed a porous structuring, made up of closely packed micron sized spheres, consistent with the appearance of apatite precipitated in previous systems. The outside and inside surfaces are separated by a distinct 2–5 μm thick interface with denser structuring. This intermediate tube wall layer may act as a substrate that supports the development and evolution the resulting microstructure observed.

16601_syst202000062-fig-0006.jpg

Figure 6Open in figure viewerPowerPointX-ray fluorescence (XRF) elemental maps of calcium orthophosphate-pyrophosphate chemical gardens grown from calcium loaded hydrogel ([Ca2+]=1 M) and [PPi]=0 % or [PPi]=12 % solutions.

Micrographs reveal changes in microstructure as the pyrophosphate content of the reactant solution is incrementally increased. Tubes grown from [PPi]=6–18 % solutions also exhibit spherical microstructuring, similar to those grown from [PPi]=0 % solutions, however, there are distinct regions of a secondary amorphous microstructure that appears to coat tube surfaces, possibly resulting from the addition precipitation of an amorphous calcium pyrophosphate phase. For some of the tubes grown from [PPi]=6 % solutions, we observed delamination of this secondary microstructure, revealing the thickness of the coating to be in the range of 3–5 μm. For tubes grown from [PPi]=12–18 %, the secondary microstructure appears to form a thin film over the underlying primary spherical microstructure, perhaps no more than 1–2 μm in thickness. Further compositional analysis of the surfaces of tubes grown from [PPi]=0 % solutions and [PPi]=12 % solutions was investigated with X-ray fluorescence (XRF) spectroscopy. As shown in Figure #syst202000062-fig-0006#6, elemental maps revealed distinct phosphorous rich regions upon the outer walls of tubes grown from [PPi]=12 %. This suggests that pyrophosphate species are a prevalent constitutional component of the secondary microstructure, whilst the underlying microstructure is compositionally comparable to that of the surface of tubes prepared with [PPi]=0 % solutions.

2724_cbic202000109-fig-0001.jpg

Figure 1Open in figure viewerPowerPointScheme of the interaction of oncocin with the 23S rRNA of the T. thermophilus ribosome. The scheme was generated with PoseView8 and edited with inkscape (http://www.inkscape.org/).

The interaction of oncocin with the Thermus thermophilus 70S ribosome has been characterized by crystal structure analysis (PDB ID: 4Z8 C). The peptide interacts exclusively with the 23S ribosomal RNA (rRNA; Figure #cbic202000109-fig-0001#1) and blocks the peptidyl transferase center and the peptide exit tunnel of the ribosome. Specific interactions mostly occur via its N-terminal part. A sequence alignment of 23S rRNA sequences shows that the oncocin binding site of T. thermophilus is strictly conserved in several species of human pathogens, i.e., E. coli BW25113, P. aeruginosa ATCC 27853, K. pneumoniae ATCC 10031, Acinetobacter baumannii ATCC 15308, and Staphylococcus aureus DSM 6247 (Figure #cbic202000109-fig-0002#2). The conserved structure of the oncocin-binding site indicated that oncocin and its optimized analogues Onc72 and Onc112 might be able to inhibit protein translation in both Gram-positive and Gram-negative bacteria. Thus, we tested the binding of Onc112 to 70S ribosome preparations obtained from five different bacteria. As we concluded that DNase treatment did not affect the fluorescence polarization assay, the new protocol was applied to all studied bacteria. The Kd values determined for the K. pneumoniae ATCC 10031 ribosome and Onc112 (Kd=77±1 nmol/L) were almost identical to E. coli, which was expected as both bacteria belong to the family of Enterobacteriaceae. In agreement with molecular modeling, Onc112 bound equally well to A. baumannii ATCC 15308 ribosomes (Kd=73±4 nmol/L), even better to the ribosome preparation of P. aeruginosa ATCC 27853 (Kd=36±2 nmol/L), and only slightly worse to the ribosome of S. aureus DSM 6247 (Kd=102±5 nmol/L). Both the molecular modeling and the binding affinities determined in vitro indicated a highly conserved binding site of Onc112 among the tested Gram-negative and Gram-positive bacteria (Figure #cbic202000109-fig-0001#1, Table 1). Thus, Onc112 represents a highly relevant scaffold to develop antibiotics targeting a broad range of pathogens.

2724_cbic202000109-fig-0002.jpg

Figure 2Open in figure viewerPowerPointSequence alignment of 23S rRNA at the oncocin binding site. The residues depicted in bold are within 5 Å of oncocin in the crystal structure of oncocin bound to the T. thermophilus ribosome. Residues that interact with oncocin via the base are underlined. Four residues differ in the human 28S rRNA (boxed in red). T.t, S.a, E.c., K. p., P.a., A.b., and H.s. denote T. thermophilus, Staphylococcus aureus, E. coli, K. pneumoniae, P. aeruginosa, A. baumannii, and Homo sapiens, respectively. See Tables S2 and S3 for alignment statistics.

The interaction of oncocin with the Thermus thermophilus 70S ribosome has been characterized by crystal structure analysis (PDB ID: 4Z8 C). The peptide interacts exclusively with the 23S ribosomal RNA (rRNA; Figure #cbic202000109-fig-0001#1) and blocks the peptidyl transferase center and the peptide exit tunnel of the ribosome. Specific interactions mostly occur via its N-terminal part. A sequence alignment of 23S rRNA sequences shows that the oncocin binding site of T. thermophilus is strictly conserved in several species of human pathogens, i.e., E. coli BW25113, P. aeruginosa ATCC 27853, K. pneumoniae ATCC 10031, Acinetobacter baumannii ATCC 15308, and Staphylococcus aureus DSM 6247 (Figure #cbic202000109-fig-0002#2). The conserved structure of the oncocin-binding site indicated that oncocin and its optimized analogues Onc72 and Onc112 might be able to inhibit protein translation in both Gram-positive and Gram-negative bacteria. Thus, we tested the binding of Onc112 to 70S ribosome preparations obtained from five different bacteria. It has to be noted that it remains an open question, how the uptake rates calculated here have to be interpreted mechanistically, as it is unknown what portion entered the cytosol and finally reached the ribosomes. For Gram-negatives, the peptides are likely partially trapped in membranes, the periplasmic space, and at the negatively charged surface. Such off-target effects will most likely differ among bacteria and are thus difficult to study. However, at least for Api88 – a C-terminally amidated version of Api137 – we showed in 2012 that it enters E. coli, K. pneumoniae, and P. aeruginosa at reasonable quantities when N-terminally labeled with carboxyfluorescein (Cf; Figures #cbic202000109-fig-0002#2 and S3). Confocal laser scanning microscopy indicated further that Cf-Api88 enters the cytoplasm for all three bacteria. However, PrAMPs will likely bind to other intracellular proteins as well. Recently, we reported that Api88, a close analogue of Api137, binds in vivo to nonribosomal proteins, such as DnaK and GroEL. DnaK was identified as an intracellular target of PrAMPs including apidaecins and oncocins, but the binding is much weaker than for ribosomes.

2724_cbic202000109-fig-0003.jpg

Figure 3Open in figure viewerPowerPointFluorescence polarization curves recorded for A) Cf-Onc112 (Cf–VDKPPYLPRPRPPRrIYNr-NH2, r: d-arginine) or B) Cf-Api137 (gu-O(Cf)NNRPVYIPRPRPPHPRL-OH, gu: N, N, N’, N’-tetramethylguanidino, O: l-ornithine) in the presence of 70S ribosome preparation obtained from five different bacteria. The data were recorded after an incubation period of 90 min at 28 °C. Curves were fitted to a dose-response curve with a variable slope parameter [y=min+(max–min) / (1+(x/IC50)−Hill slope)] by using SigmaPlot.

The Kd values determined for 5(6)-carboxyfluorescein (Cf) -labeled Onc112 and the 70S ribosome isolated from E. coli BW25113 were very similar to the values reported for E. coli BL21(DE3)RIL after an incubation time of 90 min (Supplementary material, Table S1). However, when this ribosome preparation protocol was applied to P. aeruginosa, a jellylike highly viscous sample was obtained that was unsuitable for fluorescence polarization measurements. Assuming that the high viscosity was mostly due to the presence of DNA, a DNase digest step was added, besides some other minor changes. The ribosome preparation of E. coli BW25113 obtained by the published and new protocols provided similar affinities, that is, Kd values of 36±14 and 75±4 nmol/L for Onc112 and 328±72 and 379±22 nmol/L for Api137 (Figure #cbic202000109-fig-0003#3, Tables 1 and S1). The Kd value of a scrambled Onc112 was eight times higher than for Onc112; this is in agreement with data reported for truncated and inverso peptides. Api137, which binds in reverse direction compared to Onc112 and close to the peptidyl transferase center, bound equally well to the ribosome preparations of both Enterobacteriaceae and P. aeruginosa (Kd ranging from 155 to 379 nmol/L; Figure #cbic202000109-fig-0003#3, Table 1), but less efficient than Onc112. However, the binding to A. baumannii and S. aureus ribosomes was much weaker with Kd values only in the low μmol/L-range. Assuming that the reduced binding constants of Api137 might be related to the release factors participating in its ribosome binding, the tryptic in-gel digests of all five ribosome preparations were analyzed by LC–MS. Release factors were indeed detected in the ribosome preparations of E. coli (RF1 and RF3), P. aeruginosa (RF1 and RF2), and K. pneumoniae (RF1, RF2, and RF3), but not in the digests of A. baumannii and S. aureus despite detecting similar numbers of other ribosomal proteins (Table S4). However, addition of recombinant E. coli RF1 at equimolar quantities to the A. baumannii ribosome preparation did not affect the Kd values of Api137. It should be noted that it is uncertain if recombinant E. coli RF1 can bind to A. baumannii ribosomes and if release factors affect the Kd values of Api137.

