N-Pyrazinoyl Substituted Amino Acids as Potential Antimycobacterial Agents-The Synthesis and Biological Evaluation of Enantiomers.

Tuberculosis is an infectious disease caused by Mycobacterium tuberculosis (Mtb), each year causing millions of deaths. In this article, we present the synthesis and biological evaluations of new potential antimycobacterial compounds containing a fragment of the first-line antitubercular drug pyrazinamide (PZA), coupled with methyl or ethyl esters of selected amino acids. The antimicrobial activity was evaluated on a variety of (myco)bacterial strains, including Mtb H37Ra, M. smegmatis, M. aurum, Staphylococcus aureus, Pseudomonas aeruginosa, and fungal strains, including Candida albicans and Aspergillus flavus. Emphasis was placed on the comparison of enantiomer activities. None of the synthesized compounds showed any significant activity against fungal strains, and their antibacterial activities were also low, the best minimum inhibitory concentration (MIC) value was 31.25 µM. However, several compounds presented high activity against Mtb. Overall, higher activity was seen in derivatives containing l-amino acids. Similarly, the activity seems tied to the more lipophilic compounds. The most active derivative contained phenylglycine moiety (PC-d/l-Pgl-Me, MIC < 1.95 µg/mL). All active compounds possessed low cytotoxicity and good selectivity towards Mtb. To the best of our knowledge, this is the first study comparing the activities of the d- and l-amino acid derivatives of pyrazinamide as potential antimycobacterial compounds.


Introduction
Tuberculosis (TB) is a highly contagious infection caused by Mycobacterium tuberculosis (Mtb) and was responsible for over 1.5 million deaths worldwide in 2018, which ranks it in the top 10 causes of deaths and the leading cause of dying due to a single infectious agent according to the World Health Organization (WHO) [1]. Most often, TB affects the lungs of the infected individuals, although the cases of so called extra-pulmonary TB are well documented as well. Despite great advances and many clinically successful drugs, a desired cure still eludes the reaches of modern medicine [1].
The treatment of TB, as well as other bacterial infections nowadays, is hindered by growing drug resistance (DR). For TB, the WHO differentiate multidrug-resistant tuberculosis (MDR-TB) and extensively drug-resistant tuberculosis (XDR-TB). Infections with MDR or XDR strains constitute an As a potential drug, PZA core substituted with amino acid (aa) may present numerous advantages in comparison with simple aliphatic or aromatic amides and esters. Compared to amides, small PZA-derived peptides could be more easily handled by Mtb peptidases due to their natural similarity to the peptides (compared to amides) while being stable enough in plasma medium to reach the site of infection (compared to esters). As for any other amidic-type drugs, the hydrolysis by host peptidases (amidases) diminishing the activity can occur. However, at this stage of the drug design, this issue, although important, is very troublesome to address without relevant experimental data concerning the exact MoA and the plasma stability of the proposed compounds. In aa prodrugs of antimicrobial drug dapson [29], the half-life in human plasma was shown to be in order of tens to hundreds of minutes, based on the used aa. If such stability was also observed for our title compounds, it would rank them-at least from the stability perspective-among aliphatic long carbon chain esters of POA (half-life in order of tens of minutes [24]) which are stable and active in macrophage model of Mtb infection. Similarly, they would be hydrolysed considerably quicker than aliphatic amides (half-life in order of thousands of minutes [24]). Therefore, we believe that the proposed derivatives could act as both efficient and sufficiently stable prodrugs of POA but at the same time we do not rule out the possibility that the presented derivatives can be active in their amidic, non-hydrolysed form.
To test these hypotheses, we have decided to synthesize a series of N-pyrazinoyl substituted amino acid derivatives and test their activity against a range of mycobacterial, bacterial, and fungal strains available to our research group. Furthermore, to our knowledge, we are the first to investigate the differences in the biological activities of compounds derived from l-and d-aa in both mildly acidic (pH 6) and commonly used pH 6.6.

Chemistry
Thirty-nine derivatives of PZA coupled with pure l-or d-aa or with aa enantiomeric mixtures were synthesized (for structure, see Table 1, for SMILES, calculated molecular descriptors, and selected biological data, see the Supplementary Material). In compound codes, dl signifies racemic mixtures and d/l denotes the mixture of enantiomers. Several coupling strategies were evaluated concurrently. Herein, the unsuccessful strategies (Procedures 1-4) are describe only in brief outlines.
Procedure 1: Reacting pyrazinoyl chloride, (prepared as previously described in Zitko et al. [26]), with free aa and their further esterification was found unsuitable. Multiple products (not identified) were observed, as indicated by thin-layer chromatography (TLC), and the desired product was very difficult to separate. Furthermore, due to the very low solubility of aa in organic solvents, water solutions had to be employed, resulting in the undesirable hydrolysis of the pyrazinoyl chloride, further diminishing the yield.
Procedure 2: Reacting pyrazinoyl chloride with aa esters (or their hydrochlorides) partially solved the problems with solubility in organic solvents, although, side-products remained a problem.
Procedure 3: The aminolysis of methyl pyrazinoate with aa ester (hydrochlorides) in ethanol did not work, even when heated with addition of catalytic amount of 4-(dimethylamino)pyridine (DMAP).

PC-d-Ala
compounds, which proved to be a correct assumption based on the activities discussed further below. We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].  Based on the extensive knowledge on mycobacterial esterases [37], the use of esters over free acids was also favoured to potentially increase the bioavailability of our mostly polar title compounds, which proved to be a correct assumption based on the activities discussed further below.
We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].

