The synthesis and characterization (m.p., IR and NMR spectra, elemental analyses) of salicylanilide benzoates [2-(phenylcarbamoyl)phenyl benzoates; 1] were published recently [17]; their synthetic route is depicted in Scheme 1. Yield of esters 1 ranged from 44 up to 88% [17].

Eighteen salicylanilide benzoates were evaluated against four mycobacterial strains—one tuberculous and three atypical ones (Mycobacterium avium and two strains of Mycobacterium kansasii). Results are summarized in Table 1.

All tested compounds exhibited a significant activity against drug-sensitive M. tuberculosis at micromolar concentrations (0.5–8 μmol/L) with 1m, 1o and 1r showing superiority (MICs ≤ 1 μmol/L). When we evaluated the isomers on the salicylic ring, 4-chloroderivatives were more beneficial than 5-chloro ones—only with 1c vs.1d being an exception and with no difference in the 1g vs.1h pair. The order of the moieties on the aniline ring is as follows (according to decreased potency): 4-CF₃ ≥ 3,4-dichloro > 3-CF₃ (limited data) ≈ 4-Br > 3-Cl = 3-Br > 4-Cl > 3-F > 4-F. The benzoylation of salicylanilides provided esters with noticeably improved activity when compared to the parent phenolic molecules S (Table 2) [18]—e.g., even eight-fold for 1m; no ester exhibited inferior activity than its “parent” salicylanilide, and just 1b, 1c, and 1r have identical MIC values.

M. avium showed the lowest level of susceptibility among the investigated mycobacterial strains (MICs 4–16 μmol/L). Compounds 1d, 1h, 1o and 1r are the most active esters. In general, derivatives substituted in the aniline part by 3-chloro (1a, 1b), 4-bromo (1g, 1h), and 4-trifluoromethyl (1o, 1p, 1r) moieties exhibited better activity; on the other hand, 3-fluoroderivatives (1i, 1j) offered minimal benefit. With two exceptions, molecules derived from 5-chlorosalicylic acid showed a higher or equal activity than their 4-chloro isomers. The introduction of a benzoyl fragment into salicylanilide molecules resulted in an increased activity against M. avium—only three MIC values are higher than those of the parent salicylanilides (Table 2) [18], while others are equal or mostly lower in the case of benzoates, even four times in some cases.

Both clinically isolated and collection strains of M. kansasii were inhibited by salicylanilide benzoates 1 with MICs ≤ 8 μmol/L with clear 1o superiority. 4-CF₃, 3-CF₃, 3,4-diCl and 4-Br represent the more suitable aniline substitution patterns; no substituent of the aniline part was evaluated as being significantly less beneficial than others. The influence of the halogen position on the salicylic ring is ambiguous. When concentrated on MICs towards the strain 235/80, there is a surprising fact—when 5-chloro-2-hydroxy-N-phenylbenzamides are esterified, the activity against M. kansasii did not change (or was even diminished for 1k vs.S-k), whereas masking of the phenolic group of 4-chloro-2-hydroxy-N-phenylbenzamides resulted mostly in improved in vitro activity (up to four times); only 1h retained the same MIC values. Under our experimental conditions, benzoic acid itself revealed no activity against M. tuberculosis and M. avium and only a very weak inhibition potency for M. kansasii (MICs ≥ 250 μmol/L).

The most active derivatives proved a similar efficacy when compared to INH against drug-sensitive M. tuberculosis, and all derivatives exhibited substantially higher activity against M. avium and M. kansasii 235/80; the activities of INH and benzoates 1 against M. kansasii 6509/96 are almost identical. Second-line oral drug PAS seems to be significantly less active than the newly synthesized derivatives against all tested strains.

Salicylanilide benzoates expressed predominantly lower or equal MIC values in comparison to corresponding acetates [8] and benzenesulfonates [12]. Carbamates possessed a slightly higher in vitro inhibitory activity for M. tuberculosis, whereas MIC levels against atypical strains are approximately similar [9]. Salicylanilide N-acetyl-L-phenylalanine esters demonstrated a superior activity against M. tuberculosis and somewhat worse against M. avium [10]. Benzoates surpassed the antimycobacterial activity of salicylanilides esters with different N-benzyloxycarbonyl α-amino acids [11].

In conclusion, the benzoylation of salicylanilides S led to derivatives with predominantly higher in vitro activity against all four mycobacterial strains. The aim of improving the antimycobacterial potency was successfully achieved. The reason may lay in the increased lipophilicity of synthesized esters, which facilitates passage through biomembranes. With respect to the weak intrinsic activity of benzoic acid against M. kansasii, the possibility of synergistic action of released salicylanilides and benzoic acid could be included, as it has been previously observed.