2724_cbic202000109-fig-0004.jpg

Figure 4Open in figure viewerPowerPointRelative peptide quantities determined for five bacterial cell culture supernatants after an incubation period of 30 min. Cell counts were adjusted by optical density to ∼7.5×108 (A,C) and ∼7.5×109 cells/mL (B,D). Peptides were quantified by LC–MS using multiple reaction monitoring (MRM). # indicates an Api137 quantity below the LOQ corresponding to 20 %. An alternative view of the data is shown in Figure S2.

For low cell counts, the supernatants of the Gram-negatives contained between 80±17 and 98±7 % of the l-Onc112 quantities present in the control (Figure #cbic202000109-fig-0004#4 A) indicating uptake rates below 20 %. Expectedly, lower quantities of typically around 75 % were detected in the supernatants of the high cell count experiments, except for E. coli with only 56±7 % (Figure #cbic202000109-fig-0004#4 B). In view of similar bacterial uptake rates and comparable ribosome affinities measured for four Gram-negatives, the higher antibacterial activities of Onc112 against Enterobacteriaceae compared to non-fermenters P. aeruginosa and A. baumannii cannot be explained. Even for Enterobacteriaceae, the uptake rates calculated for high cell counts of E. coli (∼44 %) and K. pneumoniae (∼25 %) are opposite to the MIC values of 8 and 2 mg/L, respectively, despite identical ribosome binding affinities. This clearly indicates that lower peptide quantities are required to inhibit the growth of K. pneumoniae. Considering the uptake of 352 and 208 ng Onc112 by E. coli and K. pneumoniae (high cell counts), a single cell contained on average 0.5 and 0.3 fg peptide, respectively. This corresponds very well to the 0.2 fg/cell recently reported for E. coli BL21AI incubated with half of the l-Onc112 concentration (i. e., 4 mg/L). [9] The uptake of Api137 for low (high) cell counts of E. coli and A. baumannii was 83±7 % (60±14 %) and 98±6 % (90±10 %), respectively, which was similar to Onc112 except for the high-cell count experiment with A. baumannii (Figure #cbic202000109-fig-0004#4C and D). For these two bacteria, the higher uptake rates in E. coli and the more than sixfold higher affinity to the E. coli ribosome resembled nicely the 32-fold lower MIC values of 4 μg/mL for E. coli and 128 μg/mL for A. baumannii. The uptake rate in K. pneumoniae appeared to be extremely low, as we did not detect reduced peptide concentrations in the supernatants, not even when high cell counts were used (101±6 %). Considering the extremely low uptake, the twofold lower MIC for Api137 against K. pneumoniae (2 μg/mL) is surprising despite the twofold higher affinity to the ribosome.

32807_chem202101866-fig-0001.jpg

Figure 1Open in figure viewerPowerPoint(a) G-quadruplex (G4) formation via stacking of Hoogsteen hydrogen bonded G-tetrads. (b) Structure and polymorphism of major type of G4s.

G-quadruplexes (G4s) are considered as promising drug targets for therapeutic applications. The four-stranded G-quadruplex structures are basically formed by guanine-rich DNA or RNA sequences in near-promoter, telomere or UTR regions. Quadruplexes play pivotal roles in the regulation of replication, transcription and translation as well as in the maintenance of telomere length homeostasis. Given the significance of quadruplexes in biological functions, much effort has been put into the development of effective G4 binders derived either from natural products or synthetic compounds. G4s are composed of three or more layers of stacked G-tetrads that are formed by four guanine residues through Hoogsteen hydrogen bonds (Figure #chem202101866-fig-0001#1a). The square planar alignment of G-tetrads provides a unique site for specific recognition of quadruplexes via π-π stacking interactions. G4 binding can also be controlled by interaction with G4 loops or grooves (Figure #chem202101866-fig-0001#1b) via electrostatic interactions or intercalation between G-tetrads or by a combination of all these modes.

32807_chem202101866-fig-0002.jpg

Figure 2Open in figure viewerPowerPointChemical structures of carbazole and its derivatives that show potent pharmacological activities such as anti-inflammatory, anti-microbial (Murrayanine),8b antiproliferative (Mukonine),9b antioxidant (Carbazomadurin A),11 and antitumor (Ellipticine,14 Celiptium,17 Alectinib17).

Given the unique feature of quadruplex recognition through G-tetrads, most of the selective G4 binders are those that possess a large flat-shaped aromatic surface that is much larger than that of a duplex binder to avoid non-specific interactions with DNA duplexes. Heteroaromatic scaffolds like carbazoles have been designed as potential quadruplex binding ligands. Carbazole, a strong pharmacophoric moiety, is a tricyclic structure consisting of two benzene rings fused on both sides of a nitrogen-containing five-membered ring (Figure #chem202101866-fig-0002#2). This heterocyclic ring system can stack upon the top or bottom of external G-tetrads of quadruplexes by π–π interactions. The core system of a carbazole can also be linked with one-to-three external side chains like cationic arms to exploit electrostatic interactions with negatively charged phosphate backbones or with some side-chain appendages for hydrogen-bonding interactions with the G4 groove/loop regions. Of important note, carbazole derivatives with appropriate functional groups can adopt crescent shaped topologies enabling a well-defined interaction with intramolecular G-quadruplexes. The structural unit is also able to display spectroscopic properties that can further provide potential means for monitoring binding interactions with quadruplexes. Based on these intriguing features, carbazoles have attracted considerable interest for targeting various G-quadruplex structures. Not surprisingly, the carbazole ring system has thus been a key chemical scaffold in a variety of biologically active compounds. Owing to a wide spectrum of bioactivity and therapeutic properties, carbazoles have been considered as a potential drug candidate for the treatment of multiple diseases like cancer, diabetes, viral and bacterial diseases, neurological disorders etc. The carbazole ring is also prevalent in several medicinally active natural products for example, murrayanine, mukonine, carbazomycins, carbazomadurin A, staurosporine, murrayafoline A (Figure #chem202101866-fig-0002#2). In addition, many carbazole derivatives have been chemically synthesized and are well established for their pharmacological activities. More interestingly, there are numerous carbazole based commercially drugs that are available in the market such as ellipticine, olivacine, datelliptium, alectinib, celiptium etc (Figure #chem202101866-fig-0002#2).

32807_chem202101866-fig-0003.jpg

Figure 3Open in figure viewerPowerPointSchematic drawing of binding modes found for carbazole ligands. They have been divided into two main categories: Binding with local and binding with global conformational changes. The first category includes ligands that induce changes around the binding pocket and they were either designed to bind to the major conformation (A) of the target sequence or to target solely in a topology selective manner. Whereas in the conformational selection mechanism, the ligand binds to a pre-existing minor conformation (b) and influences the conformational equilibrium in favour to the binding competent form. Further, the ligand can induce a change in the overall topology of the binding partner and follow an induced fit mechanism.

