Questiomycin A demonstrates antibacterial activity against Mycobacterium tuberculosis by directly targeting FabD

Abstract

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains a major global health threat, particularly due to the rise of multidrug-resistant TB (MDR-TB). Current treatment limitations highlight the urgent need for new therapeutics to improve efficacy and shorten therapy duration. Through high-throughput screening, we identified Questiomycin A (QM), a phenoxazinone compound with promising anti-tuberculosis activity against drug-sensitive and drug-resistant Mtb strains. Transcriptomic profiling indicated disruption of cell wall-associated pathways, while thermal proteome profiling showed that QM binding substantially increased the thermal stability of FabD (malonyl CoA-acyl carrier protein transacylase, MCAT), a key enzyme in mycolic acid biosynthesis. Surface plasmon resonance, enzymatic assays, and genetic manipulation confirmed QM directly binds to and inhibits FabD. This inhibition damages cell wall integrity, increases permeability, disturbs intracellular pH homeostasis, and collapses the proton motive force, ultimately leading to bacterial death. Pharmacokinetic evaluation in six-week-old male Balb/c mice following a single oral dose demonstrates rapid absorption but limited drug exposure. Together, these findings reveal that QM targets FabD to compromise cell wall homeostasis in Mtb, providing a foundation for developing novel anti-tubercular agents targeting FabD.

Antibacterial activity of Questiomycin A against Mycobacterium tuberculosis via direct targeting of FabD, effective against drug-sensitive and drug resistant strains.

Introduction

Tuberculosis (TB) is a serious chronic infectious disease caused by Mycobacterium tuberculosis (Mtb), which continues to pose a significant threat to global public health. According to the Global Tuberculosis Report 2024, 10.8 million new TB cases were reported globally in 2023, with approximately 400,000 cases of multidrug-resistant (MDR) and rifampicin-resistant (RR) TB¹. The number of deaths reached 1.25 million, reaffirming TB as the world’s leading infectious disease killer. While drug therapy remains the cornerstone of TB control, current treatment regimens are lengthy and suffer from poor patient adherence. For instance, even the newly recommended treatment for drug-sensitive TB requires the use of four drugs (isoniazid, rifapentine, moxifloxacin and pyrazinamide) over a four-month period². Furthermore, the availability of the 6-month BPaL regimen (bedaquiline, pretomanid and linezolid) remains extremely limited globally. In 2023, the treatment success rate for MDR/RR-TB was only 68%¹. Consequently, there is an urgent need for anti-tuberculosis drugs with distinct targets and mechanisms to shorten treatment duration and simplify therapy.

Natural products have historically been a rich source of anti-infective agents, offering diverse chemical scaffolds and biological activities^3–5. In this study, we screened the L6500 microbial natural product library (685 compounds, TargetMol) and identified Questiomycin A (QM), chemically named 2-amino-phenoxazine-3-one (Phx-3 or APO), as a compound with potent activity against Mtb. Although phenoxazinone compounds are known for their antibacterial, antifungal, and antiviral properties^6–11, little is known about their precise mechanism of action against mycobacteria^12–14, and their impact on Mtb, particularly in terms of specific molecular targets, has not been elucidated.

Here, we aimed to elucidate the anti-tuberculosis activity of QM against both drug-sensitive and drug-resistant Mtb. Integrated transcriptomic and phenotypic analyses revealed that QM compromises cell wall integrity, increases cell wall permeability, and leads to intracellular pH imbalance and collapse of the proton motive force (PMF), ultimately resulting in bactericidal effects. Further thermal proteome profiling (TPP) and target validation identified FabD as a direct molecular target of QM.

FabD (malonyl CoA-acyl carrier protein transacylase, MCAT) is a conserved enzyme of the bacterial type II fatty acid synthase (FAS II) system. Unlike the MAT domain embedded in the multifunctional mammalian FAS I complex, bacterial FabD functions as an independent monomer^15,16. Biochemical and structural studies in representative bacteria such as Escherichia coli^17,18 and Pseudomonas aeruginosa¹⁹ have demonstrated its essential role in fatty acid biosynthesis and highlighted its druggability. In Mycobacterium tuberculosis, FabD has been structurally characterized, revealing catalytic features distinct from canonical bacterial enzymes²⁰. However, selective small-molecule inhibitors of FabD are still lacking, leaving its potential as a therapeutic target largely underexplored. Our findings highlight FabD as a therapeutic target and support the potential of QM as a lead compound for antituberculosis drug development.

Results

In vitro anti-tuberculosis activity of QM

Through preliminary screening of a microbial natural product library (685 compounds), we identified the phenoxazine derivative Questiomycin A (QM, structure shown in Fig. 1a) as a promising anti-tubercular lead compound for further investigation. The minimum inhibitory concentrations (MICs) of QM against the Mtb H37Rv strain, 30 clinical drug-resistant isolates, and 10 clinical drug-susceptible isolates were determined using the microplate Alamar Blue assay²¹, with MIC₅₀ values ranging from 0.042 to 0.113 μg/mL and MIC₉₀ values ranging from 0.125 to 0.5 μg/mL. As shown in Table 1, no cross-resistance was observed with existing anti-TB drugs, indicating a potentially distinct mechanism of action.

Fig. 1. In vitro anti-tuberculosis activity of QM.

a Chemical structure of QM. MW: 212.2. b Time-kill curves of QM and RFP (positive control) against Mtb H37Rv. The bacterial cultures were treated with varied drug concentrations for 14 days and plated on 7H10 agar plates for CFU/mL enumeration at the indicated time points. Data represent mean ± SD (Log₁₀ CFU/mL) from three independent experiments. The dotted horizontal line represents the CFU detection limit. c Time-kill curves of QM clinical drug-resistant strains (13946-Pre-XDR, 16833-Pre -XDR and 16995-Pre-XDR). Data represent mean ± SD (Log₁₀ CFU/mL) from three independent experiments. The dotted horizontal line represents the CFU detection limit.

Table 1.

The minimum inhibitory concentrations (MICs) of QM and anti-tuberculosis drugs against Mtb H37Rv, 30 clinical drug-resistant isolates, and 10 clinical drug-sensitive isolates

	MIC (μg/mL)
Strain	QM	RFP	INH	EMB	LVX	MFX	BDQ	LZD	CFZ
H37Rv	0.25	0.02	0.04	1.25	0.31	0.16	0.04	0.31	0.16
13529-PDRa	0.25	0.02	2.5	5	1.25	2.5	0.08	0.31	0.16
28056-PDRa	0.25	0.04	>10	0.62	2.5	0.62	0.16	0.16	0.62
12585-MDRb	0.25	>10	>10	1.25	0.31	0.16	0.16	0.16	0.16
12657-MDRb	0.13	0.16	>10	10	0.31	0.16	0.31	0.31	1.25
12897-MDRb	0.25	10	5	2.5	0.31	0.16	0.02	0.16	0.16
14862-MDRb	0.25	>10	>10	>10	0.62	0.16	0.02	0.31	0.08
15171-MDRb	0.25	0.31	>10	>10	0.16	0.62	0.62	0.31	0.16
28350-MDRb	0.25	>10	>10	1.25	0.31	0.16	1.25	0.31	0.31
29118-MDRb	0.25	>10	>10	1.25	0.31	0.16	0.04	0.31	0.08
29499-MDRb	0.25	>10	10	1.25	0.31	0.16	0.08	0.31	0.16
30141-MDRb	0.25	>10	10	>10	1.25	0.31	0.04	0.16	0.16
30180-MDRb	0.25	>10	>10	>10	0.31	0.16	0.04	0.31	0.16
30450-MDRb	0.25	>10	>10	10	0.31	0.16	0.04	0.31	0.08
30459-MDRb	0.25	>10	>10	1.25	0.31	0.16	0.08	0.62	0.62
30611-MDRb	0.13	10	>10	1.25	0.16	0.08	0.02	0.16	0.08
30819-MDRb	0.25	>10	10	5	1.25	0.31	0.04	0.16	0.16
13385-Pre-XDRc	0.13	0.62	>10	>10	5	5	0.16	0.62	0.16
13946-Pre-XDRc	0.25	>10	2.5	10	>10	>10	0.08	0.31	0.08
16833-Pre-XDRc	0.13	>10	>10	>10	>10	>10	0.04	0.31	0.16
16995-Pre-XDRc	0.13	>10	>10	>10	2.5	0.62	0.04	0.31	0.16
29420-Pre-XDRc	0.25	>10	2.5	1.25	5	2.5	0.04	0.16	0.16
29704-Pre-XDRc	0.13	>10	>10	2.5	5	5	0.04	0.16	0.16
30554-Pre-XDRc	0.25	>10	>10	10	10	5	0.04	0.31	0.08
30797-Pre-XDRc	0.25	>10	>10	10	5	2.5	0.04	0.16	0.62
11492-XDRd	0.25	0.62	>10	>10	5	5	0.31	0.16	1.25
14470-XDRd	0.25	>10	>10	1.25	10	2.5	0.31	0.31	0.16
17080-XDRd	0.25	>10	>10	5	0.31	2.5	0.04	10	0.08
17851-XDRd	0.13	>10	2.5	1.25	2.5	1.25	0.31	0.31	0.31
29223-XDRd	0.25	>10	>10	>10	5	2.5	0.62	0.16	0.62
29928-XDRd	0.50	>10	5	10	>10	5	0.62	0.31	1.25
11264-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.31	0.08
16835-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.31	0.08
30037-STB	0.25	0.02	0.04	1.25	0.31	0.16	0.04	0.31	0.16
30044-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.31	0.08
30091-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.16	0.08
30093-STB	0.25	0.02	0.04	1.25	0.31	0.16	0.04	0.31	0.16
30204-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.31	0.08
30205-STB	0.13	0.01	0.04	0.63	0.31	0.16	0.04	0.31	0.08
30306-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.02	0.31	0.08
30314-STB	0.25	0.02	0.08	1.25	0.31	0.16	0.04	0.31	0.08