Code
Structure Based on the extensive knowledge on mycobacterial esterases [37], the use of esters over free acids was also favoured to potentially increase the bioavailability of our mostly polar title compounds, which proved to be a correct assumption based on the activities discussed further below.
We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].

Code
Structure We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].  Based on the extensive knowledge on mycobacterial esterases [37], the use of esters over free acids was also favoured to potentially increase the bioavailability of our mostly polar title compounds, which proved to be a correct assumption based on the activities discussed further below.
We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].

PC-d-OtBu-Thr-Me
All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.

PC-l-Tyr-Et
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.

PC-dl-Val-Et
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.

PC-d-Val-Me
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H

PC-l-Val-Me
All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d6), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-L-Ala), ester derivative (PC-L-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-D/L-Pgl-Me, PC-DL-Phe-Et) are enclosed in the Supplementary Material.
In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H Procedure 4: The use of standard coupling reagents as 1,1 -carbonyldiimidazole (CDI) in anhydrous tetrahydrofurane (THF) with aa showed good results, although in some cases starting pyrazinoic acid was still detected by TLC as a small impurity difficult to separate.
Procedure 5: At last, general procedure, as described in Section 3.1.1. (and shown in Figure 2), was achieved by adjusting Procedure 4. The method employs esters of aa instead of aa with the free carboxylic group. The increased solubility of the reactants and products in used organic solvents (THF) greatly simplified the syntheses and subsequent purifications. The products were created in good yields (positively related to their lipophilicity) and without epimerization. pyrazinoic acid was still detected by TLC as a small impurity difficult to separate. Procedure 5: At last, general procedure, as described in Section 3.1.1. (and shown in Figure 2), was achieved by adjusting Procedure 4. The method employs esters of aa instead of aa with the free carboxylic group. The increased solubility of the reactants and products in used organic solvents (THF) greatly simplified the syntheses and subsequent purifications. The products were created in good yields (positively related to their lipophilicity) and without epimerization. Based on the extensive knowledge on mycobacterial esterases [37], the use of esters over free acids was also favoured to potentially increase the bioavailability of our mostly polar title compounds, which proved to be a correct assumption based on the activities discussed further below.
We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].  Based on the extensive knowledge on mycobacterial esterases [37], the use of esters over free acids was also favoured to potentially increase the bioavailability of our mostly polar title compounds, which proved to be a correct assumption based on the activities discussed further below.
We tried to synthesize Pro and His ester derivatives. However, these proved to be readily soluble in water even as esters. Therefore, we were unable to sufficiently purify them for the biological screening. PC-MeAcr derivative was obtained as a by-product by the elimination of the hydroxyl of serine methyl ester, as previously described by Pinheiro et al. [32].
All compounds were characterized by 1 H-and 13 C-NMR spectroscopy in hexadeuterodimethyl sulfoxide (DMSO-d 6 ), IR spectroscopy using the attenuated total reflectance (ATR) method, and mass spectrometry. All analytical results were in agreements with the presumed structures.
In 1 H-NMR, the amidic hydrogen was always observed as a clearly distinguishable doublet, usually around 8.0-9.5 ppm, and the usual J coupling constant ranged between 8.4-8.5 Hz. Pyrazine hydrogens were observed as two doublets and one multiplet at 9. 1-9.4, 8.8-9.1, and 8.6-8.9 ppm, respectively. C-α hydrogen was observed as a distinguishable multiplet at 4.4-4.6 ppm in all chiral compounds. The acidic hydrogen in free carboxylic moiety was sometimes not observed due to rapid exchange with solvent, which is common in DMSO-d 6 1 H-NMR. Alcoholic OH hydrogen in Ser ester derivatives was observed as a clear doublet of doublets or a triplet at 5.3 ppm. In the free acid Ser derivative, as well as in Thr esters, the OH hydrogen was sometimes indistinguishable. The representative spectra of free aliphatic aa derivative (PC-l-Ala), ester derivative (PC-l-Val-Me), and aromatic and phenylglycine (Pgl) aa ester derivatives (PC-d/l-Pgl-Me, PC-dl-Phe-Et) are enclosed in the Supplementary Material. In the IR spectra of all the title compounds, the amidic N-H stretching was observed as a distinguishable absorption band at 3200-3300 cm −1 , amidic C=O stretching was observed at 1640-1670 cm −1 , and ester C=O stretching at 1730-1750 cm −1 . Free acids presented characteristically broad O-H stretches around 3000 cm −1 . In Ser and Thr derivatives, an aliphatic O-H stretch was visible above 3300 cm −1 .
Amino acid esterification [38] and other common peptide synthesis procedures (e.g., tert-butyl group cleavage) [39,40] used to obtain the title compounds are considered enantioselective. Acidic hydrolysis was previously described to potentially cause epimerization or even racemization of aa derivatives [41]. CDI-mediated coupling in THF had been shown to cause a low degree of epimerization (5%) on one model peptidic compound [42]. Generally, no epimerization occurred in our conditions. Epimerization was seen only in the case of Pgl derivative (PC-d/l-Pgl-Me) synthesized using the d-Pgl methyl ester hydrochloride. The composition of the mixture of enantiomers was further investigated by optical rotation measurements and chiral HPLC. Comparing the literature value of the specific rotation of l-enantiomer ([α] D = +88.8 • [43]) with the measured value of our sample [α] D = −47.3 • , it is clear that d-enantiomer dominates the mixture of enantiomers as judged from the sense of rotation. According to the chiral HPLC, the sample consists of two enantiomers represented by peaks in a ratio of approx. 2.1:1, based on the area under the curve (AUC). We thus conclude that the compound PC-d/l-Pgl-Me is a mixture of d-and l-enantiomer in ratio of 2.1:1 in favour for d-enantiomer. The epimerization of Pgl is known to occur very easily using standard coupling conditions and thus stereoselective synthesis would require specific reagents not employed in this work [44]. Chiral HPLC analysis was performed for enantiomeric mixture PC-d/l-Pgl-Me, racemic mixtures (PC-dl-Ala-Me, PC-dl-Leu-Et, PC-dl-Phe-Et, PC-dl-Val-Et) and for the samples of which we had the complete couple of antipodes. The chromatograms are presented in the Supplementary Material.

