Design and Synthesis of Highly Active Antimycobacterial Mutual Esters of 2-(2-Isonicotinoylhydrazineylidene)propanoic Acid

The combination of two active scaffolds into one molecule represents a proven approach in drug design to overcome microbial drug resistance. We designed and synthesized more lipophilic esters of 2-(2-isonicotinoylhydrazineylidene)propanoic acid, obtained from antitubercular drug isoniazid, with various alcohols, phenols and thiols, including several drugs, using carbodiimide-mediated coupling. Nineteen new esters were evaluated as potential antimycobacterial agents against drug-sensitive Mycobacterium tuberculosis (Mtb.) H37Rv, Mycobacterium avium and Mycobacterium kansasii. Selected derivatives were also tested for inhibition of multidrug-resistant (MDR) Mtb., and their mechanism of action was investigated. The esters exhibited high activity against Mtb. (minimum inhibitory concentrations, MIC, from ≤0.125 μM), M. kansasii, M. avium as well as MDR strains (MIC from 0.25, 32 and 8 µM, respectively). The most active mutual derivatives were derived from 4-chloro/phenoxy-phenols, triclosan, quinolin-8-ol, naphthols and terpene alcohols. The experiments identified enoyl-acyl carrier protein reductase (InhA), and thus mycobacterial cell wall biosynthesis, as the main target of the molecules that are activated by KatG, but for some compounds can also be expected adjunctive mechanism(s). Generally, the mutual esters have also avoided cytotoxicity and are promising hits for the discovery of antimycobacterial drugs with improved properties compared to parent isoniazid.


Introduction
Currently, tuberculosis (TB) and COVID-19 are worldwide pandemic infectious diseases that cause the most deaths in the world. Both diseases have similar symptoms such as cough, fever and difficulty breathing. However, TB has a longer incubation period with a slower onset of disease [1]. About one-quarter of the world's population is estimated to be infected with Mycobacterium tuberculosis (Mtb.), the bacteria that give rise to the disease, especially in low-resource countries including regions with high poverty, lacking basic amenities and medical facilities [2]. Anti-TB medicines have been used for decades and strains that are resistant to one or more of the medicines have been documented in every country surveyed. Worldwide, only 57% of multidrug-resistant (MDR) TB patients are 2. Materials and Methods 2.1. Chemistry 2.1.1. General All chemicals for synthesis and analysis were purchased from Merck KGaA (Darmstadt, Germany), Penta Chemicals Unlimited (Prague, Czech Republic) or Lach-Ner (Neratovice, Czech Republic) and were used as received without further purification. The structures of the prepared substances were confirmed via 1 H NMR and 13 C NMR spectroscopy analysis. NMR spectra (500 MHz for 1 H and 126 MHz for 13 C) were measured in DMSO-d 6 , DMF-d 7 or CDCl 3 as solvents at ambient temperature by a Varian VNMR S500 instrument (Varian, Palo Alto, CA, USA). The values of chemical shifts (δ) are given in parts per million (ppm) and the spectra were referenced internally, to tetramethylsilane as standard by the residual signal of protic solvent (DMSO-d 6 34.89, 163.15 for 13 C and CDCl 3 : 7.25 for 1 H, 77.10 for 13 C). The coupling constants (J) are reported in Hz. Infrared spectra were recorded by a Nicolet 6700 FT-IR spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) in the range 600-4000 cm −1 ; ATR (Ge) technique of measuring was used. Elemental analysis was performed on a Vario MICRO Cube Element Analyzer (Elementar Analysensysteme, Hanau, Germany). Both calculated and found values are given as percentages. Melting points were recorded using a Büchi B-545 apparatus (BÜCHI Labortechnik AG, Flawil, Switzerland). Retention factors (R f ) of all prepared compounds, and reaction progresses were analysed by thin layer chromatography (TLC); the plates were coated with 0.2 mm Merck 60 F254 silica gel (Merck Millipore, Darmstadt, Germany) and were visualised by UV irradiation (254 nm). Dichloromethane/methanol (S1) (4:0.3 v/v) and dichloromethane/acetone (S2) (4:0.5 v/v) mixtures were used as the eluents for TLC and for column chromatography. Merck Kieselgel 60 Å (0.040-0.063 mm) Merck KGaA (Darmstadt, Germany) was used for column chromatography.