^aPoly-resistant M. tuberculosis.

^bMultidrug-resistant M. tuberculosis.

^cPre-extensively drug-resistant M. tuberculosis.

^dExtensively drug-resistant M. tuberculosis.

STB, clinically sensitive strain of M. tuberculosis.

QM questiomycin A, First-line drugs: RFP Rifampin, INH isoniazid, EMB ethambutol; Group A drugs: LVX levofloxacin, MFX moxifloxacin, BDQ bedaquiline, LZD linezolid; Group B drug: CFZ clofazimine.

The frequency of spontaneous resistance to QM was further evaluated. No resistant colonies were recovered when wild-type Mtb H37Rv was plated on 7H10 agar containing 4× or 8× MIC of QM, corresponding to a spontaneous mutation frequency of <10⁻⁹. Under the experimental conditions employed, we were unable to isolate QM-resistant mutants, indicating that QM exhibits a low propensity to select for spontaneous resistance.

The colony-forming unit (CFU) enumeration method²² was used to determine the minimum bactericidal concentrations (MBCs) of QM against Mtb H37Rv and three clinical isolates. The MBC of QM was found to be 2–4 μg/mL. Time-kill curves²³ revealed that the bactericidal activity of QM was both time- and concentration-dependent. Treatment with 8× or 16× MIC resulted in a >3-Log reduction in CFU by day 7, while treatment with 2× or 4× MIC achieved a ~3-Log reduction by day 14 (Fig. 1b, c).

The post-antibiotic effect (PAE) is defined as the sustained inhibition of bacterial growth following brief exposure to an antimicrobial agent, even after its removal. The results demonstrated that QM exhibited PAE against Mtb H37Rv. After 2 h of drug exposure, bacterial suspensions were diluted 1000-fold in fresh medium to remove the compound, following standard PAE procedures^24,25. The duration of PAE increased in a concentration-dependent manner (Table 2). To evaluate the potential contribution of residual drug after dilution, we conducted additional experiments using sub-MIC concentrations (QM 1/5× MIC, 50 ng/mL; rifampin 1/4× MIC, 5 ng/mL). At these low concentrations, bacterial growth was delayed only minimally (QM: 6.00 ± 0.31 h; RFP: 9.84 ± 0.58 h), which is substantially shorter than the prolonged PAE observed following high-dose exposure (QM up to 54.92 h; RFP > 98 h). These results indicate that residual QM after dilution may contribute marginally but cannot account for the extended PAE, confirming that the observed effects reflect genuine post-antibiotic activity rather than residual drug effects.

Table 2.

Post-antibiotic effect (PAE) of RFP and QM against Mtb H37Rv after 2 h of exposure

RFP			QM
0.05 μg/mL	0.5 μg/mL	5 μg/mL	0.5 μg/mL	5 μg/mL	50 μg/mL
38.89 ± 0.80 h	84.01 ± 0.24 h	>98.29 h	6.33 ± 0.73 h	31.45 ± 0.12 h	54.92 ± 0.82 h

Rifampin (RFP) was used as a positive control.

Anti-tuberculosis activity of QM under different physiological conditions

To evaluate the bactericidal activity of QM under different physiological conditions, Mtb H37Rv was cultured at pH 5, 6, or 7 and treated with QM at concentrations of 8× MIC, 1× MIC, or 1/8× MIC, or with rifampicin (RFP, 1× MIC, positive control). CFU counts were determined on days 0, 1, 7, and 21. Both QM and RFP exhibited bactericidal activity at all pH values tested (Fig. 2a).

Fig. 2. Anti-tuberculosis activity of QM under different physiological conditions.

a QM’s bactericidal effect on Mtb under different pH conditions. Mtb was treated with QM at 8 × MIC, 1 × MIC, or 1/8 × MIC, or with rifampicin (RFP, 1 × MIC, positive control) at pH 5, 6, and 7. CFU counts were measured on days 0, 1, 7, and 21. The dotted horizontal line represents the CFU detection limit. b Intracellular activity of QM against Mtb H37Rv in J774A.1 macrophage. J774A.1 cells were infected at MOI 5 for 4 h, then treated with QM or RFP (0.5, 2 and 5 μg/mL, RFP as positive control) for 72 h. Intracellular bacterial burden was quantified by CFU enumeration. Data represent mean ± SD (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001; n.s. = not significant. The dotted horizontal line represents the CFU detection limit. c Antibacterial activity of QM against nutrient-starved Mtb model (NR-Mtb). BDQ and RFP were included as positive controls, and INH as a negative control. CFU counts were determined after treatment with QM, BDQ, RFP, or INH. Data represent mean ± SD (n = 4). ****P < 0.0001; n.s. = not significant. The dotted horizontal line represents the CFU detection limit.

J774A.1 cells were selected as they are a widely used macrophage model for studying the intracellular survival of Mtb²³. Cytotoxicity assays conducted after 24 hours of QM exposure revealed half-maximal cytotoxic concentration (CC₅₀) values of 27.63 μg/mL for J774A.1 murine macrophages. Infected macrophages (MOI = 5, 4 h infection) were treated with QM or RFP at 0.5, 2, and 5 μg/mL for 72 h. QM reduced intracellular bacterial load in a dose-dependent manner (ΔLog₁₀ CFU = 0.19, 0.56, and 0.84, respectively), while RFP (positive control) produced comparable reductions (ΔLog₁₀ CFU = 0.22, 0.63 and 0.93, respectively). These results indicate that QM retains inhibitory activity against intracellular Mtb, comparable to that of RFP under the same conditions (Fig. 2b).

Given the importance of non-replicating Mtb in persistent infection and treatment failure, we assessed the activity of QM against nutrient-starved, non-replicating Mtb. In this experiment, isoniazid (INH) was included as a negative control²⁶, as it lacks activity against non-replicating Mtb, while bedaquiline (BDQ) and RFP served as positive controls. Treatment with QM at 5, 10, and 20 μg/mL reduced CFU counts by 0.80, 1.40, and 2.20 ΔLog₁₀ CFU, respectively, which was comparable to BDQ at the same concentrations (0.53, 1.69, and 2.40 ΔLog₁₀ CFU). As expected, INH showed no activity in this model. In contrast, RFP exhibited pronounced bactericidal activity, with CFU reductions of 1.80, 2.64, and 3.20 ΔLog₁₀ CFU at 5, 10, and 20 μg/mL, respectively (Fig. 2c).