As most carbazole ligands have been found to bind in intermediate exchange regime, they will be classified according to their binding mode (Figure #chem202101866-fig-0003#3) in the further sections. The different binding modes found for G4 DNA targeting carbazole ligands have been divided into two main sections: Binding with local or with global conformational changes (Figure #chem202101866-fig-0003#3). Binding with local changes has been further differentiated by the purpose of ligand design and refers to ligand binding to the major conformation of a targeted sequence (Figure #chem202101866-fig-0003#3a) and binding to a specific G4 topology irrespective of the sequence (Figure #chem202101866-fig-0003#3b). Global conformational changes can either be induced by a conformational selection or induced fit mechanism. It is often difficult to differentiate whether the conformational change takes place before (conformational selection) or after (induced fit) ligand binding. In order to determine the mechanism, all rate constants at varying substrate:ligand concentrations would have to be known. Even though NMR is the best technique to evaluate these parameters, the binding modes of the here presented ligands have not been examined in detail. Often, the ligand solubility is insufficient to reach the required excess of ligand. Hence, the classification into conformational selection occurred on the basis of a pre-existing minor conformation in slow exchange that has been increasingly populated upon ligand addition. In the conformational selection binding mode, the ligand interacts to a pre-existing conformation of a biomolecule, only present in small amounts and shifts the conformational equilibrium towards the binding competent form. The carbazole ligands BTC and BTC-f (Figure #chem202101866-fig-0008#8), developed by Dash group, select the minor-populated conformer (in Figure #chem202101866-fig-0003#3 designated as b) of c-MYC G4 DNA and binds to it in a specific manner. Both these bis-triazolyl carbazole derivatives were prepared by one-pot Cu(I) catalysed Huisgen 1,3-dipolar azide-alkyne cycloaddition. BTC f with a N-alkylated NMe2 side chain in carbazole moiety showed a ΔTm value of 22.7 K in FRET based DNA melting studies at only 100 nM ligand concentration (ligand:G4=1 : 2) and a Kd of 300 nM and ligand BTC lacking the central NMe2 side chain displayed quite similar binding affinity for c-MYC (ΔTm value of 24 K at 200 nM ligand concentration, ligand:G4=1 : 1 and a Kd of 450 nM). The ligands were titrated to c-MYC22 G4 DNA, a stabilized mutant of wild-type c-MYC Pu27 that presents additional signals presumably belonging to multiple structures of the 3'-capping formed by the TAA-flanking nucleotides. Both the ligands were found to induce an increase in signal intensity of the minor conformation (b) of c-MYC22 while reducing the signals of the major conformation (A) (Figure #chem202101866-fig-0003#3 and Figure #chem202101866-fig-0009#9). Moreover, upon ligand addition, new signals appeared in the imino region (e.g., at 10.5 ppm) suggesting that the binding-competent minor conformation (b) underwent small conformational changes (B*) (like restructuring of the capping structure) in the presence of BTC and BTC-f. Interestingly, these ligands were found to stabilize the c-MYC minor conformation formation even in the absence of any added K+ ion.

32807_chem202101866-fig-0004.jpg

Figure 4Open in figure viewerPowerPointChemical structures of carbazole ligands capable of binding to the major conformation of G4 DNA.

Many carbazole ligands bind to the major conformation of the G4 target sequence without a recognizable effect on the conformational equilibrium or the overall topology. The carbazole ligands 3ao, 3ap, BMVEC and 2f (Figure #chem202101866-fig-0004#4) can be categorized within this group. Ligand 3ao (alternatively known as Tz1) and 3ap are mono-triazolyl carbazole derivatives that have been developed by in situ copper free click reaction in the presence of c-MYC G4.Au@Fe3O4 and BCL2 G4.Au@Fe3O4 nanotemplates. 3ao containing two aryl carboxamide motifs is capable of specifically recognizing c-MYC parallel G-quadruplex (Kd=170 nM) over i-motifs and BCL2 G4 while the meta-isomer 3ap prefers to bind to BCL2 G4 DNA with a dissociation constant in comparatively high micro molar regime (Kd=0.68 μM). In case of 3ao, significant line broadening of the imino signals has been observed at a ratio of [3ao]:[c-MYC]=0.25 indicating strong binding interactions. These molecules have been shown to reduce off-target effects in biological experiments; ligand 3ao and 3ap could selectively downregulate the expression of c-MYC and BCL2 genes, respectively via stabilization of promoter quadruplexes and eventually lead to apoptosis in cancer cells. Another carbazole ligand 3be generated from in situ cycloaddition with c-MYC i-motif DNA nanotemplate (c-MYC C4.Au@Fe3O4) results in general line broadening of the characteristic imino signals (15–16 ppm) for i-motif C–C+ base pairs as well as in the aromatic region of c-MYC i-motif. Tang and coworkers reported two carbazole iodides MVEC [(3-Bis-(1-methyl-4-vinylpyridium iodine) 9-ethyl-carbazole)] and BMVEC [(3,6-Bis-(1-methyl-4-vinylpyridium iodine) 9-ethyl- carbazole)] to explore the binding mode of carbazole derivatives with telomeric G4 DNA (Figure #chem202101866-fig-0004#4). BMVEC has cationic charge on the two pendant groups of pyridinium rings of 9-ethylcarbazole while MVEC has one cationic charge pendant group. CD spectroscopy showed that BMVEC significantly stabilizes both parallel Hum6 (T2AG3) (Δ=16 °C at ligand:G4=1 : 1) and antiparallel Hum22 (AT2G3T2AG3) telomeric G4 DNA (Δ=7 °C at ligand:G4=1 : 1) but MVEC prefers to bind to parallel Hum6 DNA (Δ=10 °C at ligand:G4=1 : 1). An induced CD-signal was observed near 300 nm on binding of BMVEC to basket-type antiparallel Hum22. 1D 1H NMR spectroscopy further revealed that BMVEC interacts with Hum6 G4 DNA in slow exchange on NMR time scale. The imino proton signals (10–12 ppm) of Hum6 G4-G6 gradually disappeared and a new set of imino proton resonances appeared at a 1 : 1 BMVEC:Hum6 stoichiometry. The authors conclude that at low concentration, BMVEC initially binds to the groove region and at higher concentrations, it then end stacks onto the G-tetrad near the G6 residue. NMR titrations with MVEC result in progressive upfield shifts of G4-G6 signals and line broadening of G4 signal at MVEC:Hum6 ratio of 1 : 1 indicating significant interaction with the Hum6 G4 residue in an intermediate exchange regime. In case of Hum22, BMVEC induces only significant line broadening of the imino resonance signals (10.5–12.0 ppm). Both these carbazole iodides were shown to exhibit anti-proliferative activity against hepatocellular and colorectal carcinoma cell lines. A novel multi-carbazole derivative consisting of a tri(carbazole)benzene core was identified as a promising G4-stabilizer by Mergny and Smith group. They envisaged that the hetero-polyaromatic scaffold with C3-symmetry and a crescent shaped architecture could enable effective stabilization by π-stacking interactions and the three protonated N-alkylamine side chains angled in appropriate positions on the scaffold could facilitate selective interactions with loop and groove regions of G-quadruplexes (Figure #chem202101866-fig-0004#4). The multi-carbazole ligand 2f with C3-alkyl chain length and pyrrolidine cyclic amine substituents exhibited a high degree of stabilization for hybrid and antiparallel 21-mer telomeric G4 DNA in DNA-based FRET melting studies (ΔTm=14.3 °C; ligand conc. 5 μM, ligand:G4=25 : 1) and for parallel KRAS G4 DNA (ΔTm=22.7 °C; at 5 μM ligand concentration). 1D 1H NMR spectroscopy with KRAS G4 demonstrated severe line broadening and decreased intensity of all the imino signals (10–12 ppm) of the guanine residues involved in the formation of G-quartets.

32807_chem202101866-fig-0005.jpg

Figure 5Open in figure viewerPowerPointChemical structure of carbazole 4e and Cz1.

For the ligands 4e and Cz1, a more specific interaction with c-MYC22 could be observed (Figure #chem202101866-fig-0005#5). c-MYC22 is a stabilized mutant of the purine-rich nuclease hypersensitive element NHEIII1 located upstream of the c-MYC promoter sequence.

32807_chem202101866-fig-0006.jpg

Figure 6Open in figure viewerPowerPointImino region of 1D 1H NMR spectrum of the c-MYC22 G-quadruplex DNA with increasing [ligand]:[DNA] molar ratio of (a) 4e and (b) Cz1. b) The spectra were recorded at 298 K, 600 MHz. Experimental conditions: 100 μM DNA in 25 mM Tris ⋅ HCl (pH 7.4) buffer containing 100 mM KCl in 5 % d6-DMSO/95 % H2O. c) sequence of c-MYC22 with the numbering used for assignment that has been transferred from Ambrus et al..29 d) and e) Mapping of the observed changes in 1D 1H NMR spectra upon addition of 4e and Cz1, respectively, on the solution NMR structure of c-MYC22 (PDB:1XAV). T, A and G are light blue, dark blue and grey, respectively. GH1 and GH8, TH6, AH8 and AH2 that experience chemical shift perturbation are highlighted with red, light blue and dark blue spheres, respectively. The mapping reveals that both ligands affect similar signals, which are located at the external tetrads and the groove formed by the G-stretches 8–9-10 and 13–14-15. Therefore, an identic binding site for both ligands is expected. However, Cz1 shows stronger binding as compared to 4e.

