Identification of 4-Amino-Thieno[2,3-d]Pyrimidines as QcrB Inhibitors in Mycobacterium tuberculosis

The global tuberculosis (TB) epidemic has been exacerbated by the rise in drug-resistant TB cases worldwide. To tackle this crisis, it is necessary to identify new vulnerable drug targets in Mycobacterium tuberculosis, the causative agent of TB, and develop compounds that can inhibit the bacterium through novel mechanisms of action. The QcrB subunit of the electron transport chain enzyme cytochrome bc1 has recently been validated to be a potential drug target. In the current work, we report the discovery of a new class of QcrB inhibitors, 4-amino-thieno[2,3-d]pyrimidines, that potently inhibit M. tuberculosis growth in vitro. These compounds are chemically distinct from previously reported QcrB inhibitors, and therefore, 4-amino-thieno[2,3-d]pyrimidines represent a new scaffold that can be exploited to inhibit this drug target.

I nfection with Mycobacterium tuberculosis resulted in over 9 million new cases of tuberculosis (TB) and 1.5 million deaths in 2017, making it the most deadly infectious agent in the world (1). This epidemic is exacerbated by the rise of multidrug-resistant (MDR) TB cases that are resistant to at least the two frontline antibiotics used to treat TB, isoniazid and rifampin. MDR-TB constituted 3.6% of new TB cases in 2017 and 17% of previously treated TB cases, with rates of MDR-TB being as high as 50% among previously treated TB cases in some countries (1). Furthermore, 8.5% of MDR-TB cases in 2017 were estimated to be extensively drug resistant (XDR), which are also resistant to a fluoroquinolone and a second-line injectable drug (1). This rise in drug resistance and the scarcity of drugs in the pipeline have made it clear that we are not equipped to successfully battle the ongoing TB epidemic.
Here, we present the discovery of 4-amino-thieno [2,3-d]pyrimidines as a new series of QcrB inhibitors that potently inhibit M. tuberculosis growth and that are chemically distinct from previously identified QcrB inhibitors. This work adds to the growing number of QcrB inhibitors that have recently been identified and contributes to our understanding of ways to exploit this target in the development of new chemotherapeutic strategies for TB treatment.

Identification of 4-amino-thieno[2,3-d]pyrimidines with growth-inhibitory activity in Mycobacterium smegmatis and M. tuberculosis.
In an effort to identify novel inhibitors of mycobacteria, we screened a selection of 78 small-molecule nucleotide mimetics purchased from ChemBridge Corporation for compounds that inhibit the growth of M. smegmatis in a high-throughput liquid culture assay. From these screens, we identified a 4-amino-thieno [2,3-d]pyrimidine (CB37) that inhibited the growth of M. smegmatis ( Fig. 1A and B). We hypothesized that the charged carboxylate group on CB37 may greatly reduce penetration through the cell envelope of the mycobacteria and selected a set of 9 structurally related compounds that did not contain the carboxylate group but that contained the 2-ethyl-6-methylthieno[2,3-d]pyrimidine core (nearest neighbors) to purchase and assay for inhibition of M. smegmatis ( Fig. 1; see also Fig. S1 and entries 2 to 10 in Table S1 in the supplemental material). Eight of the compounds showed either similar levels of growth inhibition against M. smegmatis as CB37 or no growth inhibition at all (Fig. S1). However, one of these compounds, CB81, showed improved growth inhibition in M. smegmatis ( Fig. 1C and D). We resynthesized CB81 and henceforth designate it CWHM-728.
To determine if CWHM-728 also had activity against M. tuberculosis, we performed zone-of-inhibition assays with the wild-type (WT) M. tuberculosis Erdman strain by spreading approximately 2.5 ϫ 10 8 CFU of bacteria on an agar plate and spotting 5 l of a 100 mM stock of CWHM-728 dissolved in dimethyl sulfoxide (DMSO) onto a disk in the center of the plate. After incubation at 37°C for 10 days, the bacteria formed a lawn and a zone absent of bacterial growth indicated growth inhibition by the compound. DMSO had no effect on M. tuberculosis growth in this assay and did not generate a zone of clearing on its own, whereas incubation of M. tuberculosis with CWHM-728 resulted in growth inhibition ( Fig. 2A). To test if CWHM-728 has a bacteriostatic or a bactericidal effect on M. tuberculosis, we cultured M. tuberculosis in liquid medium in the presence of DMSO or 5 M or 25 M CWHM-728 and enumerated the viable CFU after 14 days of incubation (Fig. 2B). While exposure to 5 M CWHM-728 for 14 days caused a slight but not statistically significant decrease in the number of CFU compared to that for the DMSO-treated control, exposure to 25 M CWHM-728 caused a significant reduction in the number of CFU per milliliter compared to that for the DMSO-treated control. However, exposure to CWHM-728 did not decrease the number of viable CFU below the number of initial CFU on day 0, indicating that over 14 days, this concentration of CWHM-728 had a bacteriostatic effect on M. tuberculosis.