In vivo pharmacokinetics of QM

The pharmacokinetic properties of QM were subsequently evaluated in Balb/c mice following a single oral administration at 10 mg/kg. QM was rapidly absorbed, reaching a maximum plasma concentration (C_max) of 43.89 ng/mL at a T_max of 0.5 h. The compound exhibited a relatively short plasma elimination half-life (t_1/2 = 1.12 h) and low systemic exposure, with an area under the plasma concentration–time curve from 0 to 24 h (AUC_0-24h) of 84.15 ng·h/mL (Table 3).

Table 3.

Pharmacokinetic parameters of QM following oral administration in mice

Parameters	Units	QM
Dose	mg/kg	10
t1/2 a	h	1.12
Tmax	h	0.5
Cmax	ng/mL	43.89
AUC(0–24h) b	ng⋅h/mL	84.15

^aPlasma elimination half-life.

^bPlasma exposure.

Transcriptomic profiling reveals potential mechanisms of action of QM

QM exhibits potent anti-tuberculosis activity. To further elucidate the underlying mechanism, we utilized transcriptomic analysis to investigate its mode of action. Comparison of the transcriptomic data from the QM (4 × MIC, 1 μg/mL, 6 h) group with the DMSO control revealed 74 Differentially expressed genes (DEGs). KEGG pathway enrichment analysis revealed that DEGs were distributed across both lipid metabolism and energy metabolism pathways; however, notable differences were observed in their enrichment patterns. Energy metabolism encompasses multiple sub-pathways, including oxidative phosphorylation and glycolysis/gluconeogenesis. Although individual DEGs were annotated to these pathways, their overall enrichment was relatively modest and did not reach the same level of statistical significance as lipid metabolism-related pathways. In contrast, lipid metabolism pathways, such as glycerolipid metabolism, exhibited the most prominent DEGs clustering and enrichment significance, suggesting a more pronounced transcriptional impact of QM on lipid metabolic processes. Consistent with these findings, Gene Ontology (GO) enrichment analysis indicated significant changes in genes encoding lipoproteins associated with the Mtb cell wall (Fig. 3a and Supplementary Fig. S1). Collectively, these results suggest that QM treatment preferentially perturbs lipid metabolism-related pathways and may thereby influence the composition and function of the Mtb cell wall.

Fig. 3. Transcriptomic analysis reveals that QM affects the cell wall biosynthesis pathway in Mtb.

a KEGG pathway enrichment analysis of DEGs following QM treatment. Bubble plot showing the top 30 enriched pathways. Each bubble represents a pathway, with bubble size corresponding to the number of differentially expressed genes (DEGs) and color indicating the q-value. The q-value represents the adjusted p-value after correction for multiple hypothesis testing. b RT-qPCR validation of selected genes associated with cell wall component biosynthesis in Mtb. MurC (UDP-N-acetylmuramate--L-alanine ligase), korA (2-oxoglutarate oxidoreductase subunit KorA), ripA (peptidoglycan endopeptidase RipA), fbpB (diacylglycerol acyltransferase/mycolyltransferase Ag85B), Rv2953 (trans-acting enoyl reductase), and fadD32 (long-chain-fatty-acid-AMP ligase FadD32) were analyzed. Gene expression was normalized to sigA. Data represent mean ± SD (n = 4). *P < 0.05; **P < 0.01; ****P < 0.0001; n.s. = not significant.

To validate the transcriptomic results, RT-qPCR was performed to assess the expression levels of genes associated with cell wall biosynthesis in Mtb following treatment with QM (Fig. 3b). The expression levels of Rv2953, involved in glycolipid biosynthesis, ripA, a peptidoglycan endopeptidase, and fbpB, a mycolyltransferase, were significantly downregulated. In contrast, fadD32, a long-chain fatty-acid–AMP ligase, exhibited an upregulation trend. Furthermore, murC, involved in early peptidoglycan synthesis, and korA, associated with the TCA cycle and cell wall biogenesis, were markedly upregulated following QM treatment. These data indicate that QM triggers a complex transcriptional response in genes associated with various components of the cell wall biosynthetic pathway, suggesting distinct regulatory mechanisms underlying the inhibition of cell wall synthesis in Mtb.

QM disrupts the integrity and permeability of the Mtb cell wall

To further support the above findings, we assessed Mtb cell wall permeability using ethidium bromide (EtBr) accumulation assays. H37Rv cells were treated with DMSO, 4 × MIC QM, or isoniazid (INH, 4 × MIC, positive control) for 48 h. QM treatment resulted in increased EtBr uptake compared with the DMSO, indicating enhanced cell wall permeability (Fig. 4a). Consistently, the positive control INH also caused a marked increase in EtBr accumulation. Transmission electron microscopy (TEM) revealed that after 48 h of 1 μg/mL (4 × MIC) QM treatment, some Mtb cells exhibited disrupted cell membranes and cell walls, with cytoplasmic leakage and occasional plasmolysis observed (Fig. 4b). As a positive control, 0.16 μg/mL (4 × MIC) INH-treated cells also displayed cell wall and membrane disruption, as shown in TEM micrographs (Supplementary Fig. S2), whereas DMSO-treated cells maintained intact cell envelope structures. To investigate whether QM compromises cell wall function, we evaluated the susceptibility of H37Rv to various stress conditions²⁷ following treatment. Compared with the control groups, no significant differences in bactericidal activity were observed under oxidative (H₂O₂) or acidic (pH 4.5) stress. However, QM-treated Mtb exhibited markedly increased sensitivity to 0.05% sodium dodecyl sulfate (SDS) exposure. These results suggest that QM disrupts Mtb cell wall permeability (Supplementary Fig. S3).

Fig. 4. QM enhances cell wall and membrane permeability, leading to disruption of the proton motive force (PMF) in Mtb.

a Ethidium bromide (EtBr) accumulation assay. Mtb H37Rv cells were treated with DMSO, 4 × MIC QM (1 μg/mL), or INH (0.16 μg/mL, positive control) for 48 h. QM treatment increased EtBr uptake, indicating enhanced cell wall permeability. Data are presented as mean ± SD from three independent experiments. b Transmission electron microscopy (TEM) images of Mtb cells following 48 h treatment with 1 μg/mL QM. DMSO-treated cells served as the control. Arrows indicate regions of cell wall damage. Scale bar = 200 nm. c Outer membrane permeability increment with QM, EDTA (1 mM) as a positive control. **P < 0.01; ***P < 0.001; n.s. = not significant. Bars represent mean ± SD from three independent experiments. d Inner membrane permeability increment with QM, EDTA (1 mM) as a positive control. ****P < 0.0001; n.s. = not significant. Data represent mean ± SD of three independent biological replicates. e DiSC₃(5) membrane potential measurement for various QM dosages. 1% SDS was used as a positive control, 0.125 μM RFP was used as a negative control. Fluorescence was monitored for 38 min. Data are shown as mean ± SD (n = 4). f Effect of QM on intracellular pH of Mtb H37Rv. CCCP as a positive control. Data are presented as mean ± SD of three independent experiments.

Outer membrane permeability, measured by 1-N-phenylnaphthylamine (NPN) uptake^28,29, was significantly enhanced in QM-treated Mtb cells (Fig. 4c). Propidium iodide (PI) staining^28,29 showed increased inner membrane permeability and entry of PI into Mtb cells, indicating membrane damage (Fig. 4d and Supplementary Fig. S4). Although QM did not induce detectable dissipation of membrane potential, as measured by 3,3-dipropylthiadicarbocyanine iodide (DiSC₃(5)) fluorescence^28,29, intracellular pH was measured using BCECF-AM in combination with the Intracellular pH Calibration Buffer Kit according to the manufacturer’s instructions. QM treatment significantly decreased intracellular pH, as detected by BCECF-AM fluorescence³⁰, suggesting that QM may disrupt the PMF (Fig. 4e, f). These findings indicate that QM impairs Mtb cell envelope integrity and homeostasis.

