Untargeted and Targeted Liquid Chromatography‐Mass Spectrometry–Based Lipidomic Profiling Revealed a Potential Biomarker Panel to Distinguish Clinical Isolates of M. tuberculosis From Nontuberculous M. kansasii

Abstract

The rising incidence of nontuberculous mycobacteria (NTM) infections, particularly Mycobacterium kansasii (M. kansasii), and their overlapping clinical features with Mycobacterium tuberculosis (MTB) are leading to diagnostic ambiguity in tuberculosis‐endemic regions. Accurate differentiation remains limited by conventional diagnostic methods, emphasizing the need for molecular‐ and lipid‐based biomarkers. In the present era of “OMICS” sciences, herein we attempted a comprehensive analysis of MTB and M. kansasii lipid profiles to elucidate species‐specific lipidomic features as potential biomarkers. Three clinical isolates each of MTB and M. kansasii were obtained from treatment‐naïve patients with microbiologically confirmed pulmonary infections. Lipid profiling was performed by integrating untargeted and targeted profiling through ultraperformance liquid chromatography coupled with tandem mass spectrometry from bacilli total lipid and mycobacterial cell wall lipid extracts. Untargeted profiling exhibited a higher abundance of fatty acyls, GLs, selected GPLs (PE, PG, PI, and PA), and saccharolipids (SLs) specifically Ac2SGL in MTB, whereas M. kansasii isolates were characterized by elevated levels of GPLs like PC and LPC, polyketides, and specific SLs such as diacylated trehalose species. Targeted quantification confirmed differential expression of GL (TG and DG) and GPL (PC, PE, LPC, PI, and PS) species, supporting their diagnostic relevance. Additionally, biomarker analysis further identified five lipid species with strong discriminative potential. Collectively, these findings support the development of a robust lipidomic biomarker panel for the accurate differentiation of MTB and M. kansasii, with potential implications for improved diagnostics and targeted therapeutic strategies after further confirmation.

1. Introduction

Tuberculosis (TB), mainly caused by Mycobacterium tuberculosis (MTB), remains one of the deadliest communicable diseases and is among the top 10 causes of death worldwide by 2023 [1. Nontuberculous Mycobacteria (NTM) are a heterogeneous group comprising more than 169 species of environmental organisms found in soil, water, dust, and diverse animals, which can lead to opportunistic infections. The clinical presentation of NTM infections often overlaps with that of MTB, complicating diagnosis and treatment strategies. A further complex attribute of NTM species from MTB is evidenced by the fact that several presumed TB cases have been shown to be due to NTM [2, 3]. Among clinically relevant NTM, Mycobacterium kansasii (M. kansasii) is notable for its high pathogenic potential and TB‐like pulmonary manifestations [4]. Current diagnostic approaches, which primarily rely on Ziehl–Neelsen (ZN) staining and acid‐fast bacilli (AFB) culture, lack the specificity to reliably differentiate between MTB and NTM, thereby contributing to the underdiagnosis and underreporting of NTM infections in TB‐endemic regions. This diagnostic ambiguity underscores the urgent need for reliable molecular or biochemical markers capable of distinguishing between these clinically similar but biologically distinct mycobacterial groups.

A defining feature of the mycobacterial genus is its complex, lipid‐rich cell envelope, which accounts for up to 60% of its dry mass. This envelope comprises various structurally diverse lipid components including mycolic acids (MAs), sulfolipids, phosphatidylinositol (PI), phosphatidylinositol mannosides (PIMs), lipomannan (LM), lipoarabinomannan (LAM), and trehalose‐containing glycolipids, all of which contribute to cell wall integrity, virulence, and immune modulation [5, 6]. Although MTB and NTM share several of these lipid components, they also exhibit species‐specific lipid signatures. For instance, sulfoglycolipids (SL‐I) and polyacyltrehaloses (PATs) are restricted to the MTB complex and play key roles in immune evasion and intracellular survival. Conversely, M. kansasii and other NTM species uniquely express glycopeptidolipids, such as C‐mycosides, which are linked to biofilm formation and modulation of host immune responses [7, 8].

Previous lipidomic studies, including those by Layre et al. [9, 10], Sartain et al. [8], and Gonzalo et al. [7, 11], have provided valuable insights into mycobacterial lipid diversity but were largely limited in scope and depth. Layre et al. [9, 10] established a chemotaxonomic classification framework for differentiating Mycobacterium tuberculosis complex (MTBC) members based on lipid composition and identifying several virulence‐associated metabolites. However, their analyses were qualitative and restricted to MTBC species. Sartain et al. [8] developed the Mtb LipidDB for M. tuberculosis H37Rv using qualitative lipidomics, yet this work focused on a single strain without interspecies comparison and quantitative validation. In the present study, we utilized this database through MS‐LAMP software for accurate annotation and verification of MTB‐specific lipid masses from our IDA‐based acquired lipidomic data. Gonzalo et al. [7] analyzed clinical MTBC isolates and selected NTM species (M. abscessus, M. avium, M. bohemicum, M. celatum, M. fortuitum, M. interjectum, M. intracellulare, M. kansasii, M. marinum, M. mucogenicum, M. nonchromogenicum, M. peregrinum, M. phlei, M. shimoidei, M. simiae, M. smegmatis, and M. xenopi). They demonstrated lipid fingerprint patterns for rapid diagnostic differentiation based on dominant lipid masses. Their subsequent study [10] further refined MTBC identification but excluded NTMs altogether. Although their work established rapid diagnostic differentiation, it was based on a few dominant mass features rather than the entire lipidome and lacked both quantitative assessment and mechanistic interpretation of lipid variation.

More recently, Wang et al. [12] conducted a serum‐based lipidomic and proteomic comparison of patients infected with MTB and NTM, demonstrating differential host lipid responses. Although the lipid composition of MTB has been well studied, the lipid biology of NTM species, particularly M. kansasii, remains underexplored [5]. Furthermore, direct comparative analyses integrating untargeted and targeted LC‐MS–based lipidomics between MTB and NTM species such as M. kansasii are scarce. To date, systematic integration of untargeted and targeted LC‐MS/MS lipidomics encompassing both total and cell wall lipid (CWL) fractions with MTB and M. kansasii has not been reported.

Given the structural and functional diversity of mycobacterial lipids, major lipid classes viz. fatty acyl (FA), glycerolipid (GL), glycerophospholipid (GPL), polyketide (PK), prenol lipid (PR), and saccharolipid (SL) were analyzed to capture the full spectrum of metabolic and physiological adaptations, distinguishing MTB and M. kansasii. Profiling these lipid classes through both untargeted and targeted LC‐MS/MS enabled a comprehensive evaluation of lipid‐based metabolic signatures, ensuring that observed differences reflect true biological variation rather than analytical bias.

Addressing this gap, the present study integrates high‐resolution untargeted and targeted LC–MS/MS lipidomics to achieve comprehensive qualitative and quantitative profiling of clinical isolates of MTB and M. kansasii. The objective was to elucidate distinct lipidomic signatures, identify differentially expressed lipid species, and explore their potential as biomarkers for species discrimination and diagnostic development. The findings advance current understanding of mycobacterial lipid biology, offering mechanistic insight into lipid‐mediated pathogenesis and providing a foundation for developing lipid‐based diagnostic and chemotaxonomic tools.

2. Methodology

2.1. Material and Chemicals

Middlebrook 7H9 broth (BD Biosciences, United States), oleic acid/albumin/dextrose/catalase (OADC) (BD Biosciences, United States), Tween‐80 (Sigma‐Aldrich, United States), glycerol (Fisher Scientific), potassium chloride (KCl) (Fisher Scientific), chloroform (CHCl₃) (Fisher Scientific and CHROMASOLV, Honeywell), methanol (CH₃OH) (Fisher Scientific and CHROMASOLV, Honeywell), n‐hexane (C₆H₁₄) (Fisher Scientific), acetonitrile (C₂H₃N) (CHROMASOLV, Honeywell), 2‐propanol (C₃H₈O) (CHROMASOLV, Honeywell), ammonium acetate (Honeywell, Fluka), Teflon capping glass vials (Borosil), Pasteur pipette, and Whatman No. 1 filter paper (Merck). SPLASH LIPIDOMIX Mass Spec Standard and natural standards viz. 1,2‐dimyristoyl‐sn‐glycerol (DAG), 1,2‐diheptadecanoyl‐sn‐glycero‐3‐phosphocholine (PC), 1‐heptadecanoyl‐2‐hydroxy‐sn‐glycero‐3 phosphocholine (LPC), 1,2‐dimyristoyl‐sn‐glycero‐3 phosphoethanolamine (PE), 1‐heptadecanoyl‐2‐(9Z‐tetradecenoyl)‐sn glycero‐3‐phospho‐(1 ^′‐myo‐inositol (PI), and 1,2‐dimyristoyl‐sn‐glycero‐3‐phospho‐L serine (PS) are from Avanti Polor, United State; Tripentadecanoin (TAG) natural standard from Sigma‐Aldrich, United States.

2.2. Clinical Sample Processing and Culture Conditions

Sputum samples were aseptically collected from patients with confirmed mycobacterial infections who had not yet started antitubercular or antibiotic therapy/treatment‐naive patients with microbiologically confirmed pulmonary mycobacterial infections. Diagnosis was established by ZN staining, AFB culture, and GeneXpert positivity. Mycobacteria were cultured from these samples under BSL‐3 conditions and considered as clinical isolates. Three independent clinical isolates each of MTB and M. kansasii were taken for this study.

Cultures were inoculated into a 30‐mL Middlebrook 7H9 medium supplemented with 0.05% Tween‐80, 0.2% glycerol, and 10% OADC enrichment and incubated at 37°C with agitation until the exponential phase (1 OD₆₀₀ cells) was reached. Bacterial cells were carefully harvested and stored at −20°C for further processing.

