Murine RAW 264.7 macrophages (ATCC TIB-71) were grown at 37°C in a humidified atmosphere containing 5% CO₂ in RPMI 1640 medium supplemented with 5% fetal bovine serum, 1 mM sodium pyruvate solution, and 2 mM GlutaGro supplement (Corning). HepG2 cells (ATCC HB-8065) were cultured in Dulbecco-modified Eagle medium (DMEM), 10% fetal bovine serum (FBS), and 1× penicillin streptomycin solution (100 U/ml). M. tuberculosis H37Rv-LP (ATCC 25618) constitutively expressing codon-optimized DsRed from plasmid pBlazeC8 (DREAM8) (Carroll et al., 2018) was cultured at 37°C in Middlebrook 7H9 medium containing 10% v/v OADC (oleic acid, dextrose, catalase) supplement (Becton Dickinson) and 0.05% w/v Tween 80 (7H9-Tw-OADC) plus 100 μg/ml hygromycin B.

Compounds were obtained in 384-well plates as 10 mM DMSO stock solutions and tested as described (Manning et al., 2017). Briefly, assay plates were prepared in clear bottom, black 384-well plates with 30 μl cRPMI and 0.6 μl compound in columns 3–22 (final concentration of 10 μM compound, 1% DMSO). 1% DMSO was used as the negative control, 10 mM isoniazid (INH) and 100 μM staurosporine (STA) were included as positive controls (maximum inhibition) for anti-tubercular activity and cytotoxicity respectively. INH and STA were also included in dose response on each plate. RAW264.7 cells were infected with M. tuberculosis DREAM8 as described (Manning et al., 2017) at an MOI of 1 for 24 h and extracellular bacteria removed by washing. Cells were recovered using Accumax, harvested, washed and resuspended in serum-free RPMI; 30 μl of infected cells were dispensed into each well at 3,300 cells/well. Plates were incubated for 72 h, 10 μl of 5X SYBR Green I was added and plates were imaged with an ImageXpress Micro High Content Screening System (Molecular Devices) using a ×4 objective and FITC and Texas Red channels. MetaXpress was used to analyze images. The integrated intensity of M. tuberculosis or macrophages was calculated for each well. Growth inhibition was calculated for each test well by normalizing to the average integrated intensity of the DMSO control wells. IC₅₀ was calculated as the compound concentration required to reduce bacterial growth by 50%. TC₅₀ was calculated as the compound concentration required to reduce macrophage viability by 50%.

Cytotoxicity against HepG2 cells was measured after 72 h. Cells were seeded in 384-well plates at 1,800 cells per well. Compounds were added as a 10-point three-fold serial dilution after 24 h (final assay concentration of 1% DMSO). CellTiter-Glo^® reagent (Promega) was added and relative luminescence units (RLU) measured. Data were normalized to the DMSO controls. Curves were fitted using the Levenberg–Marquardt algorithm; TC₅₀ was calculated as the compound concentration required to reduce cell viability by 50%.

Minimum inhibitory concentrations (IC₉₀) against M. tuberculosis were determined in liquid medium as described (Ollinger et al., 2013). Bacterial growth was measured after 5 days by OD₅₉₀. IC₉₀ was defined as the concentration of compound required to inhibit growth of M. tuberculosis by 90% and was determined using the Levenberg–Marquardt least-squares plot.

We were interested in identifying new anti-tubercular molecules with activity against intracellular bacteria. We used our previously validated high throughput high content screen to test the MyriaScreen II library of 10,000 diverse molecules which is comprised of ∼60% singletons with diverse structural groups (TimTec/Sigma Aldrich). Molecules were tested at a single concentration of 10 µM (Figure 2A). We monitored cytotoxicity to the macrophages in the same wells to exclude false positives (since the bacteria are unable to replicate if the macrophages are dead). We identified 308 compounds with >70% bacterial growth inhibition; of these we discarded 134 with significant macrophage toxicity, defined as <70% macrophage survival (Figure 1, Figure 2B). This left 174 hit compounds that were non-cytotoxic but restricted the growth of M. tuberculosis in macrophages, a hit rate of 1.7% (Figure 1).

We saw some negative inhibition values in the screen, which is not unusual. Compound fluorescence will lead to negative inhibition and is seen more often in the green end of the spectrum than the red. Cytotoxicity can lead to increased signal from the bacterial populaton, presumably because the bacteria are released into the medium and no longer intracellular; we see this effect with staurosporine. Changes in growth of RAW264.7 cells could also be a factor, since uninfected cells will grow, whereas the infected cells do not tend to grow; if compounds are very active, this could lead to increased macrophage growth.