To identify the half-maximal inhibitory concentration (IC 50 ) of CWHM-728 against M. tuberculosis, we used a microplate alamarBlue assay (MABA), which is a highthroughput assay commonly used to evaluate the efficacy of antimycobacterial compounds (24). The MABA utilizes the redox-sensitive dye resazurin, which is blue in its oxidized form but becomes reduced to the pink fluorescent compound resorufin as a result of bacterial metabolism and respiration. Fluorescence can therefore be quantified as a readout for M. tuberculosis metabolism and respiration, which serves as a proxy for M. tuberculosis growth and/or survival. We incubated M. tuberculosis in the presence of increasing concentrations of CWHM-728 and found that CWHM-728 inhibited M. tuberculosis in the MABA with an IC 50 of 2.7 Ϯ 0.84 M (Fig. 2C).
To explore chemical modifications that would improve upon the growth-inhibitory activity of CWHM-728 and develop structure-activity relationships (SAR), we used the MABA to test if the analogs that we had previously obtained in which the n-butyl side chain was replaced with other aliphatic groups had inhibitory activity against M. tuberculosis (see entries 1 to 10 in Table S1). Analogs that had side chains with ionizable groups (entries 1 and 3) and polar groups, such as hydroxyl (entry 4), had poor activity, with IC 50 values being greater than 25 M. Capping the hydroxyl group as methoxy (entries 5 and 6) restored some potency, and isopropoxy (entry 7) yielded a compound 8-fold more potent than the compound with a butyl side chain (entry 2). Cyclopropyl was tolerated (entry 8), with a submicromolar IC 50 , while slightly bulkier cyclopentyl and piperidine side chains resulted in very poor potency (entries 9 and 10).
To further extend the SAR, we synthesized analogs of CWHM-728 containing lipophilic side chains, and the resulting analogs were evaluated in the MABA (Table S2). Replacing the n-butyl side chain with t-butyl caused a nearly 10-fold decrease in potency (entry 2), but trifluoro-n-butyl and isopentyl variants (entries 3 and 4) were found to be greater than 10-fold more potent than the original n-butyl (entry 1). Capping the NH with a methyl group (entry 5) demonstrated little effect on potency. Finally, a series of phenyl and alkylphenyl side chains was prepared and showed remarkable SAR, with potency being strongly dependent on chain length (entries 6 to 9).
The most potent compound from our SAR evaluation was CWHM-1023, which shared the 4-amino-thieno [2,3-d]pyrimidine core scaffold with CWHM-728 but which contained a 3-phenylpropyl side chain (Fig. 3A). We found that CWHM-1023 inhibited M. tuberculosis growth in the zone-of-inhibition assay (Fig. 3B) and inhibited M. tuberculosis in the MABA with an IC 50 of 83 Ϯ 5.4 nM (Fig. 3C). This is a 38-fold improvement in potency compared to that of CWHM-728, which had an IC 50 of 3.2 Ϯ 0.13 M in this

Mutations in qcrB confer resistance to 4-amino-thieno[2,3-d]pyrimidines.
To identify the target of the 4-amino-thieno[2,3-d]pyrimidines, we selected for M. tuberculosis mutants that were resistant to CWHM-728 by plating approximately 2.5 ϫ 10 8 CFU of WT M. tuberculosis Erdman on 7H11 agar medium containing 10 M CWHM-728 and incubating the culture at 37°C for 12 weeks. We found that spontaneous CWHM-728-resistant mutant colonies emerged at an approximate frequency of 1 in 2.3 ϫ 10 7 . To identify the genetic basis for CWHM-728 resistance, we performed whole-genome sequencing on four CWHM-728-resistant strains and found that all four strains harbored missense mutations in the qcrB gene. Three of the strains, isolated from a single culture, had mutations resulting in an A178T amino acid change, and one strain, isolated from a second independent culture, had a mutation generating a V338G substitution (Table 1). We then sequenced the qcrB locus in 8 additional CWHM-728-resistant mutants isolated from a third independent culture and identified a missense mutation in qcrB in every isolate (Table 1). One strain harbored the A178T mutation, and 3 more had the V338G mutation that we had isolated previously. In addition, we found that the remaining isolates harbored a A178V, G175S, or G315S mutation in qcrB.