The calculated logP values (ClogP), which are the logarithms of the partition coefficients for octan-1-ol/water and reaction schemes, were acquired using the program ChemDraw Professional 18.1 (PerkinElmer Inc., Waltham, MA, USA).

Analysis of Sensitivity of Mtb. H 37 Ra Strains Overproducing InhA and KatG
InhA protein was overproduced in Mtb. H 37 Ra using pMV261-inhA construct as previously described [21]. KatG protein encoded by the gene MSMEG_6384 from Mycobacterium smegmatis was overproduced in Mtb. H 37 Ra using pVV16-katG smeg construct previously described [14].
Susceptibility of Mtb. H 37 Ra strains overproducing InhA tb or KatG smeg , and control strains carrying empty pMV261 or pVV16 vectors, to the esters 3l, 3m and 3o was evaluated using the drop dilution method. The cultures grown in 7H9 broth supplemented with albumin-dextrose-catalase and 0.05% Tween 80 were adjusted to optical density (OD) 600~0 .5 and 4 µL aliquots of 1, 10 −1 and 10 −2 dilutions of the compounds 3 were dropped on 7H11 agar supplemented with oleic acid-albumin-dextrose-catalase and various concentrations of the tested compounds dissolved in DMSO (2% final concentration). Plates were incubated for 28 days at 37 • C.

Investigation of mechanism of action
The mode of action was analysed by metabolic labelling of Mtb. H 37 Ra strain with 14 for 3l and 3m  and 0.125 and 0.625 µM for 3o. The final concentration of DMSO was kept at 2%. INH in final concentration of 3.5 µM was used as a control agent. 14 C acetate (specific activity 110 mCi/mmol, ARC) was added to each of the cultures (in final concentration 0.5 µCi/mL) after 24 h of cultivation with inhibitors and the cells were cultivated for next 24 h.
The lipids from the cells grown in 100 µL cultures were extracted with 1.5 mL chloroform/methanol (1:2, v/v) at 56 • C for 1 h, followed by the extraction with 1.5 mL chloroform/methanol (2:1, v/v) at 56 • C for 1.5 h. The organic extracts were combined, dried under N 2 and subjected to biphasic Folch washing [22]. Isolated lipids were dissolved in 50 µL of chloroform/methanol (2:1, v/v) and 10 µL was loaded on TLC silica gel plates F254 (Merck). Lipids were separated in chloroform/methanol/water mixture (20:4:0.5, v/v/v) and visualized using an Amersham TM Typhoon TM Biomolecular Imager or by autoradiography.
Fatty acid methyl esters (FAME) and mycolic acids methyl esters (MAME) were prepared from whole cells from 100 µL cultures as previously described [23]. Dried extracts were dissolved in 50 µL of chloroform/methanol (2:1, v/v) and 10 µL was loaded on TLC plates. FAME and different forms of MAME were separated by three runs in n-hexane/ethyl acetate (95:5, 3 runs) and visualized as described for lipids.
For subculturing, the cells were harvested after trypsin/EDTA (Sigma-Aldrich) treatment at 37 • C. To determine cytotoxicity of the compounds, the cells treated with the tested substances were used as experimental groups whereas untreated HepG2 cells served as control groups.