Thermal proteome profiling implicates FabD as an important protein in QM’s mechanism

TPP was performed to systematically investigate proteins whose thermal stability was altered upon QM treatment^31,32. Mtb H37Rv was treated with QM at 4 × MIC (1 μg/mL) or with DMSO as a control. After a thermal challenge across a 12-point temperature gradient, SDS-PAGE analysis identified ten higher temperatures (44–73 °C) showing a continuous and pronounced decline in overall protein abundance, which were selected for subsequent mass spectrometry analysis (Fig. 5a).

Fig. 5. Thermal proteome profiling reveals FabD stabilization and effects on mycolic acid biosynthesis.

a Thermal proteome profiling (TPP) workflow. Mtb H37Rv strains were treated with QM at 4 × MIC or with DMSO for 48 h. Bacterial lysates were prepared and subjected to a thermal shift at a temperature gradient of 37, 41, 44, 47, 50, 53, 56, 59, 63, 67, 70, and 73°C. After ultracentrifugation to remove precipitated proteins, equal volumes of the supernatants were analyzed by SDS-PAGE to assess thermal stability. M: Protein marker. D: DMSO-treated group. X: QM-treated group. b GO enrichment analysis of differentially stabilized proteins identified by thermal proteomics. c Thermal melting curve of FabD protein obtained by TPP. Data are from two independent biological replicates, showing ΔT_m values of 6.39 °C and 6.53 °C, indicating strong thermal stabilization upon QM treatment. |ΔT_m| values above 1 °C are generally considered significant in TPP analysis. d Thin-layer chromatography (TLC) analysis of mycolic acids. Mycolic acids were extracted from Mtb H37Rv strains treated with DMSO (control), QM (8 × MIC, 2 μg/mL), or isoniazid (INH, 8 × MIC, 0.32 μg/mL; positive control) at 37 °C for 24 h. After methylation, samples were separated by TLC using hexane/ethyl acetate (95:5, v/v) with triple development. Alpha Alpha-mycolic acid methyl ester; Methoxy Methoxy-mycolic acid methyl ester; Keto Keto-mycolic acid methyl ester; FAMEs Fatty acid methyl esters.

Across two independent biological replicates, a total of 2308 proteins were identified. Among these, 106 proteins exhibited reproducible thermal stability changes, defined as shifts in protein melting temperature (ΔT_m) greater than 1 °C or less than –1 °C in both replicates, indicating consistent stabilization or destabilization following QM treatment (Supplementary Fig. S5). ΔT_m reflects changes in protein thermal stability induced by compound exposure and may arise from ligand binding, ligand-induced conformational stabilization, or altered protein–protein interactions. Proteins meeting these criteria were therefore classified as QM-responsive proteins, displaying significant and reproducible ligand-associated thermal shifts. The consistent directionality of thermal stability changes across subsets of proteins enables prioritization of biological pathways systematically affected by QM and supports downstream functional validation.

GO enrichment analysis of the Cellular Component category revealed that QM-regulated proteins were predominantly associated with the cell wall and fatty acid biosynthesis pathways (Fig. 5b). Consistently, in the Molecular Function category, ATP binding was among the significantly enriched terms. Given the close relationship between ATP binding and cellular energy metabolism, we further evaluated whether QM affects intracellular ATP levels in Mtb (Supplementary Fig. S6). Treatment with 4× MIC QM (1 μg/mL) for 24 h resulted in no significant change in intracellular ATP levels compared with the DMSO control. In contrast, the positive controls, 1/2× MIC BDQ (0.02 μg/mL) and dicyclohexylcarbodiimide (DCCD, 100 μM), led to a marked reduction in intracellular ATP levels. These results indicate that, despite the enrichment of ATP-binding-related proteins, QM does not impair bacterial energy metabolism via inhibition of ATP synthesis. Among the top 10 proteins exhibiting the most pronounced thermal stabilization upon QM treatment, FabD, an essential enzyme involved in mycolic acid biosynthesis, stood out with ΔT_m values of 6.39 °C and 6.53 °C in two biological replicates (Fig. 5c).

Effects of QM on mycolic acid biosynthesis and its regulatory pathways

Mycolic acids are essential for maintaining the dense and low-permeability structure of the Mtb cell wall, which contributes to its resistance to environmental stress and antibiotic tolerance. To explore whether QM targets mycolic acid biosynthesis, we treated the bacterial cultures with QM, then extracted mycolic acids and analyzed them under identical conditions using thin-layer chromatography (TLC)³³. As shown in Fig. 5d, after methyl esterification, three mycolate subtypes (α-, methoxy-, and keto-mycolates) were detected. The results demonstrated that QM treatment markedly inhibited the synthesis of all three subtypes, suggesting impaired mycolic acid biosynthesis upon FabD inhibition.

To investigate whether QM modulates the expression of genes upstream and downstream of fabD involved in fatty acid and mycolic acid biosynthesis in Mtb, we performed RT-qPCR analysis following QM treatment (4 × MIC, 1 μg/mL, 48 h), selecting key genes in these pathways for validation. RT-qPCR results revealed that fas, a fatty acid synthase, accA3, involved in fatty acid precursor synthesis, and accD2, involved in mycolic acid precursor synthesis, were significantly upregulated, suggesting a compensatory response to impaired fatty acid synthesis. In contrast, pks13, a mycolic acid condensation enzyme, pcaA, a cyclopropane mycolic acid synthase, and cmaA2, involved in mycolic acid modification, were downregulated, indicating defects in mycolic acid condensation and modification. These results suggest that fabD inhibition triggers upstream transcriptional compensation but disrupts downstream mycolic acid biosynthesis (Supplementary Fig. S7).

Biochemical and biophysical validation of FabD as the molecular target of QM

To further validate FabD as a potential target of QM, we confirmed the expression level of FabD in the constructed recombinant strains by RT-qPCR (Fig. 6a). As shown, in the absence of QM, both the vector control strain (pMV261) and the fabD-overexpression strain (FabD-OE) exhibited comparable growth kinetics over a 7-day period, indicating that fabD overexpression alone does not impact the intrinsic growth of Mtb. Upon treatment with QM at 1 × MIC, the CFU of the pMV261 strain gradually declined, while the FabD-OE strain maintained stable CFU levels throughout the period, suggesting a reduced susceptibility to QM. Notably, under the 4 × MIC condition, although both strains exhibited a decline in viability, the FabD-OE strain showed significantly attenuated bactericidal response compared to the control (Fig. 6b). Therefore, the susceptibility analysis of the overexpression strains suggested that FabD is a potential target of QM.

Fig. 6. Confirmation of FabD as the molecular target of QM.

a RT-qPCR analysis of fabD mRNA levels in Mtb strains. The fabD overexpression plasmid (pMV261-fabD) was constructed and introduced into Mtb, with the empty vector strain (electroporated with pMV261) serving as the control. RT-qPCR analysis was performed to assess fabD mRNA expression in wild-type Mtb H37Rv (WT), the pMV261 strain (vector control), and the fabD-overexpression strain. Data are presented as mean ± SD from three independent biological replicates. b In vitro susceptibility assay of Mtb strains with altered FabD expression. CFU counts (Log10 scale) of vector control strain (pMV261) and the fabD-overexpression strain (FabD-OE) following exposure to QM. Bacteria were plated daily over a 7-day period during treatment, respectively. Data are presented as mean ± SD from three independent biological replicates. c Fluorescence quenching assay. Binding interaction between QM and FabD was evaluated by tryptophan fluorescence quenching. Increasing concentrations of QM resulted in dose-dependent quenching of FabD fluorescence, indicating direct binding. d Molecular docking analysis. Superimposition image of docked pose of the QM to FabD protein. Ligand-target interaction diagram showing interactions of QM with the active site residues of FabD protein. 2D molecular docking diagram showing the predicted interaction between QM and the FabD protein. Green spheres represent hydrogen bonds, magenta spheres represent amide–π stacking interactions, and light pink spheres represent π-alkyl interactions. e Surface plasmon resonance (SPR) analysis. Sensorgrams representing the interaction between QM (1.562–200 µM) and immobilized FabD on a BIAcore sensor chip. The K_D value was calculated using the BIA evaluation software based on global fitting of the sensorgram curves. f Enzyme activity assay. QM inhibits FabD in a dose-dependent manner. The graph shows FabD enzyme activity after incubation with increasing concentrations of QM (0.76–98 µM). Data are presented as mean ± SD (n = 3). The IC₅₀ value was calculated using nonlinear regression.