The qcrB gene is located within an operon containing ctaE, qcrC, and qcrA. The qcrCAB gene cassette encodes all 3 subunits of the cytochrome bc 1 complex (25). This complex is localized to the plasma membrane and forms a supercomplex with the aa 3 -type cytochrome oxidase encoded by ctaBCDE (26,27). The resulting cytochrome bc 1 :aa 3 oxidase complex catalyzes the terminal electron transfer reaction in the myco- bacterial ETC, transferring electrons from menaquinol, the lipid electron carrier in the membrane, to the terminal electron acceptor oxygen. We sought to confirm that mutations in qcrB were sufficient to confer resistance to 4-amino-thieno [2,3d]pyrimidines to rule out the possibility that the CWHM-728-resistant mutants harbored mutations elsewhere in the genome that contributed to resistance. For these studies, we engineered M. tuberculosis strains that expressed either WT qcrCAB (qcrCAB WT ) or qcrCAB with the A178T substitution (qcrCAB A178T ) from the chromosomal attB site and deleted the endogenous qcrCAB locus. We then monitored the activity of CWHM-728 (Fig. 4A) and CWHM-1023 (Fig. 4B) against these strains using the MABA. We found that the strain expressing qcrCAB A178T exhibited reduced sensitivity to both compounds compared to that of WT M. tuberculosis or the isogenic qcrCAB WT control strain, confirming that this single amino acid change in qcrB confers resistance to 4-aminothieno [2,3-d]pyrimidines. These data are consistent with a model in which the 4-aminothieno[2,3-d]pyrimidines target cytochrome bc 1 .

Culture no. and resistant strain Mutation in qcrB
a The CWHM-728-resistant strains isolated in this study are listed with the mutation identified in qcrB. b The mutations were identified by whole-genome sequencing. All other mutations were identified by Sanger sequencing of the qcrB locus. Based on our finding that mutations in the qcrB gene confer resistance to 4-aminothieno [2,3-d]pyrimidines, we propose that this new class of antimycobacterial compounds targets QcrB. To better understand how mutations in qcrB confer resistance to 4-amino-thieno[2,3-d]pyrimidines, we used computational modeling to predict the structure of the M. tuberculosis QcrB protein based on a previously published structure for QcrB from Rhodobacter sphaeroides, which shares 16% amino acid identity with the M. tuberculosis QcrB (28) (Fig. 5A). We docked CWHM-728 onto the predicted structure of QcrB and found that it localized near the putative menaquinol binding site of QcrB. Based on our computational model, we speculate that the A178T, A178V, or G175S mutation in QcrB would disrupt the contacts required for 4-amino-thieno [2,3d]pyrimidines to bind QcrB (Fig. 5B). The V338 and G315 residues are not located where they would directly interact with CWHM-728 but are positioned on neighboring helices within a very tightly packed region, such that changes in amino acid identity could possibly affect the position of those helices and disrupt the 4-amino-thieno[2,3d]pyrimidine binding site.
The predicted binding site for the 4-amino-thieno [2,3-d]pyrimidines is similar to what has previously been reported from similar modeling studies for the QcrB inhibitor Q203 (29). We investigated whether the 4-amino-thieno[2,3-d]pyrimidine-resistant strain expressing qcrCAB A178T was more resistant to Q203 than the isogenic qcrCAB WT control strain and found that the qcrB A178T mutation did not affect Q203 sensitivity in the MABA (Fig. S2). These data suggest that while the qcrB A178T allele is sufficient to confer resistance to 4-amino-thieno[2,3-d]pyrimidines, it does not confer crossresistance to Q203, possibly because the A187T mutation is not sufficient to disrupt the binding of Q203 to QcrB.

4-Amino-thieno[2,3-d]pyrimidines deplete ATP levels in M. smegmatis and M. tuberculosis.
