Present address: Diabetes Discovery Platform, Medical Research Council, Tygerberg, 7505, South Africa.
Carbon flux rerouting during Mycobacterium tuberculosis growth arrest
Article first published online: 6 OCT 2010
© 2010 Blackwell Publishing Ltd
Volume 78, Issue 5, pages 1199–1215, December 2010
How to Cite
Shi, L., Sohaskey, C. D., Pfeiffer, C., Datta, P., Parks, M., McFadden, J., North, R. J. and Gennaro, M. L. (2010), Carbon flux rerouting during Mycobacterium tuberculosis growth arrest. Molecular Microbiology, 78: 1199–1215. doi: 10.1111/j.1365-2958.2010.07399.x
- Issue published online: 21 NOV 2010
- Article first published online: 6 OCT 2010
- Accepted manuscript online: 28 SEP 2010 09:13PM EST
- Accepted 11 September, 2010.
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
A hallmark of the Mycobacterium tuberculosis life cycle is the pathogen's ability to switch between replicative and non-replicative states in response to host immunity. Transcriptional profiling by qPCR of ∼ 50 M. tuberculosis genes involved in central and lipid metabolism revealed a re-routing of carbon flow associated with bacterial growth arrest during mouse lung infection. Carbon rerouting was marked by a switch from metabolic pathways generating energy and biosynthetic precursors in growing bacilli to pathways for storage compound synthesis during growth arrest. Results of flux balance analysis using an in silico metabolic network were consistent with the transcript abundance data obtained in vivo. Similar transcriptional changes were seen in vitro when M. tuberculosis cultures were treated with bacteriostatic stressors under different nutritional conditions. Thus, altered expression of key metabolic genes reflects growth rate changes rather than changes in substrate availability. A model describing carbon flux rerouting was formulated that (i) provides a coherent interpretation of the adaptation of M. tuberculosis metabolism to immunity-induced stress and (ii) identifies features common to mycobacterial dormancy and stress responses of other organisms.
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
Tubercle bacilli switch between replicative (growth) and non-replicative (dormancy) states in response to the host environment. Immunity-induced cessation of bacterial growth leads to a chronic, asymptomatic infection that is maintained by persisting tubercle bacilli. When host immunity falters, tubercle bacilli can resume growth and cause disease. Dormant bacilli present a formidable challenge to tuberculosis control, because they are much less susceptible to antibacterial drugs than growing bacilli (Zhang, 2004). Thus, a molecular understanding of events occurring during the transition from growth to dormancy is important for identifying critical targets for new tuberculosis control strategies.
The metabolism of Mycobacterium tuberculosis during infection has been an area of intense interest since the origins of tuberculosis research. Seminal experiments showed that the tubercle bacillus, which is metabolically flexible (Wheeler and Ratledge, 1994), preferentially utilizes fatty acids as carbon and energy source during murine infection (Bloch and Segal, 1956). This preferential utilization is likely due to increased availability of lipids in the infected host cell (Russell et al., 2009) and is reflected in the M. tuberculosis genome as a significant enrichment for fatty acid degradation genes (Cole et al., 1998). Moreover, genes involved in fatty acid utilization (glyoxylate shunt: icl; gluconeogenesis: glpX and pckA) are required for growth and persistence in vivo (Collins et al., 2002; Liu et al., 2003; Sassetti and Rubin, 2003; Munoz-Elias and McKinney, 2005; Marrero et al., 2010). Some of these genes (icl, pckA) are also upregulated in non-growing bacilli during mouse infection (McKinney et al., 2000; Schnappinger et al., 2003; Timm et al., 2003). These observations led to the proposal that M. tuberculosis switches its carbon source from sugars to fatty acids during the persistent phase of infection (Bishai, 2000; Honer zu Bentrup and Russell, 2001; Timm et al., 2003). However, similar metabolic changes are observed in the stress response of other organisms, such as sporulation in yeast (Cortassa et al., 2000), where they are linked to the rerouting of carbon flow towards formation of storage compounds. Indeed, tubercle bacilli obtained from sputum of tuberculosis patients are rich in triacylglycerol (TAG) (Garton et al., 2008), a common storage compound in Actinobacteria (Alvarez and Steinbuchel, 2002). Therefore, it has been unclear whether the metabolic changes in the tubercle bacillus during infection reflect implementation of a developmental programme towards persistence or an adaptive response to changed nutritional conditions.
In the present work we examined the response of M. tuberculosis metabolism to the stress encountered during infection with a multi-scale approach that included studies in vivo (mouse infection), in silico and in vitro (broth culture). We interrogated two aspects of carbon metabolism of tubercle bacilli. One is central metabolism, which provides energy, reducing power and biosynthetic precursors. The other is the biosynthesis of major lipids, such as mycobacterial cell wall components, which constitute a sink for energy and biosynthetic precursors (Brennan, 2003), and triacylglycerol (TAG), which is a storage lipid (Brennan, 2003; Garton et al., 2008; Low et al., 2009). Transcription profiling of tubercle bacilli during infection of mouse lungs led to a metabolic model in which carbon flow redistributes at the phosphoenolpyruvate (PEP)/pyruvate-oxaloacetate (OAA) node as growth stops, thereby switching from the generation of energy and biosynthetic precursors to the synthesis of storage compounds. The in vivo data provided conditions to perform in silico simulations of metabolic flux patterns associated with M. tuberculosis growth arrest using a genome-scale metabolic network of the tubercle bacillus (Beste et al., 2007). We found that metabolic fluxes (i) were largely consistent with the transcription profiles of M. tuberculosis central metabolism genes seen during mouse infection and (ii) predicted that utilization of different carbon sources would result in similar flux solutions. This prediction was tested in vitro using M. tuberculosis cultures subjected to bacteriostatic stress in media containing different carbon sources. We found that transcriptional changes of key metabolic genes during M. tuberculosis growth arrest were independent of stress signal and carbon source. Thus, M. tuberculosis persistence shares critical metabolic features with bacterial and yeast sporulation, plant seed formation and animal hibernation. Our results lead to the novel view that the metabolic changes of M. tuberculosis are part of a developmental decision rather than a response to changed nutrient availability.
Results and discussion
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
To assess carbon flow distribution during the establishment of persistent infection, we used a model of respiratory infection of mice in which M. tuberculosis replicates in the lung for about 20 days, followed by stabilization of bacterial numbers due to expression of acquired cell-mediated immunity (Shi et al., 2003; 2005). We measured changes of bacterial transcript abundance in the mouse lung using qPCR with samples taken at multiple times post-infection (day 12 through day 100). We then expressed the data as a ratio of abundance at each given time point relative to that measured at day 12 (mid-log growth).
The in vivo data are presented below in two sections: (i) central metabolism and fatty acid catabolism and (ii) lipid biosynthesis. Of the lipid biosynthetic genes, we focused on those involved in the synthesis of the major storage compound TAG (Garton et al., 2008) and of the major components of the mycobacterial cell wall, such as mycolic acids and multimethyl-branched lipids. At the end of each section, the data are interpreted through a metabolic model of M. tuberculosis growth arrest. A scheme showing the main pathways of carbon metabolism and the representative genes in each pathway selected for analysis is presented in Fig. 1 and Table S1.