Based on the analysis of the results described above, we amplified the coding sequences of Rv2243 (fabD) and cloned into the pET-28a(+) vector for overexpression and purification as (His)₆-tagged protein. To validate the interaction between QM and FabD, we conducted intrinsic fluorescence quenching assays, QM dose-dependently reduced the intrinsic fluorescence of FabD, a marked decrease in fluorescence intensity (ΔAU_{326 nm} = 194,336 AU) was observed even at a low compound concentration (0.025 mg/mL) (Fig. 6c). Additionally, we also performed molecular docking studies, which revealed that QM binds to FabD with a binding energy of −6.7 kcal/mol (Fig. 6d). The compound interacts with key residues of FabD, including His90, Ser91 and Asn155²⁰, through hydrogen bonding and amide π stacking interactions. Surface plasmon resonance (SPR) assay revealed that, although the overall binding signal was relatively low, a concentration-dependent increase in response units was observed, suggesting specific binding between QM and FabD, with a calculated dissociation constant (K_D) of 318.5 μM (Fig. 6e).

Finally, to assess whether QM inhibits FabD activity, we performed a coupled enzyme assay using purified (His)₆-FabD and (His)₆-AcpM proteins³⁴ (Supplementary Fig. S8). FabD catalyzes the transfer of the malonyl group from malonyl-CoA to AcpM, releasing CoA. The released CoA is consumed by pyruvate dehydrogenase (PDH), which converts pyruvate and NAD⁺ into acetyl-CoA and NADH^20,35. NADH production was monitored by fluorescence (Ex_340 nm/Em_465 nm) as an indirect readout of FabD activity. QM inhibited FabD activity in a dose-dependent manner, with an IC₅₀ of 67.79 μM (Fig. 6f). These results indicate that QM directly inhibits FabD activity in vitro, supporting FabD as a molecular target of QM in Mtb.

Collectively, our data demonstrate that QM as a FabD inhibitor that blocks mycolic acid synthesis and disrupts the integrity and permeability of the cell wall in Mtb. As shown in Fig. 7, QM targets FabD, reduces the production of cell wall lipids, impairs membrane function, and ultimately exerts antibacterial activity.

Fig. 7. Proposed mechanism of action of QM targeting FabD and disrupting cell wall biosynthesis in Mtb.

QM directly targets FabD, an essential enzyme that catalyzes the transfer of malonyl groups from malonyl-CoA to acyl carrier protein (ACP), initiating fatty acid biosynthesis and subsequent mycolic acid production. Inhibition of FabD impairs mycolic acid synthesis, disrupting the integrity of the mycobacterial cell wall. This structural damage leads to intracellular pH imbalance and proton motive force (PMF) collapse, ultimately exerting bactericidal effects. The schematic illustrates the Mtb cell envelope, including labeled components (mycolic acid, peptidoglycan, and arabinogalactan) and relevant biosynthetic pathways. Arrows indicate regulatory effects: sharp-ended arrows denote stimulation or upregulation, while blunt-ended arrows represent inhibition or downregulation, with arrows indicating up- or downregulation of specific pathways or components.

Discussion

QM is a naturally derived phenoxazine compound with favorable pharmacological properties, including low toxicity and potent activity against Mtb. In recent years, it has attracted increasing attention in the field of antimicrobial research. Previous studies have shown that QM effectively inhibits the growth of pathogens such as Helicobacter pylori at concentrations ≥1 μg/mL and induces membrane damage, as observed by electron microscopy, suggesting that it may exert antibacterial effects by disrupting cell membrane integrity⁴. Additionally, the sensitivity of Mycobacterium scrofulaceum to Questiomycin A may be closely related to its cell wall structure¹⁴. In this study, we systematically investigated the anti-tuberculosis activity of QM against Mtb and identified its molecular target. QM exhibited potent bactericidal activity against both drug-sensitive and multidrug-resistant (MDR) Mtb strains, with MIC values lower than those of levofloxacin (LVX) and ethambutol (EMB). Importantly, QM showed no cross-resistance with first- or second-line anti-TB drugs. It remained active under stress conditions that simulate the non-replicating state of Mtb, such as nutrient deprivation, and exhibited a sustained PAE. These findings indicate that QM may target a pathway essential under both replicating and non-replicating conditions. We demonstrated that QM disrupts mycolic acid biosynthesis by directly inhibiting FabD, leading to cell wall disintegration, disturbance of intracellular pH homeostasis, and collapse of the PMF.

FabD (malonyl CoA-acyl carrier protein transacylase, MCAT), encoded by the Rv2243 gene³⁶, catalyzes the transfer of a malonyl group from malonyl-CoA to holo-ACP, producing malonyl-ACP. This product serves as the essential building block for the elongation of fatty acids in the FAS-II system³⁷. Through this critical enzymatic step, FabD supplies the substrate required for mycolic acid synthesis³⁸, which is indispensable for Mtb’s cell wall integrity and survival^39,40. Mycolic acids are long-chain (C60–C90) fatty acids that form the major lipid component of the Mtb cell envelope^41,42, covalently attached to arabinogalactan and peptidoglycan layers. This unique, waxy barrier not only protects Mtb from environmental stress but also contributes to its intrinsic drug resistance. Disruption of MA biosynthesis is a validated therapeutic strategy, as exemplified by INH, which targets InhA within the same pathway⁴³. Thus, FabD plays a vital role in maintaining Mtb cell envelope structure and function.

Given its essentiality and lack of human homologs, FabD represents a highly attractive and selective target for anti-TB drug development^34,44. Structural and computational studies have identified key catalytic residues and ligand-binding pockets within FabD, laying the foundation for rational inhibitor design²⁰. Although several FabD inhibitors, such as NPC313985 have been identified via virtual screening, their anti-TB activity remains unvalidated experimentally⁴⁴. In this work, we provide experimental evidence that QM directly inhibits FabD. The essentiality of FabD and its pathogen-specific nature make it an appealing therapeutic target, especially in the context of MDR-TB and long-term therapy⁴⁵.

Importantly, QM retained its anti-tuberculosis activity under stress conditions that simulate the non-replicating state of Mtb, in which classical antibiotics such as INH lose efficacy⁴⁶. These stress models, such as nutrient starvation, are well known to induce dormancy-associated phenotypes⁴⁷. Interestingly, lipid metabolism, including mycolic acid synthesis remains at least partially active during non-replicating persistence⁴⁸, representing a potential metabolic vulnerability. Our findings suggest that FabD retains residual activity under these conditions, and that its inhibition by QM may impair essential lipid synthesis, thereby contributing to bacterial clearance even during non-replicating persistence.

Although our data confirm that FabD is the direct molecular target of QM, its inhibitory potency appears to be moderate. This conclusion is supported by multiple in vitro assays showing relatively weak binding affinity in SPR (K_D = 318.5 μM), moderate docking energy (−6.7 kcal/mol), and an IC₅₀ of 67.79 μM in the enzyme activity assay. These results suggest that while QM interacts with FabD, further chemical optimization is warranted to improve its inhibitory strength. Nonetheless, its phenoxazine scaffold represents a promising starting point for developing more potent FabD-targeting agents.

The unique architecture of the Mtb cell envelope, composed of long-chain mycolic acids linked to arabinogalactan and peptidoglycan, is crucial for bacterial survival, virulence, and intrinsic drug resistance⁴⁹. Targeting enzymes involved in cell wall biosynthesis has led to several clinically valuable drugs. New candidate compounds such as BTZ-043 and SQ641 inhibit DprE1 and MurX, enzymes responsible for arabinogalactan and peptidoglycan synthesis⁵⁰. These agents are currently in clinical development for both drug-sensitive and drug-resistant TB. However, the emergence of resistance underscores the urgent need for novel druggable targets within the cell wall biosynthetic pathway. Our study identifies FabD as a promising target and characterizes QM as a novel inhibitor with a distinct mechanism of action.

