Table of Contents Author Guidelines Submit a Manuscript
Journal of Pathogens
Volume 2018, Article ID 1454316, 14 pages
https://doi.org/10.1155/2018/1454316
Research Article

Investigations into Isoniazid Treated Mycobacterium tuberculosis by Electrospray Mass Spectrometry Reveals New Insights into Its Lipid Composition

1Amity Institute of Biotechnology, Amity University Haryana, Gurugram, Manesar 122413, India
2Advanced Centre for Bio Separation Technology (CBST), Vellore Institute of Technology (VIT), Vellore, Tamil Nadu 632014, India
3Division of Clinical Microbiology and Molecular Medicine, Department of Laboratory Medicine, All India Institute of Medical Sciences, New Delhi 110029, India

Correspondence should be addressed to Zeeshan Fatima; moc.liamg@amitafnahseezrd

Received 9 January 2018; Revised 11 April 2018; Accepted 18 April 2018; Published 19 June 2018

Academic Editor: Mario M. D'Elios

Copyright © 2018 Rahul Pal et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Many of the earlier studies involving the effect of isoniazid (INH) treatment have solely focused on the fatty acyl (FA) category of Mycobacterium tuberculosis (MTB) lipids. This motivated us with the major interest to examine the impact of INH on various other categories of MTB lipids. Towards this, we chose to interpret our mass spectral data (LC-ESI-MS) by a standalone software, MS-LAMP, in which “Mtb LipidDB” was integrated. Analysis by MS-LAMP revealed that INH treatment can alter the composition of “glycerolipids (GLs)” and “glycerophospholipids (GPLs)” categories of MTB lipids, in addition to the variations to FA category. Interpretation by “MycoMass” database yielded similar results as that of Mtb LipidDB, except that significant alterations to polyketides (PKs) category also were observed. Probing biosynthetic pathways of certain key lipids belonging to any of GLs, GPLs, and PKs categories can be attractive target(s) for drug discovery or can be useful to identify means to overcome drug resistance or to obtain insights into the causal factors of virulence. To the best of our knowledge, this is the first report hinting at the influence of INH on GLs, GPLs, and PKs of MTB.

1. Introduction

Tuberculosis (TB) still remains serious illness globally and the foremost common cause for death among patients with AIDS [1]. Mycobacterium tuberculosis (MTB) the causal organism of TB is tough to control in the lieu of emergence of resistant strains inflicting multidrug resistance (MDR). MDR-TB attributes to isoniazid (INH) and rifampicin (RIF) resistance, while extensively drug-resistant tuberculosis (XDR-TB) involves extra resistance to fluoroquinolones along with one of the three injectable medications, kanamycin, amikacin, or capreomycin [2]. One of the strategies to combat the MDR phenomenon against MTB is to discover novel targets for currently used anti-TB drugs such as INH, RIF, and ethambutol (EMB).

Isonicotinic acid hydrazide or commonly known as isoniazid (INH) is one of the most significant anti-TB first-line drugs, which impede mycolic acid synthesis to kill actively growing intracellular and extracellular tubercle bacilli [3]. INH targets InhA (EC 1.3.1.9, enoyl-[acyl-carrier-protein] reductase) by forming adduct with NAD(H) needed for mycolic acid biosynthesis via affecting fatty acid synthesis II (FAS-II) pathway [4]. Similarly there are investigations reporting that INH forms complex with β-ketoacyl-ACP synthase (KasA) enzyme. To better understand the efficacy of INH, already several novel strategies have been adopted. For instance, genetics and biochemical approach have led to the discovery of two different target enzymes for INH within the unique type II FAS system involved in mycolic acid synthesis [5]. Similarly, quantitative proteomics reveals novel insights into INH susceptibility in mycobacteria, which is mediated by a universal stress protein. Increase in KatG levels induced by increased BCG-2013 levels underlies the phenotypic susceptibility of mycobacteria to INH [6]. In fact, the single-drug/single-target paradigm of most antibiotics may not apply to INH, as discovered by proteomic profiling, where it is believed that there are twelve INH-pyridine nucleotide metabolites, each one potentially targeting a different enzyme [7]. Even the in silico approach has shed novel insights into the mechanism of action of INH resistance by KatG gene mutations [3].

In this context, it would be of immense interest and value to comprehend the basis of resistance exhibited by MTB towards INH. Considering the fact that primary target of INH involves inhibition of mycolic acids synthesis [5, 811], understanding other classes of lipid compositional changes in MTB upon INH treatment may further help in identification of potential novel targets. Lipid biology has evolved to a greater extent during the last decade, primarily attributed to advancements in the sophistication of mass spectrometry (MS) instrumentation [12, 13]. Many studies have noted altered lipid metabolism associated with diseases including TB [1421]. Therefore, study of lipid profiles by high-throughput lipidomics approach involving MS has gained significant importance in recent times. Most importantly, a comprehensive study reporting a database of MTB lipids, “Mtb LipidDB” [18], attracted our attention, which provided the impetus for us to investigate the MTB lipidome, for which we chose to apply (reverse phase) liquid chromatography coupled to electrospray ionization MS, namely, LC-ESI-MS. Consequently, the aim of this study is to compare the lipid profile of MTB in response to INH, wherein the lipid profiling is interpreted from LC-ESI-MS data. Our LC-ESI-MS data indicate that INH treatment not only causes alterations to the biosynthesis of mycolic acids that is already known, but also significantly influences the production of other lipid categories such as glycerolipids (GLs) and glycerophospholipids (GPLs), which have not been reported yet and therefore, this could be the first such report. Furthermore, we interpreted our LC-ESI-MS data using another database, MycoMass [22], whose output was compared with the output yielded from Mtb LipidDB [18] and found that the results from both the databases were similar, though a major difference was that the output from MycoMass database showed significant alterations to polyketides (PKs) category due to INH, which was not very conspicuous in the output from Mtb LipidDB. Thus, our analysis hereby provides an affirmative hint, prompting us to propose a hypothesis that INH could significantly alter the synthesis of GLs, GPLs, and PKs of MTB, in addition to that of mycolic acids. The results of this study could thus be helpful to find novel drug targets towards the treatment of MTB.