The cells were seeded in a density 15,000 cells per well in a 96-well plate. The next day the cells were treated with each of the tested substances at a broad range of concentrations (1-1000 µM) in triplicates; the compounds were dissolved in DMSO (maximal incubation concentration of DMSO was 1% v/v). The controls representing 100% cell viability, 0% cell viability (the cells treated with 10% DMSO), no cell control and vehiculum controls were incubated in parallel, also as 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 incubation at 37 • C, absorbance of samples was recorded at 490 nm (TECAN, Infinita M200, Grödig, Austria). A standard toxicological parameter IC 50 was calculated by nonlinear regression from a semilogarithmic plot of incubation concentration versus percentage of absorbance relative to untreated controls using GraphPad Prism 8 software (GraphPad Software, Inc., La Jolla, CA, USA).
Results of the experiments are presented as inhibitory concentrations that reduce viability of the cell population to 50% from the maximal viability (IC 50 ). Parent INH was involved as a reference agent.
Regarding alcohols and thiols, mostly bioactive compounds with a known antimicrobial activity were chosen to form mutual derivatives, aiming to obtain molecules with an improved inhibition of mycobacteria. Methanol was used for the synthesis of reference ester 3a. Phenol (substrate for 3b) was first used as an antiseptic by Joseph Lister during implementation of antiseptic surgery methodology [24]. It was later described that a mixture of three positional isomers of cresol (tricresol) excels in its antiseptic effects. Owing to these and its other important biological properties, phenol has become an interesting synthetic template for several biologically active compounds, especially antibacterial chemotherapeutics [24]. In general, antimicrobial properties are shared by various phenols, and their halogenation or substitution by lipophilic groups (e.g., alkyl(oxy), aryl(oxy), halogen) boosts their activity (i.e., precursors of 3c-3j) [25]. Monoterpene phenolic derivatives of cymene such as carvacrol (3k) and thymol (precursor of 3l) have also been investigated for their potential antituberculotic properties. Andrade-Ochoa et al. [26] reported very good antimycobacterial activity of essential oils containing carvacrol and thymol. Their extensive QSAR and in vitro studies confirmed that these terpenes may be suitable candidates for antituberculotics. Quinoline and its derivatives are also in the forefront of interest as model structures for the design of antituberculotics (ester 3m). Thus, of particular interest is the discovery of bedaquiline, an innovative drug approved for the treatment of MDR-TB [27]. Darby et al. [28] described the antituberculotic activity of quinolin-8-ol against latent (nonreplicating) as well as active growing Mtb. They point out the properties of quinolin-8-ol as a suitable antituberculotic drug candidate for shortening the course of treatment for both active and latent form of tuberculosis. Pyridin-4-ylmethanol was involved as an analogue (derivative 3n) and metabolic precursor of isonicotinic acid present in INH. Another interesting example of bioactive phenol is triclosan (also called irgasan, 5-chloro-2-(2,4-dichlorophenoxy)phenol; precursor of 3o), as a widely used broad-spectrum biocidal agent and confirmed enoyl-ACP reductase inhibitor that has been highly modified, leading to many compounds with significant antimicrobial (including antimycobacterial) activity [29]. A study of Zimmermann and Curtis [30] provided a new perspective on clinically used antipyretics. Interestingly, some common antipyretics such as acetaminophen (paracetamol; ester 3p) are able to inhibit or promote bacterial growth, affect biofilm production, affect the motility, adherence, and metabolism of pathogens, and induce or reduce the frequency of mutations, leading to antimicrobial resistance. Targeting these factors using a synthetic combination of antituberculotics and antipyretics may provide important insights for the future search of new antitubercular molecules. From the point of view of anti-TB action, it has also proved advantageous to use phenol isostere. Thiophenols (here used for thioesters 3r and 3s) are frequently used in the manufacture of antibiotic drugs, such as sulphonamides. Antifungals butoconazole and thiomersal are also thiophenolic derivatives [31]. Certain improvement in the profile of a potential antituberculotic drug following the incorporation of the thiophenol-based fragment has been described by Mewada et al. [32]. In addition to phenol derivatives, other bioactive compounds with the OH group have been investigated for their potential antituberculotic properties. Metronidazole (precursor of 3q) is a synthetic drug from the group of nitroimidazoles. It is frequently used to treat protozoal and anaerobic infections. It has been demonstrated that this drug is able to suppress nonreplicating/latent mycobacterial population [33]. The efficacy against Mtb. under anaerobic conditions in vivo has been confirmed by Carroll et al. [34]; their study showed some effectiveness of orally administered metronidazole in terms of faster negative sputum cultivation.