Glycolysis and the pentose phosphate pathway (PPP). Glucose is oxidized through glycolysis and, to a lesser extent, through the PPP (Jayanthi Bai et al., 1975). Glycolysis yields central metabolites such as PEP, pyruvate and acetyl-CoA (AcCoA). PPP generates reducing equivalents for reductive biosynthetic reactions and provides ribose 5-P for nucleotide synthesis. Enumeration of transcripts encoding key glycolytic and PPP enzymes of M. tuberculosis showed that most transcripts were downregulated during growth arrest relative to log growth (up to 15-fold decrease at day 30 of infection) (Fig. 2A and B; Table S2). However, two genes encoding phosphofructokinase differed in their response. pfkA was downregulated, while pfkB was strongly upregulated: the number of pfkB transcripts peaked at day 18 (65-fold) and remained high for the remainder of infection (> 30-fold at day 100) (Fig. 2A). Since pfkB is a member of the devR/dosR regulon (Voskuil et al., 2003), pfkB upregulation agrees with our previous observation that DevR/DosR-regulated genes are induced by expression of immunity in the mouse lung (Shi et al., 2003).
Sugars are presumably scarce during infection, and most glycolytic reactions can be reversibly utilized for gluconeogenesis. However, a decreased intracellular concentration of C6 sugars should occur instead, due to the observed downregulation of most glycolytic genes and the strong upregulation of pfkB, whose product catalyses the irreversible and rate-limiting reaction of glycolysis. This scenario is consistent with bacillary growth arrest, since C6 sugars may serve as precursors for macromolecular biosynthesis during growth.
Fatty acid degradation. Fatty acids, which are the main source of carbon and energy for M. tuberculosis during infection, are primarily catabolized via successive rounds of β-oxidation. Even-chain fatty acids are degraded to AcCoA, and odd-chain fatty acids are degraded to both AcCoA and propionyl-CoA (PropCoA). It is likely that tubercle bacilli catabolize a wide range of fatty acids, since they possess a large number of duplicated genes that are annotated as encoding β-oxidation enzymes (Cole et al., 1998).
We examined four genes involved in β-oxidation, fadA4, fadB, fadE5 and fadE23, which had previously been reported to be induced in vivo (Dubnau et al., 2005) or under several stress conditions in vitro (Wilson et al., 1999; Schnappinger et al., 2003; Hampshire et al., 2004). We found that the fadB transcript decreased > 10-fold by day 30, while fadE5, fadE23 and fadA4 were moderately upregulated during growth arrest (two- to fourfold induction between days 18 and 30) (Fig. 2C and Table S2). This result shows that individual fad genes respond differently in vivo, perhaps reflecting specific fatty acid substrates in the infected cell.
Tricarboxylic acid (TCA) cycle. AcCoA derived from the catabolism of fatty acids or sugars is assimilated via the TCA cycle, which provides biosynthetic precursors and reducing equivalents for energy generation and biosynthetic reactions. In M. tuberculosis, the TCA cycle operates as a variant in which α-ketoglutarate is converted to succinate by an α-ketoglutarate decarboxylase (kgd) and succinic semialdehyde dehydrogenase (gabD1 and gabD2) (Tian et al., 2005). Transcript levels of key TCA cycle genes, including kgd, were downregulated during growth arrest (up to 25-fold by day 30; Fig. 2D and Table S2), implying that the TCA cycle in growth-arrested bacilli functions differently than in growing cells.
Glyoxylate shunt and methylcitrate cycle. M. tuberculosis utilizes two additional pathways for assimilating AcCoA and PropCoA from fatty acid and cholesterol catabolism. One is the glyoxylate shunt, which comprises the sequential activities of isocitrate lyase (icl) and malate synthase (glcB). It converts AcCoA into C4-intermediates while bypassing the decarboxylation steps of the TCA cycle. The second pathway is the methylcitrate cycle. In this cycle PropCoA is converted to pyruvate via three specific enzymatic activities, methylcitrate synthase (prpC), methylcitrate dehydratase (prpD) and 2-methyl-isocitrate lyase (icl) (Munoz-Elias et al., 2006). In M. tuberculosis, 2-methyl-isocitrate lyase and isocitrate lyase activities are encoded by a single gene (called icl or icl1, rv0467) (Gould et al., 2006).
Transcript enumeration by qPCR showed that expression of icl was moderately induced (up to threefold between days 15 and 30), while glcB was downregulated (10-fold) in growth-arrested bacilli. Levels of prpDC mRNA peaked (> 10-fold) at day 21 and remained high (fivefold at day 100) (Fig. 2E and Table S2). Thus, three genes involved in the methylcitrate cycle (prpC, prpD and icl) are upregulated. We note that the lack of coordinated regulation of icl and glcB separates M. tuberculosis from other bacteria, including closely related Actinobacteria such as Corynebacterium glutamicum (Wendisch et al., 1997). We propose that, even though glcB is downregulated, the flux through the shunt increases because isocitrate lyase is the key enzyme in the TCA cycle/glyoxylate shunt transition (Walsh and Koshland, 1985).
Gluconeogenesis. Growth on fatty acids, TCA cycle intermediates and other substrates that enter the TCA cycle via AcCoA requires the synthesis of glycolytic C3-intermediates, such as PEP and pyruvate, via the decarboxylation of C4-intermediates of the TCA cycle. The enzyme combinations required for C4-decarboxylation and C3-carboxylation vary among bacterial species. M. tuberculosis encodes pyruvate carboxylase (pca), which catalyses the pyruvate to OAA reaction, the malic enzyme (mez) for the malate to pyruvate reaction, and PEP carboxykinase (pckA) for the irreversible conversion of OAA to PEP. PEP carboxykinase activity is essential for M. tuberculosis growth and persistence in mice (Marrero et al., 2010). In addition, gluconeogenesis requires dephosphorylation of fructose 1,6-bisphosphate by a 1,6-bisphosphatase (glpX) (Movahedzadeh et al., 2004). Of the above genes, only pckA was upregulated in tubercle bacilli during mouse lung infection (up to threefold induction between days 18 and 50), as previously observed (Timm et al., 2003). Levels of the pca transcript were below detection, and transcript levels for mez and glpX were downregulated (up to 10-fold at day 30, Fig. 2F and Table S2). The combined upregulation of pckA and downregulation of glpX, which encode enzymes catalysing irreversible reactions, point towards C3-compound synthesis during M. tuberculosis growth arrest.
Glutamate and glutamine synthesis. To gain insight on how M. tuberculosis growth arrest affects carbon flux between carbon and nitrogen metabolism via the TCA cycle intermediate α-ketoglutarate, we probed the expression of genes involved in the metabolism of glutamate, which is one of the most abundant cellular metabolites in tubercle bacilli (Tian et al., 2005). Glutamate is synthesized by glutamate synthase (gltDB) from α-ketoglutarate and glutamine, the primary product of ammonium assimilation. Glutamine is synthesized by glutamine synthetase, which is encoded by the essential gene glnA1 (Tullius et al., 2003; Harth et al., 2005). Glutamate dehydrogenase (gdh) can also participate in glutamate metabolism by catalysing the reversible reaction from glutamate to α-ketoglutarate and ammonium.