2. Materials and Methods

2.1. Materials

INH (Sigma Aldrich), Middlebrook broth 7H9, Middlebrook agar, OADC and ADC (BD Difco), potassium chloride (KCl), dimethyl sulfoxide (DMSO) and chloroform (Fisher Scientific), methanol (Merck), Teflon capping glass vials (Borosil), Pasteur pipette, and Whatman No. 1 filter paper were the materials used.

2.2. Culture Conditions

MTB (clinical strain) was used for the study. Culture was maintained on Lowenstein-Jenson (LJ) slant before experiment strain was freshly streaked on agar plate after 21 days of full growth. It was again inoculated in Middlebrook 7H9 broth for 14 days as a primary () culture. Starter culture H37Rv untreated or treated with INH drug at subinhibitory conc. of 0.5 μg/ml was inoculated with primary () culture of 0.1 OD600 in a screw cap flask containing 10 mL of Middlebrook 7H9 broth supplemented with 10% albumin/dextrose/catalase (BD Difco), 0.2% glycerol (Fischer Scientific), and 0.05% Tween-80 (Himedia) in 100 mL flasks (Schott Duran). The flasks were placed at 37°C for 14 days or till its exponential phase with agitation.

2.3. Extraction of Total Lipids

Cells of MTB untreated (control) and INH treated at exponential phase were used for whole cell lipid extraction by modified Folch method [21, 23]. Briefly, the MTB treated and untreated cells were harvested at 10,000 rpm for 10 minutes. Cell was homogenized in aqueous solution for 3 minutes and suspended in CHCl3 and CH3OH in ratio, 1:2. Cells were shaken well and centrifuged at 2000 rpm at 4°C for 10-15 minutes. Supernatant was transferred to another glass vial and then remaining CHCl3 was added and filtered through Whatman No. 1 filter paper. The extract was then washed with 0.88 % KCl to remove the nonlipid contamination, which resulted in the formation of two phases. From the two phases, the lower dense layer of chloroform containing lipid was taken by glass Pasteur pipette in 5 mL glass vial having Teflon capping. The vials were stored at -20°C until further analysis. The concentration of the extracted lipids was estimated by using a method described elsewhere [24] and 5 μg/μl was used for further analysis.

2.4. Ultraperformance Liquid Chromatography-Electrospray Ionization-Mass Spectrometry (UPLC-ESI-MS)

The samples were analyzed by LC-MS (Waters, ACQ-TQD#QBP 1152), a triple quadrupole tandem mass spectrometer in polarity switching mode. LC was done on a C18 column (100 mm x 3 mm, 2.6 μm 100Å pore size) through isocratic mode of elution for 30 minutes, using 5% isopropanol, 90% methanol, and 5% ammonium acetate (5 mM, pH 6.5), at a flow rate, 0.1 ml/min. The source temperature was 120°C, the desolvation temperature was 350°C, and the cone voltage was set at 40 volts (V). The sample was introduced using an autosampler with 1 μl of sample injection volume containing 5 μg of lipids. Capillary voltage (i.e., spray voltage) was set to 3.50 kilo volts (kV). The data were recorded in the mass range, m/z 200-2000, and the data were processed within the MassLynx software, where each chromatogram was smoothened and the background was subtracted. Duplicate or triplicate LC-ESI-MS data sets were acquired and the extent of reproducibility of the data was verified. Further analysis was carried out with the reproducible LC-ESI-MS data. Analysis of the mass spectral data was performed by using a standalone software MS-LAMP, in which “Mtb LipidDB” (http://www.mrl.colostate.edu/; Sartain et al., 2011) [18] and the database of Lipid Metabolites and Pathways Strategy Consortium (LIPID MAPS; http://www.lipidmaps.org) are integrated [25]. Data acquired in positive ESI mode only were considered for interpretation by “Mtb LipidDB” of MS-LAMP, whereby the m/z values of the peaks in the mass spectra were assigned to singly protonated ion, i.e., [M+H]+ only, and by setting the mass window range for every search to 0.5. The same set of observed m/z values (that were interpreted by MS-LAMP) were analyzed using MycoMass database [22] as well, following the same search criteria; namely, only [M+H]+ ions were searched with mass window range set to 0.5.

2.5. Lipase Assay

MTB cells were grown in Middlebrook 7H9 broth in the absence (control) and presence of INH. Whole cell protein was extracted and protein concentration was determined by Lowry method as previously described [26]. Lipase activity was performed by measuring the amount of p-nitrophenol (p-NP) released from p-NP ester substrate with varying lengths of fatty acids. The total lipase activity was estimated using protein extract of MTB. The standard lipase activity assays were performed in 100 μl reaction system consisting of a final concentration of 0.5 mM p-NP esters substrate and the buffer (pH 8.0) of 80 mM H3BO3, 80 mM H3PO4, 300 mM NaCl, 0.3% Triton X-100, and 20% glycerol. The reaction mixture of purified protein was incubated at 37°C for 40 minutes and the release of p-nitrophenol was determined spectrophotometrically at 405 nm [27, 28].