Four esters with the lowest MICs (≤1 μmol/L against any mycobacterial strain) were selected for advanced biological tests. Benzoates 1m, 1n, 1o and 1r were evaluated for their in vitro activity against five MDR-TB strains and one XDR-TB strain (Table 3). All four derivatives exhibited very low MICs (0.25–2 μmol/L). Interestingly, in most cases MDR strains are even more sensitive than drug-sensitive M. tuberculosis. This susceptibility is independent on the resistance patterns indicating no cross-resistance with the conventionally used drugs.

4-Trifluoromethyl derivative 1o was assayed as the most active compound. Based on the pair 1m vs. 1n, the preferable location of the chlorine is the position 4 of the salicylic ring (compound 1m). Compound 1o, the derivative of 5-chlorosalicylic acid, exhibited a better in vitro activity than 1r, which was synthesized from 5-bromosalicylic acid. The salicylanilide benzoates 1m, 1n, 1o, and 1r exhibited a higher activity against drug-resistant strains (expressed as MICs) than salicylanilide esters with N-acetyl-L-phenylalanine [10] and similar or slightly better MIC values when compared to salicylanilide carbamates [9].

Salicylanilides and their esters were referred to share some toxic effects on eukaryotic cells [7,8,9,10,16,19]. Therefore we examined the in vitro cytotoxicity of three most anti-MDR-TB active benzoates (compounds 1m, 1o, 1r) and their parent salicylanilides (S-m, S-o, S-r) in the liver Hep G2 model. The cytotoxicity values, which are expressed as IC₅₀, i.e., concentration which decreases the viability of the cells to 50% from the maximal viability, was taken over from reference [16] (Table 4).

The esterification of 5-chloro-N-(3,4-dichlorophenyl)-2-hydroxybenzamide (S-m) and 5-chloro-2-hydroxy-N-[4-(trifluoromethyl)phenyl]benzamide (S-o) by benzoic acid led to the derivatives with significantly decreased toxicity (approximately three and seven times, respectively); contrarily, the benzoylation of S-r has resulted in a slightly higher cytotoxicity (2.71 vs. 2.34 µmol/L).

Based on the comparison of MIC and IC₅₀, it is possible to predict that the antimycobacterial activity of salicylanilide benzoates is not only the result of a general cytotoxic impact, but that they probably should have additional specific effect(s) against M. tuberculosis—e.g., recently reported inhibition of isocitrate lyase and methionine aminopeptidase [16].

However, benzoates still exhibited IC₅₀ values in micromolar range similar to the activities against atypical mycobacteria. Selectivity indexes (SI) for M. tuberculosis ranges from 2.34 to 4.80. The situation for MDR-TB and XDR-TB strains is quite advantageous with SI values of 1.17–10.16; the ratios are more favourable for 1o. Generally, only SI values about the break point of 10 could be considered to be still border sufficient, others are poor.

Although benzoic acid possesses only a very mild cytotoxicity (IC₅₀ of 2,881 µmol/L in our assay), unfortunately the benzoylation did not fill up our expectation about the significant toxicity reduction of parent salicylanilides, in contrast to improved antimycobacterial efficacy. Otherwise, esterification of salicylanilides may be a perspective way to reduce undesired cytotoxicity; it is necessary to search new acids, because benzoic acid brings a certain, but insufficient benefit.

Synthesis and characterization of the presented 2-(phenylcarbamoyl)phenyl benzoates 1a–r was published by Krátký et al. [17].

All compounds were tested against Mycobacterium tuberculosis 331/88 (H₃₇Rv) (dilution of the strain was 10⁻³) and three nontuberculous strains: Mycobacterium avium 330/88 (resistant to INH, RIF, ofloxacin and ethambutol; dilution 10⁻⁵) and two strains of Mycobacterium kansasii—235/80 (dilution 10⁻⁴) and clinically isolated strain 6509/96 (dilution 10⁻⁵). The description of the used method can be found in [12]. The following concentrations of esters were used: 1,000, 500, 250, 125, 62.5, 32, 16, 8, 4, 2, 1, 0.5, 0.25 and 0.125 µmol/L. MIC (µmol/L) is the lowest concentration at which the complete inhibition of mycobacterial growth was occurred. Isoniazid (INH) and structurally similar para-aminosalicylic acid (PAS) were chosen as reference drugs for the comparison. The most active compounds were evaluated in the similar conditions and concentrations against six MDR-TB strains (dilution 10⁻³) with different resistance patterns: 7357/1998, 234/2005, 53/2009, 9449/2006, Praha 1 and Praha 131 (XDR-TB strain). All these strains were resistant to INH, rifamycines, and streptomycin; in some cases additional resistance was present.