Carbazole ligand 4e was developed by Dash group via dynamic combinatorial approach (DCC) from an imine-based combinatorial library in the presence of c-MYC G4 DNA (NMR data not published). Both these ligands exhibited high selectivity for quadruplex over duplex DNA. The ligands showed a clear binding preference for c-MYC G4 [4e: ΔTm=23.4 °C (ligand concentration 1 μM, ligand:G4=5 : 1), Kd=1.08 μM; Cz1: ΔTm=15.8 °C (ligand concentration 1 μM, ligand:G4=5 : 1), Kd=0.21 μM] and acted as endogenous transcriptional regulators of c-MYC gene in cancer cells. These compounds were subsequently shown to induce cytotoxicity and apoptosis in cancer cell lines. In case of 4e, the imino signals of G8, G9, G13 and G14 were more affected as well as one or two signals in the overlapping region of G10, G15 and G19 (Figure #chem202101866-fig-0006#6a) (numbering according to Figure #chem202101866-fig-0006#6c). In the aromatic region, line broadening and CSPs were also observed for signals of the flanking nucleotides T1, G2, T20, A21 and A22 (data not shown). In molecular docking studies, binding to the 3'-external tetrad was observed. The NMR data provide supporting evidence that both external tetrads act as potential binding sites as well as interaction of 4 e with the groove formed by G8-G9-G10 and G13-G14-G15. Titration of Cz1 resulted in strong line broadening of the imino protons of c-MYC22 (NMR data not published). However, upon a 1 : 1 molar ratio of Cz1:c-MYC22, the binding was shifted into fast exchange resulting in CSPs and an increase in signal intensity (Figure #chem202101866-fig-0006#6b). Altogether, most affected signals were also located at the groove formed by the G-stretches 8–9-10 and 13–14-15 and the aromatic signals of T1, G4, T20 and A22 (data not shown). Thus, a similar binding site for Cz1 like 4 e is expected (Figure #chem202101866-fig-0006#6d–e). BMVC (3,6-bis(1-methyl4-vinylpyridium)carbazole diiodide) is one of the few ligands that binds in slow exchange rate to c-MYC22 and the NMR solution structures of the 1 : 1 (PDB code 6JJ0) and 2 : 1 (PDB code 6O2L) complex were reported (Figure #chem202101866-fig-0007#7, numbering according to Figure #chem202101866-fig-0006#6c). The crescent shaped ligand is highly sensitive and efficient light-up fluorescent probe for G4 DNA due to its good water solubility and biocompatibility. NMR studies revealed that the ligand first binds to the 5'-end and forms a tight complex. Moreover, at a ligand:DNA molar ratio of 1 : 1, signals of the 2 : 1 complex are also coming up with the 3' external tetrad as additional binding site. Interestingly, the 5’ (T4, G5 and A6) and 3’ flanking segments underwent large conformational changes creating additional planes that serve as potential binding sites for BMVC. BMVC also adopts a contracted conformation that allows optimal π-π stacking with both the external tetrads and 5’- and 3’ capping structures. 2D NMR analysis provided detailed binding mode of BMVC: The G2 and A3 residues at the 5’-capping segment stack over G4 of the 5’-tetrad while A3 inserts into the arc of BMVC forming a BMVC-A3 plane. The T1 stacks on the central carbazole core of BMVC further stabilizing the BMVC-A3 plane. At the 3‘-end, T20 forms a new plane with BMVC and stack upon 3’-tetrad. However, the 3'-binding site showed a weaker affinity (as the A21 and A22 residues do not participate in stacking interactions with BMVC-T20 plane) and displayed a high sensitivity to T20A mutation or truncation of the TAA-flanking sequence. These indicate that BMVC binds the MYC G4 with greater affinity (Kd=36 nM) and specificity, considerably stronger than other reported G4 ligands and represses c-MYC gene expression in cancer cells.

32807_chem202101866-fig-0007.jpg

Figure 7Open in figure viewerPowerPoint(a) Chemical structure of ligand BMVC. (b) top view onto the 5’-end of the 1 : 1 BMVC:c-MYC22 complex (PDB: 6JJ0).30 (c) side view of the 1 : 1 complex and (d) side view of the 2 : 1 BMVC : c-MYC22 complex (PDB: 6O2L),30 BMVC* indicates the weaker binding site. A-, T- and G-residues are depicted in green, blue and grey, respectively.

BMVC (3,6-bis(1-methyl4-vinylpyridium)carbazole diiodide) is one of the few ligands that binds in slow exchange rate to c-MYC22 and the NMR solution structures of the 1 : 1 (PDB code 6JJ0) and 2 : 1 (PDB code 6O2L) complex were reported (Figure #chem202101866-fig-0007#7, numbering according to Figure #chem202101866-fig-0006#6c). The crescent shaped ligand is highly sensitive and efficient light-up fluorescent probe for G4 DNA due to its good water solubility and biocompatibility. NMR studies revealed that the ligand first binds to the 5'-end and forms a tight complex. Moreover, at a ligand:DNA molar ratio of 1 : 1, signals of the 2 : 1 complex are also coming up with the 3' external tetrad as additional binding site. Interestingly, the 5’ (T4, G5 and A6) and 3’ flanking segments underwent large conformational changes creating additional planes that serve as potential binding sites for BMVC. BMVC also adopts a contracted conformation that allows optimal π-π stacking with both the external tetrads and 5’- and 3’ capping structures. 2D NMR analysis provided detailed binding mode of BMVC: The G2 and A3 residues at the 5’-capping segment stack over G4 of the 5’-tetrad while A3 inserts into the arc of BMVC forming a BMVC-A3 plane. The T1 stacks on the central carbazole core of BMVC further stabilizing the BMVC-A3 plane. At the 3‘-end, T20 forms a new plane with BMVC and stack upon 3’-tetrad. However, the 3'-binding site showed a weaker affinity (as the A21 and A22 residues do not participate in stacking interactions with BMVC-T20 plane) and displayed a high sensitivity to T20A mutation or truncation of the TAA-flanking sequence. These indicate that BMVC binds the MYC G4 with greater affinity (Kd=36 nM) and specificity, considerably stronger than other reported G4 ligands and represses c-MYC gene expression in cancer cells.

32807_chem202101866-fig-0008.jpg

Figure 8Open in figure viewerPowerPointStructures of topology selective ligand BMVC-1c-Br and specific G4-conformation selective carbazole derivatives BTC, BTC f and carbazole-thiazole orange conjugate Cz-TO.

Beside ligands that bind specifically to a particular promoter G4 there are also topology-specific ligands. The group of Ta-Chau Chang synthesized a derivative of BMVC with a high selectivity towards parallel G4s: BMVC-12C-Br (Figure #chem202101866-fig-0008#8). Owing to the presence of dodecyl alkyl chain, the derivative acts as a surfactant between oil and water. Based on this hydrophobicity, they developed a method to separate G4 of different topologies - the emulsified induced filtration (EIF). A two-phase system containing a mixture of G4s in the aqueous phase and the ligand as surfactant between the phases was used. As model systems, a mixture of the hybrid-type Tel24-M and predominantly parallel Tel19-M telomeric sequences as well as a mixture of the two promoter sequences c-MYC-2345 (parallel) and BCL2mid-M (hybrid) were used. By ultrasonic emulsification, an oil-in-water emulsion has been formed and filtrated through a membrane of mixed cellulose esters (MCEs). With this process, a transfer of the parallel G4 into the emulsion particles by binding to BMVC-12C-Br resulting in a structural separation was nicely reported by NMR. In the conformational selection binding mode, the ligand interacts to a pre-existing conformation of a biomolecule, only present in small amounts and shifts the conformational equilibrium towards the binding competent form. The carbazole ligands BTC and BTC-f (Figure #chem202101866-fig-0008#8), developed by Dash group, select the minor-populated conformer (in Figure #chem202101866-fig-0003#3 designated as b) of c-MYC G4 DNA and binds to it in a specific manner. Both these bis-triazolyl carbazole derivatives were prepared by one-pot Cu(I) catalysed Huisgen 1,3-dipolar azide-alkyne cycloaddition. BTC f with a N-alkylated NMe2 side chain in carbazole moiety showed a ΔTm value of 22.7 K in FRET based DNA melting studies at only 100 nM ligand concentration (ligand:G4=1 : 2) and a Kd of 300 nM and ligand BTC lacking the central NMe2 side chain displayed quite similar binding affinity for c-MYC (ΔTm value of 24 K at 200 nM ligand concentration, ligand:G4=1 : 1 and a Kd of 450 nM). The ligands were titrated to c-MYC22 G4 DNA, a stabilized mutant of wild-type c-MYC Pu27 that presents additional signals presumably belonging to multiple structures of the 3'-capping formed by the TAA-flanking nucleotides. Both the ligands were found to induce an increase in signal intensity of the minor conformation (b) of c-MYC22 while reducing the signals of the major conformation (A) (Figure #chem202101866-fig-0003#3 and Figure #chem202101866-fig-0009#9). Moreover, upon ligand addition, new signals appeared in the imino region (e.g., at 10.5 ppm) suggesting that the binding-competent minor conformation (b) underwent small conformational changes (B*) (like restructuring of the capping structure) in the presence of BTC and BTC-f. Interestingly, these ligands were found to stabilize the c-MYC minor conformation formation even in the absence of any added K+ ion. A fluorescent G4-binding dye - the carbazole-thiazole orange conjugate Cz−TO (Figure #chem202101866-fig-0008#8) was found to bind the BCL2 2345 G4 DNA via end-stacking as the signals of tetrad-forming guanines are more affected by binding. This observation is not comprehensible for us. However, there are some signals (approximately at 11.05, 11.3 and 11.45 ppm) increasing in intensity upon ligand addition to BCL2 2345 that were ignored by the authors. Considering the spectral changes resulted by BTC and BTC f binding, a conformational selection mechanism seems plausible. As only 2 datapoints (0 eq and 0.75 eq) were published, a detailed analysis is not possible. However, Cz-TO specifically enhances its fluorescence intensity over 70-fold in the presence of BCL2 G4 while showing less than 30-fold increase with promoter (c-MYC, c-KIT1 and VEGF) and telomeric G4s and other nucleic acids.