2.3. Extraction of Total Lipids (TL)

The modified Folch method was used for TL extraction of Mycobacterium cells [13]. Cells were homogenized by sonication in an aqueous solution for 3 min, with three cycles of 1‐min intervals, and mixed with CHCl₃ and CH₃OH in a 1:2 ratio. A total of 10 μL of Splash mix standard was added in each sample, and the solution was centrifuged at 2000 rpm at 4°C for 10–15 min. The supernatant was transferred to another glass vial, and the remaining CHCl₃ was added to obtain a final volume of CHCl₃ and CH₃OH in a ratio of 1:1. After filtration with Whatman No.1 filter paper, nonlipid contamination was removed by saturating the extract with 0.88% KCl. The lower layer was taken using a Pasteur pipette in a 5‐mL glass vial with Teflon capping and dried with liquid nitrogen, then stored at −20°C for further analysis.

2.4. Extraction of CWL

CWL were extracted using the modified Minnikin method, as described previously [14]. The harvested cells were mixed with 3 mL of hexane and kept overnight in the dark at room temperature. The next day, the suspension was centrifuged at 2500 rpm for 10 min. The upper layer was collected in a glass vial and 10 μL of Splash mix standard was added in each sample. After that, 3 mL of chloroform and after 30 min, 3 mL of methanol was added and kept at room temperature for 2 h. The upper hexane layer containing CWL was transferred to the glass vial and dried under gentle liquid nitrogen flux and then stored at −20°C for further use.

2.5. Electrospray Ionization Mass Spectrometry (ESI‐MS/MS)

The biological replicates were obtained from three TB patients, and each isolate has been analyzed in three independent technical replicates to ensure analytical reproducibility and precision in ESI‐MS/MS measurements.

2.5.1. Untargeted Mass Spectrometry

For information‐dependent acquisition (IDA), TL and CWL samples were reconstituted in CHCl₃ and CH₃OH in 1:1 and 9:1 ratios, respectively, and analyzed by UHPLC‐MS/MS (ultrahigh–performance triple quadruple tandem mass spectrometer), that is, AB Sciex 4500 QTRAP MS/MS equipped with a Sciex Turbo Spray ion source, in both positive and negative modes. UHPLC was done on the Agilent C18 column (100 × 2.1 mm, 1.8 μm) through gradient mode of elution for 41 min using Buffer A comprised of two solutions (Solution 1: water/acetonitrile (ACN) 95/5, v/v was prepared by mixing 5 mM of ammonium acetate; Solution 2, Solution 1/methanol, 90/10 v/v); and buffer B ACN/methanol/IPA 85/10/5, v/v/v at a flow rate of 0.3 mL/min. The source temperature was 500°C, the ion‐spray voltage (IV) was set to 5.50 kV, ion‐source Gasses 1 and 2 were set to 50, curtain gas was 30, and collision gas was set on the medium. 2 μL of lipid sample was injected for each run, and the data were recorded in the 200–2000‐m/z mass range. Further, the analysis was done by using the mass spectrometry‐based lipid(ome) analyzer and molecular platform (MS‐LAMP) software [13, 15].

2.5.2. Targeted Mass Spectrometry

2.5.2.1. Optimization of ESI MS/MS Conditions for Absolute Quantification of GL and GPL

For optimization, direct infusion of internal and natural standards has been done by the Harvard syringe pump in conjunction with the UHPLC‐MS/MS. Internal standard SPLASH and natural standards viz. TAG (Tripentadecanoin), DAG (1,2‐dimyristoyl‐sn‐glycerol), PC (1,2‐diheptadecanoyl‐sn‐glycero‐3‐phosphocholine), LPC (1‐heptadecanoyl‐2‐hydroxy‐sn‐glycero‐3 phosphocholine), PE (1,2‐dimyristoyl‐sn‐glycero‐3 phosphoethanolamine), PI (1‐heptadecanoyl‐2‐(9Z‐tetradecenoyl)‐sn glycero‐3‐phospho‐(1 ^′‐myo‐inositol)), and PS (1,2‐dimyristoyl‐sn‐glycero‐3‐phospho‐L serine) have been optimized. The flow rate for the syringe pump was maintained between 20 and 30 μL/min, and the MS was configured to scan Q1 in the m/z range of 100–900 Da to obtain the mass spectrum of all the standards in positive ion modes.

The UHPLC gradient optimization was done on kinetex C18 column 100Ǻ (50 × 2.1 mm, 1.7 μm) with gradient mode of elution. To separate targeted transition pairs, Buffer A comprises water with 0.5% formic acid and 10 mM of ammonium acetate, whereas Buffer B has methanol with 0.5% formic acid and 10 mM of ammonium acetate with a 0.3‐mL/min flow rate employed. A 30‐min run was performed for GLs (TG and DG), and 20 min for GPLs (PC, Lyso‐PC (LPC), PE, PI and PS) were applied. A total of 196 lipid species (125 for GL and 71 for GPL) were targeted for quantitative analysis. Oven and autosampler temperatures were adjusted to 60°C and 10°C, respectively. The source temperature was 500°C, the IS was set to 4.50 kV, ion‐source Gasses 1 and 2 were set to 50, curtain gas was 35, and collision gas was put on the medium. A total of 10 μL of each lipid sample was injected by using an autosampler. Data were acquired in positive polarity in the 200–1000‐Da mass range. The UHPLC‐MS/MS was done in the multiple reaction monitoring (MRM) mode, with the detection of the transition pairs of the individual protonated parent ions and their common daughter ions. Software Analyst v1.6.3 was used for data acquisition in wiff format.

2.5.2.2. Quality Control Measurements

Natural standards of DG, TG, PC, LPC, PE, PI, and PS, along with a pooled sample matrix, were prepared and analysed for quality control. The individual standards and mycobacterial test samples exhibited well‐resolved and distinct peaks, enabling absolute quantification of the targeted lipid species, including GL and GPL. Furthermore, the limit of detection (LOD) and limit of quantification (LOQ) were determined to validate the performance of the developed UHPLC‐MS/MS method. To assess the methodological LOD and LOQ, internal and natural standards were serially diluted within the range of 0.98–1000 ng/mL, and regression analysis was performed using Microsoft Excel.

2.6. High‐Throughput Data Analysis

2.6.1. Untargeted Data Analysis

The MS‐LAMP software were used for interpreting IDA lipid data. In the current version of the program, two databases are incorporated based on lipid databases: “M. tb Lipidome MS‐LAMP” and “General Lipidome MS‐LAMP” [15]. The LC‐MS data were acquired from the Analyst software in text format. Data sorting was done manually in .xls format where the common masses from all three replicates with a cutoff of 10,000 (E ⁴) intensity were put into “Mtb LipidDB” of MS‐LAMP to find mycobacterial specific lipid masses for both positive and negative modes. The mass window range for each search was set to 0.25 and the m/z values in the mass spectra were attributed to singly protonated ions, that is, [M⁺H]⁺ and [M⁻H]⁻, exclusively. The masses identified from both positive and negative ionization modes were combined for further analysis.

2.6.2. Targeted Data Analysis

Matrix‐matched calibration was performed independently for TL and CWL extracts to account for matrix effects and ensure accurate quantification. For this, all TL extracts were pooled to prepare a representative TL‐specific matrix, and all CWL extracts were pooled separately to prepare a CWL‐specific matrix.

A pooled natural standard mixture was prepared using known concentrations of respective TG and DG standards for GL quantification and PC, LPC, PE, PI, and PS standards for GPL quantification. These were serially diluted and subsequently spiked into the respective TL and CWL matrices prior to LC‐MS analysis.

Separate LC‐MS runs were conducted for GL and GPL standard calibration series within each matrix type (TL and CWL), generating independent calibration curves for every lipid subclass. The acquired data were integrated with MultiQuant software (Sciex) v3.0, where each identified peak was smoothened and the background was subtracted. The integrated data were then exported to MS Excel for further statistical analysis. Calibration curves were prepared through internal to natural standard ratios followed by matrix subtraction, ensuring high quantitative accuracy and reproducibility across both TL and CWL datasets. The calibration curves, established for TL and CWL matrices, respectively, for each lipid subclass, each exhibiting excellent linearity (R ² = 0.99). These matrix‐specific calibration curves were used to calculate absolute lipid concentrations, ensuring quantitative robustness and cross‐matrix comparability.

2.7. Multivariate Statistical Analysis

Untargeted sorted and processed data were uploaded to MetaboAnalyst 6.0 for univariate, multivariate, hierarchical cluster, and biomarker analysis and validation, as explained in Fatima et al. [14]. Student′s t‐test (p ≤ 0.05) was applied for univariate analysis, whereas principal component analysis (PCA) and partial least squares‐discriminant analysis (PLS‐DA) were performed for multivariate evaluation. Hierarchical clustering was visualized through dendrograms and heat maps, and biomarker screening was conducted using receiver operating characteristic (ROC) curve analysis. Venn diagrams were generated using Venny 2.1, and graphical representations (pie, 3D column, and doughnut charts) were prepared in MS Excel. Statistical robustness and model reliability were supported by high‐multivariate validation metrics (R ² > 0.9; Q ² > 0.7), significance thresholds (p ≤ 0.05, V I P ≥ 1.0), and ROC‐based validation with tenfold cross‐validation to minimize model overfitting, confirming model stability and predictive accuracy.

