Vibrational Properties of Benzoxaboroles and Their Interactions with Candida albicans’ LeuRS

Benzoxaboroles have emerged over the past decade mainly due to their growing medicinal importance. Regarding the wide application of IR spectroscopy in the pharmaceutical industry, the vibrational properties of over a dozen of benzoxaboroles were described, based on results of DFT calculations as well as IR and Raman spectra measurements. Investigated series of compounds included the currently available antifungal drug (Tavaborole, AN2690) as well as its derivatives. An intense and well-isolated band corresponding to the B-OH group stretching vibrations was present in all experimental IR spectra in the range of 1446–1414 cm−1 and can be considered as characteristic for benzoxaboroles. The vibrational properties of benzoxaboroles are shown to be affected by the formation of intramolecular as well as intermolecular hydrogen bonds, which should also influence the interactions of benzoxaboroles with biomolecules and impact on their biological functions. Docking studies of the benzoxaboroles’ adenosine monophosphate (AMP) spiroboronates into the Candida albicans leucyl-RS synthetase binding pocket showed that the introduction of an amine substituent has a strong influence on their binding. The determined values of inhibition constants manifest high potential of some of the investigated molecules as possible inhibitors of that enzyme.


IR and Raman Spectra Measurements
Transmission and reflection techniques were used to record infrared spectra. In the case of the transmission technique, the classical method of obtaining a potassium bromide pellet was used. An ATR Golden Gate accessory equipped with a diamond crystal was used in the reflection technique. The spectral resolution was set to 1 cm −1 for the transmission technique and 2 cm −1 for the ATR technique; each spectrum was averaged from 32 scans.
The Raman spectra were collected on a Nicolet Almega Raman dispersive spectrometer. Samples were placed on a glass substrate set on the Peltier cooled Linkam stage. The temperature was stabilized for 10 min before measurements. The spectra were recorded using a diode laser with a 532 nm excitation line and a high-resolution grating of 2400 lines/mm. The exposure time was 15 s for all experiments; each spectrum was averaged from 2 scans. The obtained spectra were developed by correcting the background and baseline and normalizing the scale.

Samples Preparation
Samples of benzoxaboroles were obtained according to previously reported procedures [21,25,62] and their purity was confirmed on the basis of elemental analysis.

Docking Studies
The crystal structure of editing domain of leucyl-RS synthetase from Candida albicans in complex with AN2690 (structure 5AGJ) was taken from the Protein Databank (www. rcsb.org, accessed on 28 February 2018) and the AN2690-AMP ligand was removed prior to docking studies. The protein structure was processed by AutoDock Tools (http:// autodock.scripps.edu/, accessed on 28 February 2018) to remove water molecules, connect hydrogen atoms and calculate partial charges. The charges were calculated on the basis of standard force-field parameters applied in AutoDock 4.2. Since the standard parameters in AutoDock do not cover the boron atom, the following force-field parameters were introduced: Rii (sum of vdW radii of two atoms of the same type in Angstrom) = 1.98; epsii (vdW well depth in kcal/mol) = 0.034; vol (atomic solvation volume in Angstromˆ3) = 4.6000; solpar (atomic solvation parameter) = −0.00110; Rij_hb (H-bond radius of the heteroatom in contact with a hydrogen in Angstrom) = 0.0; epsij_hb (well depth of H-bond in kcal/mol) = 0; H-bond (integer indicating type of H-bonding atom (0 = no H-bond) = 0; rec_index (initialized to −1, but later on holds count of how many of this atom type are in receptor) = −1; bond_index (used in AutoDock to detect bonds; see "mdist.h", enum {C,N,O,H,XX,P,S}) = 4 [63]. To judge the binding of the investigated ligands, the Gibbs free energies as well as the resulting inhibition constants of all the ligands were calculated in Autodock [64]. The Gibbs free energies have been determined according to a modified Wesson and Eisenberg method for protein solvation modeling [65]. In the applied method, the binding energy was determined as the difference between the energy of a protein-ligand complex and the sum of the energies of a protein and a ligand [66]. Table 1 contains wavenumbers characteristic for the benzoxaboroles' vibrations, the predicted ones based on DFT calculations for 1 and its dimer, as well as corresponding signals found in the recorded IR and Raman spectra. A scaling factor of 0.967 was applied for the calculated wavenumbers [67]. Table 1. DFT calculated and experimental IR and Raman frequencies characteristic for the benzoxaboroles' vibrations in 1 and its dimer (cm −1 ).