Chemistry
(E)-2-(2-Isonicotinoylhydrazineylidene)propanoic acid 2 was synthesized from lowcost and commercially available isoniazid 1 and pyruvic acid in boiling methanol, according to the previously described method [14,19] with almost quantitative yield (Figure 1). The crystals of the product precipitated from the reaction mixture immediately after addition of pyruvic acid. The precipitate was washed repeatedly with cold methanol. Only the more thermodynamically stable (E)-isomer was detected. (E)-2-(2-Isonicotinoylhydrazineylidene)propanoic acid 2 was esterified with various (thio)alcohols, or (thio)phenols by Steglich procedure employing N-(3-dimethylaminopropyl)-N -ethylcarbodiimide hydrochloride as a coupling agent and 4-(dimethylamino)pyridine as a catalyst. Dichloromethane was used as the solvent (Figure 1). This established method is efficient and features mild conditions.The resulting urea byproduct was removed by extraction to water and organic layer was extracted successivelly with sodium bicarbonate and saturated brine and evaporated under reduced pressure. All crude products were purified by crystallization or column chromatography.
The yields of 3 ranged from 36 to 92% depending mainly on purification process. In some cases, the crystallization process had to be repeated several times, resulting in diminished yields. A relatively low yield (36%) was obtained for the triclosan-based compound 3o, most likely due to a strong steric hindrance of the reacting hydroxyl. In addition, the crude product had to be purified several times by recrystallization. Better reaction conversions provided reactions with aliphatic substrates leading to (thio)esters 3a and 3r; these compounds did not need to be purified more than once. Unsubstituted phenol and naphthols (3b and 3i, 3j) or some phenols substituted by electron-donating substituents (3c, 3k, 3l) also gave good yields (52-81%). Compounds 3k and 3l are positional isomers; compound 3l was prepared in lower yield (52%), most likely due to a steric hindrance. However, the yields of compounds 3i, 3j, 3k and 3l were reduced by their repeated column purification of the crude product. We also observed a partial decomposition of compounds 3i, 3j, 3k and 3l upon contact with silica gel, even after its basification using triethylamine. This phenomenon was partially eliminated in these compounds by the aprotic composition of the mobile phase (S2). The synthesis of derivative 3s showed better reaction conversion due to the higher nucleophilicity of thiophenol compared to the synthesis of the phenol derivative 3b; a similar result was observed in the case of compounds 3r and 3a.
The formation of two geometric isomers was observed in the synthesis of compounds 3i and 3p. In the first case, the (Z)-isomer, identified by NMR, was not isolated because it was obtained in a very small amount and removed during purification. Compound 3p was prepared as a mixture of isomers (4:1 according to NMR) described below.
Each prepared compound was characterized by 1 H and 13 C NMR, IR spectra and melting point. The purity was assessed by TLC and elemental analysis.
The large majority of the (thio)esters 3 exhibited better antitubercular activity than parent acid 2, with an exception of compound 3n. Thus, converting free carboxyl to its more lipophilic analogues is beneficial in terms of enhanced antimycobacterial properties (up to ≥8×). Importantly, identical conclusions can also be made for parent INH 1.