Based on our model that 4-amino-thieno[2,3-d]pyrimidines target the QcrB subunit of cytochrome bc 1 , we hypothesized that CWHM-728 and CWHM-1023 inhibit the mycobacterial ETC. To test this hypothesis, we incubated M. tuberculosis in the presence of 1 M CWHM-728, 1 M CWHM-1023, or 400 nM Q203 for 24 h and measured the ATP levels in the bacteria using the BacTiter-Glo assay (Promega). We found that treatment with CWHM-728, CWHM-1023, and Q203 decreased the ATP levels in M. tuberculosis by 75%, 60%, and 58%, respectively (Fig. 6A). Similar results were observed in M. smegmatis, where 10 M CWHM-1023 caused a 41% decrease in ATP levels in M. smegmatis (Fig. 6B). These data support the possibility that 4-aminothieno [2,3-d]pyrimidines target a complex involved in energy generation, consistent with our hypothesis that these compounds inhibit QcrB in the mycobacterial ETC.
M. smegmatis and M. tuberculosis lacking cytochrome bd are hypersensitive to 4-amino-thieno[2,3-d]pyrimidines. Mycobacteria have a branched ETC that can ter- New Inhibitors of Mycobacterium tuberculosis QcrB minate either in the cytochrome bc 1 :aa 3 terminal oxidase or in the cytochrome bd terminal oxidase, encoded by cydAB (25,30). Cytochrome bc 1 :aa 3 oxidase and cytochrome bd oxidase have somewhat overlapping roles in the ETC to transfer electrons from menaquinol to oxygen, where the cytochrome bd oxidase can partially compensate for the loss of cytochrome bc 1 :aa 3 oxidase activity, as evidenced by the observation that mutants of cytochrome bd exhibit increased sensitivity to cytochrome bc 1 inhibitors (8,31). Therefore, we hypothesized that 4-amino-thieno[2,3-d]pyrimidines would exhibit enhanced activity against mycobacteria lacking cytochrome bd compared with that of the WT strains. To test this hypothesis, we generated M. tuberculosis ΔcydA and M. smegmatis ΔcydA mutants, which lack cytochrome bd, and measured the sensitivity of these mutant strains to 4-amino-thieno[2,3-d]pyrimidines compared to that of the WT strains. We found that deletion of cydA increased the sensitivity of M. tuberculosis to CWHM-728 13.8-fold (Fig. 7A), CWHM-1023 22.6-fold (Fig. 7B), and Q203 7.7-fold (Fig. 7C), as measured by a decrease in the MABA IC 50 compared to that for WT M. tuberculosis. Additionally, we found that CWHM-1023 had a 4.6-fold lower IC 50 against M. smegmatis ΔcydA than against WT M. smegmatis, which was partially complemented by expressing cydAB in this strain (Fig. 7D). These data demonstrate that genetic deletion of cytochrome bd sensitizes mycobacteria to inhibition by 4-aminothieno[2,3-d]pyrimidines, which is consistent with our hypothesis that these compounds target the QcrB subunit of cytochrome bc 1 .

DISCUSSION
There is a dire need for new anti-TB drugs that shorten treatment regimens and that are effective against MDR and XDR strains of M. tuberculosis. To begin to address this need, two anti-TB agents have recently received accelerated regulatory approval from the U.S. Food and Drug Administration (FDA) (3). These two drugs, bedaquiline (Sirturo) and delamanid, are the first antibiotics to be approved for the treatment of TB in the last 50 years (32). The discovery and development of bedaquiline in particular have raised general interest in targeting M. tuberculosis ETC and respiration as a therapeutic strategy.
Simultaneously, numerous phenotypic screens have identified QcrB inhibitors that inhibit the growth of M. tuberculosis in axenic culture and/or in macrophages (4, 8, 15,  16, 19, 21, 23, 33), raising further interest in the ETC as a drug target. The most clinically advanced of these QcrB inhibitors is the imidazopyridine amide (IPA) Q203 (telacebec), which is currently in clinical trials for the treatment of TB (5). In this study, we report the discovery of a novel class of QcrB inhibitors, 4-amino-thieno [2,3-d]pyrimidines, that are chemically distinct from the previously identified QcrB inhibitors. Therefore, the new chemical scaffold described here adds to the growing chemical space that can be exploited to target the mycobacterial cytochrome bc 1 complex. The most potent 4-amino-thieno [2,3-d]pyrimidine inhibitor of QcrB that we identified, CWHM-1023, had an IC 50 of approximately 83 nM for M. tuberculosis in the MABA assay (see Table S2 in the supplemental material). This is in comparison with the IC 50 s for the known respiration inhibitors Q203, bedaquiline, and thioridazine, which are 1.5 nM, Ͻ78 nM, and 11.2 M, respectively, against M. tuberculosis in the MABA (Table S2).