No
Vibrations Based on DFT Calculations  Table 2 shows maxima of the most characteristic bands, corresponding to the stretching vibrations of the B-OH as well as C-O groups for AN2690 and remaining investigated benzoxaboroles (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15) in DFT calculated as well as experimental IR spectra (cm −1 ). In the case of the calculated C-O stretching vibrations, several bands were observed in the corresponding region of the experimental spectra; therefore, no unambiguous assignment was possible and Table 2 contains exclusively DFT calculated data.

Docking Studies of Benzoxaboroles' AMP Spiroesters with C. albicans LeuRS
Taking all the mentioned structural issues into account (i.e., optical isomerism as well as rings' arrangement in the resulting spiroboronate), 66 different ligands for fifteen investigated compounds (1-15) were prepared. Coordinates of all the input as well as optimized structures are given in Tables S2-S30 (Supplementary Material). The results of the docking experiments for all the ligands are presented in Table S1 (Supplementary Material). Table 3 collects the structures of the lowest energy of binding among all the studied ligands, representing the strongest ligand-enzyme interactions (the lowest inhibition constant). The lowest binding energy, number of structures, mean binding energy, calculated inhibition constant, and number of hydrogen bonds formed with the ligand are included. The values reported previously for the parent drug (AN2690) [49] are given for comparison ( Table 1). The name of the ligand describes its structure; the first number reflects the benzoxaborole studies, then the symbol of the optical isomer (R or S, if applicable), and the last number stands for the arrangement of the benzoxaborole and the phenyl ring in the corresponding spiroboronate. Just to give an example, the following symbol, 3-R-1-AMP, stands for an AMP spiroboronate of benzoxaborole 3 of configuration R, in which the AMP and benzoxabororole ring are at the same side of the structure.

Vibrational Properties of Benzoxaboroles
Despite growing interest in benzoxaboroles as pharmaceutical agents, no vibrational properties of those molecules have been investigated in detail so far. Following the results reported previously for phenylboronic acids and the corresponding boroxines [68], the assignment of signals characteristic of benzoxaborole vibrations in IR spectra was carried out. The assignments were confirmed by a comparison of calculated values with experimental IR spectra (ATR as well as in a KBr pellet) and supported by the analysis of the Raman spectra, which are sensitive to vibrations with high symmetry. Since the formation of a centrosymmetric dimer is a common structural pattern for benzoxaboroles in the solid state, vibrational characteristics of a dimeric structure should be closer to those found in the solid-state samples analyzed by IR. Most of the signals obtained for 1 and its dimer are near each other and in close agreement with the experimental spectra. The most characteristic parts of the calculated spectra of 1, its dimer, and experimental spectra are shown in Figure 3. According to expectations, for signals corresponding to groups engaged in formation of a hydrogen-bonded dimer, considerable differences in frequencies determined for 1 and its dimer can be observed. It accounts for the band corresponding to the B-OH in-plane bending (No. 3, Table 1), which is centered at 918 cm −1 for the monomer and shifts to 1208 cm −1 for the dimer, whereas a signal of 1215 cm −1 is observed in the experimental spectra. A similar shift is observed for the frequency characteristic of the B-OH out-of-plane bending, which is 523 cm −1 for 1 and 767 cm −1 for the corresponding dimer (No. 8, Table 1), and the maximum of the peak in the experimental IR spectrum is at 768 cm −1 . The observed trend is similar to that found in spectra of other compounds able to form dimers or more aggregated structures through the use of hydrogen bonds, e.g., carboxylic acids. The blue shift observed for bands attributed to the C(O)OH group bending vibrations is widely considered to be a result of the stiffening of the structure due to the presence of hydrogen bonds [69]. Some spectral features are exclusive characteristics of dimers. For example, the frequency corresponding to the C-O stretching and BO 2 in-plane bending (No. 6, Table 1) at 956 cm −1 was found only for the dimer, which is in agreement with the position of this band observed in the in experimental spectrum, i.e., 976 cm −1 . Bands originating from B-OH stretching vibrations are less affected by the hydrogen bond formations; the maximum of the most characteristic and intense signal shifts from 1460 cm −1 for 1 to 1473 cm −1 for its dimer. The considerably less intense signal corresponding to the C-O bond stretching shifts from 957 cm −1 for 1 to 1038 cm −1 for the dimeric structure. This effect is similar to that described for carboxylic acids and results from a more rigid structure of the dimer due to the hydrogen bond formation [69].   [21], and hydrogen-bonded interactions with the solvent for 8 [22]. It is also in accordance 208 with the shift between calculated bands for 1 and its corresponding dimer described 209 above. In the case of the studied formyl derivatives (9 and 10), the B-OH group vibration 210 frequencies are affected by the formation of an intramolecular hydrogen bond that en-211 gages the B-OH group (Figure 4). Consequently, only one band corresponding to this 212 group is observed in DFT calculated spectra at 1464 cm −1 for both compounds (Table 2). 213 The calculated lengths of the intramolecular hydrogen bonds shown in Figure 4  Two characteristic signals, corresponding to B-OH stretching vibrations in the remaining investigated benzoxaboroles, were identified on the basis of DFT calculations and are found in the spectral ranges of 1460-1468 cm −1 and 1402-1420 cm −1 . The corresponding bands in the experimental IR spectra of the investigated compounds are shifted and occur in the ranges of 1413-1446 cm −1 and 1332-1377 cm −1 . One of the possible reasons for this shift is the formation of hydrogen bonds in the investigated solid phase samples of benzoxaboroles. Structures of some of the studied benzoxaboroles in the solid state were previously investigated by a single crystal X-Ray technique, showing dimeric structures for AN2690 [25], 1 [38], 2 [23], 3 [70], 9, [21], hydrogen-bonded infinite chains for 7 [22] and 6 [21], and hydrogen-bonded interactions with the solvent for 8 [22]. It is also in accordance with the shift between calculated bands for 1 and its corresponding dimer described above. In the case of the studied formyl derivatives (9 and 10), the B-OH group vibration frequencies are affected by the formation of an intramolecular hydrogen bond that engages the B-OH group (Figure 4). Consequently, only one band corresponding to this group is observed in DFT calculated spectra at 1464 cm −1 for both compounds ( Table 2). The calculated lengths of the intramolecular hydrogen bonds shown in Figure 4 are: 1.869 Å for 9 and 1.870 Å for 10. They both equal 1.853 Å when applying the dispersion-corrected B3LYP (empirical dispersion = D3BJ) [71].    Table S1, Supplementary Materials). 223 When comparing the structures optimized in the gas phase and those take 224 crystal structure (Table S1, Table S1). It is worth noting that among th 237 interacting ligands presented in Table 3 (Table S1), no clear 253 sion can be drawn, since in about 50% cases structures of type "1" or "2" intera 254 strongly with the enzyme. It is worth noting, however, that the majority of the mos 255 acting structures (almost 70%, Table 3) contain rings at the same side of the ligan 256