3. Results

3.1. Untargeted Lipid Profiling Reveals a Distinct Distribution of Lipid Species in MTB and M. kansasii Clinical Isolates

The untargeted IDA lipidome profile analysis, Figure S1 shows typical chromatograms achieved during data acquisition of MTB and M. kansasii samples in positive and negative ion modes. The overall identified m/z values ranged from 303.12 to 1820.88 for MTB and 301.2 to 1688.76 for M. kansasii. The distribution of these values across the six lipid categories is detailed in Table S1. The analysis of lipid categories was conducted using the lower 0.25 window range of MS‐LAMP to avoid nonspecific identification and to maintain the precision of data and its interpretation. The processed IDA data from the MTB and M. kansasii isolates were statistically analyzed using MetaboAnalyst 6.0 for distinction and grouping. The data exhibited distinct MTB and M. kansasii lipid profiles and provided a comprehensive overview of the lipid categories at the molecular level, as mentioned in the subsections below.

3.1.1. Covariance of TL and CWL Through Multivariate Statistical Analysis Between MTB and M. kansasii Isolates

Using unsupervised PCA and PLS‐DA, we evaluated the covariance in the TL extracts of MTB and M. kansasii isolates. PCA demonstrated an explained variance of 89.4% for PC1 and 8.3% for PC2, accounting for 97.7% of the total variance (Figure 1a). PLS‐DA showed a covariance of 89.2% for PC1 and 5.1% for PC2, summing to 94.3% of the total covariance (Figure 1b). The Venn diagram revealed 39% unique lipids identified in MTB isolates, whereas 26.2% in M. kansasii isolates and 34.8% common to both the isolates (Figure 1c). The list of identified lipids in the Venn diagram with their scientific names is shown in S. Sheet 1. Figure 1d illustrates the percentage distribution of identified lipids in six lipid categories viz. FA, GL, GPL, PK, PR, and SL within MTB and M. kansasii, respectively. FA, GL, and GPL constituted the primary lipid categories, accounting for 95% of all identified lipids in both MTB and M. kansasii. All lipid categories were abundantly present in MTB isolates than in M. kansasii isolates, whereas PR lipids were exclusively present in MTB isolates (Figure 1e). The list of identified lipids in the Venn diagram, with their scientific names, is shown in S. Sheet 1. Further, the lipid distribution pattern was analyzed within the lipid categories through a heat map and cluster dendrogram, as shown in Figure 1f,g, respectively. The dendrogram illustrates the lipid profile based on the hierarchical relationship between MTB and M. kansasii clinical isolates. In the TL extracts of bacilli, two replicates of MTB and M. kansasii were closely related, making the differentiation less distinct.

Figure 1.

Multivariate and statistical analyses of TL extracts from MTB and NTM (M. kansasii) clinical isolates. Score plots of PCA (a) and PLS‐DA (b) showing partial overlap yet distinct clustering trends between MTB (red) and M. kansasii (green), indicating compositional divergence in TL lipid profiles. (c) Venn diagram showing exclusive and shared lipids identified across both isolates. (d) Pie charts represent the relative percentage distribution of identified major lipid categories in MTB and NTM (M. kansasii) isolates. (e) Venn diagrams showing unique and common lipids among lipid categories, that is, fatty acyls (FA), glycerolipids (GL), glycerophospholipids (GLP), polyketides (PK), prenol lipids (PR), and saccharolipids (SL). (f) The clustering heat map generated from MetaboAnalyst 6.0 indicating relative abundance patterns of lipid categories in MTB and NTM (M. kansasii) isolates, where brown color indicates higher and blue color indicates low abundance. (g) Hierarchical clustering dendrogram demonstrating clear species‐level grouping and intraspecies similarity based on TL profiling.

Similarly, the TL extract, unsupervised PCA and PLS‐DA were performed to evaluate the covariance of the first five components in the CWL extract of MTB and M. kansasii isolates. PCA revealed an explained variance of 69% for PC1 and 21.6% for PC2, accounting for 90.6% of the total variance (Figure 2a). PLS‐DA showed a covariance of 68.6% for PC1 and 8.2% for PC2, summing to 76.8% of the total covariance (Figure 2b). The Venn diagram represented 28.9% unique lipids identified to MTB isolates conversely 34.6% in M. kansasii isolates, and 36.5% common to both isolates (Figure 2c). The list of identified lipids in Venn diagram with their scientific names is appended in S. Sheet 2. The percentage distribution of all six lipid categories in MTB and M. kansasii is illustrated in Figure 2d, where FA, GL, and GPL constituted 96% of identified lipids in MTB and 94% in M. kansasii. MTB isolates contained a higher percentage of FA than the M. kansasii isolates, whereas other lipid categories, viz. GL, GPL, PK, PR, and SL were higher in M. kansasii than MTB isolates (Figure 2e). The list of identified lipids in the Venn diagram with their scientific names is shown in S. Sheet 2. Additionally, a heat map (Figure 2f) and cluster dendrogram (Figure 2g) were utilized to identify patterns and relationships between MTB and M. kansasii isolates based on their lipid compositions. Figure 2f also shows that FAs were elevated in MTB isolates, whereas GL, GPL, PK, PR, and SL lipids were high in M. kansasii isolates. Unlike the bacilli TL extract, CWL extract displayed clear differentiation between MTB and M. kansasii isolates as depicted in Figure 2g. Interestingly, the hierarchical clustering dendrogram generated by Metaboanalyst 6.0 shows distinct clustering of MTB and M. kansasii isolates based on the identified lipid moieties.

Figure 2.

Multivariate and statistical analyses of CWL extracts from MTB and NTM (M. kansasii) clinical isolates. Score plots of PCA (a) and PLS‐DA (b) showing distinct species‐level grouping based on cell wall lipid composition. (c) Venn diagram showing exclusive and shared lipids identified in MTB and NTM (M. kansasii) isolates. (d) Pie charts representing the relative percentage distribution of identified major lipid categories in cell wall lipid extracts of both isolates. (e) Venn diagrams showing unique and common lipids among lipid categories, that is, fatty acyls (FA), glycerolipids (GL), glycerophospholipids (GLP), polyketides (PK), prenol lipids (PR), and saccharolipids (SL). (f) The clustering heat map generated from MetaboAnalyst 6.0 indicates the relative abundance pattern of lipid categories, where brown color indicates higher and blue color indicates low abundance. (g) Hierarchical clustering dendrogram demonstrating distinct species‐specific grouping, confirming compositional divergence between MTB and M. kansasii based on CWL profiling.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

3.1.2. Abundance of Lipid Subclasses in TL and CWL Extracts of MTB and M. kansasii Isolates

Figure 3a illustrates the comparative abundance of various lipid subclasses in the TL extract of MTB and M. kansasii isolates, whereas Figure 3b presents various lipid subclasses for the CWL extract. Table S2 provides a list of the identified numbers of these lipid subclasses in TL and CWL extracts of MTB and M. kansasii isolates. In the FA lipid category of TL extract, MTB exhibits higher levels of branched fatty acids, keto‐MA, methoxy‐MA, alpha‐MA, DIMA, TMM, and DIMB compared with M. kansasii, whereas GMM was equally present in both isolates. Within the GL category, MG, DG, and TG subclasses were higher in MTB compared with M. kansasii. For the GPL category, MTB isolates showed higher levels of Lyso‐PG, Lyso‐PI, Lyso‐PIM1, PG, CL, PI, Ac1PIM1, and Ac1PIM3 than M. kansasii; Lyso‐PIM3, Lyso‐PIM4, Lyso‐PIM5, Lyso‐PIM6, PIM5, and Ac2PIM2 were exclusive to MTB isolates; conversely, M. kansasii isolates had marginally higher levels of Lyso‐PE, PE, PIM1, and PIM2, whereas Lyso‐PIM2, PIM3, PIM4, and Ac1PIM2 were equally distributed in both isolates. In the PK lipid category, mannosyl‐β1‐phosphomycoketides were found only in MTB, whereas nonribosomal peptides/PK hybrids were high in M. kansasii isolates. Ubiquinones were marginally higher in MTB in the PR category, and bactoprenols were exclusive to M. kansasii. In the SL category, Ac2SGL was prevalent in MTB, whereas DAT1 and DAT2 were equally distributed in both isolates.

The CWL extract of MTB showed higher levels of branched FA, methoxy‐MA, DIMA, TMM, and GMM compared with M. kansasii; alpha‐MA was significantly abundant in M. kansasii, whereas keto‐MA and DIMB were equally present in both isolates. In the GL category, MG was higher in MTB, TG was higher in M. kansasii, and DG was equally present in both isolates. For the GPL category, MTB isolates exhibited marginally higher levels of Lyso‐GP, Lyso‐PI, Lyso‐PIM1, Lyso‐PIM5, PI, PIM1, Ac1PIM2, and Ac1PIM3 than M. kansasii; M. kansasii showed higher levels of Lyso‐PE, PG, PIM2, PIM3, and Ac1PIM1, whereas Lyso‐PIM4, PIM4, PIM5, and Ac1PIM4 were exclusive to M. kansasii, whereas Lyso‐PIM2, Lyso‐PIM3, CL, PE, and Ac2PIM2 were equally present in both isolates. In the PK lipid category, mannosyl‐β1‐phosphomycoketides were marginally higher in MTB, whereas nonribosomal peptides/PK hybrids were abundant in M. kansasii isolates. For the PR category, ubiquinones were slightly higher in M. kansasii, whereas bactoprenols and bactoprenol diphosphates were found only in MTB. In the SL category, Ac2SGL was equally present in both isolates, whereas DAT1 and DAT2 were exclusive to M. kansasii.

