Mycobacterium tuberculosis H37Ra was purchased from the American Type Culture Collection (ATCC25177). It was grown on Löwenstein–Jensen slopes (BioMaxima, Lublin, Poland) for 2 weeks, then transferred to fresh Middlebrook 7H9 broth supplemented with 10% ADC and 0.2% glycerol (MilliporeSigma, St. Louis, United States). Thymoquinone of ≥98% purity was purchased from BIOKOM SA. (Janki, Poland). The purity was confirmed by HPLC analysis. Reference antibiotics, such as isoniazid, ethambutol, rifampicin, streptomycin, and ciprofloxacin were purchased from MilliporeSigma (St. Louis, United States), whilst resazurin (alamarBlue) solution was purchased from Invitrogen (Carlsbad, United States).

For the sigma factor expression analysis, 2 mL of samples were taken after 3 and 48 h from high-density cultures exposed to TQ. The total RNA was isolated according to the manufacturer’s directions with the FastRNA Pro Blue Kit (MP Biomedicals, Santa Ana, United States, MP Biomaterials). The qPCR reactions were performed as described before (Sawicki et al., 2022). For the relative quantification of transcripts, targets were normalized to 16S rRNA. Primers used in the analyzes (Millipore, Sigma-Aldrich) are listed in Table 1. Relative mRNA quantification was calculated according to the delta–delta mathematical model.

For metabolite analysis, raw data obtained from LC–MS acquisition were converted in Mass Hunter Qualitative Analysis software (version B.07.00; Agilent Technologies, Santa Clara, CA, United States) to mzDATA format. Open-source XCMS online software (version 3.7.1) for feature extraction was used. CentWawe and Obiwarp methods for feature detection and retention time correction were applied. The intensity threshold was set at 500; features were retained only if present in at least six replicates, and in pairwise analysis, they were considered statistically significant at a value of p of <0.01 (Welch t-test). MS-LAMP software¹ with an integrated Mycobacterium tuberculosis (M. tb) Lipidome database was used to annotate mycobacterial lipids as singly protonated ions (M + H) + with an allowed 0.05 m/z mass difference, whilst other compounds were tentatively identified in MetaboAnalyst5.0² with the MS Peaks to Pathways module with a mass accuracy of 5 ppm. Only compounds confirmed by EIC chromatograms and MS/MS spectra were further submitted for enrichment analysis.

The pairwise comparison of the test and control groups in the metabolomic experiment was based on the Welch t-test. The results were considered statistically significant at a p-value of <0.01. Other comparisons were based on the mean ± standard deviation of the mean (SD). The observations of lipid levels and gene expression results were calculated as fold changes. In the metabolomic experiment, 6–10 replicates were run, whilst other experiments were performed in triplicate.

In the first step, we determined the susceptibility of mycobacteria to thymoquinone. The observed MIC value equaled 32 μg/mL. This value was obtained for a standardized amount of bacteria per mL (1.5 × 10⁶ CFU/mL); however, metabolomic experiments require high-density cultures (10⁹ CFU/mL), as indicated by Halouska et al. (2013). The stressing compound should be used in a dose large enough to affect the cellular metabolome relative to untreated cells but should not induce cell death (Halouska et al., 2013). The drug concentration between 1 and 24 times the MIC should be used to identify an optimal drug dosage (Halouska et al., 2013). Hence, we used the AlamarBlue assay to experimentally determine the effective sub-inhibitory dose of TQ. AlamarBlue is a tetrazolium compound reduced by metabolically active cells to a purple formazan product (Amin et al., 2009). The highest concentration of TQ, which did not cause the change from pink to blue in the high-density culture, was 256 μg/mL, and this concentration was used in the metabolomic experiment. In the metabolomic experiment, bacteria should be allowed to grow with the stressing agent for at least one generation before harvesting (Halouska et al., 2013). In this study, M. tuberculosis was incubated with TQ for 24 and 48 h. These correspond to approximately one and two replication cycles of slowly growing mycobacteria, which divide once per 18–24 h (Wanger et al., 2017).