Despite QM’s potent anti-tuberculosis activity and its ability to target FabD, its pharmacokinetic profile remains suboptimal. The short half-life and low peak concentration limit overall drug exposure, thereby constraining its therapeutic potential in the current form. These pharmacokinetic challenges highlight the need for rational structural optimization to improve its drug-likeness and bioavailability. Nevertheless, the identification of FabD as its molecular target and the elucidation of its mechanism of action provide a strong foundation for future drug development aimed at targeting FabD.

In conclusion, this study identified the phenoxazine compound QM as a novel FabD-targeting lead with potent anti-tuberculosis activity. These findings expand our understanding of lipid metabolism in Mtb pathophysiology and pave the way for the development of FabD-targeted therapeutics against drug-resistant and persistent tuberculosis.

Methods

Strains

E. coli DH5α was obtained from Takara (Kyoto, Japan). M. tuberculosis H37Rv (ATCC 27294) was obtained from the American Type Culture Collection. Clinical isolates of M. tuberculosis were obtained from the National Clinical Laboratory for Tuberculosis, Beijing Tuberculosis and Thoracic Tumor Research Institute.

Compound library for screening

The compound libraries used in this study included the L6500 Microbial Natural Product Library (685 compounds) purchased from TargetMol (Boston, MA, USA).

Minimum inhibitory concentration (MIC)

The assay was performed against Mtb H37Rv, drug-sensitive clinical isolates, and drug-resistant clinical isolates. Rifampin, isoniazid, and other standard anti-TB drugs were used as references. Compounds were tested in a two-fold serial dilution ranging from 0.001 to 10 μg/mL. The minimum inhibitory concentration (MIC) was defined as the lowest concentration preventing a color change from blue to pink.

Minimal bactericidal concentration (MBC)

The CFU enumeration method was used to determine MBCs against Mtb H37Rv and three clinical isolates. Cultures were treated with compounds in 24-well plates for 7 days. Aliquots were plated on 7H10 agar and incubated for 28 days. The MBC was defined as the lowest drug concentration resulting in a ≥ 99.9% reduction in CFU compared to the initial inoculum.

Time-kill curve assay

H37Rv cultures (2 × 10⁵ CFU/mL) were treated with QM at concentrations ranging from 0.5× to 16 × MIC in 24-well plates. Drug-free cultures served as controls. At days 0, 3, 7, 10, and 14, 10-fold serial dilutions were plated on 7H10 agar. CFU counts were recorded after 28 days, and killing kinetics were plotted as Log₁₀ CFU/mL versus time.

Post-antibiotic effect (PAE) assay

QM was diluted in 7H9 broth to final concentrations ranging from 0.5 to 50 μg/mL. Rifampicin (RFP) served as a positive control at 0.05–5 μg/mL. H37Rv in log-phase was diluted to 3 × 10⁵ CFU/mL and mixed 1:1 with drug solutions in 96-well plates, followed by incubation at 37 °C for 2 h. After incubation, the bacterial suspensions were diluted 1000-fold to remove residual antibiotics. The treated (T) and untreated control (C) groups were then incubated at 37 °C. At days 0–6, aliquots were plated on 7H10 agar. After 3 weeks, colony-forming units (CFU) were enumerated. The PAE was calculated as the time required for bacteria in the treated group to increase by one Log₁₀ unit minus that of the control group: PAE (h) = T − C.

Pharmacokinetic studies

All animal procedures were approved by the Beijing Tuberculosis and Thoracic Tumor Research Institute and were conducted in accordance with institutional guidelines for animal care and use. Six-week-old male Balb/c mice weighing 16–19 g were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd.

Mice were housed under standard laboratory conditions with a 12 h light/12 h dark cycle, a controlled temperature of 26 ± 0.5 °C, and relative humidity of 50% ± 10%, with free access to food and water. Prior to oral dosing, mice in the gavage group were fasted for 12 h.

QM was formulated in 0.5% (w/v) carboxymethyl cellulose and administered by oral gavage at a dose of 10 mg/kg (n = 3). Blood samples were collected from the retro-orbital plexus under light anesthesia with isoflurane at 0.5, 1, 2, 6, 8, and 24 h post-dose. Blood samples were anticoagulated with sodium heparin, and plasma was obtained by centrifugation. Plasma samples (50 μL) were mixed with 100 μL of acetonitrile for protein precipitation prior to analysis. The mice were euthanized by CO₂ asphyxiation.

Plasma concentrations of QM were quantified using a validated liquid chromatography–tandem mass spectrometry (LC–MS/MS) method. Analyte quantitation was performed on a Quantum Access mass spectrometer operated in positive ion mode. The LC–MS/MS system consisted of an Agilent UPLC coupled to an API 4000 triple quadrupole mass spectrometer (AB Sciex, USA). Data acquisition and processing were carried out using Agilent software (version 1.5.2).

Chromatographic separation was achieved on a Zorbax SB-C18 column (50 × 2.1 mm, 3.5 μm) using an isocratic mobile phase consisting of methanol/water containing 0.1% formic acid at a flow rate of 0.3 mL/min. Pharmacokinetic parameters were calculated by noncompartmental analysis using WinNonlin software (version 6.3; Pharsight Corporation, Mountain View, CA, USA).

Intracellular activity in macrophages

Logarithmically growing Mtb H37Rv was used to infect J774A.1 macrophages for 4 h. After washing with PBS, infected cells were treated with test compounds or control drugs for 72 h. Following lysis with 0.1% SDS buffer (Sigma-Aldrich), 10-fold serial dilutions of the bacterial suspension were plated on 7H10 agar. Plates were incubated at 37 °C, and CFUs were counted after 3–4 weeks.

Effect of QM on Mtb under different pH conditions

Mtb H37Rv strains were cultured in Middlebrook 7H9 medium supplemented with 10% OADC and 0.05% Tween-80 to logarithmic phase (optical density, OD₅₇₀ = 0.4–0.6). Cells were harvested by centrifugation (3500 × g, 10 min, 4 °C), washed twice with PBS, and resuspended in 7H9 medium supplemented with 10% OADC and 0.02% Tyloxapol, adjusted to pH 5.0, 6.0 or 7.0. The starting OD₅₇₀ of each culture was adjusted to 0.04. Under different pH conditions, QM was added to the cultures at final concentrations of 0.125×, 1×, and 8× MIC. RFP was used as a positive control at a final concentration of 1× MIC. Cultures were incubated at 37 °C for 21 days. At designated time points (days 0, 1, 7, and 21), aliquots were collected, serially diluted, and plated on Middlebrook 7H10 agar for CFU enumeration. Plates were incubated at 37 °C for 3–4 weeks. All experiments were performed in biological duplicates with technical triplicates.

Nutrient starvation model and bactericidal activity of QM

Log-phase Mtb H37Rv cultures were harvested by centrifugation at 3000 × g for 10 min and washed three times with sterile 1 × PBS to remove residual media. The bacterial pellet was then resuspended in 1 × PBS containing 0.05% Tween-80 and homogenized thoroughly to ensure even dispersion. The bacterial suspension was incubated under nutrient-limited conditions at 37 °C for 15 days to induce a dormant-like state. During incubation, OD₅₇₀ was periodically monitored to confirm the establishment of metabolic quiescence. After 15 days, bacterial viability was assessed by plating on 7H10 agar to validate dormancy.

QM, isoniazid, rifampin and bedaquiline were administered at final concentrations of 20, 10, and 5 μg/mL, respectively. A DMSO-treated group served as a negative control. After 4 days of drug treatment at 37 °C, 100 μL from each well was plated on 7H10 agar. All biological replicates were performed in quadruplicate.