3.1.3. Distribution of Identified Unique and Common Lipid Species in TL and CWL Extracts of MTB and M. kansasii Isolates

The distribution of identified lipid molecules in MTB and M. kansasii isolates was analyzed through the Venn diagram (Figure 4). In the TL extract, 13.7% of lipids were unique to MTB, whereas 9.5% were unique to M. kansasii, and 4.4% were shared between the clinical isolates. In the CWL extract, 10.1% of lipids were exclusive to MTB, whereas 11.2% were exclusive to M. kansasii, and 2.3% were common to both the clinical isolates. Additionally, 5% of lipids were shared between MTB TL and CWL extracts, whereas 4.2% were shared between M. kansasii TL and CWL extracts. A total of 14.9% of the lipids were common across all TL and CWL extracts. Further, 6% of lipids were shared between M. kansasii CWL and MTB TL extracts, 2.1% between MTB′s CWL and M. kansasii′s TL extracts, 3.7% among MTB and M. kansasii′s TL extracts and MTB′s CWL extracts, 3.6% among MTB and M. kansasii′s TL extracts and M. kansasii′s CWL extracts, 5% among MTB and M. kansasii′s CWL extracts and MTB′s TL extract, and 4.2% among MTB and M. kansasii′s CWL extracts and M. kansasii′s TL extract. The list of identified lipids in the Venn diagram with their scientific names is enclosed in S. Sheet 3.

Figure 4.

A comparative Venn diagram of TL and CWL extracts from MTB and NTM (M. kansasii). The four‐way Venn diagram shows the distribution of unique and shared lipid numbers with percentages across TL and CWL extracts of both isolates.

3.2. MRM‐Based Targeted Lipidomics Exhibits Absolute Quantification of GL and GPL Lipid Species Among MTB and M. kansasii Isolates

The data achieved from untargeted lipid profiling indicated significant variations in GL and GPL lipid classes. Therefore, the quantification of TG and DG molecular species from the GL lipid category and PC, LPC, PE, PI, and PS from the GPL lipid category was done through MRM‐based targeted lipid profiling. The optimized LC‐MS parameters for internal standard Splash and their respective natural standards, viz. TG, DG, PC, LPC, PE, PI, and PS are mentioned in Table S3. The Q1 was scanned in the m/z range of 100–900 Da to obtain the mass spectrum of all the standards in positive ion modes (Figure S2a). The MS2 scan of each parent ion was performed to obtain the optimum fragmentation pair, as shown in Figure S2b. The XICs for all standards are illustrated in Figure S2c. The LOD and LOQ ranged from 45.76 to 109.86 ng/mL and 138.66 to 332.99 ng/mL, respectively. LOD and LOQ for individual natural standards are provided in Table S4. For the absolute quantification of various GL and GPL lipid species from TL and CWL extracts, we incorporated 125 transition pairs for GL, including 42 for TG and 83 for DG, as well as 71 transition pairs for GPL, consisting of 18 for PC, 10 for LPC, seven for PE, 24 for PI, and 12 for PS molecular species. These transition pairs are detailed, along with their respective lipid IDs, in Table S5. However, in the TL extract, a total of 99 GL lipid molecules were successfully quantified, comprising 41 TG and 58 DG, along with 65 GPL lipid species, including 18 PC, eight LPC, seven PE, 23 PI, and nine PS. In contrast, the CWL extract yielded 71 GL lipid species, consisting of 27 TG and 44 DG, and 29 GPL lipid species, including 11 PC, nine LPC, seven PE, and two PI. The absolute concentrations of these lipid species, expressed in ng/mL of 1 OD cells, are provided in Table S6. The percentage distribution of differentially expressed GL and GPL lipid species in TL and CWL extracts from both clinical isolates was analyzed and illustrated through pie charts (Figure S3). Figure S4a,b illustrates the relative abundance in the percentage of all quantified GL and GPL lipid species in MTB and M. kansasii clinical isolates among TL and CWL extracts, respectively.

3.2.1. Distribution of Quantified GL and GPL Lipid Species in TL and CWL Extracts of MTB and M. kansasii Isolates, Along With Differential Expression

Further, the p values of all quantified lipids were calculated and those lipids having p values of ≤ 0.05 were only considered as statistically significant. Table 1 provides the absolute concentration of significant lipid molecular species from GL and GPL categories. The percentage distribution of significantly differentially expressed GL and GPL lipid molecules in TL and CWL extracts from both MTB and M. kansasii clinical isolates is illustrated in doughnut charts (Figure 5a,b). TL extracts exhibited a high percentage of DGs followed by TG, PC, PE, PS, PI, and LPC, whereas CWL extracts also showed the highest percentage of DG than TG, PE, and LPC. The PC and PI lipid molecules were not significantly identified. Figure 5c,d illustrates the percentage distribution of significantly quantified GL and GPL lipid species in both MTB and M. kansasii clinical isolates. In TL extract, a total of 78, including 33 TG and 45 DG lipid species in the GL lipid category and 33 GPL lipid species comprising 11 PC, four LPC, seven PE, five PI, and six PS were found significant. In contrast, a sum of 22 GL having seven TG and 15 DG lipid species and eight GPL including three LPC, and five PE, having p values of ≤ 0.05 in CWL extract. The data in Figure 5 revealed that TL extracts contained a higher number of significant GL and GPL lipid moieties compared with the CWL extracts.

Table 1.

Absolute concentrations of quantified GL and GPL lipid species based on ≤ 0.05 p values.

(A) TL extract
Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)	Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)	Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)
TG (42:0/FA14:0)	3.55	1.26	DG 12:0–12:0	0.63	0.45	PC 24:0	0.02	0.28
TG (42:0/FA16:0)	6.35	1.63	DG 12:0–14:0	0.55	0.32	PC 30:0	1.26	0.64
TG (44:0/FA14:0)	3.85	0.99	DG 12:0–14:0	0.49	0.27	PC 32:0	4.01	1.61
TG (44:0/FA16:0)	5.96	1.18	DG 12:0–16:0	0.24	0.14	PC 33:0	0.43	0.25
TG (44:0/FA18:0)	2.91	0.64	DG 12:0–18:4	0.02	0.01	PC 34:0	7.07	0.38
TG (46:0/FA14:0)	3.32	0.75	DG 12:0–20:0	0.37	0.24	PC 34:1	69.88	5.05
TG (46:0/FA16:0)	5.76	1.21	DG 12:0–22:0	0.20	0.12	PC 35:1	16.25	1.51
TG (46:0/FA18:0)	2.27	0.44	DG 12:0–22:6	0.87	1.66	PC 36:0	1.24	0.90
TG (47:1/FA14:0)	0.63	0.12	DG 14:0–14:0	0.21	0.10	PC 36:1	13.63	9.87
TG (47:1/FA16:0)	1.03	0.23	DG 14:0–14:1	0.05	0.02	PC 38:1	0.81	0.55
TG (47:1/FA16:1)	1.15	0.26	DG 14:0–16:0	0.38	0.23	PC 40:1	0.06	0.05
TG (47:1/FA17:0)	0.29	0.06	DG 14:0–16:1	0.16	0.34	LPC 14:0	0.30	0.03
TG (47:1/FA18:1)	0.69	0.13	DG 14:0–18:0	0.45	0.20	LPC 17:0	0.86	4.95
TG (48:0/FA14:0)	1.76	0.38	DG 14:0–20:0	0.16	0.08	LPC 20:5	1.45	6.40
TG (48:0/FA16:0)	6.23	1.36	DG 14:0–22:0	0.08	0.04	LPC 22:0	0.04	0.15
TG (48:0/FA18:0)	2.73	0.61	DG 14:0–22:6	1.04	1.21	PE 32:0	25.13	13.94
TG (49:2/FA14:0)	0.10	0.02	DG 14:1–18:0	0.11	0.21	PE 33:1	24.57	16.94
TG (49:2/FA16:0)	0.35	0.07	DG 14:1–20:0	0.96	0.04	PE 34:0	21.31	5.38
TG (49:2/FA16:1)	0.86	0.18	DG 14:1–22:6	0.03	0.02	PE 35:1	55.14	13.35
TG (49:2/FA17:0)	0.18	0.03	DG 16:0–16:0	7.72	5.25	PE 36:1	7.05	2.86
TG (49:2/FA18:1)	0.62	0.11	DG 16:0–16:1	0.73	0.41	PE 37:2	16.73	1.62
TG (49:2/FA18:2)	0.31	0.06	DG 16:0–18:0	4.85	2.60	PE 38:1	0.44	0.04
TG (50:0/FA14:0)	0.58	0.12	DG 16:0–20:0	0.34	0.22	PI_26:0	0.68	0.44
TG (50:0/FA16:0)	4.01	0.85	DG 16:0–20:5	0.50	0.12	PI_34:2	0.28	0.14
TG (50:0/FA18:0)	3.63	0.76	DG 16:0–22:6	1.68	1.01	PI_35:0	0.83	1.23
TG (51:0/FA17:0)	0.08	0.02	DG 16:1–16:1	0.46	0.14	PI_35:2	0.10	0.04
TG (51:0/FA18:0)	0.23	0.06	DG 16:1–18:0	1.59	5.60	PI_38:2	0.02	0.01
TG (51:3/FA16:1)	0.14	0.03	DG 16:1–20:5	0.14	0.03	PS_32:0	0.01	0.02
TG (51:3/FA17:0)	0.02	0.00	DG 16:1–22:0	0.01	0.03	PS_32:1	0.01	0.00
TG (52:0/FA16:0)	1.15	0.27	DG 16:1–22:6	0.09	0.03	PS_34:1	0.02	0.01
TG (52:0/FA18:0)	2.42	0.58	DG 18:0–18:0	3.17	1.99	PS_35:1	0.00	0.00
TG (52:0/FA20:0)	0.09	0.02	DG 18:0–20:0	0.11	0.04	PS_35:2	0.01	0.00
TG (55:7/FA22:6)	0.01	0.00	DG 18:0–22:6	0.91	0.46	PS_36:1	0.00	0.01
			DG 18:1–18:1	5.38	1.53
			DG 18:1–20:0	0.03	0.01
			DG 18:1–22:0	0.03	0.02
			DG 18:1–22:6	1.13	0.50
			DG 18:2–18:2	0.31	0.09
			DG 18:2–22:6	0.12	0.04
			DG 18:3–20:5	0.01	0.01
			DG 18:3–22:6	0.23	0.14
			DG 20:0–20:5	0.06	0.01
			DG 20:0–22:6	0.13	0.03
			DG 20:1–20:5	0.03	0.01
			DG 22:6–22:6	0.41	0.17
(B) CWL extract
Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)	Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)	Analyte (lipid species)	MTB (ng/mL)	NTM (M. kansasii) (ng/mL)
TG (42:0/FA14:0)	37.67	304.30	DG 12:0–14:0	3.20	3.23	LPC 14:0	8.47	1.36
TG (47:1/FA16:0)	5.28	21.61	DG 12:0–14:0	2.02	2.26	LPC 16:0	47.20	177.25
TG (47:1/FA17:0)	1.95	6.18	DG 12:0–16:0	2.98	2.49	LPC 18:0	20.20	51.76
TG (47:1/FA18:1)	3.27	12.04	DG 12:0–20:0	12.51	6.54	PE 32:0	0.73	0.18
TG (49:2/FA14:0)	0.25	1.40	DG 12:0–22:0	5.15	3.68	PE 33:1	0.90	0.18
TG (49:2/FA16:1)	2.72	11.33	DG 14:0–16:1	6.66	3.08	PE 34:0	0.61	0.07
TG (49:2/FA18:2)	0.80	3.80	DG 14:0–20:0	5.88	3.58	PE 35:1	1.75	0.10
			DG 14:0–22:6	4.17	7.25	PE 37:2	0.30	0.03
			DG 16:0–16:0	310.36	162.73
			DG 16:0–16:1	11.35	14.14
			DG 16:0–18:0	324.84	217.13
			DG 16:1–16:1	8.37	5.20