Untargeted LC–MS analysis was performed to identify dysregulated metabolites. The changes in the levels of lipids were presented as fold changes in relation to the control group and were considered statistically significant with a value of p of <0.01. The exposure of mycobacteria to TQ resulted in the downregulation of almost all detected lipid classes. The exceptions were individual glycerolipids, glycerophospholipids, and several lysoforms, which are monoacylated glycerophospholipids (Figures 1A–D). The downregulation was observed after the first and second day of the exposure, suggesting a slight (small fold change) but long-lasting influence of TQ. Additionally, the hydrolysis of the bacterial cell wall and the methylation of liberated mycolic acids, which were visualized on TLC plates, revealed that TQ does not influence the mycomembrane, as mycolic acid fractions in the test and control samples after both incubation times were similar (Figure 1E).

Untargeted LC–MS analysis enabled the description of other compounds that were influenced by bacteria after their exposure to TQ. Molecules upregulated in test samples were nicotinic acid ribonucleotide, L-valine, acetyl phosphate, 3-phospho-D-glycerate, and adenosine 5′-monophosphate, whereas for nicotinamide adenine dinucleotide, 5′-methylthioadenosine, nicotinamide, and N-acetyl-L-citrulline, downregulation was noticed (Figure 2). The changes in the intensities of these metabolites were presented as box plots and were considered statistically significant with a value of p of <0.01. Indicated compounds were subjected to enrichment analysis, which showed in what metabolic processes they are involved (Figure 3). Nicotinic acid and nicotinamide metabolism had the highest enrichment ratio (almost 35) and statistical significance. It was followed by cysteine and methionine metabolism, and valine, leucine, and isoleucine biosynthesis. A lower score but still high significance was obtained for purine metabolism (Figure 3).

To track mycobacteria response to TQ at the transcriptional regulatory level, we measured the shifts in transcripts of 13 genes encoding sigma transcription factors. This group of regulatory genes controls the remodeling of metabolism in the adaptive bacterial response to environmental conditions (Rodrigue et al., 2006). Therefore, the evaluation of expression levels of sigma factors provides a global view of the bacterial response to compounds inhibiting their growth. The bacteria harvest time for gene expression analysis was set at 3 and 48 h. The first time point gave an insight into the rapid bacterial response at the genetic level. These rapid transcriptomic changes preceded the metabolic rearrangement. The second analysis performed after 48 h of incubation with TQ proved that transcriptomic changes are dynamic in time and can be correlated with metabolomic dysregulations. Bacteria exposed to TQ responded in a time-dependent manner. After 3 h, a significant upregulation of sigE was observed. The fold change of that gene was up to five times higher than the fold change of other monitored genes. Later, after 48 h, all the sig genes were significantly overexpressed, with noticeable dominance of the sigF transcript (Figure 4).

The LC-DAD-MS analysis showed a similar profile of TQ stability in CB and CBI (Figures 5A–E). Regardless of the presence of bacteria, TQ content was decreasing over time, which was proven by lowered signal intensity on DAD and EIC chromatograms and by observed changes in UV spectra. HPLC analysis confirmed diminished TQ concentration and peak purity, especially after 48 h (Table 2). The pseudomolecular ion of TQ, m/z 165.08, was observed at 20.1 min (Figure 5D). Its usual degradation products, hydrothymoquinone, thymoquinone dimer, or thymol, were not confirmed. Interestingly, two pseudomolecular ions of m/z 166.08 were detected. In the CB sample, m/z 166.08 was present at 16.7 min, whilst in the CBI it was accompanied by a second ion at 3.6 min (Figure 5F). Despite the same m/z value and molecular formula (C9H11NO2), the observed ions were characterized by different MS/MS spectra and fragmentation patterns (Figure 5F). Isomer 2-amino-5-(propan-2-yl)cyclohexa-2,5-diene-1,4-dione was present in culture broth without bacteria, whilst 2-(1-aminoethyl)-5-methylcyclohexa-2,5-diene-1,4-dione appeared in the inoculated sample, suggesting that this isomer may be a product of cellular biotransformation.