Antimycobacterial Activity Screening
Mycobacterial screening of the synthesized compounds was done using microdilution assay (Microplate Alamar Blue Assay, MABA [45]) on M. smegmatis DSM 43465 (ATCC 607), Mycobacterium aurum DSM 43999 (ATCC 23366), and the avirulent strain of Mtb H37Ra ITM-M006710 (ATCC 9431)-for results, see Tables 2 and 3. The minimum inhibitory concentration (MIC) values were measure in µg/mL. To account for molecular weight, the inhibitory activity (expressed as MIC) of the discussed compounds were recalculated to µM concentration. Fast-growing M. smegmatis and M. aurum are used as surrogate model organisms in antimycobacterial research [46,47]. Mtb H37Ra is commonly used as an accessible valid alternative for the virulent strain [48]. The selected title compounds were also tested in pre-screening against a virulent strain of Mtb H37Rv CNCTC My 331/88 (ATCC 27294), M. kansasii CNCTC My 235/80 (ATCC 12478), and M. avium ssp. avium CNCTC My 80/72 (ATCC 15769). The activity of PC-MeAcr was in accordance with previously published results (MIC on Mtb H37Rv was 50 µg/mL in our study as well as in the study by Pinheiro et al. [32]). Although PC-MeAcr might have been considered cytotoxic due to the acrylate group, no significant cytotoxicity was noted on the human HepG2 cell line, as presented in Table 4. For the full results of this pre-screening, see Table S2 in the Supplementary Material.
For the purposes of this article, only compounds with MIC values lower than 31.25 µg/mL (roughly corresponding to 100 µM based on molecular weight) were considered as active.
Notable activity was detected only against Mtb H37Ra at pH 6. No significant activity was detected against M. smegmatis at neither pH 6 nor pH 6.6, low activity was noted only for PC-MeAcr (MIC = 31.25 µg/mL, pH 6.6). Only one compound (PC-d/l-Pgl-Me) showed notable activity against M. aurum (MIC = 31.25 µg/mL, pH 6). If not stated otherwise, the following discussion concerns activity against Mtb H37Ra at pH 6.
Limited data on similar aa derivatives of PZA are available in the literature. In the work of Panda et al. [33], the l-derivatives of non-esterified aa (Leu, Val, allo-Ile, Met, Phe, Trp) were tested. These compounds were more or similarly active at pH 7 compared to isoniazid (INH) (> 70% growth inhibition at 30 µg/mL). However, as a different Mtb strain and methodology was used in the study of Panda et al. [33], direct comparison to our study is not applicable. PC-Asp-diEt and some atypical aa derivatives of the structure presented in Figure 1 were tested previously by Kushner et al. [30] with no activity, which is in accordance with our results presented in Tables 2 and 3. pH Dependence Based on one of the proposed MoA of our compounds, that is POA-releasing prodrugs, all derivatives were tested at mildly acidic pH 6 as required for the in vitro activity of PZA/POA, and the results were compared to the activities at the commonly used pH 6.6 [22]. The specific pH value for the acidic measurement was selected to balance the activity of tested compounds with the growth rate of cultivated mycobacteria [49]; furthermore, a similar pH is also found in macrophages, where the intracellular pathogen of Mtb resides [50], making these conditions closer to reality. The necessity of acidic conditions for the in vitro activity of PZA and POA is a well-known paradigm. A similar pH-activity dependence is known also for POA esters [22]. However, Pinheiro et al. [32] and den Hertog et al. [51] have shown that the activity of PZA and POA may also be observed under non-acidic conditions. According to den Hertog et al. [51], the in vitro activity of PZA is probably tied to specific metabolic and energetic effects occurring in acidic medium, although they concluded that the same effects could also be induced by other stressors, e.g., low temperature or starvation. In general, these implications do not affect the correctness of the in vitro measurements presented in this manuscript or elsewhere but rather propose new ways to investigate the MoA of PZA.
Based on the results presented in Tables 2 and 3, the impact of the pH value on the activity of our compounds against Mtb was substantial. At least an 8-fold decrease in MIC was observed for all active compounds, corresponding to the change from hundreds of µg/mL to tens of µg/mL (or lower). The effect of pH on the activity against other mycobacterial strains was less profound, similar to PZA itself, which could be suggesting PZA-linked MoA. Table 2. Antimycobacterial activities of the tested compounds at pH 6.6 presented as the minimum inhibitory concentration (MIC) (µg/mL), the most active compound is in bold. The chemical structures of the derivatives are presented in Table 1.