Figure 5.

Distribution of significant quantified lipids of GL (TG and DG) and GPL (PC, LPC, PE, PI, and PS) subclasses in TL and CWL extracts of both isolates. (a) and (b) Doughnut charts show relative distribution percentage of statistically significant (p ≤ 0.05) quantified GL and GPL lipid species in TL and CWL extracts, respectively. Bar graphs (c) and (d) exhibit the comparative abundance of individual TG, DG, PC, LPC, PE, PI, and PS lipid species between MTB and NTM (M. kansasii) isolates in TL and CWL extracts, respectively.

3.3. Screening and Establishment of Lipid Biomarkers of MTB and M. kansasii

Firstly, the data were sorted and processed for both the extracts of MTB and M. kansasii isolates and analyzed by using classical univariate ROC curve analysis to identify biosignatures. This process involved calculating the area under the curve (AUC) and identifying the sensitivity and specificity at the optimal cutoff point based on the Youden index. To strengthen the reliability of our results, we utilized the bootstrap method to determine the 95% confidence interval (CI) for the AUC. An AUC value of 1.0 was considered perfect for both lipid extracts. All statistical analyses were conducted using MetaboAnalyst 6.0 web‐based software, with statistical significance defined as a p value of ≤ 0.05. In TL extracts, a total of 36 lipid molecules were identified comprising six FAs, 13 GLs, and 17 GPLs, whereas 38 lipids screened out from the CWL extracts including 14 FAs, 16 GLs, six GPLs, and two PKs with 1.0 AUC. Among these five lipid species, namely MG (RCO2H = 18 : 1), TG (R1CO2H + R2CO2H + R3CO2H = 50 : 1), TG (R1CO2H + R2CO2H + R3CO2H = 51 : 0), hydroxyphthioceranic acid (C33), and Lyso‐PG (RCO2H = 18 : 1) were found to be common between both TL and CWL extracts with same trend (Figure 6). MG (RCO2H = 18 : 1) was found exclusively in both extracts of MTB isolates, whereas TG (R1CO2H + R2CO2H + R3CO2H = 50 : 1), TG (R1CO2H + R2CO2H + R3CO2H = 51 : 0), hydroxyphthioceranic acid (C33), and Lyso‐PG (RCO2H = 18 : 1) were present only in M. kansasii isolates.

Figure 6.

Box plots and ROC curves displaying discriminative lipid biomarkers identified in TL and CWL extracts of MTB and NTM (M. kansasii). (a) (Panels A, B, and C) TL‐derived lipid biomarkers from the fatty acyl (FA), glycerolipid (GL), and glycerophospholipid (GPL) categories, respectively, each showing perfect discrimination between MTB and M. kansasii (AUC = 1.0; 95% CI). (b) (Panels D, E, F, and G) corresponding CWL biomarkers for FA, GL, GPL, and polyketide (PK) classes, also achieving AUC = 1.0 with 95% confidence intervals, confirming their strong diagnostic potential.

(a)

(b)

Subsequently, targeted lipid profiling was evaluated to assess the differential regulation of quantified lipid species, with statistical significance defined at p ≤ 0.05. In the TL extract, 33 TG and 39 DG lipid species were downregulated, whereas six DG lipid species were upregulated in M. kansasii. Furthermore, 10 PC, one LPC, seven PE, four PI, and four PS lipid species were downregulated, whereas one PC, three LPC, one PI, and two PS lipid species were upregulated in M. kansasii as depicted in Figure 7a. Figure 7b illustrates the lipid changes with p values of ≤ 0.05 in the CWL extract, where seven TG, three DG, and and LPC lipid species were significantly upregulated, whereas 12 DG, one LPC, and five PE lipid species were downregulated in M. kansasii. Nine DGs, one LPC, and five PEs lipid molecular species were observed in both TL and CWL extracts; however, PC, PI, and PS lipid species were exclusively found in the TL extract. A comprehensive list of the differentially expressed lipids in TL and CWL extracts is provided in Table 2. Lipid biomarkers identified with names from untargeted and targeted biomarker analysis are provided in Table 3.

Figure 7.

Histograms depicting differential regulation of significant quantified lipid species from GL and GPL categories in TL and CWL extracts of MTB and NTM (M. kansasii) isolates. Histograms represent the relative percentage of upregulation (pink color) and downregulation (blue color) of significant quantified lipid species from GL and GPL categories in TL (a) and CWL (b) extracts, respectively.

Table 2.

Up and down percentages of quantified lipid species from GL and GPL lipid categories.