2.6. Criteria for Interpretation and Analysis of UPLC-ESI-MS Data

(i) Criterion based on peak intensity values: in this study, only those peak m/z values in the mass spectra have been considered for interpretation, whose intensities are higher than 5000 (namely, threshold value to ignore the peaks with poor signal-to-noise ratios). The peak m/z values selected in this manner were searched in the two databases: Mtb LipidDB and MycoMass, so as to identify the prospective lipids.

(ii) Criterion based on values of mass window range used for database search: during the course of analysis using MS-LAMP (in which Mtb LipidDB is integrated), when the mass window range for searching was set to any value less than 0.5, e.g., 0.25, very few lipids were obtained in the output from “Mtb LipidDB” (see Supplementary Material). However, while searching by using values of the mass window range > 0.5 in MS-LAMP, the mass errors (M: difference between the observed or queried m/z value and the m/z value available in the database for the respective lipid) estimated for the lipids obtained in the output were found to be quite high and consequently such outputs were not considered for further interpretation. Moreover, it must be noted that the data herein have been recorded in a mass spectrometer (triple quadrupole; see Materials and Methods, Section 2.4) that is capable of offering unit resolution only, for which the mass window range = 0.5 is an optimum or suitable value to carry out searches using MS-LAMP. Likewise, for the searches performed using MycoMass also, the outputs obtained by setting the mass window range = 0.5 only were considered for further analysis. Thus, the mass errors (M) of all the lipids obtained in the output herein are ≤ 0.5 and only those lipids have been considered for further interpretation. Furthermore, as mentioned in Section 2.4, all the m/z values have been ascribed to [M+H]+ only.

3. Results

3.1. Analysis of LC-ESI-MS Data Using Mtb LipidDB Through MS-LAMP

Figure 1 depicts representative chromatogram (inset) and mass spectra of untreated (control) and INH-treated samples. Mass spectra corresponding to different retention times in the chromatograms acquired from “control” and INH-treated samples are shown in Supplementary Material (Figures S1S11). As already described (see Materials and Methods), only data acquired in positive ESI mode were analyzed. 51 and 104 observed m/z values were chosen from untreated (control) and INH-treated sample, respectively, for further interpretation, among which 16 m/z values were observed in both the samples, which is depicted in the Venn diagram (Figure 2(a)). Strikingly, only 15 out of 35 and 31 out of 88 observed m/z values were interpretable by “Mtb LipidDB” (that is integrated in MS-LAMP) (Figure 2(b)). These 15 and 31 m/z values corresponded to 22 and 54 lipids, respectively (Figure 2(c)). Table 1 enlists the molecular formula of the potential lipids identified from “Mtb LipidDB”, corresponding to every 15 (control) and 31 (INH-treated) observed m/z values, which were queried in MS-LAMP, and Table 2 shows the names of the respective lipids. With regard to the 16 observed m/z values that are commonly detected in both “control” and “INH-treated” samples, only 5 m/z values were interpretable by MS-LAMP, which corresponded to 5 lipids (Table 3).

Table 1: Molecular formula of the lipids corresponding to observed m/z values, obtained from Mtb LipidDB (MS-LAMP).
Table 2: Names of the lipids corresponding to the observed m/z values, obtained from Mtb LipidDB (MS-LAMP).
Table 3: Molecular formula and its respective name of the lipids corresponding to the 5 observed m/z values out of 16, which were noted in both “control” and “INH-treated” samples, obtained from Mtb LipidDB (MS-LAMP).
Figure 1: Representative mass spectra from control and INH-treated samples (see Supplementary Figures S1S11). Inset depicts zoomed-in portion of the chromatogram showing peak at a particular retention time, corresponding to the respective mass spectrum shown here.
Figure 2: Venn diagrammatic representation of (a) number of m/z values observed in “Control” and “INH-treated” MTB samples; (b) number of m/z values for which lipids were identifiable in Mtb LipidDB (MS-LAMP); (c) “number of lipids” interpreted from Mtb LipidDB in “Control” and “INH-treated” MTB samples, by querying the observed m/z values (see Figure 2(b)) in MS-LAMP.
3.2. Analysis of LC-ESI-MS Data Using MycoMass Database

We also analyzed the LC-ESI-MS data with another database, MycoMass, reported by Layre et al. 2011 [22], which contains 5399 molecules, namely, nearly twice the number of lipids present in Mtb LipidDB (2518 lipids) reported by Sartain et al. 2011 [18]. Table 4 contains the list of observed m/z values that were not interpretable by Mtb LipidDB, but several of those were interpretable using MycoMass database. As expected, the number of prospective lipids identified from MycoMass is greater than those resulted from Mtb LipidDB, simply because MycoMass contains nearly twofold more lipids than Mtb LipidDB.

Table 4: Number of lipids in each category in Mtb LipidDB and MycoMass databases.

4. Discussion

Lipid homeostasis holds a consequential role in MTB pathogenicity and nearly 60% of its cellular dry weight is due to lipids [18]. The lipids’ diverse structural architecture and lipophobicity form an obstruction for host antibodies, sterilizers, and antimycobacterial drugs [29] and thus are beneficial for growth and survival of MTB. INH, a known anti-TB drug, targets several genes, namely, KATG, NDH, MSH, and NAT, but the primary target with dominant phenotype remains INHA, which encodes enoyl-Acyl-Carrier-Protein (ACP) termed as inhA reductase. The consensus on INH target demonstrates the inhibition of C26 FA elongation by FA synthase II complex, which restricts the formation of mycolic acids [9]. To obtain insights and to have a better understanding into mechanism of action of INH, we therefore decided to compare the lipidome profile of INH exposed MTB cells with the unexposed MTB cells.