Additionally, we evaluated antimycobacterial activity of the corresponding (thio)alcohols and phenols used for the preparation of esters 3 ( Table 2). The activities of phenol and its methyl, methoxy and halogen derivatives were negligible as well as for paracetamol and metronidazole (MIC ≥1000 µM). Both naphthols were more active (250-1000 µM). Terpenes exhibited MIC values of 64-500 µM, while thymol showed a mild superiority to isomeric carvacrol. Quinolin-8-ol was identified to be the most potent against all three strains, closely followed by triclosan and then 4-phenoxyphenol with lower inhibition of M. avium, especially. These results confirmed justification of the esters design and mutual drugs concept used in this study. To highlight connection of two bioactive scaffold in one molecules entity, it is necessary to remark that the activity of esters is usually markedly higher than in the case of both components separately, thus indicating additivity and/or synergism after their conjugation. Having results of initial antimycobacterial screening in our hands, we selected some esters against six MDR-and one XDR-TB strains with different resistance profiles. Thus, we selected the most anti-TB active esters (i.e., with both MIC of <0.5 µM) covering various representative types of (thio)phenolic derivatives (3b, 3d, 3g, 3i-3l, 3o and 3r) and esters of established drugs (3k-3m, 3o-3q). Importantly, the esters were able to inhibit the growth of resistant strains, albeit with higher MIC values of 8-125 µM ( Table 3). The susceptibility profile of all esters is independent on the resistance pattern; Praha 1 strain was the least sensitive. However, the results indicate a cross-resistance to the parent INH. Quinolin-8-yl ester 3m was the most active derivative with uniform MIC of 8 µM corresponding with its MIC for Mtb. H 37 Rv, thus suggesting adjunctive mechanism(s) of action in addition to an "isoniazid-like" one. For example, quinolin-8-ol has been identified as a metal chelator [28]. Other more potent esters were 2-isopropyl-5-methylphenyl (i.e., derivative of thymol, 3l), phenyl 3b and especially 4-methoxyphenyl 3d and 4-chlorophenyl 3g esters, thus highlighting importance of the phenyl ring substitution by a small lipophilic group. In these types of compounds, the highly active INH-based scaffold can eliminate INH-susceptible mycobacteria at very low concentrations, and, in the case of resistance, the pathogen is killed by the second scaffold (quinolin-8-ol, thymol, etc.). Additivity or synergism are also conceivable.

Investigation of Mechanism of Action
The highly active esters 3 (obtained from thymol, quinolin-8-ol and triclosan, 3l, 3m and 3o, respectively) were used for experimental evaluation of their mechanism of action, especially with respect to their intrinsic antimycobacterial activity. Keeping in mind their design and retained INH scaffold, we considered enoyl-ACP reductase (InhA) as a potential main target. That is why we determined MIC values of these selected esters for Mtb. H 37 Ra overproducing InhA together with their activity against KatG overproducing Mtb. H 37 Ra strain. The gene katG used for this construct was of Mycobacterium smegmatis origin.
To confirm identical mechanism of action with INH (InhA inhibition), the mycobacterial strain overproducing InhA as the target molecule should be significantly more resistant to the action of the esters due to an excess of the cellular target. On the other hand, the strain overproducing KatG activator should be more susceptible due to the facilitated bioactivation of prodrugs into the active form.
The MIC found are reported in Table 4. It covers results for H 37 Rv and all H 37 Ra strains involved in this study, i.e., those without additional genes but carrying an empty vector (pMV261, pVV16), strain overproducing InhA (pMV261-inhA) and strain with KatG overproduction (pVV16-katG). Obviously, the Mtb. H 37 Ra overproducing InhA tb cells are more resistant to the investigated esters 3 (0.25 vs. 5 µM for 3l and 3m, 0.325 vs. 6 µM for 3o). Thus, we experimentally confirmed that these compounds inhibit InhA protein as their main target. In addition, the strain overproducing KatG smeg exhibited significantly lower MIC values than the control strain, especially for thymol derivative 3l, i.e., confirming enhanced mycobacterial inhibition depending on facilitated activation mediated by KatG analogously to our previously reported 2-(2-isonicotinoylhydrazineylidene)propanamides [14]. The esters 3m and 3o also exhibited inhibition of INH-resistant strain of M. avium. This nontuberculous species lacks the enzyme KatG [35]; therefore, this activity must be due to a facilitated nonenzymatic release of isonicotinoyl radical and/or the effect of the second component of esters. It has been described that convenient substitution on hydrazide N 2 can enhance its activation to promote its antimycobacterial properties [10,35]. This phenomenon can explain lower MIC values in comparison with the parent INH. An easier conversion to active form also offers a potential advantage for inhibiting strains with katG mutations that prevent INH bioactivation. The dual antimycobacterial activity, dependent on (thio)alcohol/phenol compounds used for esterification, is another advantage overcoming resistance to INH.