RNA extraction, RT-qPCR and RNA sequencing

Mtb H37Rv treated with DMSO or QM (4 × MIC) for 6 h and 48 h. Cells were harvested after treatment with RNA protect bacterial reagent (Qiagen, Germany), lysed in RLT buffer + β-mercaptoethanol, and disrupted with zirconia beads. Total RNA was extracted using RNeasy Mini Kit (Qiagen, Germany) and eluted with DEPC water. RNA quality was assessed by OD₂₆₀/OD₂₈₀. cDNA synthesis was performed using the PrimeScript RT reagent kit with gDNA Eraser (Takara, Japan), and qPCR was carried out using TB Green Premix Ex Taq. Breifly, cDNA was synthesized from 1 μg RNA. qPCR was performed using 2 μL of cDNA and gene-specific primers listed in Table S1. Samples were run on Applied Biosystems StepOnePlus Real-Time PCR System equipment at 95°C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. Expression levels of the target genes were normalized to the reference gene sigA, and relative quantification was calculated using the 2^−ΔΔCt method.

Qualified RNA was purified using the RNAClean XP Kit and DNase-treated (Qiagen). Purified RNA was quantified using NanoDrop and Agilent 4200 systems. After rRNA depletion, libraries were constructed via cDNA synthesis, end repair, adaptor ligation and PCR enrichment. Library quality was assessed with Qubit 2.0 and Agilent 4200. Paired-end sequencing was performed on the Illumina NovaSeq 6000 platform. DEGs were identified and analyzed using GO and KEGG enrichment. GO terms with q-value ≤ 0.05 were considered significant. Enrichment was visualized using bubble plots, with Rich Factor indicating the degree of enrichment.

Ethidium bromide permeability assay

To assess cell wall permeability, Mtb H37Rv strains were treated with DMSO (control) or QM (4 × MIC) for 48 h. Bedaquiline (BDQ, 4 × MIC) was included as a negative control. After treatment, log-phase cultures were harvested, washed twice with 1 × PBS, and resuspended in PBS to an OD₅₇₀ of 0.5. Cell suspensions were transferred into black-walled 96-well plates, and EtBr was added to a final concentration of 20 μM. Fluorescence was recorded every minute for 30 min at 37 °C using a microplate reader (excitation: 530 nm; emission: 590 nm). Increased EtBr fluorescence was used as an indicator of enhanced cell wall permeability. Experiments were performed in triplicate.

Transmission electron microscopy (TEM)

Log-phase Mtb H37Rv cultures were treated with QM (4 × MIC) or DMSO (control) for 48 h. Cells were fixed in 2.5% glutaraldehyde at 4 °C for 12 h, post-fixed with 1% osmium tetroxide (OsO₄; Ted Pella, USA), and washed with PBS. Samples were dehydrated in graded ethanol, embedded in resin, and sectioned (70–90 nm). Ultrathin sections were stained with uranyl acetate and lead citrate, then examined using a transmission electron microscope (HITACHI, Japan) at 80 kV.

Outer membrane permeability assay

Outer membrane permeability of Mtb was assessed using the fluorescent probe 1-N-phenylnaphthylamine (NPN, 30 μM). Log-phase cultures (OD₅₇₀ = 0.5) were treated with QM or 1 mM EDTA (positive control) at 37 °C for 48 h. Cells were harvested, washed with GHEPES buffer (pH 7.2), resuspended to OD₅₇₀ = 0.4, and fluorescence was measured (excitation/emission: 350/420 nm). Experiments were performed in triplicate.

Cell membrane integrity assay

Propidium iodide (PI, 1 μM) staining was used to evaluate the membrane integrity of Mtb after 48 h treatment with DMSO (control), QM (1× and 2× MIC), or 1 mM EDTA. Treated cells were washed with PBS, resuspended to OD₅₇₀ = 0.4, and fluorescence was recorded (excitation/emission: 535/615 nm). All experiments were conducted in triplicate.

Membrane depolarization assay

Membrane potential disruption was analyzed using 3,3-dipropylthiadicarbocyanine iodide (DiSC₃(5), 2 μM). Mtb cells (OD₅₇₀ = 0.6) were washed and resuspended in GHEPES buffer (pH 7.2), and baseline fluorescence (excitation/emission: 622/670 nm) was monitored before and after QM addition. SDS (1%) and rifampicin (0.125 μM) were used as positive and negative controls, respectively. Fluorescence was recorded every 2 min for 38 min; assays were performed in quadruplicate.

Intracellular pH measurement using BCECF-AM

Mtb H37Rv was treated with QM (0.25×, 1×, and 4× MIC) at pH 6.0 for 24 h at 37 °C. Cells were harvested, washed with PBS (50 mM, pH 6.0), and incubated with 10 μM BCECF-AM (Merck, Cat. no. 14562) at 37 °C for 30 min in the dark with shaking. Carbonyl cyanide m-chlorophenyl hydrazone (CCCP, 10 μM) served as a positive control. After staining, cells were washed to remove excess dye, resuspended, and transferred to black 96-well plates for fluorescence measurement.

To generate a calibration curve, intracellular pH was clamped using the Intracellular pH Calibration Buffer Kit (Cat. no. P35379, Thermo Fisher Scientific). For this, Mtb cells were loaded with 10 μM BCECF-AM in the presence of PowerLoad™ reagent and incubated for 30 min at 37 °C. Calibration buffers at pH 4.5, 5.5, 6.5 and 7.5 were used to equilibrate intracellular and extracellular pH. Fluorescence intensity was measured at excitation/emission wavelengths of 490/535 nm and 430/535 nm using a microplate reader. The ratio of fluorescence intensity at different pH values was used to construct a standard curve, from which the intracellular pH of experimental samples was interpolated. Data represent the average of three technical replicates.

Thermal proteome profiling

Mtb H37Rv cells were treated with QM (4 × MIC) or DMSO (control) for 48 h. Samples were divided into 12 groups (each in duplicate) and heated at 12 temperatures (37–73 °C, 3 °C intervals) for 3 min, followed by equilibration at room temperature and cooling on ice. Cells were lysed with lysozyme (50 μg/mL), 0.8% NP-40, and PMSF, then mechanically disrupted. Lysates were centrifuged (14,000 × g, 10 min, 4 °C), and supernatants were collected for analysis.

Protein concentrations were determined by BCA assay. Equal amounts of protein were analyzed by SDS-PAGE for quality control, and the last 10 temperature points were selected for further analysis. Proteins were digested using magnetic beads and trypsin, TMT-labeled, pooled, desalted and fractionated under basic pH. Peptides were dried and stored at −20 °C until LC-MS/MS analysis.

Samples were analyzed using a Q Exactive HF-X mass spectrometer equipped with a nano-ESI source. MS1 scans were acquired at 60,000 resolution (m/z 350 – 1800), followed by HCD fragmentation (NCE 34%) and MS2 scans at 45,000 resolution for the top 40 precursors.

Data were processed using Proteome Discoverer 3.1 against the Mycobacterium tuberculosis UniProt database (20240708). Protein abundances were normalized to the lowest temperature, and melting curves were generated. Melting temperatures (T_m) were calculated using the TPP R package.

$$f\left(T\right)=\frac{1-plateau}{1+e^{-\left(\frac{a}{T}-b\right)}}+plateau$$

where T represents the temperature, and a, b, and Plateau are constants. At the lowest temperature (T_min), f(T) was set to 1. The protein melting temperature (T_m) was defined as the temperature at which half of the protein abundance is denatured, f(T_m) = 0.5.

The coefficient of determination (R²) was used to assess the goodness of fit of the melting curves. The melting temperature (T_m) corresponds to the temperature at which the curve reaches 0.5, and the slope (Slope) represents the steepest point of the curve. In addition, the plateau of the fitted melting curve was determined from the lower asymptote of the curve. Differential proteins were identified from independent biological replicates based on predefined ΔT_m thresholds (|ΔT_m| > 1 °C), consistent T_m shift direction, treatment-induced changes exceeding replicate variability, and curve quality (R² > 0.8, plateau < 0.3).

Gene Ontology (GO) enrichment analyses (biological process, molecular function, cellular component) and protein–protein interaction analyses were performed to evaluate biological relevance.

Mycolic acid extraction and thin-layer chromatography (TLC) analysis

Mtb H37Rv was cultured in 7H9 medium in the presence of QM (at 4× MIC), isoniazid (INH, positive control) or DMSO (control) at 37 °C for 24 h. Cells were harvested by centrifugation and washed once with 1×PBS.