32807_chem202101866-fig-0009.jpg

Figure 9Open in figure viewerPowerPointImino region of 1D 1H NMR spectrum of the c-MYC22 G-quadruplex DNA with increasing [Ligand]:[DNA] molar ratio of (a) BTC and (b) BTC f. The spectra were recorded at 298 K, 600 MHz. Experimental conditions: 100 μM DNA in 25 mM Tris ⋅ HCl (pH 7.4) buffer containing 100 mM KCl in 10 % D2O/90 % H2O. Both ligands bind to a pre-existing minor conformation and induce a change in conformational equilibrium of c-MYC22. The numbering is according to Figure 6c). a) is adapted with permission from Ref. [36]. Published by The Royal Society of Chemistry. b) is adapted with permission from Ref. [37]. Copyright 2015, Springer Nature.

In the conformational selection binding mode, the ligand interacts to a pre-existing conformation of a biomolecule, only present in small amounts and shifts the conformational equilibrium towards the binding competent form. The carbazole ligands BTC and BTC-f (Figure #chem202101866-fig-0008#8), developed by Dash group, select the minor-populated conformer (in Figure #chem202101866-fig-0003#3 designated as b) of c-MYC G4 DNA and binds to it in a specific manner. Both these bis-triazolyl carbazole derivatives were prepared by one-pot Cu(I) catalysed Huisgen 1,3-dipolar azide-alkyne cycloaddition. BTC f with a N-alkylated NMe2 side chain in carbazole moiety showed a ΔTm value of 22.7 K in FRET based DNA melting studies at only 100 nM ligand concentration (ligand:G4=1 : 2) and a Kd of 300 nM and ligand BTC lacking the central NMe2 side chain displayed quite similar binding affinity for c-MYC (ΔTm value of 24 K at 200 nM ligand concentration, ligand:G4=1 : 1 and a Kd of 450 nM). The ligands were titrated to c-MYC22 G4 DNA, a stabilized mutant of wild-type c-MYC Pu27 that presents additional signals presumably belonging to multiple structures of the 3'-capping formed by the TAA-flanking nucleotides. Both the ligands were found to induce an increase in signal intensity of the minor conformation (b) of c-MYC22 while reducing the signals of the major conformation (A) (Figure #chem202101866-fig-0003#3 and Figure #chem202101866-fig-0009#9). Moreover, upon ligand addition, new signals appeared in the imino region (e.g., at 10.5 ppm) suggesting that the binding-competent minor conformation (b) underwent small conformational changes (B*) (like restructuring of the capping structure) in the presence of BTC and BTC-f. Interestingly, these ligands were found to stabilize the c-MYC minor conformation formation even in the absence of any added K+ ion.

32807_chem202101866-fig-0010.jpg

Figure 10Open in figure viewerPowerPointInduced-fit binding mechanism of G4-interactive ligand. Ligand BMVC-8C3O induces conformational transition from (a) hybrid-I and (b) hybrid-II telomeric G4 topology to (d) parallel topology.

Ligand BMVC-8C3O was shown to convert the overall topology of its G4 binding partner (Figure #chem202101866-fig-0010#10). This is typical for an induced fit binding mode. The ligand was prepared by Chang and co-workers via covalent attachment of a tetraethylene glycol (8C3O) terminating in a methyl-piperidinium cation with the carbazole diiodide derivative BMVC. Upon interaction with BMVC-8C3O, the hybrid type Tel23 (hybrid I) (Figure #chem202101866-fig-0010#10a) and Tel25 (hybrid II) (Figure #chem202101866-fig-0010#10b) G4s topologically convert from their apo form to a parallel structure (Figure #chem202101866-fig-0010#10d).

32807_chem202101866-fig-0011.jpg

Figure 11Open in figure viewerPowerPointChemical structures of other G4-interactive carbazole ligands: the carbazole derived (a) fluorescent probes and (b) anti-cancer agents.

The fused heteroaromatic nitrogen containing ring system of carbazole is an effective electron donating optical chromophore. The large π-conjugated scaffold also displayed good Stokes shifts and high quantum yields owing to intramolecular charge transfer (ICT) along with excellent biocompatibility and stability. Thus, carbazoles are widely used as fluorescent turn on probes for sensitive and efficient monitoring of cellular G4s. A number of carbazole based fluorescent ligands with superior selectivity over the duplex DNA were reported (Figure #chem202101866-fig-0011#11a). The triethylene glycol conjugated carbazole derivative Cz-TEG and the groove binder cationic pyridinium containing analogue Cz-Py exhibit superior G4 selectivity and low cytotoxicity that are useful for G4-DNA sensing (Figure #chem202101866-fig-0011#11a). A benzindole substituted carbazole cyanine dye 9E PBIC was reported as a selective fluorescent probe for parallel c-MYC G4 DNA. The dye selectively discriminates parallel c-MYC 2345 G4 over other parallel, antiparallel and hybrid G4 and ss/ds DNAs by showing a 100-fold fluorescence enhancement. Another light up probe, BPBC composed of benzimidazole and carbazole moieties was reported for the selective recognition of parallel G4s. For visualization of cellular G4s, a two-photon fluorescent probe Cz-BT was synthesized by conjugating a benzothiazole moiety with the carbazole core with a diethyl amine side chain at the 9th position. As the probe requires higher wavelength excitation (>700 nm), it could easily avoid the interference of autofluorescence within cellular environment. Later, a bis-carbazolyl methane-based cyanine derivative (Bis-Cz-Cy) was developed for two-photon excited fluorescence (TPEF) microscopy imaging of G4s within cellular environment. The probe shows excitation maxima at 820 nm and thus is practically more effective for non-invasive live cell imaging due to deep penetration, low photobleaching and high spatial resolution. Another two-photon excited ligand EBMVC-B was employed with a modified G-rich oligonucleotide (5'-TGAG3AG4-3'-3'-T-5') to develop an intermolecular G4-ligand complex that acts as a fluorescent sensor of blood potassium levels (Figure #chem202101866-fig-0011#11a). The G-rich oligonucleotide contains an inverted thymine nucleotide whose 3’-terminus was connected to the 3’-terminus of the upstream nucleotide blocking nuclease activity in biological fluids. The oligonucleotide folds into an intermolecular G4 specifically in the presence of K+ ions and subsequently interacts with the G4-binder EBMVC-B producing fluorescence turn on signals in response to K+ ions. Interestingly, the G4-EBMVC-B complex selectively detects K+ ions with no interference from other competitive ions like Na+, Li+, Ca2+, Mg2+, NH4+, Zn2+, Cu2+ etc. under physiological condition. The fluorescent carbazole iodide derivative BMVC exhibits high sensitivity for quadruplexes with anti-anti-anti-anti and anti-anti-syn-syn arrangements (anti-parallel topology) while showing weak fluorescence responses for duplexes and quadruplexes with anti-syn-anti-syn arrangement. Further, another BMVC analogue containing two ortho-pyridinium groups instead of para-pyridinium, o-BMVC has been developed showing a large contrast in fluorescence decay time, binding affinity, and fluorescent intensity between G4 structures. Owing to its longer fluorescence decay times, the ligand is able to specifically visualize the location of G4 foci in living cells. By staining several cancer cells and normal cells with o-BMVC, it could be demonstrated that significantly more G4 foci are present in cancer cells. In further studies with patient cells, o-BMVC appears applicable as biosensor for human head and neck cancer. Several derivatives of carbazoles have been explored as effective G4 targeting anticancer agents (Figure #chem202101866-fig-0011#11b). The group of Muniyappa and Bhattacharya introduced carbazole-benzimidazole conjugates (CMP and CHP) for selective inhibition of telomerase activity. Both these conjugates demonstrated fluorescence light-up in the presence of Hum21 telomeric G4 DNA and structural inversion from the K+-stabilized hybrid G4 structure into a stable, telomeric parallel G4 DNA while showing distinct induced circular dichroism (ICD) signal with Na+ stabilized anti-parallel Hum21. The bis-benzimidazole ligands could inhibit telomerase function and induce cytotoxicity in telomerase positive cancer cells by selectively penetrating the nucleus of cancer cell lines over the normal telomerase negative primary cells. Inspired from these findings the group again developed six new carbazole based benzimidazoles for selective inhibition of telomerase activity (Figure #chem202101866-fig-0011#11b). Among these series of carbazoles, the dimeric bis-benzimidazoles with hydroxyethyl substituted piperazines (Cz-BzI 5–6) demonstrated significantly high binding and stabilization capability for telomeric G4 DNA and induced cancer cell specific apoptosis. In a subsequent study, the group synthesized a series of other carbazole based benzimidazoles for targeting G4 structures in oncogene promoter regions. All the ligands have been found to stabilize the G4 DNA of c-MYC, c-KIT1, c-KIT2, VEGF and BCL2 gene promoters and repress the expression of the oncogenes in cancer cells but they fail to discriminate among diverse G4 topologies. Chen, Huang and Tan group developed a four-leaf clover-like ligand IZCZ-3 for specific stabilization of parallel topology of promoter c-MYC G4 (Figure #chem202101866-fig-0011#11b). The ligand was synthesized by conjugating a carbazole moiety with a triaryl-substituted imidazole in a one-pot condensation reaction. The ligand IZCZ-3 imparted high stability to c-MYC G4 and showed little effect on other telomeric and promoter quadruplexes HRAS, c-KIT1, BCL2, KRAS, RET and PDGFA as well as i-motif, triplex, duplex and hairpin DNA. Molecular docking studies revealed that IZCZ-3 stacked perfectly on the terminal G-tetrad and the central imidazole moiety was found to be located onto the ion channel of the c-MYC G4. The conjugate efficiently inhibits c-MYC transcription via binding and stabilizing the promoter c-MYC G4 in cells leading to cell cycle arrest at G0/G1 phase and apoptosis. It also inhibits tumor growth in a human cervical squamous cancer xenograft.