Among eight lipid categories defined by LIPID MAPS (http://www.lipidmaps.org), the lipids identified thus far in MTB encompass only six categories: fatty acyls (FAs), glycerolipids (GLs), glycerophospholipids (GPLs), prenol lipids (PLs), polyketides (PKs), and saccharolipids (SLs) [18, 22]. Therefore, the results obtained from MS-LAMP covered these six categories of lipids only. As already described (cf. vide supra, Section 2.6), the mass window range = 0.5 was found to be an optimum value to carry out searches in MS-LAMP. Further, the data herein have been analyzed qualitatively only.

Out of 22 potential lipids interpreted from the “control” sample, 5 belonged to FAs category, 12 are GLs, and 5 are GPLs (Tables 1 and 2). On the other hand, the 54 prospective lipids found from the INH-treated sample comprise 11 FAs, 21 GLs, and 18 GPLs (Figure 3 and Tables 1 and 2). Thus, the output from MS-LAMP obtained upon querying LC-ESI-MS data acquired from “control” and “INH-treated” MTB samples suggests that INH treatment triggers synthesis of not only more lipids of every lipid category (Figures 2 and 3), but also entirely different set of lipids in each lipid category that were not found in the untreated/control sample (Tables 1 and 2). Significant alterations to the lipids belonging to GL and GPL category in this investigation are notable, since studies thus far have only reported about changes happening to mycolic acids (FA category) due to INH treatment [5, 9, 10]. These different sets of lipids could be involved in the mechanism of drug resistance, which indeed opens up new vistas for more investigations.

Figure 3: Comparison of variations in every lipid category between “Control” and “INH-treated” MTB cells: number of lipids in each category is interpreted from Mtb LipidDB [18], by querying the m/z values observed from the respective samples, in MS-LAMP (see Section 2.6).

Mycolic acid (MA) and phthiocerol dimycocerosates (DIM) belong to pathogenic FA class and it was observed that both were not detected in the presence of INH (Tables 1 and 2). DIM is one of the factors responsible for causing virulence with unresolved mechanisms; however some data showed that DIM shields MTB from oxidative stress, e.g., due to NO (nitric oxide), and modulates the production of key inflammatory cytokines such as TNF-α, which regulate the immunogenic responses. INH inhibits this class and makes the bacilli incapable to grow in oxidative stress [30]. It is also documented that mycolates play a significant role in cell wall permeability [31]. Therefore, INH affects the DIM level and thus makes the MTB cell more permeable. Further, it is interesting that the level of mycocerosic acid in “INH-treated” sample seems to be relatively higher as compared to the “control” sample (Tables 1 and 2). Mycocerosic acid is esterified with long-chain diol, phthiocerol, to generate phthiocerol dimycocerosate (DIM). Thus, it may be speculated that INH inhibits DIM synthesis at post-mycocerosic acid step, namely, inhibiting the esterification reaction between phthiocerol and mycocerosic acids in the biosynthetic pathway [32]. Similarly, we could detect alterations in fatty acyl chain lengths of hydroxyphthioceranic acid and phthioceranic acid, in response to INH, and found that in both classes of molecules the chain length was reduced (Tables 1 and 2). This observation prompts us to hypothesize that synthesis of sulfolipid, wherein hydroxyphthioceranic acid and phthioceranic acid are its major constituents, is affected, which are uniquely present in MTB and play critical roles in the host-pathogen interaction [32].

The GLs category contains three different subclasses depending on the number of fatty acids attached to glycerol backbone, namely, monoacylglycerol (MG), diacylglycerol (DG), and triacylglycerol (TG). GLs can be associated with either MA [33] or non-MA too [34]. When MTB bacilli are in dormant stage or stress condition they accumulate TGs. When MTB requires energy, the TGs are catabolized by lipase enzymes that are present on the outer wall to generate FAs that could be used for energy production [21, 27]. Herein, we found that the lipase activity was significantly decreased in the presence of INH (Figure 4), which means low burning and less utilization of accumulated TGs that are required by the MTB under stress conditions, such as presence of INH in this case [35]. Towards this, the output from MS-LAMP (Tables 1 and 2; Figure 3) provides supportive evidence in that relatively higher levels of DG and TG are observed in INH-treated sample, when compared to those levels in “control” sample, wherein relatively more MG are observed (Tables 1 and 2).

Figure 4: Lipase activity in response to INH. Mean lipase activity estimated as described in methods of three independent experiments SD of control and INH-treated cells are depicted as bar graph, where y-axis shows O.D. at 405 nm. depicts value < 0.5.

GPLs form another important category of lipids, which appeared to be more responsive in presence of INH. Moreover, it is already known that virulent strains of MTB possess higher contents of GPLs [36]. Increased levels in major GPLs were detected in response to INH, which include phosphatidyl ethanolamine (PE), phosphatidylinositol (PI), phosphatidyl glycerol (PG), and phosphatidylinositol mannoside (PIMs) (Tables 1 and 2). MTB changes their lipid composition in response to various stress conditions in order to adapt to the changed environment. There are reports showing that, under INH stress, genes associated with lipid biosynthesis including GPL undergo mutation resulting in alteration of lipid composition, which further leads to the development of resistance to INH [3739]. PIMs constitute a subclass of GPLs and are important components of the MTB cell envelope. Polar PIM species can also serve as membrane anchors for lipomannan (LM) and lipoarabinomannan (LAM). Synthesis of PIMs occurs sequentially in MTB where PIM1 helps in synthesis of PIM2 and likewise till PIM6, as revealed by extensive genetic and biochemical studies [40]. PIMs are substrates for heavy mannosylation to form LMs and additional arabinosylation to produce LAMs, which have roles in pathogenicity [21, 40]. It was found that acylated phosphatidylinositol mannosides, Ac1PIM1, were either drastically decreased or absent in presence of INH; hence we could hypothesize that acylation of PIMs is perhaps inhibited by INH and the role of acylation of PIMs towards drug resistance requires further probing.