Lipophilicity and Sidechain Effects
Mycobacterial membrane is very lipophilic. Compared to other bacteria, it contains unusually high concentrations of specific lipidic molecules, e.g., mycolic acids [52]. Consequently, highly lipophilic low molecular weight compounds are expected to penetrate the membrane through a simple passive diffusion more easily. Nevertheless, the exact physico-chemical requirements relevant to the Mtb membrane are still unknown [53]. To assess the impact of the lipophilicity of the synthesized compounds on their activities, we investigated the relationship between calculated logP and respective MIC values at pH 6 (as no significant activity was recorded pH 6.6). All compounds were categorized considering their chemical nature (free carboxylic acid/ester) and enantiomer aspects (as discussed later). The logP (o/w) chemical descriptor, hereafter referred to as logP, was calculated by Molecular Operating Environment (MOE) 2019.0101 (Chemical Computing Group (CCG), Quebec, Canada) [54]. The resulting scatter plot of logP and antimycobacterial activity presented as log(1/MIC [M]) (a higher value means better activity) is shown in Figure 3. The calculated logP and MIC (µM) values at pH 6 are contained in Table 3.
Molecules 2020, 25, x 12 of 30 Figure 3. Scatter plot representing antimycobacterial activity and calculated lipophilicity of the compounds. The colours denote stereochemistry; the shape of the symbols distinguishes between esters and free acids. Gly and MeAcr derivatives are presented as green triangles. The MIC values of the majority of the derivatives in the non-active region were not determined precisely (see Table 3).
When comparing the activities of our title compounds, higher activity is seen mostly in derivatives of aa with hydrophobic aliphatic or aromatic side chains.
In a homologous series of Gly, Ala, Val, Leu, and Ile derivatives, high activity was noted only for derivatives of Ala and Leu. Interestingly, no activity was seen in the similar lipophilic ester of Val or in the derivatives of mutual chain isomers Leu and Ile. The highest activity in aliphatic aa derivatives was seen in Met derivative PC-L-Met-Me, having comparable activity with PC-L-Ala-Et (MIC = 3.91 µg/mL). The activity of tested polar aliphatic aa derivatives (Cys, Ser, Thr) seemed tied mainly to the chemically protected form (active PC-L-OBn-Ser-Me vs. inactive PC-L-Ser-Me with unsubstituted hydroxyl), supposedly due to a higher lipophilicity of the protected alternative.
High activities were noted for the majority of aromatic aa derivatives. The highest activity of all tested compounds was seen in PC-D/L-Pgl-Me (MIC < 1.95 µg/mL), and derivatives of Tyr were active as well. All observed results may be tied to the generally higher lipophilicity of aliphatic and aromatic compounds; however, enzymatic preferences towards the hydrolysis of specific amidic bonds may also play a significant role, as discussed later.

Stereochemical Aspects of Activity
Enzymatic substrate specificity is the cornerstone of catalysis, and enzymes with one-enantiomer specificity are known in microorganisms [56,57]. We were therefore curious to investigate the differences in activities of the L-and D-enantiomers of the title compounds, as they may suggest either a specific MoA of the unhydrolyzed amide, or a more facile hydrolysis of one enantiomer.
As may be seen from Figure 3, D-enantiomers (shown in red) were in general inactive. However, as discussed above, caution is needed when comparing enantiomers, as other factors, such as the overall lipophilicity, may also influence the activity. Therefore, it would be safe to compare the activities of the same esters, while comparing activity of different esters of the same aa derivative should be judicious.
Most illustrative examples following the above-mentioned rule (D-isomers are inactive) are once again represented by derivatives of Ala. A comparison of PC-L-Ala-Et with PC-D-Ala-Et (line 3) clearly shows the preference of the L-enantiomer (MIC = 3.91 µg/mL vs. ≥ 125 µg/mL, respectively). . Scatter plot representing antimycobacterial activity and calculated lipophilicity of the compounds. The colours denote stereochemistry; the shape of the symbols distinguishes between esters and free acids. Gly and MeAcr derivatives are presented as green triangles. The MIC values of the majority of the derivatives in the non-active region were not determined precisely (see Table 3).
Clearly, for the same aa derivatives, the least active compounds in categories free acid-methyl ester-ethyl ester are embodied by derivatives with free carboxylic moiety (represented by gems in Figure 3). Furthermore, ethyl esters (represented by squares) were generally more active than methyl esters in our study. This general observation can be attributed to their higher lipophilicity (as indicated by increased logP in Table 3) compared to the free acids. As an illustrative example, in l-Ala derivatives, the MIC value of the ethyl ester is at least 64-times lower compared to that of the free acid (PC-l-Ala-Et MIC = 3.91 µg/mL vs. PC-l-Ala MIC ≥ 250 µg/mL), with the logP difference of approximately 0.5 (−0.53 for ethyl ester vs. −1.02 for free acid)-line 1 in Figure 3. In Leu and Glu esters, high activity was noted only for more lipophilic ethyl esters, while the less lipophilic methyl esters were inactive (PC-dl-Leu-Et vs. PC-d-Leu-Me-line 2 and PC-l-Glu-diEt). In fact, the majority of methyl esters (represented by circles) are located below the activity cut-off in Figure 3, with the exception of highly lipophilic compounds, e.g., PC-l-OBn-Ser-Me, PC-d/l-Pgl-Me. The effect of growing activity with the lengthening of the ester carbon chain had previously also been observed in derivatives of POA [21,23] or other antimycobacterial compounds [55] and it has been attributed to the above-mentioned better penetration inside the mycobacteria. However, as discussed later, other effects-as stereochemistry-seem to play an important role in structure-activity relationships in our compounds. Therefore, the effect of ester moiety on activity stays inconclusive.
When comparing the activities of our title compounds, higher activity is seen mostly in derivatives of aa with hydrophobic aliphatic or aromatic side chains.
In a homologous series of Gly, Ala, Val, Leu, and Ile derivatives, high activity was noted only for derivatives of Ala and Leu. Interestingly, no activity was seen in the similar lipophilic ester of Val or in the derivatives of mutual chain isomers Leu and Ile. The highest activity in aliphatic aa derivatives was seen in Met derivative PC-l-Met-Me, having comparable activity with PC-l-Ala-Et (MIC = 3.91 µg/mL). The activity of tested polar aliphatic aa derivatives (Cys, Ser, Thr) seemed tied mainly to the chemically protected form (active PC-l-OBn-Ser-Me vs. inactive PC-l-Ser-Me with unsubstituted hydroxyl), supposedly due to a higher lipophilicity of the protected alternative.
High activities were noted for the majority of aromatic aa derivatives. The highest activity of all tested compounds was seen in PC-d/l-Pgl-Me (MIC < 1.95 µg/mL), and derivatives of Tyr were active as well. All observed results may be tied to the generally higher lipophilicity of aliphatic and aromatic compounds; however, enzymatic preferences towards the hydrolysis of specific amidic bonds may also play a significant role, as discussed later.