We analyzed the 174 resulting hits and prioritized hit compounds based on their physicochemical properties including molecular weight (MW), calculated partition coefficient (cLogP), novelty, structural diversity, chemical tractability, and commercial availability. We selected 43 compounds for further analysis (Figure 1, Figure 3); we purchased new batches of compounds and determined the activity using a dose response format (Figure 2C). Seven compounds exhibited cytotoxicity against eukaryotic cells that hampered our ability to determine their anti-bacterial activity and were excluded from further study (Table 1). The IC₅₀ of the remaining 36 hits was determined (Figure 3C); of these all but one (35/36) had confirmed activity with IC₅₀ < 50 µM against intracellular M. tuberculosis (Table 1). The majority of the hits (Mori et al., 2015) had an IC₅₀ <5 μM, and 6 compounds were potent with an IC₅₀ < 1 µM (Figure 3C). The majority of the compounds had no cytotoxicity (TC₅₀ > 50 µM); a few compounds were cytotoxic at higher concentrations, but retained good selectivity (Table 1). The IC₅₀ for INH against M. tuberculosis was 0.23 ± 0.08 µM (n = 192) and for STA against the RAW264.7 cells was 0.0034 ± 0.0013 µM (n = 112).

We further evaluated cytotoxicity using the HepG2 cell line cultured with either glucose or galactose in the medium (Table 2). HepG2 cells cultured with galactose rely on mitochondrial oxidative phosphorylation rather than glycolysis which increases their susceptibility to mitochondrial toxicants allowing us to identify compounds with mitochondrial toxicity (Kamalian et al., 2015). The majority of the compounds showed no difference in cytotoxicity between the two conditions with the exception of compound TPN-0073571 which had increased cytotoxicity against HepG2 cells in galactose (ratio of TC_50glu/TC_50gal of ∼4) suggesting that this compound may impair mitochondrial function. Interestingly the compound showed similar cytotoxicity against the murine macrophages (TC₅₀ = 15 ± 5 µM). Overall, cytotoxicity against the two cell lines (murine RAW 264.7 and human HepG2 cells) was comparable (Table 2).

We wanted to determine if our compounds were selective for intracellular bacteria, so we determined the IC₉₀ against M. tuberculosis in axenic culture. Only 12 compounds had IC₅₀ > 20 µM against extracellular M. tuberculosis, with six of these having an IC₉₀ of >20 µM (Table 2). None of the compounds were more active against M. tuberculosis growing in axenic culture, with the extracellular/intracellular IC₅₀ ratios ranging from 4 to 28.

We performed structural clustering on the set of 174 hits in order to identify common and potentially privileged scaffolds. Compounds were grouped into series sharing the same scaffold/chemotype, each featuring 2 to 8 compounds with 8 singletons. Scaffolds used in the substructure searching, along with their frequency in the hit list and the entire library are depicted in Table 3.

The prioritization of compounds was not based solely on biological potency, but focus was placed on the quality/structural novelty of each chemotype, along with their physicochemical properties. We highlighted five chemotypes that were of interest for further development. A representative compound from each series is shown in Figure 4.

We identified 15 compounds with a benzene amide ether scaffold in the primary screen, of which 10 were active (defined as >70% inhibition of bacterial replication) (Table 3). The BAE series has not previously been identified in phenotypic screens against replicating M. tuberculosis, although one compound from this series (6) had activity against non-replicating M. tuberculosis (Christophe et al., 2009). In addition, compound 12 was identified as an inhibitor of M. tuberculosis with an IC₉₀ of 6.9 μg/ml (20). We purchased 13 commercially-available compounds and determined intracellular activity against M. tuberculosis, although we did not see activity up to 20 μM. Eleven compounds had activity against intracellular M. tuberculosis (Table 4). Two compounds (4 and 7) were cytotoxic to macrophages and we could not calculate activity.

NC–not calculated due to cytotoxicity. cLogP was calculated using Collaborative Drug Discovery (https://www.collaborativedrug.com/).

Eight compounds showed good intracellular activity (IC₅₀ <10 µM). There was a ten-fold difference between the most potent compound (3, IC₅₀ = 1.4 μM) and the least potent compound (13, IC₅₀ = 13 μM). Compounds generally showed low cytotoxicity against both RAW264.7 cells and HepG2 cells with either TC₅₀ >100 μM or a selectivity index of >50 (TC₅₀/IC₅₀). Compound 3 and compound 12 were the most potent (IC₅₀ = 1.4 µM and IC₅₀ = 2.4 µM respectively). Although, compound 3 had some cytotoxicity, the selectivity index was 40. Compound 12 was not cytotoxic, with TC₅₀ >100 µM and a selectivity index of >42. None of the compounds were active against replicating M. tuberculosis in axenic medium (IC₉₀ >20 μM) demonstrating clear selectivity for intracellular bacteria.

We were able to determine some elements of the structure activity relationship (SAR) for this series using the limited compound set available. Compounds with a sulfur atom as a linker were generally more potent than those with an oxygen as a linker, with a range of 1.4–6.3 µM compared to 4.9–13 µM. The closest comparison was between 13 with the oxygen linker (IC₅₀ = 4.9 µM) and compounds 3 and 12 (IC₅₀ = 1.4 and 2.4 µM respectively). In addition, the two molecules with significant toxicity to macrophage cells contained the oxygen linker (4 band 7). Four compounds (1, 3, 5 and 12) with one or two substitution of halogen atoms on the amide phenyl ring (R₂) exhibited good intracellular activity.