We found that glnA1 and gdh were downregulated (approximately fivefold) by day 30 (Fig. 2G and Table S2). In contrast, gltDB was induced (up to threefold) between days 18 and 30, suggesting that M. tuberculosis directs its carbon flux from α-ketoglutarate towards formation of glutamate during growth arrest. This direction of the carbon flux is supported by the decreased expression of kdg in the TCA cycle variant (see above). Glutamate may function as a nitrogen reserve to be utilized for bacterial re-growth when environmental conditions become favourable. It may also serve as a ‘compatible solute’ to keep metabolic enzymes functional under stress conditions (Kempf and Bremer, 1998). Moreover, glutamate contributes to mycobacterial cell wall structure (Harth et al., 2000). The observed downregulation of glutamine synthetase (glnA1) in non-growing bacilli is consistent with a reduced requirement for glutamine, which is a precursor of the cell wall component poly-l-glutamine (Tullius et al., 2003). Thus, the ratio of poly-l-glutamate to poly-l-glutamine in the M. tuberculosis cell wall is expected to increase during growth arrest.
Metabolic scheme I: accumulation of C3 compounds
The changes in mRNA abundance reported above suggest that as tubercle bacilli stop replicating, they respond to the decreased demand for energy and biosynthetic precursors by downregulating glycolysis, PPP and the TCA cycle. The resulting reduced synthesis of NADH from central metabolism fits well with the reduced activity of the electron transport chain postulated in our previous work (Shi et al., 2005). In the context of decreased TCA cycle activity, C4 compounds (malate and OAA) are mostly generated via the increased activity of the glyoxylate shunt, which also serves an anaplerotic function (Gottschalk, 1985). OAA is then converted to PEP by PEP carboxykinase. Carbon from PropCoA is directed towards pyruvate formation via the methylcitrate cycle. In addition to the above pathways, PEP and pyruvate are formed via the concerted upregulation of glycolytic phosphofructokinase (pfkB) and downregulation of gluconeogenic fructose 1,6-bisphosphatase (glpX), which would inhibit the late steps of gluconeogenesis. Thus, the flux of carbon during growth arrest of M. tuberculosis is expected to be towards the generation of the C3 compounds PEP and pyruvate.
To explore the metabolic fate of C3 compounds during growth arrest, we probed two aspects of lipid biosynthesis: (i) synthesis of TAG, a storage lipid implicated in the stress response and long-term survival of M. tuberculosis (Daniel et al., 2004; Garton et al., 2008; Low et al., 2009), and (ii) biosynthesis of major cell wall components, such as mycolic acids and multimethyl-branched lipids, which utilizes significant amounts of ATP, reducing power and biosynthetic precursors generated from central metabolism.
TAG synthesis. TAGs are synthesized from glycerol 3-P and short- or intermediate-chain acyl primers. Genes involved in synthesis of monoacylglycerol 3-P (plsB2) and diacylglycerol 3-P (rv2182c) were expressed at an unchanged or slightly reduced (approximately twofold) level during growth arrest (Fig. 2H and Table S2). Of the multiple TAG synthases encoded by M. tuberculosis, the TAG synthase encoded by tgs1 (TGS1) accounts for most of the TAG synthetic activity, at least when examined in vitro (Daniel et al., 2004; Sirakova et al., 2006). This gene, which is regulated by dosR/devR, was strongly upregulated in growth-arrested bacilli (20- to 60-fold between days 18 and 100) (Fig. 2H and Table S2), suggesting that TAG accumulates in non-growing bacilli in mouse lungs, as seen with non-growing bacilli in vitro (Garton et al., 2002). Much smaller changes were seen with other TAG synthase genes: rv3371 increased and rv3734 decreased by two- to threefold at day 30.
Mycolic acid synthesis. Mycolic acids are long-chain, high-molecular-weight α-alkyl, β-hydroxyl fatty acids that are essential components of the mycobacterial cell wall. They are synthesized by the type I and type II fatty acid synthase systems (FAS-I and FAS-II) (Bhatt et al., 2007). FAS-I, which is a single, multi-domain polypeptide encoded by fas, mediates de novo synthesis of intermediate-length, saturated α-chain of mycolic acids (C26) and synthesis of short-chain fatty-acyl-CoA primers (C16). FAS-II is a multi-enzyme complex that elongates the acyl-CoA primers generated by FAS-I to produce the full-length meromycolate chain of mycolic acids (Bhatt et al., 2007). FAS-I and FAS-II are linked by the activity of a β-ketoacyl-AcpM synthase III (fabH), which catalyses the condensation between FAS-I acyl-CoA primers and malonyl-AcpM (the product of malonyl-CoA activation by malonyl-CoA:AcpM transacylase). The fabH gene product also funnels precursors to FAS-II (Choi et al., 2000).
We observed that fas (FAS-I system) was upregulated (up to 10-fold) between days 15 and 30 of infection. In the FAS-II system, kasA, one of the kas operon genes (Gupta and Singh, 2008), was modestly induced (twofold), while expression of the unlinked fabG1 decreased ∼ 10-fold. In addition, fabH was strongly downregulated (25-fold) (Fig. 2I and Table S2). The downregulation of fabH strongly suggests uncoupling of the FAS-I- and FAS-II-mediated reactions. The potential consequences of such uncoupling on the fate of the FAS-I product and on the structure of the cell wall of non-growing bacilli are discussed below (Metabolic scheme II: TAG accumulation and altered cell wall composition).
Synthesis of multimethyl-branched lipids. Multimethyl-branched lipids, which include phthiocerol dimycocerosate (PDIM) and sulpholipids (SL), are uniquely found in the cell envelope of pathogenic mycobacteria, where they have been associated with virulence and modulation of host immune responses (Karakousis et al., 2004). These compounds are synthesized by specialized polyketide synthases that elongate straight-chain fatty acids by stepwise addition of short acyl chains (Jackson et al., 2007). Precursors for the synthesis of PDIM and SL are straight-chain fatty-acid-acyl primers and methylmalonyl-CoA (MMA), the product of PropCoA activation by a PropCoA carboxylase (Gago et al., 2006).
Expression of accD5, which encodes the PropCoA carboxylase β chain, was decreased at day 21, concurrent with bacterial growth arrest. Levels of mutAB, which encodes the MMA-succinyl-CoA mutase that catalyses the interconversion between MMA and succinyl-CoA (Savvi et al., 2008), were below qPCR detection. Genes involved in the synthesis of PDIM were downregulated (4- to 10-fold), while pks2, which encodes a polyketide synthase involved in the synthesis of the main sulpholipid SL-1, was moderately upregulated (> 2-fold) by day 30 (Fig. 2J and Table S2). These data are consistent with PDIM synthesis decreasing and, albeit modestly, SL-1 synthesis increasing as tubercle bacilli enter the non-replicative state.
Metabolic scheme II: TAG accumulation and altered cell wall composition
Based on the results reported above, we propose that the strong downregulation (25-fold) of the fabH product, which bridges the FAS-I and FAS-II systems during mycolic acid synthesis, leads to uncoupling of the FAS-II system (elongation) from FAS-I (de novo synthesis). Consequently, the FAS-I product would be rerouted to other pathways, including TAG synthesis. Indeed, TAG, which is likely produced by de novo synthesis (Sirakova et al., 2006), accumulates in tubercle bacilli as they stop growing in response to stress in vitro, concurrent with decreased fatty acid incorporation into phospholipids and loss of acid-fastness (Deb et al., 2009).
Our data are also consistent with the possibility that the cell wall of non-growing tubercle bacilli differs structurally from that of growing bacilli. Once uncoupled from FAS-I, the FAS-II system would elongate the existing meromycolate chain of mycolic acid rather than the newly made FAS-I products. Additionally, the slight upregulation of SL-1 synthetic genes and the concurrent downregulation of PDIM biosynthetic genes suggest that MMA, a precursor for both PDIM and SL-1, could be rerouted from PDIM to SL-1, in agreement with biochemical data showing that blocked synthesis of PDIM leads to increased relative abundance and mass of SL-1 (Jain et al., 2007). Peptidoglycan, another cell wall component, is also remodelled in non-replicating tubercle bacilli (Lavollay et al., 2008).