(A) TL extract
Analyte (lipid species)	p	Down (%)	Up (%)	Analyte (lipid species)	p	Down (%)	Up (%)	Analyte (lipid species)	p	Down (%)	Up (%)
TG (42:0/FA14:0)	0.00023	64.35		DG 12:0–12:0	0.029352	28.22		PC 24:0	0.000621		93.97
TG (42:0/FA16:0)	0.000586	74.32		DG 12:0–14:0	0.001311	41.05		PC 30:0	0.001168	49.47
TG (44:0/FA14:0)	2.03E‐05	74.32		DG 12:0–14:0	0.003985	44.41		PC 32:0	0.000229	59.94
TG (44:0/FA16:0)	2.95E‐05	80.28		DG 12:0–16:0	0.000361	41.24		PC 33:0	0.005001	41.32
TG (44:0/FA18:0)	5.89E‐05	78.15		DG 12:0–18:4	0.016275	43.44		PC 34:0	0.000435	94.58
TG (46:0/FA14:0)	1.78E‐05	77.43		DG 12:0–20:0	0.035837	35.91		PC 34:1	2.07E‐05	92.77
TG (46:0/FA16:0)	3.85E‐07	78.95		DG 12:0–22:0	0.006121	41.43		PC 35:1	0.000113	90.69
TG (46:0/FA18:0)	9.98E‐06	80.66		DG 12:0–22:6	0.000854		47.74	PC 36:0	0.025692	27.84
TG (47:1/FA14:0)	5.64E‐07	80.74		DG 14:0–14:0	0.00148	52.57		PC 36:1	0.003165	27.60
TG (47:1/FA16:0)	0.000156	77.37		DG 14:0–14:1	0.000356	66.86		PC 38:1	0.005026	31.98
TG (47:1/FA16:1)	3.43E‐05	77.38		DG 14:0–16:0	0.005746	40.62		PC 40:1	0.013603	21.65
TG (47:1/FA17:0)	4.37E‐07	79.93		DG 14:0–16:1	0.000258		52.50	LPC 14:0	0.00375	90.46
TG (47:1/FA18:1)	2.75E‐06	81.61		DG 14:0–18:0	0.000455	55.88		LPC 17:0	0.004508		82.58
TG (48:0/FA14:0)	1.86E‐05	78.35		DG 14:0–20:0	0.010097	49.06		LPC 20:5	0.014907		77.35
TG (48:0/FA16:0)	0.000205	78.20		DG 14:0–22:0	0.001954	54.22		LPC 22:0	0.030272		76.38
TG (48:0/FA18:0)	1.84E‐05	77.65		DG 14:0–22:6	0.037084		13.70	PE 32:0	6.62E‐05	44.53
TG (49:2/FA14:0)	2.39E‐05	79.99		DG 14:1–18:0	0.023294		47.91	PE 33:1	0.000385	31.02
TG (49:2/FA16:0)	7.99E‐06	79.89		DG 14:1–20:0	0.000115	95.45		PE 34:0	5E‐05	74.77
TG (49:2/FA16:1)	7.62E‐06	78.95		DG 14:1–22:6	0.000856	49.34		PE 35:1	4.45E‐06	75.78
TG (49:2/FA17:0)	2.35E‐05	84.26		DG 16:0–16:0	0.021656	31.97		PE 36:1	0.001004	59.50
TG (49:2/FA18:1)	1.93E‐06	82.95		DG 16:0–16:1	0.000384	43.31		PE 37:2	1.11E‐05	90.32
TG (49:2/FA18:2)	1.57E‐05	82.08		DG 16:0–18:0	0.005168	46.41		PE 38:1	3.51E‐05	90.65
TG (50:0/FA14:0)	6.25E‐07	79.67		DG 16:0–20:0	0.002361	33.51		PI_26:0	0.004772	34.85
TG (50:0/FA16:0)	6.68E‐06	78.84		DG 16:0–20:5	0.000192	76.61		PI_34:2	0.033126	48.54
TG (50:0/FA18:0)	2.49E‐06	78.96		DG 16:0–22:6	0.000556	39.97		PI_35:0	0.020753		33.05
TG (51:0/FA17:0)	9.39E‐05	79.74		DG 16:1–16:1	3.95E‐05	69.58		PI_35:2	0.029939	55.93
TG (51:0/FA18:0)	9.88E‐05	73.33		DG 16:1–18:0	3.36E‐05		71.65	PI_38:2	0.002626	73.60
TG (51:3/FA16:1)	1.47E‐05	77.43		DG 16:1–20:5	1.2E‐05	79.91		PS_32:0	0.024165		45.71
TG (51:3/FA17:0)	5.98E‐06	79.11		DG 16:1–22:0	0.029591		67.68	PS_32:1	0.028797	68.55
TG (52:0/FA16:0)	1.46E‐05	76.15		DG 16:1–22:6	0.00079	63.11		PS_34:1	0.037013	63.97
TG (52:0/FA18:0)	1.51E‐05	76.04		DG 18:0–18:0	0.009176	37.13		PS_35:1	0.042851	74.80
TG (52:0/FA20:0)	1.78E‐05	78.54		DG 18:0–20:0	0.002667	60.38		PS_35:2	0.038263	75.60
TG (55:7/FA22:6)	1.33E‐05	88.30		DG 18:0–22:6	0.000155	50.15		PS_36:1	0.009635		35.64
				DG 18:1–18:1	1.34E‐05	71.53
				DG 18:1–20:0	0.003263	60.95
				DG 18:1–22:0	0.045712	31.72
				DG 18:1–22:6	0.000636	56.18
				DG 18:2–18:2	2.88E‐05	71.81
				DG 18:2–22:6	0.000245	61.87
				DG 18:3–20:5	0.046063	58.91
				DG 18:3–22:6	0.003382	40.46
				DG 20:0–20:5	0.001287	76.80
				DG 20:0–22:6	7.23E‐05	80.28
				DG 20:1–20:5	0.017775	69.93
				DG 22:6–22:6	3.76E‐05	59.89
(B) CWL extract
Analyte (lipid species)	p	Down (%)	Up (%)	Analyte (lipid species)	p	Down (%)	Up (%)	Analyte (lipid species)	p	Down (%)	Up (%)
TG (42:0/FA14:0)	0.036155		87.62	DG 12:0–16:0	0.030818	16.27		LPC 14:0	0.00061	83.92
TG (47:1/FA16:0)	0.003706		75.55	DG 12:0–20:0	0.005633	47.74		LPC 16:0	0.038686		73.37
TG (47:1/FA17:0)	0.006948		68.52	DG 12:0–22:0	0.002278	28.47		LPC 18:0	0.041112		60.97
TG (47:1/FA18:1)	0.015121		72.89	DG 14:0–16:1	0.002179	53.80		PE 32:0	8.42E‐07	74.62
TG (49:2/FA14:0)	0.024774		82.40	DG 14:0–20:0	0.022196	39.05		PE 33:1	3.5E‐07	80.41
TG (49:2/FA16:1)	0.013818		75.99	DG 14:0–22:6	0.021517		42.43	PE 34:0	1.67E‐05	87.77
TG (49:2/FA18:2)	0.026838		78.93	DG 16:0–16:0	0.00364	47.57		PE 35:1	3.58E‐06	94.19
				DG 16:0–16:1	0.042504		19.73	PE 37:2	0.00041	91.12
				DG 16:0–18:0	0.005507	33.16
				DG 16:1–16:1	0.044248	37.85
				DG 16:1–18:4	0.029947	65.11
				DG 18:0–20:0	0.025454		38.91
				DG 20:0–20:5	0.007493	66.02
				DG 20:0–22:0	6.55E‐05	88.05
				DG 20:1–20:5	0.003728	26.11

Table 3.

Biomarker panel depicting category‐wise scientific names constructed from lipids differentially identified between MTB and NTM (M. kansasii) isolates.

(A) Untargeted IDA based
Common lipid signatures in TL and CWL extracts
Hydroxyphthioceranic acid (C33)	MG (RCO2H = 18:1)		Lyso‐PG (RCO2H = 18:1)
	TG (R1CO2H + R2CO2H + R3CO2H = 50:1)
	TG (R1CO2H + R2CO2H + R3CO2H = 51:0)
Exclusive lipid signatures in TL extract
FA	GL		GPL
Hydroxyphthioceranic acid (C34)	MG (RCO2H = 17:0)		Lyso‐PE (R1CO2H = 16:0)
Phthioceranic acid (C36)	MG (RCO2H = 15:0)		Lyso‐PE (R1CO2H = 16:1)
Phthioceranic acid (C45)	TG (R1CO2H + R2CO2H + R3CO2H = 50:2)		Lyso‐PE (R1CO2H = 18:0)
Mycocerosic acid (C29)	DG (R1CO2H + R2CO2H = 28:0)		Lyso‐PE (R1CO2H = 17:0)
Mycolipenic acid (C29)	DG (R1CO2H + R2CO2H = 29:0)		Lyso‐PI (RCO2H = 19:2)
	DG (R1CO2H + R2CO2H = 29:1)		Lyso‐PG (RCO2H = 18:0)
	DG (R1CO2H + R2CO2H = 33:0)		Lyso‐PG (RCO2H = 19:0)
	DG (R1CO2H + R2CO2H = 39:2)		Lyso‐PG (RCO2H = 19:1)
	DG (R1CO2H + R2CO2H = 52:0)		Lyso‐PG (RCO2H = 20:0)
	DG (R1CO2H + R2CO2H = 30:1)		PG (R1CO2H + R2CO2H = 31:2)
			PE (R1CO2H + R2CO2H = 31:0)
			PE (R1CO2H + R2CO2H = 37:0)
			Lyso‐PIM1 (RCO2H = 19:2)
			PI (R1CO2H + R2CO2H = 34:4)
			Ac1PIM1 (R1CO2H + R2CO2H + R3CO2H = 48:3)
			PIM2 (R1CO2H + R2CO2H = 37:0)
Exclusive lipid signatures in CWL extract
FA	GL		GPL		PK
Mycocerosic acid (C30)	MG (RCO2H = 17:1)		Lyso‐PG (RCO2H = 17:0)		MPM (C31)
Mycocerosic acid (C33)	MG (RCO2H = 27:0)		Lyso‐PE (R1CO2H = 19:0)		MPM (C31)/PE (R1CO2H + R2CO2H = 33:2)
Phthioceranic acid (C33)	MG (RCO2H = 27:0)/PE (R1CO2H + R2CO2H = 33:2)		PE (R1CO2H + R2CO2H = 33:2)
Mycocerosic acid (C34)	DG (R1CO2H + R2CO2H = 30:0)		Lyso‐PG (RCO2H = 17:0)/PE (R1CO2H + R2CO2H = 33:2)
Phthioceranic acid (C34)	DG (R1CO2H + R2CO2H = 39:1)		Lyso‐PE (R1CO2H = 18:0)
Mycolipanolic acid (C28)	DG (R1CO2H + R2CO2H = 33:1)
Hydroxyphthioceranic acid (C48)	DG (R1CO2H + R2CO2H = 43:1)
Mycocerosic acid (C27)	DG (R1CO2H + R2CO2H = 34:1)
Mycolipenic acid (C28)	DG (R1CO2H + R2CO2H = 32:1)
Mycocerosic acid (C34)/PE (R1CO2H + R2CO2H = 33:2)	DG (R1CO2H + R2CO2H = 35:1)
Phthioceranic acid (C34)/PE (R1CO2H + R2CO2H = 33:2)	DG (R1CO2H + R2CO2H = 43:1)/PE (R1CO2H + R2CO2H = 33:2)
Mycolipanolic acid (C28)/PE (R1CO2H + R2CO2H = 33:2)	DG (R1CO2H + R2CO2H = 32:1)/PE (R1CO2H + R2CO2H = 33:2)
Mycolipenic acid (C28)/PE (R1CO2H + R2CO2H = 33:2)	DG (R1CO2H + R2CO2H = 30:1)
(B) Targeted MRM based
Common lipid signatures in TL and CWL extracts		Exclusive lipid signatures in TL extract		Exclusive lipid signatures in CWL extract
GL	GPL	GL	GPL	GL	GPL
DG 12:0–16:0	LPC 14:0	TG (42:0/FA16:0)	PC 24:0	DG 16:1–18:4	LPC 16:0
DG 12:0–20:0	PE 32:0	TG (44:0/FA14:0)	PC 30:0	DG 20:0–22:0	LPC 18:0
DG 12:0–22:0	PE 33:1	TG (44:0/FA16:0)	PC 32:0
DG 14:0–20:0	PE 34:0	TG (44:0/FA18:0)	PC 33:0
DG 14:0–22:6	PE 35:1	TG (46:0/FA14:0)	PC 34:0
DG 16:0–16:0	PE 37:2	TG (46:0/FA16:0)	PC 34:1
DG 16:0–18:0		TG (46:0/FA18:0)	PC 35:1
DG 16:1–16:1		TG (47:1/FA14:0)	PC 36:0
DG 20:0–20:5		TG (47:1/FA16:1)	PC 36:1
		TG (48:0/FA14:0)	PC 38:1
		TG (48:0/FA16:0)	PC 40:1
		TG (48:0/FA18:0)	LPC 17:0
		TG(49:2/FA16:0)	LPC 20:5
		TG(49:2/FA17:0)	LPC 22:0
		TG(49:2/FA18:1)	PE 36:1
		TG(50:0/FA14:0)	PE 38:1
		TG(50:0/FA16:0)	PI_26:0
		TG(50:0/FA18:0)	PI_34:2
		TG(51:0/FA17:0)	PI_35:0
		TG(51:0/FA18:0)	PI_35:2
		TG(51:3/FA16:1)	PI_38:2
		TG(51:3/FA17:0)	PS_32:0
		TG(52:0/FA16:0)	PS_32:1
		TG(52:0/FA18:0)	PS_34:1
		TG(52:0/FA20:0)	PS_35:1
		TG(55:7/FA22:6)	PS_35:2
		DG 12:0‐12:0	PS_36:1
		DG 12:0‐14:0
		DG 12:0‐14:1
		DG 12:0‐18:4
		DG 12:0‐22:6
		DG 14:0‐14:0
		DG 14:0‐14:1
		DG 14:0‐16:0
		DG 14:0‐18:0
		DG 14:0‐22:0
		DG 14:1‐18:0
		DG 14:1‐20:0
		DG 14:1‐22:6
		DG 16:0‐20:0
		DG 16:0‐20:5
		DG 16:0‐22:6
		DG 16:1‐18:0
		DG 16:1‐20:5
		DG 16:1‐22:0
		DG 16:1‐22:6
		DG 18:0‐18:0
		DG 18:0‐22:6
		DG 18:1‐18:1
		DG 18:1‐20:0
		DG 18:1‐22:0
		DG 18:1‐22:6
		DG 18:2‐18:2
		DG 18:2‐22:6
		DG 18:3‐20:5
		DG 18:3‐22:6
		DG 20:0‐22:6
		DG 22:6‐22:6