Interactions of Benzoxaboroles' AMP Spiroesters with C. albicans LeuRS
There are several issues to consider when analyzing the results of docking studies. First of all, a large variation in the results of the evaluated inhibition constants is generally observed. The values determined for all the 66 various ligands studied for the investigated benzoxaboroles (1-15) range from 0.28 nM for 3-R-1-AMP to 396.82 nM for 14-S-2-AMP (see Table S1, Supplementary Materials).
When comparing the structures optimized in the gas phase and those taken from crystal structure (Table S1, structures taken from crystal are denoted with "c"), there are no significant differences in the determined inhibition constants in the case of benzoxaboroles 7-9. However, compound 6 exhibits significant differences in the gas-phase optimized structure and that taken from the crystal., The reason for this could be the differences in B-OH group arrangement in benzoxaborole 6, which is anti in the optimized structure ( Figure 3a) and syn in the crystal phase [21]. In the cases of structures 7-9, the conformation is identical in both the solid and gas phases, which results in similar inhibition constants determined for both input structures [21,22].
One of the crucial issues to consider is the influence of optical isomerism on interactions with the enzyme. The differences in the determined inhibition constants for optical isomers are significant in most of the studied cases. For example, it is 4.39 nM for 2-R-1-AMP and 53.52 nM for 2-S-1-AMP. At the same time, enantiomer R displays generally lower inhibition constant values (see Table S1). It is worth noting that among the most-interacting ligands presented in Table 3, only one structure is enantiomer S (ligand 2-S-2-AMP), whereas all the remaining structures are enantiomers R. These results may denote some chirality of the enzyme's active site [72]. In most cases, both enantiomers in a pair form different numbers of hydrogen bonds; however, in about one-third of them they are exactly the same. Formation of hydrogen bonds between the enzyme and the docked-in ligands is therefore not highly stereospecific. Nevertheless, enantiomers R generally tend to form more hydrogen bonds with the binding site of the enzyme (it is 44% and 24% for R and S, respectively). It was, however, impossible to identify any specific interactions that clearly differentiate the enantiomers' binding.
The other factor influencing the inhibition constant is the arrangement of the benzoxaborole and AMP parts in the structure of the resulting spiroester ligand. Two structural patterns, differing in the reciprocal setting of the benzoxaborole and adenosine rings, can be distinguished. Those are denoted with "1" (both rings at the same side of the spironoronate) and "2" (rings at the opposite side of the spiroboronate). In all but two cases of pairs of type "1" and "2" spiroboronates, the differences in the determined values of inhibition constant are significant. When analyzing all the results (Table S1), no clear conclusion can be drawn, since in about 50% cases structures of type "1" or "2" interact more strongly with the enzyme. It is worth noting, however, that the majority of the most-interacting structures (almost 70%, Table 3) contain rings at the same side of the ligand (type "1"), whereas in the experimental AN2690-AMP complex they are at the opposite sides of the ligand (type "2").
The formation of hydrogen bonds between the ligand and enzyme strengthens their mutual interactions, resulting in lower inhibition constants. While for both the parent drug (AN2690) as well as for 1 the number of hydrogen bonds is five, for the studied 3-amino-benzoxaboroles (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15) it varies from one (nine ligands, 13% of all structures) to five (five ligands, 7% of all structures). Most of the docked-in ligands form, however, three hydrogen bonds (24 ligands, 36% of all structures) and some of them form two hydrogen bonds (16 ligands, 24% of all structures). It is worth noticing that, similar to the case of the AN2690-AMP ligand, the vast majority of hydrogen bonds with the enzyme are formed by the AMP part of the studied ligands and not by the benzoxaborole fragment.
The results of docking studies show that the introduction of a spatially structured amine substituent considerably influences the interactions between the enzyme and the tested substances. While the unsubstituted benzoxaborole (1), also known as AN2679, reveals a binding mode analogous to that of AN2690 (Figure 5a,b), most of the ligands studied adopt different positions within the enzyme's binding pocket. Only several studied ligands (namely: 1-4, 10 and 12) adopt positions analogous to that known for AN2690-AMP (Table S1). However, all but one of them are amongst the most-interacting ligands given in Table 3. It does not apply exclusively to 12-R-1-AMP, which represents another mode of binding with an inhibition constant of 2.76 (Table 3). It is worth noting that some of the studied derivatives display the in silico determined inhibition constants close to that of the parent drug (Tavaborole, AN2690), suggesting their high antifungal potential. However, several structural issues of the studied 3-aminobenzoxaboroles, such as optical isomerism, various possible arrangements of both components of the spiroester ligands, and a variety of adopted binding modes with the enzyme, may lower the real content of the possibly active structure. This may result in moderate in vitro activity of the studied 3-amino-benzoxaboroles, especially when applied as a racemic mixture. This is in line with results of anti C. albicans activity of 2-15 published recently, showing a considerably lower activity of 3-amino-benzoxaboroles in comparison with AN2690 [21].  (Table S1). However, all but one of them are amongst the most-interacting ligands 274 given in Table 3. It does not apply exclusively to 12-R-1-AMP, which represents another 275 mode of binding with an inhibition constant of 2.76 (Table 3). It is worth noting that some