Carbon flow redistribution
Collectively, the transcription profiling data presented above lead to a model of carbon flow redistribution during growth arrest of M. tuberculosis (Fig. 3). The model predicts that, when tubercle bacilli stop replicating, the main function of central metabolism shifts from providing energy and biosynthetic precursors for bacterial growth to accumulating the storage compounds TAG and glutamate. A key functional shift occurs at the metabolic node that mediates interconversion of C4 compounds and C3 compounds, as seen in the response to stress by many microorganisms (Cortassa et al., 2000; Sauer and Eikmanns, 2005). The proposed accumulation of glutamate and TAG is another common theme of metabolic adaptation to stress (Kempf and Bremer, 1998; Murphy, 2001; Alvarez and Steinbuchel, 2002; Waltermann and Steinbuchel, 2005). These storage compounds may serve as carbon and energy sources for long-term persistence and re-growth (Low et al., 2009). Aspects of the model, such as upregulation of the glyoxylate shunt and glutamate synthesis, are supported by the results of isotopomer-assisted metabolite analysis of hypoxic adaptation in Mycobacterium smegmatis (Tang et al., 2009).
In silico simulation of M. tuberculosis metabolic state during growth and growth arrest
The metabolic model sketched above is based on transcriptional data obtained from key genes involved in central and lipid metabolism. To begin to assess how transcriptional data correlate with metabolic reactions, we took an in silico approach utilizing metabolic networks and flux balance analysis (FBA). With this approach, simulations are performed in which input is nutritional substrate and output is the biomass composition of an in silico‘cell’. Biomass is formed through reactions ordered in a metabolic network. The outcome of the simulations is the estimation of fluxes through each reaction in the metabolic network required to obtain maximal biomass. We used these methods to predict metabolic flux patterns in growing and non-growing tubercle bacilli and to assess whether the predicted fluxes were compatible with the bacterial mRNA abundance data obtained in vivo. Even though the in vivo transcriptional profiling interrogated only selected metabolic pathways, we used a genome-scale metabolic network of M. tuberculosis (GSMN-TB) [nearly 900 reactions (Beste et al., 2007)] for the simulations in silico, because regulation of metabolism is a global phenomenon rather than one that depends on particular rate-limiting steps (Fell, 1992).
We first constructed two in silico cells by adjusting biomass composition. One cell represented growing M. tuberculosis, while the other represented the more minimal cell composition predicted for non-growing M. tuberculosis (Experimental procedures and Tables S3 and S4A and B). We then used FBA to predict optimal fluxes through each reaction in the network in the models for growing and non-growing cells, when glucose and phospholipid, alone and in combination, were used as carbon sources. Ratios of optimal fluxes were calculated to identify metabolic flux differences between growing and non-growing bacilli. The flux solutions, which are shown in Table S4C–H for all (∼ 100) central metabolism reactions in the GSMN-TB network, were then compared with the qPCR data.
For critical pathways of central metabolism, the flux solutions associated with the comparison between growing and non-growing cells were consistent with the in vivo data (Fig. 4). First, increased flux in the glyoxylate shunt (Rn 103) and methylcitrate cycle (Rn 82–85) was predicted under all nutritional conditions. This result suggests that increased utilization of these pathways is a response to reduced growth rate rather than to carbon source changes. Second, fluxes through the PPP were predicted to decrease, as expected from the mRNA studies. Third, when glucose and phospholipid were used together as substrate, fluxes through the reactions mediated by PEP carboxykinase (pckA) and the malic enzyme (mez) (Rn 72 and 101) towards C3-compound formation were predicted to increase. Fourth, the only glycolytic reaction predicted to increase was that mediated by phosphofructokinase (pfkA,B) (Rn 41), while the gluconeogenic reaction catalysed by fructose 1,6-bisphosphatase (glpX) (Rn 43) was predicted to decrease. The latter two sets of results agree with the PEP and pyruvate synthesis proposed in the metabolic model in Fig. 3.
Differences between in vivo data and in silico predictions were also observed. In general, a lower degree of concordance was seen between experimental data and in silico fluxes when glucose and phospholipids were used separately. For example, fluxes through several TCA cycle steps (Rn 55, 56, 58, and 65) were predicted to be higher in non-growing bacilli in silico (Fig. 4), but the corresponding genes were not upregulated in vivo. Other differences involved obtaining the same metabolic outcome in silico and in vivo, but with different routes/reactions being utilized. For example, the mouse infection data showed upregulation of icl (the first gene in the glyoxylate shunt), while in silico analysis revealed an increase in flux associated with the second reaction (Rn 103, mediated by malate synthase). Also, non-growing cells showed upregulation of pckA, but not mez, in vivo, while fluxes through both reactions were predicted to increase in silico.
In conclusion, the results of the in silico simulations were consistent with the main aspects of the metabolic model derived from the qPCR data, although the same metabolic outcome was sometimes obtained via different routes in silico and in vivo. This result presumably indicates a need for introducing additional constraints in silico to fully recapitulate the in vivo state.
Adaptation of carbon metabolism of M. tuberculosis to stress in vitro
The finding that metabolic shifts accompanying growth arrest were largely independent of nutrient availability in silico led us to test experimentally whether these metabolic shifts were related to bacterial growth rate or to nutrient availability. To address this question, we compared transcription profiles associated with growth arrest of M. tuberculosis in the mouse lung with those of M. tuberculosis cultures subjected to bacteriostatic treatments in media containing different carbon sources.
Metabolic characteristics of growth-arrested bacilli in enriched culture media. We first examined whether transcript abundance changes in the metabolic genes examined in vivo were also seen when M. tuberculosis cells were cultured in standard rich media (containing glucose, fatty acids and amino acids; see Experimental procedures) and subjected to either of two bacteriostatic stressors, gradual O2 depletion (Wayne and Hayes, 1996) and treatment with nitric oxide (NO) released from diethylenetriamine/NO (DETA/NO) (Voskuil et al., 2003).
We found a significant overlap in gene expression profiles between non-growing bacilli in vivo and in vitro. Key carbon metabolic pathways/reactions upregulated in mouse lungs were also induced in growth-arrested cultures (Fig. 5 and Tables S2 and S5A and B). These included glycolysis (pfkB), the glyoxylate shunt (icl), gluconeogenesis (pckA), synthesis of short- to intermediate-chain-length fatty acids (fas), elongation of meromycolate of mycolic acid (kasA) and synthesis of TAG (tgs1) (Fig. 5).
The induction of tgs1 should result in accumulation of TAG in non-growing bacilli. When we tested this possibility in the gradual O2 depletion model by cytological staining and microscopy, we found that hypoxic cultures gradually accumulated neutral lipids while losing acid-fastness (Fig. 6), as previously seen with M. tuberculosis cultures subjected to multiple stress conditions (Deb et al., 2009). Accumulation of TAG provides a phenotypic correlate of tgs1 mRNA increase.