Stereochemical Aspects of Activity
Enzymatic substrate specificity is the cornerstone of catalysis, and enzymes with one-enantiomer specificity are known in microorganisms [56,57]. We were therefore curious to investigate the differences in activities of the l-and d-enantiomers of the title compounds, as they may suggest either a specific MoA of the unhydrolyzed amide, or a more facile hydrolysis of one enantiomer.
As may be seen from Figure 3, d-enantiomers (shown in red) were in general inactive. However, as discussed above, caution is needed when comparing enantiomers, as other factors, such as the overall lipophilicity, may also influence the activity. Therefore, it would be safe to compare the activities of the same esters, while comparing activity of different esters of the same aa derivative should be judicious.
Most illustrative examples following the above-mentioned rule (d-isomers are inactive) are once again represented by derivatives of Ala. A comparison of PC-l-Ala-Et with PC-d-Ala-Et (line 3) clearly shows the preference of the l-enantiomer (MIC = 3.91 µg/mL vs. ≥ 125 µg/mL, respectively). Other l-derivatives being more active were noted as well, although the activities might be biased by the ester differences, as in the case of PC-l-Glu-diEt vs. PC-d-Glu-diMe (MIC = 31.25 µg/mL vs. MIC = ≥ 500 µg/mL, line 4) or PC-l-Tyr-Et vs. PC-d-Tyr-Me (MIC = 7.81 µg/mL vs. MIC = 125 µg/mL). The activity of racemic or enantiomeric mixtures (painted in grey), as for PC-d/l-Pgl-Me and PC-dl-Leu-Et, is not possible to attribute to the exact enantiomer; although, based on the overall trends, we propose l-Leu derivative to be more active. As Pgl is not a natural amino acid, the activity could by tied to both enantiomers. We were unable to test this hypothesis as the pure l-or d-enantiomers of Pgl derivative were not synthesized. Finally, both l-and d-Trp derivatives showed similar, although very low activity against Mtb H37Ra (one or two dilution steps below the cut-off value). However, based on the observed cytotoxicity of PC-l-Trp-Et (l-enantiomer of Trp) on HepG2 cell lines (IC 50 = 146.6 µM, Table 4), and the unchanged MIC value depending on the pH, we attributed the antimycobacterial activity of both Trp derivatives to their non-specific toxicity of unknown nature.
The only case of d-enantiomer being more active than l-enantiomer (in the case of the Pgl derivative, we are unable to distinguish the active and inactive enantiomer) was in O-tert-butyl-protected derivatives of Thr (PC-d-OtBu-Thr-Me, MIC = 62.5 µg/mL vs. PC-l-OtBu-Thr-Me, MIC ≥ 500). The activity was low, however. The reason for this preference is unknown. Nevertheless, ether-protected derivatives are in fact non-natural aa. Observed activity is still very interesting, as we noted that in bacteria the same d-enantiomer showed no activity while the l-enantiomer was active against Staphylococcus aureus strains (discussed further in Section 2.2.2.).

Antibacterial Activity Screening
All compounds were tested by the microdilution broth method [58] against Staphylococcus aureus subsp. aureus CCM 4223 (ATCC 29213), Staphylococcus aureus subsp. In previous studies, the activities of the methyl esters of PC-Gly, PC-l-Ala, PC-l-Ser, and PC-l-Tyr against some bacterial strains were evaluated, yet no activity was noted [31], in agreement with the observations in this article. The activities of some free acid compounds reported in the study of Panda et al. [33] could be tied to the methodology differences and strain susceptibility, as no activity was seen when corresponding esters were used in our studies. For the full results (Table S4) and methodology, see the Supplementary Material.

Antifungal Activity Screening
All compounds were tested by the microdilution broth method [59,60] against Candida albicans CCM 8320 (ATCC 24433), Candida krusei CCM 8271 (ATCC 6258), Candida parapsilosis CCM 8260 (ATCC 22019), Candida tropicalis CCM 8264 (ATCC 750), Aspergillus fumigatus ATCC 204305, Aspergillus flavus CCM 8363, Lichtheimia corymbifera CCM 8077, and Trichophyton interdigitale CCM 8377 (ATCC 9533). No compound showed any significant activity against tested fungal strains up to the highest tested concentration, which was 500 µM for the majority of compounds, 250 µM for PC-l-SBn-Cys, and 125 µM for PC-l-OBn-Ser-Me. Three compounds (PC-d-Asp-diMe, PC-d-Ile-Me and PC-l-Ile-Me; the same as for antibacterial testing) were insoluble in the medium and therefore were not tested. No differences in MIC values read after 24 and 48 h (72 h) of incubation were noted. For the full results (Table S6) and methodology, see the Supplementary Material.