Further, we performed comparative analysis of lipid categories identified from MycoMass database with MS-LAMP’s output. The search in MycoMass database was also done in the same manner as the search was done with Mtb LipidDB using MS-LAMP, whereby the window range was set to 0.5 (see Section 2.6). As already described (cf. vide supra, Section 3.2), the number of lipids obtained by searching MycoMass database is greater than that yielded from Mtb LipidDB (MS-LAMP), since the number of lipids in MycoMass database is more than that of Mtb LipidDB. MycoMass database consists of maximum number of “SLs” and interestingly, Mtb LipidDB has more FAs than MycoMass (Table 5). The number of GLs in MycoMass is about thrice the number of GLs in Mtb LipidDB; however, the number of GPLs in both the databases is somewhat comparable (Table 5). Therefore, when the outputs corresponding to every lipid category resulting from each of these two databases were compared, some interesting inferences emerged.

Table 5: Table showing list of observed m/z values, for which lipids were not identifiable in Mtb LipidDB (MS-LAMP).

Table 6 depicts the comparison of the outputs yielded from both databases, for the observed m/z values from “control” and “INH-treated” samples. Similar to the output of Mtb LipidDB (MS-LAMP), the results from the MycoMass database also indicated variations to the category of GLs and GPLs, due to INH treatment (Table 6). However, in the case of output from MycoMass database, significant alterations to polyketides (PKs) upon INH treatment too were noticeable, which was not so conspicuous in the MS-LAMP’s output. This can be comprehended from Table 5, wherein it is very clear that MycoMass database contains 893 PKs, whereas only 21 PKs are present in Mtb LipidDB. Despite the significant differences between these two databases, certain observed m/z values, 15 out of 35 (control), 5 out of 16 (control and INH-treated), and 28 among 88 (INH-treated), have been interpreted by or found in both databases (Table 6). Among the 22 lipids that were interpreted from Mtb LipidDB (MS-LAMP) using the 15 m/z values observed in the “control” sample (see Figure 2), 16 lipids are also found in MycoMass database. Furthermore, 61 out of 88 m/z values observed in the data of “INH-treated” sample have been interpreted by MycoMass database, which correspond to a total of 158 lipids. Among these 158 lipids, 42 lipid molecules are interpretable by both databases. This suggests that the 116 lipid molecules identified from the data of “INH-treated” sample are unique to the MycoMass database. All these are pictorially illustrated in the form of Venn diagram shown in Figure 5 (also see Supplementary Tables S1S3).

Table 6: Comparison of number of lipids in each category identified from Mtb LipidDB and MycoMass databases in “control” and “INH-treated” samples. Number of observed m/z values interpreted and not interpreted by each of these two databases in “control” and “INH-treated” samples is also compared.
Figure 5: Venn diagrammatic comparison of outputs from Mtb LipidDB (MS-LAMP) and MycoMass. (a) No. of observed m/z values interpreted from or found in the respective databases. (b) No. of lipids interpreted from the observed m/z values, using Mtb LipidDB (MS-LAMP) and MycoMass (compare this with Figure 2); blue color depicts the output from MycoMass database, and black color is for the results obtained from Mtb LipidDB (MS-LAMP), whereas green color depicts the output obtained from both Mtb LipidDB and MycoMass databases.

A closer evaluation and comparative analysis of outputs from both databases showed that 10 and 24 m/z values observed in “control” and in “INH-treated” samples, respectively, were not found in both databases (Table 6). Likewise, it was not possible to ascribe any lipid for the 7 observed m/z values that were commonly found in the data from both “control” and “INH-treated” samples (Table 6). This suggests that these 41 m/z values (Supplementary Table S4) in the mass spectral data might be due to new lipid molecules, which may have to be discovered in MTB, hinting the need to update these two databases. Alternatively, the strains of MTB utilized to create these two databases are perhaps different from the “clinical strain” that has been used in this investigation, which can be the reason for the absence of these m/z values in these two databases. Figure 6 depicts bar graph representation of the data shown in Table 6, whereby the outputs from Mtb LipidDB (MS-LAMP) and MycoMass are compared.

Figure 6: Comparison of the output obtained from MycoMass with that of Mtb LipidDB (MS-LAMP) for the “Control” and “INH-treated” MTB samples (see Section 2.6, Figure 3, and Table 6).