Only two sets of genes showed different profiles under the three conditions (mouse, hypoxia and NO treatment). One pertained to four fad genes, which responded differently in the three conditions, suggesting that fad gene responses are condition-specific. The other was prpCD in the methylcitrate cycle, which was induced during mouse infection and under hypoxia, but not with NO exposure (Figs 2 and 5, and Tables S2 and S5A and B). This result may reflect different levels of the accessory sigma factor E, which regulates the prpCD response (Manganelli et al., 2001) under several stress conditions (P. Datta and M.L. Gennaro, unpubl. data).
The observation that the vast majority of metabolic genes probed gave the same response to three bacteriostatic conditions favours the possibility that these responses are associated with growth arrest rather than with nutrient availability (the latter presumably varies between the in vivo situation and the culture conditions utilized in vitro). This possibility was tested by utilizing fatty-acid-free culture media (below).
Metabolic characteristics of growth-arrested bacilli in media containing defined carbon sources. We next carried out transcription profiling of M. tuberculosis cultures grown in two media containing defined carbon sources but no fatty acid. One culture medium was simplified 7H9 medium in which glucose and glutamate serve as carbon and energy source. The other was LIM minimal medium in which glucose was the sole carbon and energy source (media composition is found in Table S6A and B). Potential sources of fatty acids, such as bovine serum albumin (BSA) and Tween-80, were excluded (see Experimental procedures).
When M. tuberculosis cultures grown in the above media were treated with bacteriostatic concentrations of NO, we found that key metabolic genes, such as icl, pckA and tgs1, were induced (Table 1). These data clearly establish that the upregulation of pathways, such as the glyoxylate shunt, gluconeogenesis and TAG synthesis, are not necessarily linked with a change of substrate from sugar to fatty acids during infection, but with the change of bacterial growth rate associated with the response to stress.
|Time||Fold change of mRNA copy number|
Activity of key central metabolic enzymes during growth arrest in vitro. We also measured changes in enzyme activities occurring during hypoxia to determine how they correlated with expression of the corresponding genes under the same conditions. We first tested activity of three enzymes encoded by genes seen as upregulated in the hypoxic time-course. The activity of isocitrate lyase and PEP carboxykinase increased during hypoxic stress (Fig. 7A–D). The activity of methylcitrate synthase, which is encoded by prpC, was much lower than that of citrate synthase (∼ 35- to 80-fold) (Table S7), likely reflecting the relative low level of prpC expression. We then tested activity of the products of three genes that were downregulated during hypoxia. We found that malate dehydrogenase activity decreased during the hypoxic course (Fig. 7E and F). A decrease was also observed during hypoxic stress for glutamine synthetase activity (∼ 10-fold), which was more pronounced than the 5-fold decrease in mRNA abundance of glnA1, the major glutamine synthetase gene of M. tuberculosis (Tullius et al., 2003; Harth et al., 2005) (Fig. 7G and H). The pattern seen with glutamine synthetase and glnA1 was also seen with glutamate synthase activity and gltB expression. In this case, enzyme activity was low (< 0.005 ± 0.001 mmoles min−1 mg−1) during logarithmic growth and was further reduced to below detection (< 0.0001 mmoles min−1 mg−1) during hypoxic stress (Table S7). Taken together, we found that the changes in enzyme activity correlated well with changes in RNA abundance with single isoform enzymes.
When multiple enzyme isoforms existed or when genes encoding isoforms of the same enzyme responded differently to the hypoxic stress, comparisons between enzymatic activity and RNA abundance were more complex. We examined two such cases. In one example, the observed reduction in the activity of citrate synthase, which is expressed from citA and gltA2, tracked changes of mRNA abundance for citA, but not for gltA2 (Fig. 7I and J). In another example, we observed no significant increase in phosphofructokinase activity (which is presumably expressed from pfkA and pfkB) during hypoxic conditions, even though pfkB is strongly induced (up to ∼ 100-fold) (Fig. 7K and L). The phosphofructokinase activity, which catalyses the irreversible, rate-limiting step of glycolysis, is highly regulated (Thomas et al., 1972; Doelle, 1975; Kotlarz et al., 1975), and the assay conditions may not reflect critical requirements for activity of this enzyme, such as the concentration of regulatory metabolites or relative oxygen tension.
Collectively, with the exception of phosphofructokinase, the changes in enzyme activities expressed by genes critical to our metabolic model largely support the mRNA abundance data.
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
The model proposed in Fig. 3 provides an integrated view of components of the response of M. tuberculosis metabolism to stress conditions imposed by host immunity. Key aspects of the model are: (i) the PEP/pyruvate-OAA node is a critical switch of carbon flow distribution, as seen in other bacterial species under stress (Sauer and Eikmanns, 2005), and (ii) carbon flow is rerouted towards formation of storage compounds, such as TAG and glutamate, as also seen in the stress response of many biological systems, as diverse as sporulating bacteria and fungi, plant seeds, and hibernating mammals (Cortassa et al., 2000; Murphy, 2001; Alvarez and Steinbuchel, 2002; Waltermann and Steinbuchel, 2005). Thus, our metabolic model characterizes mycobacterial dormancy as a variation of the broader stress–response theme developed in nature.
Our results redefine a central aspect of M. tuberculosis biology. Our finding that metabolic activities central to the model are upregulated in response to stress regardless of carbon source or stress signal contradicts the view that these activities are induced in response to change in nutrient availability from sugar to fat (e.g. Honer zu Bentrup and Russell, 2001; Timm et al., 2003). The preferential utilization of fatty acids by tubercle bacilli explains the observed requirement for glyoxylate shunt and gluconeogenic genes, such as icl and pckA, during infection, regardless of growth rate (Munoz-Elias and McKinney, 2005; Marrero et al., 2010). Instead, upregulation of the same (and other) metabolic pathways associated with stress-induced bacteriostasis is best interpreted in the context of cell fate decisions (transition from growth to dormancy) rather than as an adaptive response to changed nutrient availability.
A limitation of our metabolic model is that it is largely based on transcript abundance data. However, critical aspects of the model are consistent with in silico simulations. Moreover, our experiments with hypoxic cultures show high concordance between expression and activity of key central metabolism enzymes affected during stress-induced bacteriostasis in vitro, suggesting that a similar concordance exists also in vivo. Furthermore, the accumulation of TAG seen in hypoxic cultures provides a phenotypic correlate to a key prediction of the model. Further ways to test the model will include direct measurements of carbon flux distribution using 13C-labelled substrates (Eisenreich et al., 2010).
In conclusion, we propose that the carbon flow redistribution associated with M. tuberculosis growth arrest is a developmental response comparable to that seen with other biological examples of adaptation to stress.
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
C57BL/6 mice at 8–10 weeks of age were infected with M. tuberculosis H37Rv with ≈ 2 × 102 bacterial colony-forming units (cfu) per mouse, as described (Shi et al., 2003). At specified times (days 12, 15, 18, 21, 30, 50 and 100), lungs from three to four mice were harvested and snap-frozen in liquid nitrogen for subsequent RNA extraction.