4. Discussion

Differentiating MTB from NTM remains a critical clinical challenge due to overlapping clinical manifestations and limitations in current diagnostic tools [5, 16]. In this study, M. kansasii isolates, a clinically significant NTM species due to their close phenotypic and genomic resemblance to MTB and its emerging prevalence in TB‐endemic regions [4, 6] were studied. The comparative lipidomic profiling with MTB isolates using two distinct extraction methods: TL and CWL. TL extraction captures structural and storage lipids, whereas CWL extraction targets cell envelope‐associated lipids, including virulence factors essential for mycobacterial survival and pathogenicity [8, 17]. Molecular‐level characterization through dual extraction coupled with UHP‐LC MS/MS revealed distinct lipid compositions, leading to the identification of lipid‐metabolite signature panels that highlight key alterations in MTB and M. kansasii, offering valuable insights into their lipid biology [7].

Untargeted lipidomics using IDA in both positive and negative ion modes has proven effective for lipid profiling of MTB and M. kansasii [18]. Here, the data from both ionization modes were combined for comprehensive lipidomics analysis of the mycobacterial lipidome (Figure S1). Previous studies have reported several shared features between these two species, including the virulence locus (ESX‐1) and antigens (CFP‐10, ESAT‐6), as well as essential metabolic genes [6, 19, 20]. Consistently, our biostatistical analysis of TL extracts revealed moderate similarity in lipid classes distribution among MTB and M. kansasii (Figures 1 and 3a; Tables S1a and S2a; S. Sheet 1), suggesting comparable membrane architectures that may influence host–pathogen interactions and antimicrobial susceptibility.

Both MTB and M. kansasii comprise some common alpha and keto mycolates with varying chain length and abundance, reflecting a fundamental evolutionary divergence in pathogenic potential. MTB has specific long‐chain mycolates that enhance cell wall hydrophobicity, promoting immune evasion and intracellular persistence [6, 21, 22]. Whereas M. kansasii produces unique, shorter, branched, cis, and nonketo mycolates, indicative of its environmental adaptation and relatively reduced virulence [6, 22]. These compositional differences extend to associated lipid families: MTB mycolates are coupled with sulfoglycolipids and phthioceranate‐derived lipids that promote aerosol stability and transmission, whereas M. kansasii retains more hydrophilic lipooligosaccharides and phenolic glycolipids, maintaining a cell envelope better suited for environmental survival than for host infection [21, 23].

However, the detailed CWL data further demonstrated clear species‐specific distinctions (Figures 2 and 3b; Tables S1b and S2b; S. Sheet 2), consistent with earlier lipidomic and genomic reports [6, 9, 21]. MTB isolates showed enrichment in GLs, GPLs, and FAs, supporting intracellular survival and pathogenicity [13, 17], 24]. Conversely, M. kansasii displayed a broader lipid diversity dominated by PKs, PRs, and SLs, consistent with its aquatic origin and opportunistic ecological strategy. The predominance of PK‐derived stress‐response lipids reflects retention of diverse pks gene clusters that promote environmental resilience and metabolic flexibility rather than specialized virulence [6, 25–27]. Importantly, the evolutionary preservation of these PKS pathways enables M. kansasii to synthesize lipid‐derived secondary metabolites that enhance stress tolerance and ecological persistence, distinguishing it from obligate pathogenic mycobacteria such as MTB, which has streamlined its lipid metabolism toward host‐dependent pathogenicity and virulence optimization [6, 27]. Since lipidomic data are complex and extensive, a Venn diagram (Figure 4 S. Sheet 3) was used to compare and illustrate the shared and unique lipids with each type of lipid extract of MTB and M. kansasii isolates. This visualization simplified the untargeted datasets, providing a clear overview of relationships between lipid sets and facilitating data interpretation [28]. A total of 116 lipids (14.9%) are common across all four datasets, indicating a conserved lipidome core essential for mycobacterial viability regardless of extraction type [6]. However, distinct lipid subsets defined the ecological and pathogenic divergence between the two species. These compositional distinctions in lipid profiles may influence pathogenic potential, immune evasion, and antibiotic response, providing targets for species‐specific therapeutic interventions and foundation for species‐level differentiation [8, 17].

Figure 3.

Histograms showing distribution of lipid subclasses among TL and CWL extracts of the MTB and NTM (M. kansasii). (a) Histograms illustrate compositional differences in lipid subclasses across six categories, that is, FA, GL, GLP, PK, PR, and SL, in TL (a) and CWL (b) extracts of MTB and NTM (M. kansasii) isolates.

(a)

(b)

The MRM‐based targeted UHPLC–MS/MS approach enabled precise and high‐throughput quantification of mycobacterial lipid species within complex biological matrices by utilizing stable isotope‐labeled and natural standards [29, 30]. Optimization of LC‐MS parameters using the SPLASH lipidomics mix and natural standards for TG, DG, PC, LPC, PE, PI, and PS (Table S3; Figure S2) ensured accurate quantification across lipid subclasses. The strong linearity of calibration curves (R ² = 0.99) and validated LOD/LOQ values (Table S4) confirmed high analytical precision and reproducibility, strengthening the reliability of observed lipid variations between MTB and M. kansasii.

Targeted quantification of GL and GPL subclasses validated the differential lipidomic patterns identified through untargeted profiling, reinforcing the species‐specific lipid distinctions. Integrating untargeted discovery with targeted validation enhanced the molecular coverage, reliability, and biological relevance of our results, confirming that observed lipid variations represent true compositional differences rather than analytical artifacts (Sa et al. 2024; [12, 31, 32]). Quantification based on consistently detected mycobacteria‐specific transition pairs provided a robust foundation for defining lipid signatures associated with virulence and metabolic adaptation, offering deeper mechanistic insights into the functional lipid architecture of MTB and M. kansasii (Table S5).

The targeted lipidomic analysis of GLs revealed higher concentrations of mostly TG and DG species in MTB compared with M. kansasii (Tables 1 and 2; Figures 5, S3, and S4). In MTB, TG enrichment and predominance of long‐chain species such as TG (51:3/FA 18:2) and TG (55:7/FA 22:6) indicate metabolic adaptation for intracellular survival and dormancy. The Tgs1‐mediated biosynthetic pathway promotes TG accumulation in intracellular lipid inclusions (ILIs) under hypoxia or nutrient deprivation, supporting dormancy, antibiotic tolerance, and long‐term survival in the host, consistent with previous studies [5, 33–35].

Conversely, M. kansasii showed a broader but lower abundance TG profile, with short‐chain species such as TG (42:0/FA 14:0) and TG (42:0/FA 16:0) being relatively enriched. This compositional diversity likely reflects its metabolic flexibility and adaptation to fluctuating environmental conditions rather than host‐dependent persistence, consistent with its opportunistic lifestyle and reduced dependence on lipid‐based energy storage [6].

Likewise, elevated levels of DGs in MTB may link to active lipid flux through the diacylglycerol acyltransferase (DGAT/Tgs) pathway, which acylates DG to form TG [36]. Under stress, Tgs1 and Tgs2 enzymes drive TG synthesis, whereas LipY lipase mediates TG hydrolysis back to DG, maintaining energy balance and supporting cell wall maintenance during reactivation [17, 33, 36]. This bidirectional TG–DG turnover reflects a dynamic lipid recycling mechanism that underpins metabolic plasticity and virulence adaptation in MTB.

In contrast, M. kansasii exhibited a more balanced DG profile, with species such as DG (18:1–18:4) enriched, suggesting less reliance on the DGAT‐mediated storage pathway and greater lipid compositional flexibility for environmental adaptation. Collectively, these findings suggest that MTB′s DG–TG metabolic axis is tightly regulated through DGAT and LipY activity to sustain intracellular survival and virulence, whereas M. kansasii maintains a metabolically adaptable but less virulence‐oriented lipid network [13, 33, 36].