2985_chem201901961-fig-0001.jpg

Figure 1Open in figure viewerPowerPointAI-2 signaling and inhibition of signal synthesis through LuxS by the fimbrolide (1). The AI-2 signal is R-THMF in enterobacteria and the boric acid ester of S-THMF for Vibrio species.

Although AI-2 is the most common quorum-sensing signal used by many different species and produced by gram-negative as well as gram-positive bacteria, only a few approaches have been reported in which AI-2 signaling has been targeted for swarming inhibition. For E. coli, swarming-cell differentiation has been shown to be regulated by the central FlhC2D2 master regulator the transcription of which is presumably activated by AI-2 through the two-component system QseBC (Figure #chem201901961-fig-0001#1). The FlhC2D2 regulator in turn activates the fliA gene which encodes a sigma factor specific for flagellar operons. In pathogenic E. coli strains, AI-2 plays an important role for virulence and a nanoemulsion of 2.5 % limonene was found to interfere with AI-2 quorum sensing of E. coli O157:H7 (EHEC). Hereby, both swimming and swarming motilities were repressed. The biosynthesis of the AI-2 signal is carried out through cleavage of S-ribosylhomocysteine by LuxS (Figure #chem201901961-fig-0001#1). For signal detection, AI-2 is phosphorylated and derepresses transcription of target genes through binding to LsrR. Fimbrolides, a class of halogenated furanones, are important inhibitors of the LuxS signal synthase and thereby of quorum sensing by AI-2. Fimbrolides have been initially discovered as natural products from the marine red alga Delisea pulchra and a great diversity of natural and synthetic derivatives has been investigated. A furanone (1) inhibited biofilm formation and swarming but not swimming motility in E. coli and strongly antagonized the quorum sensing by AI-2. The same furanone also inhibited swarming of B. subtilis.

2985_chem201901961-fig-0002.jpg

Figure 2Open in figure viewerPowerPointAHL-based quorum sensing in enterobacteria (left) and Pseudomonas aeruginosa (right) and corresponding inhibitors that lead to inhibition of swarming motility.

Halogenated furanones have been additionally described to target the LuxE subunit of the luciferase complex of Vibrio and N-acyl-homoserine lactones (AHL)-based quorum sensing through destabilization of homologues of the LuxR-regulator. AHLs are the largest class of quorum-sensing signals in gram-negative bacteria that are produced through N-acylation of S-adenosyl-l-methionine (SAM) and cyclization to γ-lactones by homologues of the synthase LuxI (Figure #chem201901961-fig-0002#2, left). The signals are detected by binding to homologues of the transcription factor LuxR. In many species, AHLs have major impact on swarming regulation because they are regulators of, for example, the biosynthesis of the surfactant serrawettin through LuxR in Serratia spp. (Figure #chem201901961-fig-0002#2, left). Serrawettin promotes swarming motility by reduction of surface tension. Consequently, targeting AHL-based quorum sensing has been of central interest for swarming inhibition. Two differently brominated furanones (1) and (2) of D. pulchra inhibited AHL-dependent swarming motility of the enterobacterium Serratia liquefaciens which was restored in an AHL-negative mutant by supplementation with N-butanoyl-l-homoserine lactone (C4-HSL). The mechanism of swarming inhibition involves the blockage of the biosynthesis of the surfactant serrawettin W2 as mentioned above through binding to LuxR. Surprisingly, only one of four brominated furanones isolated from D. pulchra inhibited swarming of the uropathogen P. mirabilis. All four furanones (1–4) inhibited swarming of different uncharacterized environmental strains of bacteria isolated from rock surfaces as well as from samples of D. pulchra. Targeting AHL receptors (LuxR homologues) has been maybe the most frequently employed strategy for interfering with AHL-based quorum sensing. Especially AHL signal analogs that mimic the native AHLs are promising candidates for inhibitors. For example, AHL signaling can be inhibited by synthetic N-acyl cyclopentylamides (Figure #chem201901961-fig-0002#2, left). A mutant strain of enterobacterium Serratia marcescens that was unable to produce AHLs was nonmotile in a swarming assay. Exogenous supply of N-hexanoyl-l-homoserine lactone (C6-HSL) restored the swarming phenotype and competition with 50 μm N-nonanoyl cyclopentylamide (5) resulted in complete swarming inhibition. Some species such as the human pathogen P. aeruginosa even comprise more than one AHL-based quorum sensing system. In P. aeruginosa, the LuxI/LuxR homologues RhlI/RhlR and LasI/LasR utilize the signals N-butanoyl-l-homoserine lactone (C4-HSL) and N-(3-oxododecanoyl)-l-homoserine lactone (3-oxo-C12-HSL), respectively (Figure #chem201901961-fig-0002#2, right). These AHL-based quorum-sensing systems are hierarchically interconnected by the master regulator LasR with further quorum-sensing and two-component systems to control virulence in P. aeruginosa. Recently discovered clinical isolates of P. aeruginosa from cystic fibrosis patients revealed an exceptional plasticity in the hierarchical regulation of quorum sensing whereby the RhlI/RhlR system could compensate the loss of functional LasR. The production of the swarming surfactant rhamnolipid which Pseudomonas requires to lower surface tension is RhlR regulated by transcription of the rhl genes. The Meijler group developed synthetic AHLs with an isothiocyanate (ITC) warhead mimicking 3-oxo-C12-HSL of P. aeruginosa. These compounds and especially a β-fluorinated derivative ICT-F (6) covalently blocked the LasR receptor at Cys79 and inhibited swarming motility by 44 % at 150 μm and by 34 % at 20 μm and also reduced pyocyanin production (Figure #chem201901961-fig-0003#3 a). In contrast, the brominated ITC-Br (7) did not bind covalently and was a LasR agonist that increased swarming motility up to 2.5-fold at 20 μm of ITC-Br in P. aeruginosa PA14. High-throughput screening of a compound library against reporter strains revealed the plant-produced flavonoids phloretin, chrysin, and naringenin as potent inhibitors of the LasR and RhlR quorum-sensing receptors of P. aeruginosa. Additionally, also flavonoids like quercetin (8), baicalein, and pinocembrin exhibited inhibitory activity whereby the presence of a specific pattern of two hydroxyl-groups on the flavonoid A-ring appeared to be required for activity (Figure #chem201901961-fig-0002#2). Flavonoids were found to be allosteric inhibitors of these quorum-sensing receptors and prevented their binding as transcription factors to DNA. Two of the most active compounds, phloretin (9) and 7,8-dihydroxyflavone were finally tested on quorum-sensing-controlled behaviors of P. aeruginosa and completely abrogated swarming at 100 μm. The flavonoid quercetin (8) considerably reduced swarming motility of P. aeruginosa and Yersinia enterocolitica at 132 μm. In Proteus vulgaris, 50 μm of quercetin (8) not only inhibited the production of N-octanoyl-l-homoserine lactone (C8-HSL) by 81 % and caused an almost equal reduction in swarming area, but also supposedly interfered with swarming by binding to the sigma factor FliA which regulates flagellar operons (Figure #chem201901961-fig-0002#2, left). A virtual docking-approach against the AHL receptor LasR identified salicylic acid and chlorzoxazone as potential quorum-sensing inhibitors of P. aeruginosa which was confirmed biochemically through LasR and additionally RhlR and resulted in inhibition of swarming of S. liquefaciens in the millimolar range. Also complex natural-product mixtures and extracts have been found to exhibit quorum-sensing inhibiting activities affecting swarming behavior. For example, propolis—bee glue—antagonized AHL-based quorum-sensing signaling in RhlR- and LasR-dependent reporter strains and reduced swarming activity of P. aeruginosa. Some signals may even lead to crosstalk between different quorum-sensing systems. An example are diketopiperazines (DKPs), cyclic dipeptides involved in trans-kingdom interactions of bacteria with eukaryotes and inter-species signaling between gram-negative and gram-positive bacteria. DKPs such as cyclo(ΔAla-l-Val), cyclo(L-Pro-l-Tyr) (10), and cyclo(l-Phe-l-Pro) were isolated from culture supernatants of various gram-negative bacteria including Pseudomonads, P. mirabilis, Citrobacter freundii, and Enterobacter agglomerans and recombinant LuxR-based AHL biosensor assay revealed that they compete with the site of AHL binding and thereby antagonize quorum sensing. Cyclo(l-Pro-l-Tyr) (10) reduced swarming of wild type S. liquefaciens as well as of a ΔswrI mutant for which swarming motility depends on external supply of N-butanoyl-l-homoserine lactone (C4-HSL) (Figure #chem201901961-fig-0002#2, left). In many cases, however, the cellular targets of quorum-sensing inhibitors or their compound classes have not yet been clearly identified. Hereby, phenotypic or transcriptional analyses have often tentatively pointed to interference with AHL-based quorum sensing as likely mechanism of swarming inhibition. An AHL-derived N-decanoyl-l-homoserine benzyl ester (11) for example inhibited swarm expansion and dendritic swarming pattern between 50 and 100 μm and reduced expression of both las and rhl genes as well as production of virulence factors including rhamnolipids (Figure #chem201901961-fig-0002#2, right). At 136 μm and higher concentrations, curcumin (12) inhibited swarming motility of E. coli, P. aeruginosa PAO1, P. mirabilis, and S. marcescens and interfered with AHL-based quorum sensing in a violacein assay (Figure #chem201901961-fig-0003#3 b). At high concentrations of around 1.5 mm, caffeine inhibited AHL production in P. aeruginosa and reduced swarming motility and zingerone inhibited swarming, swimming, and twitching motility at 5 mm and also decreased the production of AHLs. Many further natural products and synthetic compounds have been postulated to inhibit quorum sensing of P. aeruginosa at relatively high concentrations through LasR whereby swarming motility, but not growth, was inhibited. Examples are, trans-anethole with a reduction of swarming motility by 64 % at 6 mm or pyridoxal lactohydrazone with a reduction of swarming motility by about 35 % at 32 μm and ≈70 % at 126 μm.