M. tuberculosis responds to different environmental stimuli through the transcription of specific gene sets, which create a three-level network hierarchy of 13 sigma factors grouped into 5 communities (Chauhan et al., 2016). The bacteria’s exposure to TQ first caused a rapid (after 3 h) and selective shift in the number of transcripts of a major stress-responsive sigma factor, the sigE gene (Jensen-Cain and Quinn, 2001; Raman et al., 2001). sigE is induced via surface stress (Flentie et al., 2016) and upregulates mprA and mprB genes responsible for sensing cell envelope integrity and transferring stress information to the bacterial cytoplasm (Bretl et al., 2014). Additionally, sigE participates in the response to xenobiotic stimuli, oxidative stress, and regulation of fatty acids and lipid metabolism (Karp et al., 2019). The early upregulation of sigE expression may suggest that TQ rapidly disturbs the integrity of the mycobacterial cell envelope. The later (after 48 h) overexpression of all sigma factor genes caused by TQ indicates the activation of different genetic pathways aiming to overcome the influence of the stressing agent. Under the influence of TQ bacteria increased the transcription of genes (sigC, sigD, sigE, sigG, sigH, sigI, sigJ, sigK, sigL, and sigM) coding sigma factors controlling cell envelope synthesis, secretory functions, and periplasmic protein repair and degradation increased (Lee et al., 2008). Interestingly, the expression of sigF after 48 h was much higher than that of other genes. In M. tuberculosis, this gene was associated with a stationary phase/stress response related to nutrient depletion (Demaio et al., 1996). The observed significant upregulation of sigF suggests that TQ caused the exhaustion of energy resources in bacterial cells.

The characteristic feature of mycobacteria is the very high content of lipids in their cell envelope. These abundant and diverse lipids comprise approximately 40% of the dry cell mass, and together with sugar conjugates, they form an impermeable barrier, strongly contributing to bacterial resistance (Crellin et al., 2013). To investigate the potential influence of TQ on the state of this barrier, we performed a metabolomic analysis of the bacterial lipid profile. The observed downregulation of phosphatidylethanolamines, phosphatidylinositols, and phosphatidylglycerols (Figure 1), which constitute the plasma membrane (Crellin et al., 2013), suggests somewhat diminished production of these molecules caused by TQ. Additionally, only a few “lyso” forms, being partial degradation products of phosphoglycerolipids (Sahonero-Canavesi et al., 2019), were upregulated, indicating that rather reduced metabolic activity than significant degradation of the cell membrane can be considered. Interestingly, the upregulation or downregulation of the mycolic acid fraction was not observed under TQ influence. These long-chain α-branched, β-hydroxylated fatty acids covalently linked to the arabinogalactan layer (Crellin et al., 2013) had a profile similar to the control (Figure 1), meaning that TQ does not target mycomembrane. The changes observed in plasma membrane lipids may be associated with their quite rapid turnover. For example, the diacylglycerophosphoinositol pool is renewed within 7 h within mycobacterial cells (Haites et al., 2005). Therefore, the fast cycle of degradation and renewal gives a better chance to observe even slight dysregulation due to the affected metabolic activity of the cell.

TQ caused significant upregulation of nicotinic acid ribonucleotide (nicotinic acid mononucleotide, NaMN) and significant downregulation of nicotinamide adenine dinucleotide (NAD). In M. tuberculosis, NaMN is synthesized from aspartate, and after the addition of the adenyl group from ATP, it forms nicotinic acid adenine dinucleotide (NaAD). NaAD can be further converted into nicotinic acid adenine dinucleotide phosphate (NAADP), a Ca²⁺-mobilizing second messenger synthesized in response to extracellular stimuli, or into NAD and nicotinamide adenine dinucleotide phosphate (NADP), ubiquitous co-enzymes. NaMN is a unique checkpoint intermediate shared by all de novo, salvage, and recycling pathways (Rodionova et al., 2015). The upregulation of this intermediate may suggest an increased need for NAADP or NADP production under the influence of TQ. Importantly, the level of nicotinamide in test samples was very slightly but statistically significantly downregulated, confirming rather increased production reactions than the degradation of NAD because nicotinamide is a degradation product of NAD (Kasarov and Moat, 1972; Rodionova et al., 2015). Moreover, lowered levels of NAD in test samples indicate the depletion of NAD in bacterial cells after exposure to TQ. In M. tuberculosis, the result of NAD depletion is a sharp suppression of metabolic flux in multiple NAD(P)(H)-dependent pathways, including the synthesis of many amino acids (serine, proline, and aromatic amino acids) and fatty acids, the central metabolism of carbon, and energy production (Rodionova et al., 2014). In this study, lowered NAD levels were correlated with the downregulation of cell membrane lipids, confirming diminished metabolic activity as a result of TQ influence. Because accumulated NaMN was not converted into NAD, it can be hypothesized that TQ may possibly target metabolic steps between NaMN and NAD.