Altogether, the observations made in this study reinforce our hypothesis that INH treatment alters the lipid makeup of MTB and this change is not limited to FAs alone. Variations noted with GLs, GPLs, and PKs of MTB upon INH treatment could indeed open up new perspectives for further investigations, thereby helping in dissecting the evolution of drug resistance mechanisms and in exploring novel drug targets (Figure 7). Though it is not clear, as to how and in what order these changes are interconnected, the data generated from this study seemingly suggest that membrane remodeling is essential for the survival of MTB under stress. The specific links between INH resistance and named lipid moieties in this study may provide diagnostic and therapeutic targets that may be instrumental to combat MDR-TB. It needs to be noted that the inferences arrived in this investigation are actually based on lipid mass fingerprinting only, which was accomplished by conventional ESI MS and database searches involving Mtb LipidDB and MycoMass. In order to validate these inferences for the sake of proving the hypothesis proposed herein, thin layer chromatography (TLC) and tandem mass spectrometry (MS/MS) of some standard compounds need to be investigated (that may be pursued in future) and the results of this study could be the basis for choosing the standard compounds. Additionally, an aspect emerging from this study, which may be useful in future, is that many of the observed m/z values could not be ascribed to any lipid in the currently available versions of both the databases, namely, Mtb LipidDB (MS-LAMP) as well as MycoMass, indicating that such observed m/z values could be due to new lipid molecules that are yet to be discovered in MTB. This means that the versions of both the databases that we have used in this study may not be adequate and perhaps require to be updated. Furthermore, it deserves special mention that several genetic and host factors do interplay and hence must be considered while designing strategies aimed to search therapeutic targets. Extensive lipidomic analysis needs to be performed, in order to pinpoint the commonalities among various MTB strains, in terms of lipidome variation. Nevertheless, this study could be a starting resource point to link the changes due to INH resistance.

Figure 7: Model depicting that INH treatment alters the lipid makeup of MTB which is not limited to FAs. Variations noted with GLs, GPLs, and PKs of MTB upon INH treatment could open up new perspectives to explore novel drug targets (✓ denotes known lipid class (FA) while ? denotes possible additional lipid classes (GLs, GPLs, and PKs) affected by INH as proposed from this study).

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Authors’ Contributions

Rahul Pal, Saif Hameed, and Varatharajan Sabareesh contributed equally to this work.

Acknowledgments

Zeeshan Fatima thanks Science and Engineering Research Board (SERB), New Delhi (SR/FT/LS-173/2010), for the financial assistance in the form of Young Scientist Award. The authors thank Dr. Sanjeev Kanojiya, Council of Scientific and Industrial Research (CSIR), Central Drug Research Institute (CDRI), Lucknow, India, for assisting them in mass spectrometry experiments. Varatharajan Sabareesh wishes to acknowledge Dr. Sheetal Gandotra, Scientist, CSIR-Institute of Genomics and Integrative Biology (IGIB), New Delhi, India, for informing about MycoMass database.

Supplementary Materials

Figures  S1–S6: mass spectra acquired through UPLC-MS (see Experimental/Materials and Methods) of extract obtained from “untreated MTB cells”. Those m/z values that are interpretable using Mtb LipidDB database have been encircled (see Tables 1 and 2). Figures  S6–S11: mass spectra acquired by UPLC-MS of extract obtained from “treated MTB cells”. Those m/z values that are interpretable using Mtb LipidDB database have been encircled (see Tables 1 and 2). Table  S1: lipids, interpreted from/found in Mtb LipidDB (MS-LAMP) and MycoMass databases, for the observed m/z values in the “control (untreated)” MTB sample. Table  S2: lipids interpreted from/found in Mtb LipidDB (MS-LAMP) and MycoMass databases, for the observed m/z values in the “INH-treated” sample. Table  S3: lipids, identified from/found in Mtb LipidDB (MS-LAMP) and MycoMass databases, for the observed m/z values in both Control and INH-treated MTB samples. Table  S4: observed m/z values not found in (not interpretable by) both Mtb LipidDB and MycoMass databases. (Supplementary Materials)