GPLs, including PC, LPC, PE, PI, and PS, are crucial for mycobacterial membrane integrity, fluidity, and immune modulation [13, 35]. Here, GPL profiling revealed distinct species‐specific patterns between MTB and M. kansasii (Figures 5, S3b, and S4b; Tables 1 and 2), reflecting their divergent physiological and pathogenic adaptations. In TL extracts, MTB exhibited higher concentrations of PCs (notably PC 34:0), LPCs (specifically, LPC 14:0), PEs (PEs 32:7 and 38:1), and PI species that collectively reinforce membrane stability and intracellular persistence [37, 38]. PI serves as precursors for PIMs, LM, and LAM, which regulate phagosome maturation and attenuate proinflammatory immune responses through pattern recognition receptor signaling [5]. PE‐containing lipids contribute to the structural integrity of the mycobacterial envelope and facilitate nutrient assimilation within foamy macrophages [39], whereas PS enrichment suggests apoptotic mimicry contributing to immune evasion [40]. LPC, derived from PC hydrolysis, exhibits paradoxical activity: exogenous LPC can promote bactericidal macrophage signaling through the G2A–PKA–PI3K–p38 MAPK axis, yet MTB suppresses endogenous LPC accumulation to avoid this host defense mechanism [41]. Thus, the elevated levels of these phospholipids in MTB represent a strategically regulated lipid composition that balances membrane functionality with host immune manipulation to sustain chronic infection.

Conversely, M. kansasii showed relatively elevated levels of PC 24:0, LPC 17:0, and PI 31:2, particularly in CWL extracts, indicating adaptive membrane organization optimized for environmental stress endurance rather than intracellular residence [6]. PS was absent in CWL extracts of both MTB and M. kansasii, consistent with its localization to the cytoplasmic membrane rather than the cell wall, as reported previously [8, 21, 40].

Furthermore, the ROC curve was utilized to identify and validate discriminant potential lipid biomarkers derived from UHPLC–MS/MS datasets of MTB and M. kansasii isolates. The ROC framework enabled quantitative evaluation of biomarker performance based on sensitivity, specificity, and a high AUC, providing a robust measure of diagnostic accuracy [42–44]. The ROC framework enabled quantitative evaluation of biomarker performance based on sensitivity, specificity, and the AUC, providing a robust measure of diagnostic accuracy [42–44]. Using MetaboAnalyst 6.0, ROC modeling achieved an AUC of 1.0 (95% C I = 1.0, 1 sensitivity, 1 specificity) across both TL and CWL datasets, reflecting complete discrimination between species. Although a perfect AUC may suggest potential model inflation, rigorous PLS‐DA cross‐validation (R ² = 0.90, Q ² = 0.78 for TL; R ² = 0.83, Q ² = 0.76 for CWL) confirmed the robustness and predictive reliability of the models, excluding overfitting. Thus, the perfect AUC values reflect genuine lipidomic separation rather than overfitting, further supported by tenfold cross‐validation.

The preliminary biomarker panels derived from untargeted datasets, comprised 36 and 38 discriminant lipid species found in the bacilli TL and CWL extracts, respectively (Table 3a). These represent statistically significant differentiators between MTB and M. kansasii, serving as a discovery‐level dataset rather than a finalized diagnostic panel. TL profiles were enriched in FA, GL, and GPL classes, whereas CWL additionally featured PK lipids, reflecting the distinct metabolic architectures and cell envelope specializations in the two species. After exhaustive multivariate and univariate analyses, five lipid species comprising one FA, three GLs, and one GPL from these panels consistently showed strong discrimination between MTB and M. kansasii and were identified in both TL and CWL extracts (Figure 6; Table 3a). Additionally, differential expression analysis (p ≤ 0.05) of targeted datasets revealed up‐ and down‐regulated GL and GPL species (Figure 7; Table 3b), which emerged as biologically meaningful discriminators further supporting their diagnostic potential. These lipids are mechanistically linked to membrane signaling, host immune modulation, and intracellular persistence [5, 13, 38]. In contrast, the predominance of MG and PK lipids in M. kansasii reflects metabolic versatility and environmental adaptation rather than virulence specialization [6]. Thus, rather than proposing all 36–38 lipids as biomarkers, this study identifies and highlights a mechanistically defined subset of functionally relevant lipid species that differentiate MTB and M. kansasii at the species level and hold potential for diagnostic refinement.

However, this study was limited by the small number of clinical isolates analyzed, owing to BSL‐3 facility constraints and restricted patient sample availability, which may not fully capture intraspecies heterogeneity. Although the lipidomic findings were statistically validated and demonstrated strong discriminatory potential, further evaluation using larger, geographically diverse cohorts and functional assays is required to confirm the diagnostic and mechanistic relevance of the proposed biomarkers. Additionally, integrating host–pathogen lipid interaction studies would further strengthen the translational applicability of these findings.

Collectively, this integrated untargeted and targeted UHPLC–MS/MS analysis establishes that MTB and M. kansasii possess distinct lipidomic blueprints reflecting their ecological and pathogenic specialization and strategies. MTB′s enrichment in long‐chain TGs, DGs, and immunomodulatory GPLs underlines its lipid‐dependent adaptation for intracellular persistence and immune evasion, whereas M. kansasii retains a metabolically versatile but less virulent lipid repertoire dominated by short‐chain TGs and PKs. These mechanistically defined lipid signatures not only provide new insights into mycobacterial physiology but also form a basis for developing species‐specific lipid biomarkers and diagnostic tools for improved clinical differentiation of MTB and M. kansasii infections.

5. Conclusion

MTB and M. kansasii comprise complex and unique lipid profiles. Through advanced lipidomic techniques, we were able to identify specific lipid species that are either unique or common between MTB and M. kansasii. Our comparative analysis of the lipidome profiles of MTB and M. kansasii using TL and CWL extractions revealed distinct lipid compositions at the molecular level. Untargeted lipidomics via IDA in both positive and negative ion modes allowed a comprehensive analysis of lipid profiles, aiding in identifying new biomarkers and understanding lipid diversity and metabolism in these pathogens (Figure 8). Targeted lipidomics or MRM‐based methods enabled absolute quantification of specific molecular analytes, revealing significant changes in two major lipid categories, that is, GL and GPL. The absolute quantification of GL and GPL molecular species from both TL and CWL extracts of MTB and M. kansasii clinical isolates provided further insights into their differential distribution and abundance. Additionally, the biomarker analysis revealed distinct lipid signatures that can aid in distinguishing MTB from M. kansasii and are a significant addition for developing novel lipid‐based diagnostic biomarkers and therapeutic targets.

Figure 8.

Schematic workflow of integrated lipidomic workflow and potential biomarker panel for MTB and M. kansasii differentiation. The schematic summarizes the experimental design integrating untargeted and targeted LC–MS/MS workflows, multivariate analysis, and ROC validation to identify species‐specific lipid signatures distinguishing MTB from M. kansasii.

Author Contributions

Meenakshi Chugh: data curation, formal analysis, investigation, methodology, validation, visualization, and roles/writing—original draft. Saif Hameed: conceptualization, project administration, resources, supervision, validation, and writing—review and editing. Jitendra Singh: resources, supervision, validation, and writing—review and editing. Zeeshan Fatima: conceptualization, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization, roles/writing—original draft, and writing—review and editing.

Funding

This study was supported by the Indian Council of Medical Research (5/8/5/13/ITRC/Diag/2022/ECD‐1) and the Department of Biotechnology, Ministry of Science and Technology, India (BT/PR/23016/NER/95/2017).

Ethics Statement

Clinical sputum samples for MTB and M. kansasii were obtained under approval from the Institutional Ethics Committee of AIIMS, Bhopal (Approval No.: IHEC‐LOP/2018/EF0104) which is a part of a multiconsortium grant where the project coordinator had no objections. Written informed consent was obtained from each patient prior to the sample collection, in accordance with the study′s inclusion criteria and institutional ethical guidelines. All subsequent lipid extraction and LC–MS/MS‐based lipidomic analyses were performed at Amity University Haryana, Gurgaon, following approval from its institutional ethics committee (Approval No.: IEC‐AIB/AUH/2018‐3).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting Information

Additional supporting information can be found online in the Supporting Information section.

Supporting information

Supporting Information 1 Table S1: Overall and category‐wise minimum and maximum queried and identified m/z values in TL and CWL extracts. Table S2: Number of lipid subclasses for each lipid class among MTB and NTM isolates. Table S3: Optimized MRM transitions and mass spectrometry parameters for each GL and GPL species. Table S4: Details of the calibration curves of internal and natural GL and GPL standards, including LOD and LOQ. Table S5: Optimized transition pairs including precursor ion (Q1), product ion (Q3), and respective IDs. Table S6: Absolute concentrations for all quantified lipid species from GL and GPL categories. Figure S1: IDA chromatograms of MTB and NTM isolates in pos and neg ionization mode. Figure S2: Q1 scan, MS2 scan, and XICs of TG, DG, PC, LPC, PE, PI, and PS, TG natural standards for quantitative analysis. Figure S3: Distribution percentages of quantified lipid species of TG, DG, PC, LPC, PE, PI, and PS. Figure S4: Relative distribution of quantified lipid species of TG, DG, PC, LPC, PE, PI, and PS in percentage among MTB and NTM isolates.

Supporting Information 2 Excel Sheet: S. Sheet 1: List of exclusive and common lipids in Figure 1c,e. S. Sheet 2: List of exclusive and common lipids in Figure 2c,e. S. Sheet 3: List of exclusive and common lipids in Figure 4.

Chugh, Meenakshi , Hameed, Saif , Singh, Jitendra , Fatima, Zeeshan , Untargeted and Targeted Liquid Chromatography‐Mass Spectrometry–Based Lipidomic Profiling Revealed a Potential Biomarker Panel to Distinguish Clinical Isolates of M. tuberculosis From Nontuberculous M. kansasii , Pulmonary Medicine, 2026, 5535650, 25 pages, 2026. 10.1155/pm/5535650

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding authors. The data are not publicly available due to privacy or ethical restrictions.