Treatment of M. tuberculosis cultures in vitro
Unless otherwise specified, M. tuberculosis H37Rv was grown at 37°C in Dubos Tween-Albumin broth (Becton Dickinson), which contains glucose, asparagine and amino acids from a pancreatic digest of casein, and Tween-80 to prevent cell clumping. Tween-80 releases oleic acids that can also be utilized by M. tuberculosis as carbon source (Cox et al., 1978). Gradual oxygen depletion and treatment with DETA/NO were performed as described (Wayne and Hayes, 1996; Shi et al., 2005). For culturing M. tuberculosis with defined carbon sources, two additional culture media were used: a modified Middlebrook 7H9 broth with glutamic acid and glucose as carbon source (Table S6A), and LIM minimal medium containing glucose as sole carbon source (Table S6B). Sources of fatty acids were eliminated in the media by replacing the surfactant Tween-80 with the non-metabolizable Tyloxapol and by dispensing of BSA. Mid-log cultures grown in these media were treated with 100 µM DETA/NO and culture aliquots were harvested at hours 0.5, 1 and 2 by centrifugation and snap-frozen in dry-ice/ethanol for subsequent RNA extraction.
Enumeration of bacterial transcripts
RNA extraction, RT-PCR and real-time PCR for bacterial transcript enumeration from infected mouse lungs and cultures were performed, as described (Shi et al., 2003). Briefly, total RNA extraction from lungs or cultures utilized rapid mechanical lysis of M. tuberculosis in a guanidinium thiocyanate-based buffer. Reverse transcription was performed with gene-specific primers (Table S8) and ThermoScript reverse transcriptase (Invitrogen). Quantification of M. tuberculosis mRNAs was carried out by real-time PCR using gene-specific primers, molecular beacons (Table S8) and AmpliTaq Gold polymerase (Applied Biosystems) in a Stratagene Mx4000 thermal cycler. Measurements of mRNA copy numbers per cell were obtained by dividing mRNA copy number by the copy number of 16S rRNA in the same sample, since 16S rRNA copy numbers closely parallel cfu determinations throughout mouse lung infection (R2 = 0.9539) (Shi et al., 2003) and in hypoxic cultures in vitro (Desjardin et al., 2001).
In silico simulation of M. tuberculosis metabolic pathways at different growth states
A GSMN-TB (Beste et al., 2007) and FBA were used to assess selected metabolic pathways for growing and non-growing bacilli in silico. GSMN-TB contains nearly 900 reactions, including the variant TCA cycle containing a-ketoglutarate decarboxylase (kgd) and succinic semialdehyde dehydrogenase (Tian et al., 2005). In the metabolic model, input was defined as substrate utilized; output was the generation of biomass of tubercle bacilli. Biomass of ‘growing cells’ was defined as consisting of DNA, RNA, protein, and major components of the cell wall, as previously described (Beste et al., 2007), while biomass composition for ‘non-growing cells’ comprised trehalose dimycolate, TAG and polyglutamate/glutamine to reflect a shift towards the minimal cell wall composition deduced from the transcription-derived model in Fig. 3 (detailed biomass composition in Tables S3 and S4A and B). Flux distributions were obtained for maximal biomass production at each growth state by using FBA, as previously described (Beste et al., 2007). Briefly, the FBA method uses linear algebra to calculate the flux in each metabolic reaction of the model that optimizes some ‘objective function’, in our case, the production of biomass. From the perspective of the FBA analysis, the difference between growing and non-growing cells was the different biomass composition. Therefore, FBA was used to calculate flux values that would optimize production of biomass from growing and non-growing cells. The ratio between these predicted flux values (for growing and non-growing cells) provided a predicted difference of metabolic flux in each reaction between the two states.
Fluorescent staining with Auramine-O and Nile Red
Fluorescent acid-fast staining dye Auramine-O was used in combination with lipid staining dye Nile Red (9-Diethylamino-5H-benzo-α-phenoxazine-5-one). M. tuberculosis cultures were taken at multiple time points during hypoxia and heat-killed prior to staining. About 300 µl of cell suspensions were added onto a poly-lysine-coated chambered borosilicate cover glass system and incubated for 30 min at room temperature. After removing excess cell suspension, chambers were air-dried before staining. Acid-fast staining was performed by using Mycobacteria Fluorescent stain kit (Fluka). Briefly, each chamber was filled with 0.3% fluorescent Auramine-O (in 3% phenol solution), incubated in the dark for 15 min, washed with distilled water and treated with decolorizing solution for 2 min. After washing with distilled waster, cells were stained with Nile Red dye (10 µg ml−1 in ethanol), incubated in the dark for 15 min, washed with distilled water, treated with counterstain (0.1% potassium permanganate) for 3 min, washed with distilled waster and air-dried. Each chamber was examined with 100 × oil immersion lens using a fluorescent microscope AXIOVERT 200 M (Zeiss). The filter sets used for visualization of cells stained with Auramine-O and Nile Red were Fluorescein (excitation wavelength 475 nm and emission wavelength 530 nm) and Texas Red (excitation wavelength 590 and emission wavelength 630 nm) respectively. Scanned images were analysed and projected by using the OPEN LAB software (Perkin Elmer).
Preparation of cell-free extracts and enzyme assays
Mycobacterium tuberculosis was grown to either mid-log phase with an approximate OD580 of 0.4 under aerobic conditions and to multiple time points of non-replicating persistence following gradual oxygen depletion in 500 ml flasks. Cultures were harvested by placing them on ice for 2 h and then washing them twice by centrifugation with cold 10 mM phosphate buffer (pH 7.0) containing 0.05% Tween-80. Cells were disrupted in 100 mM phosphate buffer (pH 7.0) containing 1 mM ethylenediaminetetraacetic acid (EDTA), 1 mM dithiothreitol (DTT) and 1 mM phenylmethylsulphonyl fluoride (PMSF) using a Mini-Bead Beater (Bio Spec Products, Bartlesville, OK) with six repetitions of 30 s bursts with cooling on ice between each cycle. Extracts were clarified by centrifugation, and they were sterilized by filtration through a 0.45 µm pore SpinX filter (Corning, New York). The protein concentration was determined with Bio-Rad protein assay reagents (Bio-Rad Laboratories, Hercules, CA, USA).
Isocitrate lyase. Isocitrate lyase activity was measured as the amount of glyoxylphenylhydrozone produced from glyoxylate (Hoyt et al., 1991). Fifty micrograms of cell extract protein was added to a cuvette containing 50 mM MOPS buffer (pH 7.5), 5 mM MgCl2 and 4 mM phenylhydrazine hydrochloride. The reaction was started with the addition of trisodium isocitrate to 2 mM, and the absorbance at 324 nm monitored. A control reaction without isocitrate was subtracted as background. An extinction coefficient 1.7 × 104 M−1 cm−1 was used to calculate the rate of glyoxylate phenylhydrazone formation.
PEP carboxykinase. GTP-dependent PEP carboxykinase was measured with a modified protocol (Mukhopadhyay et al., 2001). Fifty micrograms of cell extract protein was added to a reaction mix containing 200 mM HEPES buffer (pH 7.2), 200 µM MnCl2, 4 mM MgCl2, 10 mM PEP, 100 mM NaHCO3, 200 µM NADH and 3 units of malic dehydrogenase from Thermus flavus. The reaction was initiated with the addition of GDP to 2 mM and followed by measuring absorbance at 340 nm. An extinction coefficient of 6223 M−1 cm−1 was used to calculate the rates of NADH oxidation. A control reaction without PEP was included and subtracted as background.
Malate dehydrogenase. Malate dehydrogenase activity was measured by the oxidation of NADH in the presence of OAA according to published protocol with minor modifications (Tian et al., 2005). Reaction mixture contains 50 µg of protein, 50 mM HEPES (pH 8.0), 0.4 mM NADH and 0.4 mM OAA. A control reaction without OAA was included and subtracted as background.