2985_chem201901961-fig-0003.jpg

Figure 3Open in figure viewerPowerPointa) Covalent inhibition of LasR by the 3-oxo-C12-HSL analogue ICT-F (6) causing reduction in swarming motility in P. aeruginosa. Quorum sensing and swarming inhibitors b) curcumin and c) phytol which causes down-regulation of flhDC expression.

Some species such as the human pathogen P. aeruginosa even comprise more than one AHL-based quorum sensing system. In P. aeruginosa, the LuxI/LuxR homologues RhlI/RhlR and LasI/LasR utilize the signals N-butanoyl-l-homoserine lactone (C4-HSL) and N-(3-oxododecanoyl)-l-homoserine lactone (3-oxo-C12-HSL), respectively (Figure #chem201901961-fig-0002#2, right). These AHL-based quorum-sensing systems are hierarchically interconnected by the master regulator LasR with further quorum-sensing and two-component systems to control virulence in P. aeruginosa. Recently discovered clinical isolates of P. aeruginosa from cystic fibrosis patients revealed an exceptional plasticity in the hierarchical regulation of quorum sensing whereby the RhlI/RhlR system could compensate the loss of functional LasR. The production of the swarming surfactant rhamnolipid which Pseudomonas requires to lower surface tension is RhlR regulated by transcription of the rhl genes. The Meijler group developed synthetic AHLs with an isothiocyanate (ITC) warhead mimicking 3-oxo-C12-HSL of P. aeruginosa. These compounds and especially a β-fluorinated derivative ICT-F (6) covalently blocked the LasR receptor at Cys79 and inhibited swarming motility by 44 % at 150 μm and by 34 % at 20 μm and also reduced pyocyanin production (Figure #chem201901961-fig-0003#3 a). In contrast, the brominated ITC-Br (7) did not bind covalently and was a LasR agonist that increased swarming motility up to 2.5-fold at 20 μm of ITC-Br in P. aeruginosa PA14. In many cases, however, the cellular targets of quorum-sensing inhibitors or their compound classes have not yet been clearly identified. Hereby, phenotypic or transcriptional analyses have often tentatively pointed to interference with AHL-based quorum sensing as likely mechanism of swarming inhibition. An AHL-derived N-decanoyl-l-homoserine benzyl ester (11) for example inhibited swarm expansion and dendritic swarming pattern between 50 and 100 μm and reduced expression of both las and rhl genes as well as production of virulence factors including rhamnolipids (Figure #chem201901961-fig-0002#2, right). At 136 μm and higher concentrations, curcumin (12) inhibited swarming motility of E. coli, P. aeruginosa PAO1, P. mirabilis, and S. marcescens and interfered with AHL-based quorum sensing in a violacein assay (Figure #chem201901961-fig-0003#3 b). At high concentrations of around 1.5 mm, caffeine inhibited AHL production in P. aeruginosa and reduced swarming motility and zingerone inhibited swarming, swimming, and twitching motility at 5 mm and also decreased the production of AHLs. Many further natural products and synthetic compounds have been postulated to inhibit quorum sensing of P. aeruginosa at relatively high concentrations through LasR whereby swarming motility, but not growth, was inhibited. Examples are, trans-anethole with a reduction of swarming motility by 64 % at 6 mm or pyridoxal lactohydrazone with a reduction of swarming motility by about 35 % at 32 μm and ≈70 % at 126 μm. In S. marcescens, production of its red pigment prodigiosin is under control of AHL-based quorum sensing. Methanolic extracts of the benthic brown alga Padina gymnospora inhibited production of this pigment and activity guided fractionation led to α-bisabolol as active compound. Furthermore, α-bisabolol inhibited extracellular protease, biofilm formation and swarming motility at and above 450 μm suggesting interference with AHL-based quorum sensing as mechanism. Swarming was abolished completely at 1.8 mm without inhibiting growth. At much lower concentrations between 17 and 34 μm, phytol (13) reduced virulence factor production of S. marcescens and strongly inhibited swarming motility (Figure #chem201901961-fig-0003#3 c). The activity of phytol was presumably mediated through quorum-sensing inhibition because it resulted in transcriptional down-regulation of many quorum-sensing-controlled genes including the swarming differentiation master-regulator genes flhC and flhD. Finally, treatment of rats with phytol in an acute pyelonephritis model even ameliorated the infection with S. marcescens.

2985_chem201901961-fig-5001.jpg

Scheme 1Open in figure viewerPowerPointThe native metabolites HHQ (15) and PQS (14) and synthetic derivatives with swarming inhibitory activity.

In addition to its multiple quorum-sensing systems, P. aeruginosa also comprises a large diversity of distinct two-component systems regulating virulence. Each of them is composed of a histidine kinase (HK) sensing external stimuli and a response-regulator protein that alters gene expression upon phosphorylation by the kinase. The many two-component systems for Pseudomonas have been shown to be intricately involved in swarming regulation for example through the action of the response regulator GacA, which is activated by the HK GacS. GacA is connected to swarming through the RhlI/RhlR system through several regulatory steps. Benzothiazole-based histidine kinase inhibitors (Rilu-1 (18), Rilu-4 (19), and Rilu-12 (20)) reduced PQS signaling, decreased rhamnolipid production and drastically impaired swarming motility at 200 μm (Figure #chem201901961-fig-0004#4). Gene-expression analysis suggested that these benzothiazoles inhibited the sensory kinase GacS whereby the transcription of the response regulator gacA and also the flagellar regulator fleQ was decreased. In some cases also chemoattractants may be important for swarming motility. This was demonstrated for P. mirabilis on minimal medium, in which swarming depended on the amino acid l-glutamine as signal lead to swarmer-cell differentiation and up-regulation of the expression of flagellin (fliC) and hemolysin (hpmA). The glutamine-analogue γ-glutamyl hydroxamate interfered with this signaling and inhibited swarming at 10 mm.

2985_chem201901961-fig-0004.jpg

Figure 4Open in figure viewerPowerPointInhibition of swarming by histidine kinase inhibitors targeting the two-component system GacSA.