References

  1. J. Zwang, M. Garenne, K. Kahn, M. Collinson, and S. M. Tollman, “Trends in mortality from pulmonary tuberculosis and HIV/AIDS co-infection in rural South Africa (Agincourt),” Transactions of the Royal Society of Tropical Medicine and Hygiene, vol. 101, no. 9, pp. 893–898, 2007. View at Publisher · View at Google Scholar · View at Scopus
  2. H. K. Khoharo and I. A. Shaikh, “Drug resistance patterns in pulmonary tuberculosis,” Journal of the Pakistan Medical Association, vol. 61, no. 3, pp. 229–232, 2011. View at Google Scholar · View at Scopus
  3. L. Jena, P. Waghmare, S. Kashikar, S. Kumar, and B. C. Harinath, “Computational approach to understanding the mechanism of action of isoniazid, an anti-TB drug,” International Journal of Mycobacteriology, vol. 3, no. 4, pp. 276–282, 2014. View at Publisher · View at Google Scholar · View at Scopus
  4. N. A. Kruh, R. Rawat, B. P. Ruzsicska, and P. J. Tonge, “Probing mechanisms of resistance to the tuberculosis drug isoniazid: Conformational changes caused by inhibition of InhA, the enoyl reductase from Mycobacterium tuberculosis,” Protein Science, vol. 16, no. 8, pp. 1617–1627, 2007. View at Publisher · View at Google Scholar · View at Scopus
  5. R. A. Slayden, R. E. Lee, and C. E. Barry, “Isoniazid affects multiple components of the type II fatty acid synthase system of Mycobacterium tuberculosis,” Molecular Microbiology, vol. 38, no. 3, pp. 514–525, 2000. View at Publisher · View at Google Scholar · View at Scopus
  6. X. Hu, X. Li, L. Huang et al., “Quantitative proteomics reveals novel insights into isoniazid susceptibility in mycobacteria mediated by a universal stress protein,” Journal of Proteome Research, vol. 14, no. 3, pp. 1445–1454, 2015. View at Publisher · View at Google Scholar · View at Scopus
  7. A. Argyrou, L. Jin, L. Siconilfi-Baez, R. H. Angeletti, and J. S. Blanchard, “Proteome-wide profiling of isoniazid targets in Mycobacterium tuberculosis,” Biochemistry, vol. 45, no. 47, pp. 13947–13953, 2006. View at Publisher · View at Google Scholar · View at Scopus
  8. K. Takayama, L. Wang, and H. L. David, “Effect of isoniazid on the in vivo mycolic acid synthesis, cell growth, and viability of Mycobacterium tuberculosis,” Antimicrobial Agents and Chemotherapy, vol. 2, no. 1, pp. 29–35, 1972. View at Publisher · View at Google Scholar · View at Scopus
  9. K. Takayama, H. K. Schnoes, E. L. Armstrong, and R. W. Boyle, “Site of inhibitory action of isoniazid in the synthesis of mycolic acids in Mycobacterium tuberculosis.,” Journal of Lipid Research, vol. 16, no. 4, pp. 308–317, 1975. View at Google Scholar · View at Scopus
  10. A. Somoskovi, L. M. Parsons, and M. Salfinger, “The molecular basis of resistance to isoniazid, rifampin, and pyrazinamide in Mycobacterium tuberculosis,” Respiratory Research, vol. 2, no. 3, pp. 164–168, 2001. View at Publisher · View at Google Scholar · View at Scopus
  11. E. K. Schroeder, O. N. de Souza, D. S. Santos, J. S. Blanchard, and L. A. Basso, “Drugs that inhibit mycolic acid biosynthesis in Mycobacterium tuberculosis,” Current Pharmaceutical Biotechnology, vol. 3, no. 3, pp. 197–225, 2002. View at Publisher · View at Google Scholar · View at Scopus
  12. S. J. Blanksby and T. W. Mitchell, “Advances in Mass Spectrometry for Lipidomics,” Annual Review of Analytical Chemistry, vol. 3, no. 1, pp. 433–465, 2010. View at Publisher · View at Google Scholar
  13. E. Ryan and G. E. Reid, “Chemical Derivatization and Ultrahigh Resolution and Accurate Mass Spectrometry Strategies for "shotgun" Lipidome Analysis,” Accounts of Chemical Research, vol. 49, no. 9, pp. 1596–1604, 2016. View at Publisher · View at Google Scholar · View at Scopus
  14. X. Han, D. M. Holtzman, D. W. McKeel Jr., J. Kelley, and J. C. Morris, “Substantial sulfatide deficiency and ceramide elevation in very early Alzheimer's disease: potential role in disease pathogenesis,” Journal of Neurochemistry, vol. 82, no. 4, pp. 809–818, 2002. View at Publisher · View at Google Scholar · View at Scopus
  15. M. Jain, C. J. Petzold, M. W. Schelle et al., “Lipidomics reveals control of Mycobacterium tuberculosis virulence lipids via metabolic coupling,” Proceedings of the National Acadamy of Sciences of the United States of America, vol. 104, no. 12, pp. 5133–5138, 2007. View at Publisher · View at Google Scholar · View at Scopus
  16. X. Han, S. Rozen, S. H. Boyle et al., “Metabolomics in Early Alzheimer's Disease: Identification of Altered Plasma Sphingolipidome Using Shotgun Lipidomics,” PLoS ONE, vol. 6, no. 7, p. e21643, 2011. View at Publisher · View at Google Scholar
  17. E. Marien, M. Meister, T. Muley et al., “Non-small cell lung cancer is characterized by dramatic changes in phospholipid profiles,” International Journal of Cancer, vol. 137, no. 7, pp. 1539–1548, 2015. View at Publisher · View at Google Scholar · View at Scopus
  18. M. J. Sartain, D. L. Dick, C. D. Rithner, D. C. Crick, and J. T. Belisle, “Lipidomic analyses of Mycobacterium tuberculosis based on accurate mass measurements and the novel "Mtb LipidDB",” Journal of Lipid Research, vol. 52, no. 5, pp. 861–872, 2011. View at Publisher · View at Google Scholar · View at Scopus
  19. C. Syme, S. Czajkowski, J. Shin et al., “Glycerophosphocholine Metabolites and Cardiovascular Disease Risk Factors in Adolescents: A Cohort Study,” Circulation, vol. 134, no. 21, pp. 1629–1636, 2016. View at Publisher · View at Google Scholar · View at Scopus
  20. L. Lu, A. Koulman, C. J. Petry et al., “An unbiased lipidomics approach identifies early second trimester lipids predictive of maternal glycemic traits and gestational diabetes mellitus,” Diabetes Care, vol. 39, no. 12, pp. 2232–2239, 2016. View at Publisher · View at Google Scholar · View at Scopus