Citrate synthase. The assay was carried out according to published protocols (Munoz-Elias et al., 2006). Briefly, 50 µg of cell extract protein was added to a 1 ml cuvette containing 50 mM HEPES buffer (pH 8.0), 100 µM 5,5-dithiobis-(2-nitrobenozate) and 400 µM OAA (pH 8.0). The reaction was initiated with the addition of AcCoA to 70 µM and thionitrobenzene formation was followed by measuring absorbance at 412 nm. An extinction coefficient of 13 600 M−1 cm−1 was used to calculate the rates. A control reaction without OAA was subtracted as background.
Methylcitrate synthase. The assay for methylcitrate synthase activity was the same as for citrate synthase with the exception that AcCoA was replaced with PropCoA.
Phosphofructokinase. Phosphofructokinase activity measurement was based on an assay involving aldolase, triose phosphate isomerase and glycerol phosphate dehydrogenase (Alves et al., 1997). Fifty micrograms of cell extract protein was added to a reaction mix (pH 7.5) containing 50 mM Tris, 5 mM MgCl2, 3 mM NH4Cl, 1 mM KCl, 7 mM fructose 6-P, 160 µM NADH, 1 unit of fructose bisphosphate aldolase, 5 units of 3-P isomerase and 1 unit of α-glycerol 3-P dehydrogenase. The reaction was initiated with the addition of ATP to 2.5 mM, and NADPH oxidation was monitored by measuring the absorbance at 340 nm. A control reaction without 6-P fructose was included and subtracted as background.
Glutamate synthase. The assay was carried out according to published protocols (McCarthy, 1983). Fifty micrograms of cell extract protein was added to a reaction mix containing 50 mM HEPES buffer (pH 8.0), 5 mM glutamine and 160 mM NADH. NADH oxidation was measured by monitoring absorbance at 340 nm. A control reaction without glutamine acid was included and subtracted as background.
Glutamine synthetase. The assay measures production of γ-glutamylhydroxamate by glutamine synthetase (Tullius et al., 2001). Components were added to a final concentration of 20 mM Imidazole (pH 7.0), 20 mM arsenate, 60 mM hydroxylamine, 1 mM MnCl2, 400 mM ADP and 50 µg of cell extract protein. The reaction was initiated with the addition of alpha-ketoglutarate to 5 mM. At 30 min intervals a 500 µl sample was removed and added to 125 µl of Stop Solution (8% w/v trichloroacetic acid, 3.3% w/v FeCl3 and 2 N HCl). The absorbance at 540 nm was measured and compared with a standard curve.
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
We thank Ron LaCourse for mouse infections; Michelle Thissen for technical help with the enzyme assays; Gyanu Lamichhane, Lenny Mindich, Issar Smith and Xilin Zhao for critical reading of this manuscript. We give special thanks to Karl Drlica for the generous contribution of his time to discussing this work. L.S. is the recipient of a CFAR grant from New York University. C.P. is the recipient of a scholarship from the NIH Fogarty International Center. The work was supported by grants from the NIAID (R.J.N., L.S. and M.L.G.), the Medical Research Services of the US Department of Veterans Affairs (C.D.S.), the Biotechnology and Biological Sciences Research Council (J.J.McF.) and the Futura Foundation (M.L.G.).
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
- 2002) Triacylglycerols in prokaryotic microorganisms. Appl Microbiol Biotechnol 60: 367–376. , and (
- 1997) Identification of ATP-dependent phosphofructokinase as a regulatory step in the glycolytic pathway of the actinomycete Streptomyces coelicolor A3(2). Appl Environ Microbiol 63: 956–961. , , , and (
- 2007) GSMN-TB: a web-based genome-scale network model of Mycobacterium tuberculosis metabolism. Genome Biol 8: R89. , , , , , , et al. (
- 2007) The Mycobacterium tuberculosis FAS-II condensing enzymes: their role in mycolic acid biosynthesis, acid-fastness, pathogenesis and in future drug development. Mol Microbiol 64: 1442–1454. , , , , and (
- 2000) Lipid lunch for persistent pathogen. Nature 406: 683–685. (
- 1956) Biochemical differentiation of Mycobacterium tuberculosis grown in vivo and in vitro. J Bacteriol 72: 132–141. , and (
- 2003) Structure, function, and biogenesis of the cell wall of Mycobacterium tuberculosis. Tuberculosis (Edinb) 83: 91–97. (
- 2000) Identification and substrate specificity of beta-ketoacyl (acyl carrier protein) synthase III (mtFabH) from Mycobacterium tuberculosis. J Biol Chem 275: 28201–28207. , , , and (
- 1998) Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 393: 537–544. , , , , , , et al. (
- 2002) Production of avirulent mutants of Mycobacterium bovis with vaccine properties by the use of illegitimate recombination and screening of stationary-phase cultures. Microbiology 148: 3019–3027. , , , , , , and (
- 2000) Dynamics of metabolism and its interactions with gene expression during sporulation in Saccharomyces cerevisiae. In Advance in Microbial Physiology. Poole, R.K. (ed.). San Diego, San Francisco, New York, Boston, London, Sydney, Tokyo: Academic Press, pp. 75–115. , , , and (
- 1978) Rapid Tween 80 hydrolysis test for mycobacteria. J Clin Microbiol 7: 104–105. , , , , and (
- 2004) Induction of a novel class of diacylglycerol acyltransferases and triacylglycerol accumulation in Mycobacterium tuberculosis as it goes into a dormancy-like state in culture. J Bacteriol 186: 5017–5030. , , , , , , and (
- 2009) A novel in vitro multiple-stress dormancy model for Mycobacterium tuberculosis generates a lipid-loaded, drug-tolerant, dormant pathogen. PLoS ONE 4: e6077. , , , , , , et al. (
- 2001) Microaerophilic induction of the alpha-crystallin chaperone protein homologue (hspX) mRNA of Mycobacterium tuberculosis. J Bacteriol 183: 5311–5316. , , , , and (
- 1975) ATP-sensitive and ATP-insensitive phosphofructokinase in Escherichia coli K-12. Eur J Biochem 50: 335–342. (
- 2005) responses of mycobacterium tuberculosis to growth in the mouse lung. Infect Immun 73: 3754–3757. , , , and (
- 2010) Carbon metabolism of intracellular bacterial pathogens and possible links to virulence. Nat Rev 8: 401–412. , , , and (
- 1992) Metabolic control analysis: a survey of its theoretical and experimental development. Biochem J 286 (Part 2): 313–330. (
- 2006) Biochemical and structural characterization of an essential acyl coenzyme A carboxylase from Mycobacterium tuberculosis. J Bacteriol 188: 477–486. , , , , and (
- 2002) Intracellular lipophilic inclusions of mycobacteria in vitro and in sputum. Microbiology 148: 2951–2958. , , , , and (
- 2008) Cytological and transcript analyses reveal fat and lazy persister-like bacilli in tuberculous sputum. PLoS Med 5: e75. , , , , , , et al. (
- 1985) Bacterial Metabolism, 2nd edn. New York: Springer-Verlag. (
- 2006) Dual role of isocitrate lyase 1 in the glyoxylate and methylcitrate cycles in Mycobacterium tuberculosis. Mol Microbiol 61: 940–947. , , , , and (
- 2008) Deciphering kas operon locus in Mycobacterium aurum and genesis of a recombinant strain for rational-based drug screening. J Appl Microbiol 105: 1703–1710. , and (