In addition to fatty acids, also other surface-active substances are known to inhibit swarming. The swarming inhibiting effect against P. mirabilis in the case of homologous sodium alkylsulfates increased with chain length from hexyl- (20–30 mm) to tetradecyl sulfate (0.1–0.5 mm) without impaired growth. At 0.5 mm, sodium tetradecyl sulfate completely inhibited swarming of P. mirabilis and impaired swarming already at 0.1 mm supposedly either by inhibition of formation of flagella or lysis of existing flagella. The effect of 58 chemical substances including detergents and surfactants was tested against Bacillus swarming. Sodium dodecyl sulfate and bile salts such as sodium taurocholate and sodium desoxycholate strongly inhibited or completely blocked swarming of different strains of B. subtilis, Bacillus alvei, Bacillus coagulans, and Bacillus circulans in the lower millimolar range, whereas polysorbates (Tween 20–80) even promoted swarming. Bile salts also inhibited swarming of enterobacteria such as P. mirabilis. Rhamnolipids of P. aeruginosa are a class of native surfactants with dual roles in reducing surface tension and modulating tendril formation. Although a rhlA mutant deficient in biosynthesis of all rhamnolipids as well as their β-d-(β-d-hydroxyalkanoyloxy)alkanoic acid (HAA) precursor is unable to swarm, the rhlB and rhlC mutants exhibit altered, irregular tendril patterns (Figure #chem201901961-fig-0005#5 a). Purified rhamnolipids even can inhibit swarming of wild-type P. aeruginosa, demonstrating their important roles in spatial modulation of motility in swarming colonies. A library of synthetic farnesyl-modified disaccharides mimicking rhamnolipids of P. aeruginosa PAO1 was explored for effects on swarming motility and quorum sensing. Many of these compounds promoted swarming at low concentrations and inhibited swarming at higher concentrations. While the farnesylated disaccharides SFβM (33) and SFβC (34) completely inhibited swarming of wild type P. aeruginosa PAO1 already at 20 and 25 μm, respectively, the closely related compound DβC (35) with a dodecyl chain rescued a rhlA mutant at 20 μm and did not inhibit swarming of wild type PAO1 up to 85 μm (Figure #chem201901961-fig-0005#5 b). This indicates that also the lipid component has major impact for controlling motility. A sulfate functionalized saturated farnesol (36) even inhibited swarming completely between 5 and 10 μm (Figure #chem201901961-fig-0005#5 b). It was proposed that different saccharide or lipid-binding receptors in the outer membrane may have been responsible for these activities. Similar to some fatty acids discussed before, these rhamnolipid mimetics may thus act on regulatory level.

2985_chem201901961-fig-5002.jpg

Scheme 2Open in figure viewerPowerPointAntibiotics inhibiting swarming at sublethal concentrations.

Each bacterial flagellum consists of a long helical protein filament which connects through a hook to the basal body in the cell envelope. Rotation of the motor complex in the membrane is powered by the transport of protons or sodium ions across the membrane. The rotor is surrounded by a ring of membrane-anchored stator complexes that comprise the corresponding ion channels and their interactions with the rotor generate the torque for the rotation of the flagellum (Figure #chem201901961-fig-0006#6). Most bacterial species possess multiple stator systems which can engage in highly dynamic rotor–stator interactions tuning the flagellar motor. The incorporation and exchange of stators in the motor complex depends on diverse environmental factors like the level of viscous drag or sodium-ion concentration but is also regulated by the intracellular second messenger cyclic diguanylate (c-di-GMP). In P. aeruginosa, motility is mediated by one rotor with two sets of stators, MotAB and MotCD. Although MotCD is required for swarming, the MotAB stator represses swarming motility. Under high c-di-GMP concentrations stator selection is in favor of MotAB and thereby c-di-GMP inhibits swarming. Also in other species elevated c-di-GMP levels lead to inhibition of motility. Intracellular c-di-GMP levels are controlled by multiple diguanylate cyclases (DGCs) which produce c-di-GMP from two molecules of GTP and phosphodiesterases (PDEs) that hydrolyze c-di-GMP (Figure #chem201901961-fig-0006#6). Different DGCs and PDEs may hereby control c-di-GMP on local and global scale in the cell and integrate diverse signals and stimuli. In a positive feedback regulation, disengaged MotCD stators further stimulate DGC activity, thereby block motility and support biofilm formation. Inhibitors of DGCs and PDEs can be designed to modulate c-di-GMP levels. Zheng et al. reported a benzoisothiazolinone derivative (37) which was found by in silico screening against the structure of an E. coli PDE. This compound inhibited selectively c-di-GMP hydrolysis of the locally acting PDE RocR of P. aeruginosa with a Ki of 83 μm, but did not inhibit three other PDEs of P. aeruginosa whereby global cellular c-di-GMP levels remained unaffected (Figure #chem201901961-fig-0006#6). Inhibition of RocR at 100 μm completely suppressed swarming but did not increase biofilm production. Another strategy to interfere with swarming motility involves direct blocking of the corresponding flagellar motor. Phenamil (38) and amiloride (39) are inhibitors of Na+-driven motors and have been used to dissect motor functions in different bacterial models such as Vibrio and Bacillus (Figure #chem201901961-fig-0006#6). Both compounds are pyrazine derivatives that block the Na+-channels of the stator complexes and thus prevent generating torque for flagellar rotation. High-throughput screening for swarming inhibitors of V. cholerae resulted in a 2,4-diamino quinazoline (40) and derivatives which inhibited swarming with IC50 values in the single-digit micromolar range (Figure #chem201901961-fig-0006#6). These compounds blocked Na+-driven flagellar motors of different Vibrio species but had no effect on the proton-driven flagellar motors of E. coli and the lateral flagella of V. parahaemolyticus.

2985_chem201901961-fig-5003.jpg

Scheme 3Open in figure viewerPowerPointExamples of secondary plant metabolites with swarming-inhibitory activity.

Competitive chemical interactions of bacteria play an important role in multi-species communities in many different environments. Thus, many species may have evolved small molecules to modulate population behaviors of their competitors to their own benefit. This includes interference with swarming motility. For example, the marine bacterium Marinobacter litoralis inhibited swarming of P. aeruginosa by its lipopolysaccharide (LPS) whereas LPS from other species did not affect motility. In another study, the methanol extracts of 72 Actinomycetes isolated from marine invertebrates were screened for activity against P. aeruginosa. Extracts of two strains inhibited at 0.1 mg mL−1 swarming of P. aeruginosa by 90 and 85 %, the major active component of which was cinnamic acid. In addition to small molecules, proteins also may contribute to competitive interactions. This was observed for the soil bacterium and human pathogen Burkholderia pseudomallei that secreted a protein factor to inhibit swarming of Burkholderia thailandensis by damaging or processing of its flagella. Also the competition for resources can influence bacterial motility. Essential trace elements such as ferric iron are highly embattled in the microbial world and bacteria compete for ferric iron by deploying siderophores as high-affinity iron chelators. Availability of ferric iron also controls swarming behavior of V. parahaemolyticus and V. alginolyticus. Although in V. parahaemolyticus iron limitation is essential for swarmer-cell differentiation, V. alginolyticus requires bioavailability of ferric iron for swarming. To sequester ferric iron from the environment, V. alginolyticus encodes many different iron-siderophore receptors in its genome that allow the bacterium to engage in piracy of siderophores produced by other species. A strain of Shewanella algae which was co-isolated with V. alginolyticus from the same seaweed sample evaded this siderophore piracy by producing avaroferrin (41) (Figure #chem201901961-fig-0007#7 a)—a chimera of the homodimeric macrocyclic hydroxamate siderophores putrebactin and bisucaberin. In a disc-diffusion assay on agar, avaroferrin (50 nmol) led to the formation of a zone with inhibited swarming motility of V. alginolyticus whereas the homodimeric siderophores were considerably less active. Other siderophores were inactive (>500 nmol), whereas deferasirox, an artificially optimized iron chelator for which no receptor in V. alginolyticus is available was a potent swarming inhibitor like avaroferrin. These results suggested that evasion of siderophore piracy by the chimeric siderophore of S. algae limited ferric iron uptake and thereby stalled swarming of V. alginolyticus. This mechanism was confirmed by exploiting the promiscuity of the central NRPS-independent siderophore (NIS) synthetases giving access to non-natural ring-size engineered siderophores, which inhibited swarming of V. alginolyticus with potency comparable to avaroferrin. In contrast, S. marcescens swarms only under limitation of ferric iron which is sensed by a two-component system through the endogenously produced iron chelator 2-isocyano-6,7-dihydroxycoumarin (42) (ICDH-Coumarin) (Figure #chem201901961-fig-0007#7 b).

2985_chem201901961-fig-5004.jpg

Scheme 4Open in figure viewerPowerPointDrugs with off-target effects that inhibit swarming behavior.