The bacterial pellet was resuspended in 2 mL of tetrabutylammonium hydroxide (25% w/v) and incubated overnight at 100 °C in a metal heating block to ensure complete hydrolysis of mycolic acids. After cooling to room temperature, 4 mL of dichloromethane, 300 µL of methyl iodide (CH₃I), and 2 mL of distilled water were added to methylate the released mycolic acids. The mixture was vortexed and incubated for 1 h at room temperature with continuous rotation.

Samples were centrifuged at 3500 rpm for 10 min to separate phases. The lower organic phase, containing the methylated mycolic acids, was collected while the upper aqueous phase was discarded. The organic phase was washed twice by adding 4 mL of distilled water, vortexing for 15 min, and centrifuging at 3500 rpm for 10 min. After each wash, the aqueous phase was removed.

The organic phase was dried under nitrogen using a TurboVap evaporator (~30 min) to obtain a yellow oily residue. This residue was dissolved in 3 mL of diethyl ether, sonicated in a water bath at room temperature for 10 min, and centrifuged at 3500 rpm for 10 min. The supernatant containing the methylated fatty acids/mycolic acids was transferred to pre-weighed 2 mL microcentrifuge tubes. Solvent was evaporated under a fume hood, and the final dry weight of the extract was determined by subtracting the empty tube weight from the total.

Each extract was resuspended in chloroform/methanol (2:1, v/v) to a final concentration of 33 μg/μL. A total of 6 μL (~200 μg) was spotted onto silica gel TLC plates. One-dimensional TLC was performed to separate the methylated mycolic acids (MAMEs) using hexane/ethyl acetate (95:5, v/v) as the mobile phase. The TLC plates were developed three times in this solvent system. The resulting profiles were compared among DMSO-, QM- and INH-treated samples to evaluate the effect of compound treatment on mycolic acid composition.

Plasmid construction

To express FabD and AcpM in E. coli, the corresponding genes were cloned into the pET-28a(+) vector, resulting in the recombinant plasmids pET28a-FabD and pET28a-AcpM, both harboring an N-terminal 6×His-tag.

For overexpression of FabD, the gene was amplified from the genomic DNA of Mtb H37Rv using primers listed in Table S2 and cloned into the integrative vector pMV261, which carries the Hsp60 promoter and a kanamycin resistance marker, resulting in the recombinant plasmid pMV261-FabD. The plasmid was introduced into Mtb H37Rv by electroporation.

Expression and purification of recombinant proteins

Recombinant His-tagged FabD and AcpM were overexpressed in E. coli C41(DE3) and purified using Ni²⁺ affinity chromatography. Cultures harboring pET28a-FabD or pET28a-AcpM were grown at 37 °C in LB medium to an OD₆₀₀ of 0.8, and protein expression was induced with 0.5 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) for 8 h at 30 °C. Cells were harvested and lysed by sonication, and the supernatants containing His-FabD or His-AcpM were loaded onto Ni²⁺ His Trap chelating column. Proteins were eluted using a stepwise imidazole gradient in elution buffer (50 mM Tris-HCl, 10 mM reduced glutathione, pH 8.0). For AcpM, after initial growth to OD₆₀₀ of 0.75 in LB medium, cells were transferred to fresh Terrific Broth medium containing kanamycin and induced with 1 mM IPTG overnight at 16 °C. All purified proteins were analyzed by SDS-PAGE followed by Coomassie blue staining. Protein concentrations were measured using the Bradford method.

Intrinsic fluorescence quenching studies

The FabD protein was diluted with 100 mM potassium phosphate buffer to final concentrations of 0%, 25%, 50%, 75%, and 100% of the original protein stock. Each concentration was assayed in triplicate. Assuming an excitation wavelength of 280 nm, the emission spectrum of the protein was recorded, and the maximum emission wavelength was determined to be 320 nm. Subsequently, the excitation spectrum was recorded with emission fixed at 320 nm, identifying an optimal excitation wavelength of 278 nm. For fluorescence quenching studies, 96 µL of FabD solution (fixed concentration) was mixed with 4 µL of QM at various concentrations (0.025–0.4 mg/mL), with DMSO as a control. Fluorescence intensity was recorded at Ex = 278 nm and Em = 308–398 nm (2 nm step), data represent the mean ± SD from three independent biological replicates.

Surface plasmon resonance (SPR) assay

Binding of QM to FabD was evaluated using a Biacore T200 at 25 °C in PBS-T buffer (PBS + 0.05% Tween 20 + 0.1% DMSO). His-FabD was immobilized on a CM5 chip via amine coupling in 10 mM sodium acetate (pH 4.5). QM (1.562 – 200 μM) was injected sequentially, and the equilibrium dissociation constant (K_D) was calculated using Biacore Evaluation Software.

Molecular docking studies

The crystal structure of Mtb FabD (PDB ID: 2QC3) was retrieved from the Protein Data Bank. Structural preparation and docking were performed using AutoDock Vina. The binding mode of QM with FabD was visualized using UCSF Chimera.

Coupled enzyme assay

FabD activity was measured using a coupled enzyme assay in 96-well microplates. The final reaction volume was 100 μL. FabD was diluted in 50 mM phosphate buffer (pH 6.8) with 1 mM EDTA (ethylenediaminetetraacetic acid), 1 mM DTT (dithiothreitol), and 0.1 mg/mL BSA (bovine serum albumin). Malonyl-CoA was prepared at 4× the final concentration in the same buffer. AcpM and PDH (pyruvate dehydrogenase, Maclin, China) were mixed at 4× their final concentrations with 8 mM pyruvate, 1 mM NAD (nicotinamide adenine dinucleotide), and 0.8 mM TPP (thiamine pyrophosphate). Final concentrations: 50 mM phosphate buffer (pH 6.8), 1 mM EDTA, 1 mM DTT, 2 mM pyruvate, 0.25 mM NAD, 0.2 mM TPP, 0.03 nM FabD, 60 μM AcpM, 15 U/100 μL PDH, 25 μM malonyl-CoA. All solutions and plates were pre-equilibrated to 37 °C. Reactions were initiated by adding 25 μL malonyl-CoA after pre-incubation of FabD (46.25 μL) with DMSO or compound (3.75 μL) at 37 °C for 30 min. Fluorescence of NADH was measured at 340 nm excitation and 465 nm emission. The inhibition rate was calculated as: Inhibition (%) = (F_DMSO − F_compound)/F_DMSO ×100%, where F_DMSO and F_compound represent the fluorescence intensity of the DMSO and compound-treated samples, respectively.

Statistics and reproducibility

Graphs and statistical analyses were performed using GraphPad Prism software version 9.5.1 (San Diego, CA, USA). Data are presented as means ± standard deviations (SD) from at least three independent experiments. Statistical significance was assessed using Student’s t-test or one-way ANOVA followed by Tukey’s post hoc test. A p-value ≤ 0.05 was considered statistically significant. The number of replicates and specific statistical tests are indicated in the respective figure legends.

Ethics statement

The care of all animals was in accordance with institutional guidelines for animal care and use, and all animal protocols were approved by the Animal Ethics Committee of the Beijing Tuberculosis and Thoracic Tumor Research Institute (License SYXK (Jing)2025-0026) and complied with international laws and policies (EEC Council Directive 86/609, 1987). The approval number for the preliminary pharmacokinetic studies was XK2025-096.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.Y., Z.W. and L.W. prepared samples. L.X., M.X., X.L., Q.J. and W.Z. performed research; B.W. and C.L. assisted with data collection; L.X. analyzed data, and with Y.L. wrote the manuscript. All authors reviewed and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Ophelia Bu.

Data availability

The RNA-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE311988. Source data underlying graphs can be obtained from Supplementary Data 1. All gene expression data from transcriptomic analysis comparing QM-treated and DMSO-treated Mycobacterium tuberculosis are provided in Supplementary Data 2. All proteins identified by TPP comparing QM-treated and DMSO-treated Mycobacterium tuberculosis are provided in Supplementary Data 3. Uncropped and unedited blot/gel images with molecular weight markers are shown in the Supplementary Information. All other data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09947-7.