Cytotoxicity Screening Results
The cytotoxicity of selected active compounds was measured using the standard human hepatocellular carcinoma cell line HepG2. The used CellTiter 96 assay is based on the reduction of tetrazolium dye MTS in living cells to formazan, which is then determined colorimetrically. The reduction of the reagent is related to availability of NADH and NADPH. The decline in levels of these metabolically important compounds in the cell leads to the decreased production of formazan.
The parameter IC 50 was used as a measure of cytotoxicity, which allows for the quantitative comparison of the toxicity among tested compounds (Table 4). Overall, the toxicity of the compounds was very low, some compounds had the IC 50 values above the tested range (PC-l-Ala-Et). The highest cytotoxicity illustrated by its lowest IC 50 value was noted for the aromatic compound PC-l-Trp-Et (IC 50 = 146.6 µM).
The selectivity index (SI) for Mtb was calculated for the most active compounds according to the equation SI = IC 50 (µM)/MIC(µM), as used by Bagla et al. [61], and the results are presented in Table 4. The tested compounds were more active against Mtb (SI > 1) than their cytotoxic effect against human cells, represented by the HepG2 cell line cytotoxicity experiment. Two of the most active compounds (PC-l-Ala-Et and PC-d/l-Pgl-Me) had SI > 40, which suggests a good cytotoxicity profile for potential further development. Table 4. Calculated selectivity index (SI) of the most active compounds against Mtb at pH 6 and pH 6.6. The compounds with a good cytotoxicity profile (SI > 10) are in bold.

General Procedure of Coupling Reaction
To a round bottom flask, 248 mg (2 mmol) of pyrazinoic acid (in the codes of final compounds denoted as PC) was suspended in anhydrous THF (3 mL), 324 mg (2 mmol) of CDI was added and the mixture was heated to 60 • C and stirred for 1 h. Afterwards, the flask was cooled down to laboratory temperature and 2 mmol of amino acid or amino acid ester (mostly in hydrochloride form) was added to the mixture with 300 µL (2.14 mmol) of triethylamine. After the addition, the mixture reacted for 2 h at laboratory temperature and was evaporated on a rotary evaporator. The solid residue was suspended in water (10 mL) and extracted to ethyl acetate (EtOAc) using repetitive extractions (typically twice with 15 mL) in separatory funnel. After each separation, the presence of the product in the phases was monitored on a TLC (silica, methanol-EtOAc 1:1). Finally, the organic layers were combined and washed with 10 mL of 0.01 M HCl and 10 mL of saturated NaCl solution. At the end, the organic phases were mixed with approximately 50 mg of activated carbon, stirred for 10 min, filtered and evaporated to dryness. The final product was left in the desiccator overnight and purified further using flash chromatography if necessary.

General Procedure for Esterification of Amino Acids
Methyl esters of d-Ile and O-benzyl-Ser derivatives were prepared as described in Li and Sha [38] using 2 mmol of amino acid, 0.5 mL of trimethylsilyl chloride (TMSCl), and 10 mL of methanol. The reaction mixture was evaporated to dryness and stored in a desiccator.

Deprotection of O-Tert-Butyl-Protected Hydroxy Amino Acids
Deprotection followed a common deprotection protocol.
A total of 0.5 mmol of O-tert-butyl-protected derivative was suspended in 1 mL of deprotection solution (0.95 mL trifluoracetic acid, 0.025 mL triisopropyl silane, 0.025 mL distilled water) and stirred for 1 h. Subsequently, the mixture was evaporated to dryness and the product was purified using flash chromatography.

General Procedure for Hydrolysis of Esters
The preparation of l-Ala derivative was achieved by acidic hydrolysis of the parent ester. A total of 0.5 mmol (approx. 100 mg) of ester was heated in 10 mL of 0.02 M solution of hydrochloric acid at 45 • C until the spot of the ester disappeared on the TLC plate. Then, the mixture was left to evaporate, and crude product was recrystallized from water.

Flash Chromatography
The compounds were adsorbed on silica gel and purified on a PuriFlash XS420 chromatographic system (Interchim, Montluçon, France) using a gradient elution. We used a combination of silica gel and DIOL-silica gel columns (Interchim) to achieve better separation of the compounds. As the mobile phase, a mixture of hexane and EtOAc was used.

HPLC Analysis and Optical Rotation Measurements
The chemical and enantiomeric purity of the compounds were tested on a Shimadzu chromatograph composed of a DGU-20A3 degasser, an LC-20AD binary gradient pump, an SIL-20AC autosampler with a 100-µL sample loop, a CTO-20AC thermostated column oven, and an SPD-M20A photodiode array detector (UV-PDA) controlled by a CBM-20A system controller from computer software LabSolutions (Shimadzu, Kyoto, Japan). The chiral separation was done on a Lux Amylose-1 (amylose tris (3,5-dimethylphenyl)carbamate) (250 × 4.6 mm) column, particle size 5 µm (Phenomenex, Germany). The mobile phase consisted of methanol or acetonitrile-50 mM aqueous NaClO 4 or methanol-0.2% (v/v) acetic acid, in specified volumetric ratios. The exact conditions are summarized in Table S1 of the Supplementary Material. The flow rate was 1 mL/min and the injection volume was 1 µL. UV detection was performed in single wavelength mode (270 nm). For the collection of the UV spectra, a photodiode array scan mode (in the range 190-400 nm with a 1 nm step) was used.
Optical rotation measurements were used for the investigation of PC-d/l-Pgl-Me derivative and for further investigations of compounds without exact antipodes (different esters of d-and l-aa). Polarimeter Krüss P3000 (A. Krüss Optronic GmbH, Hamburg, Germany) was used. Compounds PC-d/l-Pgl-Me and PC-l-Met-Me were dissolved in chloroform (to compare the obtained specific rotation with literature), and other compounds were dissolved in methanol to final concentrations c = 5 to 20 mg/mL (based on observed rotation and/or available amount of samples). The optical rotation was measured in a 5 cm long glass cuvette using a 589 nm wavelength at laboratory temperature (25 ± 1 • C). All measurements were done in triplicates and the average values are reported.