To confirm inhibition of mycolic acid synthesis by the investigated esters 3, Mtb. H 37 Ra cells were treated with different concentrations of the compounds and metabolically labelled with 14 C acetate to monitor the synthesis of mycolic acids and lipids. Lipid analysis revealed that the treatment with esters 3 resulted in the specifically decreased production of trehalose monomycolates (TMM) and trehalose dimycolates (TDM; Figure 2A). Analysis of fatty acids isolated from whole cells confirmed the reduced synthesis of all forms of mycolic acids in contrast to the synthesis of standard fatty acids that were not affected ( Figure 2B). These findings confirm the ability of the novel esters 3 to inhibit the production of mycolic acids in mycobacterial cells and thus definitely indicate InhA enzyme as their primary molecular target.

Cytotoxicity and Selectivity
All the (thio)esters 3 and their precursors were also screened for in vitro toxicity on the standard liver cancer cell line HepG2. The assay used (CellTiter 96) is based on the reduction of tetrazolium compound MTS in living cells to formazan dye, which is determined spectrophotometrically. The decline in formazan production is due to a decrease in number of viable cells.
With a focus on selectivity expressed by indexes SI (defined as IC 50 /MIC; Table 5), all compounds showed securely high values for both Mtb. H 37 Rv and M. kansasii (SI of ≥493 and ≥26.4, respectively). Due to substantially negligible toxicity, some of the compounds were also selective for MDR and XDR strains despite their lower activity. In sum, the presented esters 3 are nontoxic for liver cells that are target tissue for INH-related toxicity [36], and highly selective for mycobacteria.

Conclusions
By modifying the established antitubercular isoniazid scaffold, we have uniquely combined this molecule with other bioactive compounds (some of which have their own antimycobacterial properties) and their analogues containing hydroxyl and thiol groups to prepare mutually active derivatives. The molecules were linked with bifunctional pyruvic acid to form hydrazone and ester bonds. In addition to dual activity, the esters are more lipophilic when compared to the parent drug, which facilitates crossing through biological barriers, including the mycobacterial cell wall.
We have synthesized nineteen esters by coupling (E)-2-(2-isonicotinoylhydrazineylidene)propanoic acid activated by EDC/DMAP with different bioactive alcohols, phenols, and thiols in moderate to very good yields. The esters exhibited outstanding antimycobacterial activity against Mtb. H 37 Rv and M. kansasii. Some derivatives were also unusually active against highly resistant M. avium. The effect on MDR and XDR-TB strains was comparatively lower, but some compounds were still active depending on the second component favouring 4-chlorophenol (3g) and quinolin-8-ol (3m). It indicates an additional mechanism of action other than inhibition of InhA that was experimentally confirmed. Due to the presence of INH scaffold, the esters selectively abolished biosynthesis of all types of mycolic acids.
Hepatotoxicity is a common side effect of antimycobacterial drugs, so we evaluated toxicity of the esters to HepG2 cells. Both IC 50 and SI values indicate low toxicity and sufficient selectivity of examined compounds due to negligible cytotoxicity.
Taken mutually and together with their improved physicochemical properties, (E)-2-(2-isonicotinoylhydrazineylidene)propanoates presented here are promising hybrid antimycobacterial derivatives in terms of lower MIC, broad spectrum of activity, limited toxicity, and dual activity depending on both bioactive components.  . We thank Williams R. Jacobs, Jr. for pMV261-inhA construct and Petr Pávek, for consulting the biological evaluation procedure.