Conclusions
Despite considerable differences in the structure of investigated benzoxaboroles, a well-isolated band corresponding to the B-OH group stretching vibrations is present in experimental IR spectra of all the studies compounds in the range of 1446-1414 cm −1 and can be considered as characteristic for benzoxaboroles. Several structural issues of the studied possible inhibitors of C. albicans LeuRS influence their interaction with the enzyme. The majority of the most-interacting structures are R enantiomers, forming spiroesters with both rings at the same side of the molecule. Some of these spiroesters tend to adopt positions analogous to those of Tavaborole (AN2690), resulting in low inhibition constants and tight interactions with the enzyme. It is worth noting, however, that the majority of the most-interacting ligands adopt a conformation different than that of the parent drug determined in structural studies. Therefore, the strength of their interactions with the enzyme may not be relevant for the function of those compounds as antifungal agents.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/sym13101845/s1. Results of the docking studies for 66 various spiroester structures (Table S1). Optimized structures of the investigated compounds 2-16 ( Figures S1-S29), coordinates of the input and optimal structures of 2-16 (Tables S2-S30). The best structures obtained in the docking studies to Candida albicans leucyl t-RNA synthetase ( Figures S30-S96). Coordinates of the best structures obtained in the docking studies to Candida albicans leucyl t-RNA synthetase (Tables S31-S97

Data Availability Statement:
The data presented in this study are available in this paper as well as in Supplementary Material.