Mycobacteria have a branched ETC, whereby electrons can be shuttled either to cytochrome bc 1 or to cytochrome bd to be transferred to the terminal electron acceptor, O 2 (30). Cytochrome bc 1 is proposed to be bioenergetically more efficient than cytochrome bd under standard aerobic conditions (25). However, cytochrome bd is induced under conditions of low oxygen and contributes to mycobacterial fitness under microaerobic conditions, suggesting that cytochrome bd can support mycobac- terial survival under certain conditions (30). As such, while cytochrome bc 1 is important for the growth of M. tuberculosis under laboratory conditions (25,(34)(35)(36), M. tuberculosis is able to survive during inhibition of cytochrome bc 1 (37,38), which is consistent with our data (Fig. 2B) and previously reported data indicating that QcrB inhibitors lack early bactericidal activity and are bacteriostatic for several weeks (19,20,31). Furthermore, it was reported that treatment of an M. tuberculosis mutant lacking cytochrome bd with QcrB inhibitors results in bacterial death (19,31). Conversely, it was found that overexpression of cytochrome bd in M. tuberculosis can enable the bacteria to grow, despite cytochrome bc 1 inhibition (8,39). Together, these data indicate that cytochrome bd can compensate for cytochrome bc 1 inhibition.
Expression of cytochrome bc 1 and cytochrome bd has also been shown to change throughout the course of infection. In mice infected with M. tuberculosis, it was found that qcrC expression was the highest during acute infection (day 15) and that following 20 days of infection qcrC was transcriptionally downregulated 2-to 2.5-fold. In contrast, cydA was transcriptionally upregulated after 20 days, with the highest level of expression being detected at 30 days postinfection, when there was a 7-fold increase over the day 15 expression levels (40). These findings suggest that M. tuberculosis may rely more heavily on cytochrome bc 1 during acute stages of infection, whereas cytochrome bd is activated at later stages of infection. These expression patterns may explain why it has been found that QcrB inhibitors work well to halt M. tuberculosis growth in the mouse when administered during the acute phase of infection (4,12,16,19) but have a variable impact on the bacterial burden when treatment starts after the mice have been infected for more than 2 weeks, with most studies reporting modest decreases in the bacterial burden, if any decrease at all (4,12,19,31). Some of the variability of QcrB inhibitor efficacy during the later stages of infection likely also depends on the timing of the treatment and the dosing of the compound.
Given the respiratory flexibilty in M. tuberculosis, interest has increased in using combinations of antibiotics that target different components of the ETC, a strategy that has been shown to lead to synergy in bactericidal effects (38,41,42). In particular, our studies with 4-amino-thieno [2,3-d]pyrimidines in the ΔcydA mutants imply that inhibition of QcrB would have an enhanced impact on M. tuberculosis in combination with inhibitors of cytochrome bd, as previously observed for Q203 (42). Together, these findings indicate that 4-amino-thieno [2,3-d]pyrimidines target QcrB and highlight the idea that the concurrent inhibition of both cytochrome bc 1 and cytochrome bd is more effective than inhibition of cytochrome bc 1 alone. In addition to potentiating other respiration inhibitors, a QcrB inhibitor has recently been shown to enhance pyrazinamide as well as rifampin monotherapy in mice, indicating that QcrB inhibitors may prove to be a useful addition to the current standard of care (12).

MATERIALS AND METHODS
Bacterial strains and growth conditions. M. tuberculosis Erdman was cultured in Middlebrook 7H9 liquid medium supplemented with 60 l/liter oleic acid, 5 g/liter bovine serum albumin, 2 g/liter dextrose, and 0.003 g/liter catalase (OADC) plus 0.5% glycerol and 0.05% Tween 80. For solid medium, Middlebrook 7H10 or 7H11 agar medium supplemented with OADC and 0.5% glycerol was used. M. smegmatis mc 2 155 was cultured in LB medium supplemented with 0.5% glycerol, 0.5% dextrose, and 0.05% Tween 80. Genetic deletion mutant strains of M. tuberculosis and M. smegmatis (Table 2) were generated using specialized transduction with the conditionally replicating phage phAE87 as previously described (43,44). When appropriate, mycobacterial strains were selected on 20 g/ml kanamycin and/or 50 g/ml hygromycin.