It seems that lipophilicity did not correlate with anti-tubercular activity. Although the most active compounds (3 and 12) had cLogP of 3.97 and 4.55 respectively, compound 13 with cLogP of 3.32 was only two-fold less active and compounds with cLogP <3 still retained good activity (IC₅₀ <10 µM). An electron-withdrawing moiety on the para position of the amide phenyl group (R₂) appears to be an essential requirement for good anti-mycobacterial activity. The amide phenyl ring (R₂) is not critical to maintain antimycobacterial activity, since compounds 8 and 12 with saturated rings. Both compounds 8 and 10 with adamantane and piperidine groups respectively, had good activity (IC₅₀ = 7.6 µM and IC₅₀ = 6.6 µM). Further exploration of the length of the carbon chain or inclusion of other heteroatoms would expand the SAR, but such molecules are not available commerically and would require chemical synthesis.

We identified 6 compounds with a phenyl thiourea scaffold in the primary screen, of which 4 were active (Table 3). We purchased 5 commercially-available compounds and determined intracellular activity against M. tuberculosis. Four compounds had potent activity against intracellular M. tuberculosis (Table 5). One compound (18) was cytotoxic and we could not calculate activity. One compound (17) had significant cytotoxicity against macrophages (TC₅₀ of 1.8 µM) and was also toxic for HepG2 cells (TC₅₀ of 5.5 µM). The remaining three compounds had excellent potency (0.30–1.3 µM) and good selectivity. Compound 14 with a guanidine functionality as the core had higher cytotoxicity (SI = 14) than the two potent molecules (15 and 16) with the thiourea core (SI = >250 and >150 respectively).

NC–not calculated due to cytotoxicity. cLogP was calculated using Collaborative Drug Discovery (https://www.collaborativedrug.com/).

In contrast to the BAE series, the PTU molecules were also active against extracellular bacteria; 3 of the 5 molecules had IC₉₀ < 20 µM. In general, PTUs were 5–10-fold more active against intracellular bacteria. To the best of our knowledge, this is the first report of these compounds as antitubercular agents although other thiourea derivatives have antitubercular activity (Liav et al., 2008; Plutín et al., 2016; Ghorab et al., 2017; Doğan et al., 2020).

We identified 6 compounds with a phenyl pyrazole scaffold in the primary screen, of which 5 were active (Table 3). We purchased 4 commercially-available compounds and determined intracellular activity against M. tuberculosis. All four compounds (19–22) had activity against intracellular M. tuberculosis (Table 6). Although the compounds demonstrated some cytotoxicity against HepG2 cells and/or macrophages, the selectivity index was good (>5). These sets of compounds have 1,3,4-trisubstituted pyrazoles where C-4 bears a hydroxymethyl group. Compound 19 was the most potent with good intracellular activity (IC₅₀ = 2.9 µM) and >30-fold selectivity against HepG2 cells. Compound 21 also showed weak activity (IC₅₀ = 18.5 µM) against axenically-grown M. tuberculosis.

We identified 8 compounds with a thiophene carboxamide scaffold in the primary screen, of which all 8 were active (Table 3). We purchased 8 commercially-available compounds (23–30) and determined intracellular activity against M. tuberculosis. Five compounds had activity against intracellular M. tuberculosis with good acitivity of IC₅₀ < 10 µM (Table 7). Three compounds (23–25) were cyotoxic and we could not calculate activity.

Three compounds (26, 27 and 29) showed good intracellular potency with IC₅₀ < 5 μM. All except 26 and 28 were cytotoxic against macropahges with IC₅₀ ranging from 9 to 67 μM; compound 27 was the most toxic and also showed cytotoxicity against HepG2 cells (IC₅₀ = 4.7 μM). However, despite this, compounds 26, 28, 29, and 30 al l had good selectivity (SI > 10)—values for HepG2 were >80, 20, >60 and 15 respectively. Two compounds (24 and 25) were more cytotoxic against HepG2 cells in galactose media (ratio ∼4) suggesting that they may impair mitochondrial function. However, these were also toxic against macrophages and were not considered active. None of the compounds had any activity against extracellular M. tuberculosis.

We identified 8 compounds with a thiophene carboxamide scaffold in the primary screen, of which 2 were active (Table 3).

We purchased 7 commercially-available compounds (31–37) and determined intracellular activity against M. tuberculosis. Analogs varied in terms of substitution on the pyridine ring but all possessed the amino group at the C-3 position and a carboxamide substitution on the thiophene ring. Four compounds had activity against intracellular M. tuberculosis (Table 8), but only two had activity <10 μM (compounds 34 and 36). Compound 34 bearing a ketone moiety had weak activity. Two compounds (31 and 35) were not active. Compounds were selective, but had no activity against axenically cultured bacteria.