TQ caused a very pronounced downregulation of 5′-methylthioadenosine (MTA), a by-product of polyamine synthesis (spermine and spermidine). The reaction consumes S-adenosylmethionine (originating from methionine) and releases MTA. Because MTA is a potent inhibitor of polyamine biosynthesis and transmethylation reactions, it has to be removed from the methionine and adenine salvage pathways. Thus, MTA is used to recycle sulphur (in methionine) and AMP from adenine (Avila et al., 2004; Karp et al., 2019). If there is a lot of MTA, the need to convert adenosine to AMP is less (Albers, 2009). The detected lowered levels of MTA after bacteria exposure to TQ may mean increased polyamine biosynthesis or MTA extensive degradation. The co-observed raised level of AMP suggests a second scenario. The increasing levels of AMP (and ADP) are naturally accompanied by decreasing levels of ATP (Carling et al., 2011). Furthermore, acetyl phosphate, which was detected to be upregulated in test samples, is a product of ATP degradation (Karp et al., 2019). These indicate that TQ causes energy depletion in mycobacterial cells. This state was observed after 24 h and was persistent the next day, meaning that bacteria were unable to activate catabolic mechanisms, which generate ATP. This may be a consequence of NAD depletion and vice versa because the conversion of NaMN into NaAD needs ATP.

Additionally, the observed downregulation of lipids in the test samples was consistent with L-valine accumulation. L-Valine is a branched-chain amino acid that is degraded to produce isobutyryl-CoA, which is involved in branched-chain fatty acid biosynthesis. L-Valine degradation also yields propionyl-CoA, which supplements polyketide biosynthesis (Karp et al., 2019; Borah et al., 2021). The NAD depletion-induced decreased metabolic activity, including diminished flux for lipid biosynthesis, may contribute to the L-valine accumulation under TQ influence.

In the scientific literature, the alterations in the ATP levels and the NAD/NADH and NADP/NADPH ratios were correlated with the disruption of the mycobacterial membrane upon the action of bactericidal peptide laterosporulin10 (Baindara et al., 2016). It was confirmed that latersporulin10 kills the Mtb H37Rv strain very efficiently, and the mechanism of action was related to the disruption of the cell membrane (Baindara et al., 2016). In this study, some correlation between targeted cellular membranes and disturbed energetic homeostasis can be observed. As indicated above, the early upregulation of sigE expression may suggest that TQ rapidly affects the integrity of the mycobacterial cell envelope. Furthermore, the downregulation of lipids in the cell membrane confirms the possibility that electron transfer reactions within the cytoplasmic membrane might be altered, resulting in energy depletion. This state was long-lasting, as confirmed by the high upregulation of sigF after 48 h, which is associated with a stationary phase/stress response related to nutrient depletion (Demaio et al., 1996). Therefore, our study is in agreement with the statement that changes in energy pools are correlated with the rearrangement/disruption of the bacterial membrane.

The results show that regardless of the presence of bacterial cells, the TQ content decreases over time in the culture broth (Table 2). After the second day, TQ concentration lowered significantly, meaning that TQ is not stable in incubation conditions (elevated temperature, aeration, and light exposure). The lowering of the TQ content was manifested by the changed sample color, which became violet-brown, suggesting the formation of some metabolite. Indeed, mass spectrometry analysis revealed the presence of intense ions with the m/z value being one unit higher than the TQ ion. The detected derivative is nitrogen-substituted TQ analog (C9H11NO2) (Figure 5) The observed m/z value of 166.08 indicates that the amine group replaced the methyl group. The addition of an amine group in place of oxygen would produce an ion with a m/z of 166.12; however, this was not recorded. The nitrogen substitution occurred very fast because the metabolite was observed at the initial incubation time. Previously, this TQ derivative was obtained by Johnson-Ajinwo et al. (2018) by 2 h incubation at room temperature in the presence of diethylamine/ethylamine, water, and methanol (Johnson-Ajinwo et al., 2018). In our study, TQ was added to bacterial broth containing monosodium glutamate, which can be a source of organic nitrogen (Dubos and Middlebrook, 1947). The sample inoculation with bacteria resulted in the formation of an isomeric TQ analog, which was observed at the other retention time (Figure 5). This suggests that bacteria are able to change the position of the amine group in the TQ structure, converting 2-amino-5-(propan-2-yl)cyclohexa-2,5-diene-1,4-dione into 2-(1-aminoethyl)-5-methylcyclohexa-2,5-diene-1,4-dione. Nevertheless, since after 24 h the concentration of TQ in the test sample was more than 97% of its initial value, we can assign the observed metabolic perturbations to TQ influence. However, we also cannot exclude any possible activity of its detected nitrogen derivatives.