  21. R. Pal, S. Hameed, P. Kumar, S. Singh, and Z. Fatima, “Comparative lipidomics of drug sensitive and resistant Mycobacterium tuberculosis reveals altered lipid imprints,” 3 Biotech, vol. 7, no. 5, article no. 325, 2017. View at Publisher · View at Google Scholar · View at Scopus
  22. E. Layre, L. Sweet, S. Hong et al., “A comparative lipidomics platform for chemotaxonomic analysis of mycobacterium tuberculosis,” Chemistry & Biology, vol. 18, no. 12, pp. 1537–1549, 2011. View at Publisher · View at Google Scholar · View at Scopus
  23. R. Pal, S. Hameed, S. Sharma, and Z. Fatima, “Influence of iron deprivation on virulence traits of mycobacteria,” The Brazilian Journal of Infectious Diseases, vol. 20, no. 6, pp. 585–591, 2016. View at Publisher · View at Google Scholar · View at Scopus
  24. E. Van Handel, “Rapid determination of total lipids in mosquitoes,” Journal of the American Mosquito Control Association, vol. 1, no. 3, pp. 302–304, 1985. View at Google Scholar
  25. V. Sabareesh and G. Singh, “Mass spectrometry based lipid(ome) analyzer and molecular platform: A new software to interpret and analyze electrospray and/or matrix-assisted laser desorption/ionization mass spectrometric data of lipids: A case study from Mycobacterium tuberculosis,” Journal of Mass Spectrometry, vol. 48, no. 4, pp. 465–477, 2013. View at Publisher · View at Google Scholar · View at Scopus
  26. O. H. Lowry, J. Nira, A. Rosebrough, F. Lewis, and J. R. Rose, “Folin phenol reagent protein measurement,” The Journal of Biological Chemistry, vol. 193, pp. 265–275, 1951. View at Google Scholar
  27. J. Cao, G. Dang, H. Li et al., “Identification and Characterization of Lipase Activity and Immunogenicity of LipL from Mycobacterium tuberculosis,” PLoS ONE, vol. 10, no. 9, p. e0138151, 2015. View at Publisher · View at Google Scholar
  28. C. Li, Q. Li, Y. Zhang et al., “Characterization and function of Mycobacterium tuberculosis H37Rv Lipase Rv1076 (LipU),” Microbiological Research, vol. 196, pp. 7–16, 2017. View at Publisher · View at Google Scholar · View at Scopus
  29. P. Singh, R. N. Rao, J. R. Reddy et al., “PE11, a PE/PPE family protein of Mycobacterium tuberculosis is involved in cell wall remodeling and virulence,” Scientific Reports, vol. 6, no. 1, 2016. View at Publisher · View at Google Scholar
  30. S. Dhandayuthapani, Y. Zhang, M. H. Mudd, and V. Deretic, “Oxidative stress response and its role in sensitivity to isoniazid in mycobacteria: characterization and inducibility of ahpC by peroxides in Mycobacterium smegmatis and lack of expression in M. aurum and M. tuberculosis.,” Journal of Bacteriology, vol. 178, no. 12, pp. 3641–3649, 1996. View at Publisher · View at Google Scholar
  31. M. Jackson, C. Raynaud, M.-A. Lanéelle et al., “Inactivation of the antigen 85C gene profoundly affects the mycolate content and alters the permeability of the Mycobacterium tuberculosis cell envelope,” Molecular Microbiology, vol. 31, no. 5, pp. 1573–1587, 1999. View at Publisher · View at Google Scholar · View at Scopus
  32. T. D. Sirakova, A. K. Thirumala, V. S. Dubey, H. Sprecher, and P. E. Kolattukudy, “The Mycobacterium tuberculosis pks2 Gene Encodes the Synthase for the Hepta- and Octamethyl-branched Fatty Acids Required for Sulfolipid Synthesis,” The Journal of Biological Chemistry, vol. 276, no. 20, pp. 16833–16839, 2001. View at Publisher · View at Google Scholar · View at Scopus
  33. C. Deb, J. Daniel, T. D. Sirakova, B. Abomoelak, V. S. Dubey, and P. E. Kolattukudy, “A novel lipase belonging to the hormone-sensitive lipase family induced under starvation to utilize stored triacylglycerol in Mycobacterium tuberculosis,” The Journal of Biological Chemistry, vol. 281, no. 7, pp. 3866–3875, 2006. View at Publisher · View at Google Scholar · View at Scopus
  34. J. Bacon, L. G. Dover, K. A. Hatch et al., “Lipid composition and transcriptional response of Mycobacterium tuberculosis grown under iron-limitation in continuous culture: Identification of a novel wax ester,” Microbiology, vol. 153, no. 5, pp. 1435–1444, 2007. View at Publisher · View at Google Scholar · View at Scopus
  35. J. S. Schorey and L. Sweet, “The mycobacterial glycopeptidolipids: structure, function, and their role in pathogenesis,” Glycobiology, vol. 18, no. 11, pp. 832–841, 2008. View at Publisher · View at Google Scholar · View at Scopus
  36. A. Huazano-García and M. G. López, “Metabolism of short chain fatty acids in colon and faeces of mice after a supplementation of diets with agave fructans,” in Lipid Metabolism, R. Valenzuela Baez, Ed., vol. 8, pp. 163–182, InTech, Rijeka, Croatia, 2013. View at Publisher · View at Google Scholar
  37. M. Alberghina, “Relationship between lipid composition and antibiotic-resistance to isoniazid, streptomycin, p-aminosalicylic acid, ethambutol, rifampicin in mycobacteria.,” Italian Journal of Biochemistry, vol. 25, no. 2, pp. 127–151, 1976. View at Google Scholar · View at Scopus
  38. L. Jena, T. Nayak, S. Deshmukh, G. Wankhade, P. Waghmare, and B. C. Harinath, “Isoniazid with Multiple Mode of Action on Various Mycobacterial Enzymes Resulting in Drug Resistance,” Journal of Infectious Diseases & Therapy, vol. 4, no. 5, 2016. View at Publisher · View at Google Scholar
  39. Y. S. Morita, Y. Yamaryo-Botte, K. Miyanagi et al., “Stress-induced Synthesis of Phosphatidylinositol 3-Phosphate in Mycobacteria,” The Journal of Biological Chemistry, vol. 285, no. 22, pp. 16643–16650, 2010. View at Publisher · View at Google Scholar
  40. K. Athenstaedt and G. Daum, “Phosphatidic acid, a key intermediate in lipid metabolism,” European Journal of Biochemistry, vol. 266, no. 1, pp. 1–16, 1999. View at Publisher · View at Google Scholar · View at Scopus