NMR, IR Analysis, and Melting Point Measurements
The 1 H-and 13 C-NMR spectra of the compounds were recorded on Varian VNMR S500 (Varian, Palo Alto, CA, USA) at 500 MHz for 1 H and 126 MHz for 13 C. Chemical shifts were reported in ppm (δ) and were referred indirectly to tetramethylsilane via the signal of the solvent (2.49 for 1 H and 39.7 for 13 C in DMSO-d 6 ). The infrared spectra were recorded with spectrometer FT-IR Nicolet 6700 (Thermo Scientific, Waltham, MA, USA) using the attenuated total reflectance (ATR) methodology on germanium crystal. Melting points (mp) were measured on Stuart SMP30 (Bibby Scientific Limited, Staffordshire, UK) using the capillary method. For compounds gained in small amounts, mp was measured using Kofler melting point apparatus with a microscope.

LC-MS and GC-MS Analysis
Advantage Max ion-trap LCQ mass spectrometer with ESI ion source (Thermo Finnigan, San Jose, CA, USA) was used for confirmation of the molecular weight of compounds containing free carboxylic group in negative mode (direct injection). All spectra were processed with Thermo Finnigan Xcalibur software (Thermo Finnigan, San Jose, CA, USA).
The GC-MS analysis was performed on a GCMS-QP2010 Plus system operating in EI mode at 70 eV (Shimadzu, Tokyo, Japan). The chromatographic separation was carried out on a Rtx-5MS column (30 m × 0.25 mm × 0.25 µm,Restek,USA). The column temperature was increased by 20 • C/min, from 130 to 320 • C, and this temperature was maintained for 2 min. The injector temperature was 240 • C and the carrier gas flow rate (helium) was set to 1 mL/min. The detection range was m/z 110-700, and the ion source temperature was held at 220 • C. The sample (1 µL) was introduced into injector in split mode (ratio 1:1). The data were recorded and analysed by the GCMS Solution Software ver. 4.45 (Shimadzu,Tokyo,Japan).

H-NMR
The mycobacterial strains were cultured on Middlebrook 7H9 agar and suspensions were prepared in Middlebrook 7H9 broth. The final density was adjusted to value 1.0 according to McFarland scale and diluted in the ratio 1:20 (for fast growing mycobacteria) or 1:10 (for Mtb H37Ra) with broth.
The tested compounds were dissolved in dimethyl sulfoxide (DMSO) (Sigma-Aldrich), then Middlebrook 7H9 broth was added to obtain a concentration of 2000 µg/mL. The standards used for activity determination were isoniazid (INH), rifampicin (RIF), and ciprofloxacin (CPX) (Sigma-Aldrich). The final concentrations were reached by binary dilution and the addition of mycobacterial suspension and were set as 500, 250, 125, 62.5, 31.25, 15.625, 7.81, and 3.91 µg/mL. The final concentration of DMSO in any well did not exceeded 2.5% (v/v) and did not affect the growth of mycobacteria. Positive (broth, DMSO, bacteria) and negative (broth, DMSO) controls were included.
The plates were sealed with polyester adhesive film and incubated in dark at 37 • C without agitation. The addition of 0.01% solution of resazurin sodium salt followed after 48 h of incubation for M. smegmatis, after 72 h of incubation for M. aurum, and after 120 h of incubation for Mtb H37Ra, respectively. Stain was prepared by dissolving resazurin sodium salt (Sigma-Aldrich) in deionised water to get 0.02% solution. Then, 10% aqueous solution of Tween 80 (Sigma-Aldrich) was prepared. Both liquids were mixed up making use of the same volumes and filtered through a 0.2 µm syringe membrane filter. The microtitration panels were then incubated for further 2.5 h for the determination of activity against M. smegmatis, 4 h for M. aurum, and 18 h for Mtb H37Ra, respectively.
Antimycobacterial activity was expressed as the minimum inhibitory concentration (MIC). The MIC (in µg/mL) was determined on the basis of stain colour change (blue colour-active; pink colour-not active). The MIC values for standards are presented in the respective tables. All experiments were conducted in duplicate.

Antibacterial and Antifungal Activity Screening
For details, refer to the Supplementary Material.

Cytotoxicity Screening
The human hepatocellular liver carcinoma cell line HepG2 purchased from Health Protection Agency Culture Collections (ECACC, Salisbury, UK) was cultured in MEM (Minimum Essentials Eagle Medium) (Sigma-Aldrich) supplemented with 10% foetal bovine serum (PAA Laboratories GmbH, Pasching, Austria), 1% l-glutamine solution (Sigma-Aldrich), and non-essential amino acid solution (Sigma-Aldrich) in a humidified atmosphere containing 5% CO 2 at 37 • C. For subculturing, the cells were harvested after trypsin/EDTA (Sigma-Aldrich) treatment at 37 • C. To evaluate cytotoxicity, the cells treated with the tested substances were used as experimental groups, whereas untreated HepG2 cells served as controls.
The cells were seeded in a density of 10,000 cells per well in a 96-well plate. During the next day, the cells were treated with each of the tested substances dissolved in DMSO. The tested substances were prepared at different incubation concentrations (1-1000 µM) in triplicates according to their solubility. Simultaneously, the controls representing 100% cell viability, 0% cell viability (the cells treated with 10% DMSO), no cell control, and vehiculum controls were also prepared in triplicates. After 24 h of incubation in a humidified atmosphere containing 5% CO 2 at 37 • C, the reagent from the kit CellTiter 96 AQueous One Solution Cell Proliferation Assay (CellTiter 96; PROMEGA, Fitchburg, WI, USA) was added. After 2 h of incubation at 37 • C, the absorbance of the samples was recorded at 490 nm (TECAN, Infinita M200, Austria). A standard toxicological parameter IC 50 was calculated by nonlinear regression from a semilogarithmic plot of incubation concentration versus the percentage of absorbance relative to untreated controls using GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA). The compounds were tested in two separate experiments in cell line passage 3 and 23.