Initial screen. Logarithmically growing M. smegmatis strain csm208 was inoculated into 96-well dishes containing 100 M compounds in 200 l LB at a starting optical density at 600 nm (OD 600 ) of 0.2 and incubated with shaking at 37°C in a Tecan M200 Pro plate reader, with OD 600 measurements being taken in each well every 15 s.
Microplate alamarBlue assays (MABAs). Logarithmically growing M. tuberculosis was inoculated into 7H9 medium in 96-well plates with wells containing increasing concentrations of compound. M. tuberculosis was inoculated at an OD 600 of 0.0008, corresponding to approximately 4 ϫ 10 5 CFU/ml in 200 l per well. The plates were incubated at 37°C in 5% CO 2 for 1 week, at which point 32.5 l of a mixture containing an 8:5 ratio of 0.6 mM resazurin (Sigma) dissolved in 1ϫ phosphate-buffered saline to 20% Tween 80 was added, and the production of fluorescent resorufin was measured after incubation at 37°C in 5% CO 2  Selection and sequencing of resistant mutants. Resistant mutants were selected on 7H11 agar medium containing 10 M CWHM-728. Approximately 2.5 ϫ 10 8 CFU of WT M. tuberculosis Erdman was spread per plate, and the plate was incubated at 37°C for 12 weeks. Genomic DNA was isolated using cetyltrimethylammonium bromide-lysozyme lysis, followed by chloroform-isoamyl alcohol extraction and isopropanol precipitation, as previously described (45). Whole-genome sequencing was performed by use of an Illumina HiSeq sequencer with 50-bp single-end reads. The identification of single nucleotide polymorphisms (SNPs) was done using SeqMan NGen software (DNASTAR). The genomes were assembled and compared to the genomic DNA from the WT parental control strain to identify SNPs that may be responsible for resistance to CWHM-728. The mutations identified in qcrB by whole-genome sequencing were subsequently confirmed by Sanger sequencing. For Sanger sequencing of the qcrB locus, the genomic region was PCR amplified using primers erdqcrBfwxbaI430 (GTCTAGAATGAGTCCGA AACTGAGTCCGCC) and RK27 (GAAGCTTTCGCCGGGCTAGTGCTCGCCGTC) and then sequenced using the same two primers (GeneWiz).
Computational modeling. A homology model of M. tuberculosis QcrB was built using the crystal structure of Rhodobacter sphaeroides QcrB (PDB accession number 2QJP) (28) as the template structure, using the sequence alignment from Ko and Choi (29). The model, which was built in the Molecular Operating Environment (MOE 2016.08; Chemical Computing Group, Montreal, Quebec, Canada), was placed into a membrane environment (46), and energy was minimized. The geometry of the [2Fe-2S] center and the disulfide bond pattern adjacent to the inhibitor-binding site were assumed to match those of the template structure. CWHM-728 was docked into the homology model using the GOLD (v5.5) program (47).
ATP measurements. M. tuberculosis or M. smegmatis was inoculated into 7H9 medium with or without the test compounds at an OD 600 of 0.1 and incubated with shaking at 37°C for 24 h (M. tuberculosis) or 12 h (M. smegmatis). An aliquot of the culture was heat inactivated at 95°C for 20 min and diluted 1:100. Diluted samples were mixed with the BacTiter-Glo assay (Promega) reagent at a 1:1 ratio, and luminescence was quantified on a Tecan M200 Pro plate reader (integration ϭ 1 s). Relative luminescence units (RLU) were normalized to the number of log 10 CFU in the sample to account for differences in bacterial number.
Compounds  CWHM-1023 to confirm the purity and identity of the synthesized compound ( Fig. S3 and S4). Q203 was acquired from Enamine (catalog number EN-300-218150), and both bedaquiline (catalog number 465749185) and thioridazine (catalog number 1662504) were purchased from Sigma-Aldrich; all three compounds were tested in the MABA for comparative purposes. Data availability. The sequence data are publicly available in the NCBI Sequence Read Archive under accession no. PRJNA561987.