Conclusions
To conclude, thirty-nine N-pyrazinoyl aa derivatives were synthesized in this study. Twenty-four of these compounds (mostly d-enantiomers) had not been previously described in the literature (to date 10th February 2020; racemic mixtures and enantiomeric mixtures were considered unique compounds). The in vitro antimicrobial activity of all compounds was investigated, focused on antimycobacterial, antifungal, and antibacterial activity. No activity was detected against any fungal strains. For antibacterial activities, low activity against Staphylococcus aureus was noted for PC-l-OtBu-Thr-Me (MIC = 31.25 µM). The main biological evaluation was thus focused on antimycobacterial testing. The high activity of some compounds was mainly observed against Mtb H37Ra. Significant activity was noted only in the mildly acidic medium of pH 6, whereas, in the commonly utilized pH 6.6, the compounds were generally inactive. The mixture of enantiomers PC-d/l-Pgl-Me presented the best activity (MIC < 1.95 µg/mL, < 7.3 µM, pH 6), which-in molar concentrations-represents four times higher activity than the first-line antitubercular drug PZA. Esters PC-l-Ala-Et, PC-l-Met-Me, PC-l-OBn-Ser-Me and PC-l-Tyr-Et were up to two times more active than PZA (in molar concentrations at pH 6). The derivatives with the free carboxylic group were generally inactive. Based on the observed structure-activity relationships (SAR), we conclude that the activity is tightly linked to the lipophilicity of the compounds (more lipophilic preferred) as well as to the stereochemistry (l-isomers preferred). For the first time, PZA derivatives containing d-aa were investigated. However, their activity was significantly lower compared to the l-enantiomers.
The difference in antimycobacterial activity between l-and d-isomers might indicate the presence of a specific subcellular target for unhydrolyzed product (pharmacodynamic aspect) or simply the faster hydrolysis of the amidic bond of l-isomers by mycobacterial amidases to release the active POA from its prodrug transport form (pharmacokinetics aspect). The exact active species (parent compounds vs. hydrolysed POA) and its MoA on the molecular level are to be elucidated. However, based on the recent biochemical and crystallographic investigations of PanD [12], the most prominent target of PZA, we conclude that our compounds do not act in their non-hydrolysed form against this target, as the binding cavity at the active site of PanD is too small to accommodate anything bigger than POA or 6-Cl-POA.
The work presented in this article opens the door for further medicinal chemistry optimizations of the most active compounds (PC-l-Ala-Et, PC-dl-Leu-Et, PC-l-Met-Me, PC-d/l-Pgl-Me, PC-l-OBn-Ser-Me) and encourages future research. For the next step, we propose the assessment of derivatives combining aa with antimycobacterially active 5-chloropyrazinamide (5-Cl-PZA) as a simplest PZA derivative with known MoA [66]. Our preliminary results show that the l-Leu methyl ester derivative of 5-Cl-PZA is active against Mtb H37Ra (MIC at pH 6.6 was 62.5 µg/mL, at pH 6 it was 3.91 µg/mL). As 5-Cl-PZA is more lipophilic than PZA, even the methyl ester is probably sufficient to effectively penetrate mycobacterial membrane. We thus concluded that the lipophilicity of the aa part of our compounds indeed plays a role in their activity. For a conclusive description of the structure-activity relationships (SAR) of the proposed compounds, more derivatives need to be synthesized focusing mainly on the differences between the activity of different enantiomers with the same ester. This could bring new interesting knowledge of Mtb enzymatic patterns usable also in other applications.
The design of aa-derived small molecules-notably as hydrolysable prodrugs-could be an interesting therapeutic approach for mycobacterial infections. Mtb as any other bacterium is dependent on nitrogen uptake to synthesize vital biomolecules. However, Mtb has been found to prefer proteinogenic aa as the source of nitrogen over ammonium chloride when cultured in vitro [67,68]. Therefore, if the appropriate mechanisms and enzymes involved in these metabolic pathways were identified, the use of aa derived prodrugs could 'trick' Mtb to cleave off the aa and at the same time release the active form of the drug inside the cell. The plausibility of this strategy can be rationalized based on the recent discoveries of several selective Mtb proteases essential either for the survival or virulence of Mtb [69,70], which are proposed as good druggable targets. The ability to uptake all proteinogenic aa has been demonstrated, which could help smaller aa-containing (pro)drugs to pass through the mycobacterial membrane barrier [68]. Penetration through the Mtb membrane barrier could perhaps also be managed by one of the known Mtb ATP-binding cassette (ABC) peptide importers, e.g., dipeptide importer DppABCD or oligopeptide importer OppABCD [71]. Notably, the specificity of the latter has been shown to be very broad [72], which could support our speculations.