- 2004) Stationary phase gene expression of Mycobacterium tuberculosis following a progressive nutrient depletion: a model for persistent organisms? Tuberculosis (Edinb) 84: 228–238. , , , , , , et al. (
- 2000) Treatment of Mycobacterium tuberculosis with antisense oligonucleotides to glutamine synthetase mRNA inhibits glutamine synthetase activity, formation of the poly-l-glutamate/glutamine cell wall structure, and bacterial replication. Proc Natl Acad Sci USA 97: 418–423. , , , , and (
- 2005) All four Mycobacterium tuberculosis glnA genes encode glutamine synthetase activities but only GlnA1 is abundantly expressed and essential for bacterial homeostasis. Mol Microbiol 58: 1157–1172. , , , and (
- 2001) Mycobacterial persistence: adaptation to a changing environment. Trends Microbiol 9: 597–605. , and (
- 1991) Purification and characterization of Acinetobacter calcoaceticus isocitrate lyase. J Bacteriol 173: 6844–6848. , , and (
- 2007) Long-chain multiple methyl-branched fatty acid-containing lipids of Mycobacterium tuberculosis: biosynthesis, transport, regulation and biological activities. Tuberculosis (Edinb) 87: 78–86. , , and (
- 2007) Lipidomics reveals control of Mycobacterium tuberculosis virulence lipids via metabolic coupling. Proc Natl Acad Sci USA 104: 5133–5138. , , , , , , et al. (
- 1975) Pathways of carbohydrate metabolism in mycobacterium tuberculosis H37Rv1. Can J Microbiol 21: 1688–1691. , , , and (
- 2004) Mycobacterium tuberculosis cell envelope lipids and the host immune response. Cell Microbiol 6: 105–116. , , and (
- 1998) Uptake and synthesis of compatible solutes as microbial stress responses to high-osmolality environments. Arch Microbiol 170: 319–330. , and (
- 1975) Regulation of the amount and of the activity of phosphofructokinases and pyruvate kinases in Escherichia coli. Biochim Biophys Acta 381: 257–268. , , and (
- 2008) The peptidoglycan of stationary-phase Mycobacterium tuberculosis predominantly contains cross-links generated by l,d-transpeptidation. J Bacteriol 190: 4360–4366. , , , , , , et al. (
- 2003) pckA-deficient Mycobacterium bovis BCG shows attenuated virulence in mice and in macrophages. Microbiology 149: 1829–1835. , , and (
- 2009) Triacylglycerol utilization is required for the re-growth of in vitro hypoxic non-replicating Mycobacterium bovis Bacillus Calmette-Guerin. J Bacteriol 191: 5037–5043. , , , , , , and (
- 1983) Continuous culture of Mycobacterium avium limited for ammonia. Am Rev Respir Dis 127: 193–197. (
- 2000) Persistence of Mycobacterium tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase. Nature 406: 735–738. , , , , , , et al. (
- 2001) The Mycobacterium tuberculosis ECF sigma factor sigmaE: role in global gene expression and survival in macrophages. Mol Microbiol 41: 423–437. , , , and (
- 2010) Gluconeogenic carbon flow of tricarboxylic acid cycle intermediates is critical for Mycobacterium tuberculosis to establish and maintain infection. Proc Natl Acad Sci USA 107: 9819–9824. , , , , and (
- 2004) The Mycobacterium tuberculosis Rv1099c gene encodes a GlpX-like class II fructose 1,6-bisphosphatase. Microbiology 150: 3499–3505. , , , , , and (
- 2001) A GTP-dependent vertebrate-type phosphoenolpyruvate carboxykinase from Mycobacterium smegmatis. J Biol Chem 276: 16137–16145. , , and (
- 2005) Mycobacterium tuberculosis isocitrate lyases 1 and 2 are jointly required for in vivo growth and virulence. Nat Med 11: 638–644. , and (
- 2006) Role of the methylcitrate cycle in Mycobacterium tuberculosis metabolism, intracellular growth, and virulence. Mol Microbiol 60: 1109–1122. , , , and (
- 2001) The biogenesis and functions of lipid bodies in animals, plants and microorganisms. Prog Lipid Res 40: 325–438. (
- 2009) Foamy macrophages and the progression of the human tuberculosis granuloma. Nat Immunol 10: 943–948. , , , , and (
- 2003) Genetic requirements for mycobacterial survival during infection. Proc Natl Acad Sci USA 100: 12989–12994. , and (
- 2005) The PEP-pyruvate-oxaloacetate node as the switch point for carbon flux distribution in bacteria. FEMS Microbiol Rev 29: 765–794. , and (
- 2008) Functional characterization of a vitamin B12-dependent methylmalonyl pathway in Mycobacterium tuberculosis: implications for propionate metabolism during growth on fatty acids. J Bacteriol 190: 3886–3895. , , , , , and (
- 2003) Transcriptional adaptation of mycobacterium tuberculosis within macrophages: insights into the phagosomal environment. J Exp Med 198: 693–704. , , , , , , et al. (
- 2003) Expression of Th1-mediated immunity in mouse lungs induces a Mycobacterium tuberculosis transcription pattern characteristic of nonreplicating persistence. Proc Natl Acad Sci USA 100: 241–246. , , , , and (
- 2005) Changes in energy metabolism of Mycobacterium tuberculosis in mouse lung and under in vitro conditions affecting aerobic respiration. Proc Natl Acad Sci USA 102: 15629–15634. , , , , , , and (
- 2006) Identification of a diacylglycerol acyltransferase gene involved in accumulation of triacylglycerol in Mycobacterium tuberculosis under stress. Microbiology 152: 2717–2725. , , , , , , and (
- 2009) Central metabolism in Mycobacterium smegmatis during the transition from O2-rich to O2-poor conditions as studied by isotopomer-assisted metabolite analysis. Biotechnol Lett 31: 1233–1240. , , , , , and (
- 1972) Effect of oxygen on several enzymes involved in the aerobic and anaerobic utilization of glucose in Escherichia coli. J Bacteriol 112: 1099–1105. , , , and (
- 2005) Variant tricarboxylic acid cycle in Mycobacterium tuberculosis: identification of alpha-ketoglutarate decarboxylase. Proc Natl Acad Sci USA 102: 10670–10675. , , , , and (
- 2003) Differential expression of iron-, carbon-, and oxygen-responsive mycobacterial genes in the lungs of chronically infected mice and tuberculosis patients. Proc Natl Acad Sci USA 100: 14321–14326. , , , , , , et al. (
- 2001) High extracellular levels of Mycobacterium tuberculosis glutamine synthetase and superoxide dismutase in actively growing cultures are due to high expression and extracellular stability rather than to a protein-specific export mechanism. Infect Immun 69: 6348–6363. , , and (
- 2003) Glutamine synthetase GlnA1 is essential for growth of Mycobacterium tuberculosis in human THP-1 macrophages and guinea pigs. Infect Immun 71: 3927–3936. , , and (
- 2003) Inhibition of respiration by nitric oxide induces a Mycobacterium tuberculosis dormancy program. J Exp Med 198: 705–713. , , , , , , and (
- 1985) Branch point control by the phosphorylation state of isocitrate dehydrogenase. A quantitative examination of fluxes during a regulatory transition. J Biol Chem 260: 8430–8437. , and (
- 2005) Neutral lipid bodies in prokaryotes: recent insights into structure, formation, and relationship to eukaryotic lipid depots. J Bacteriol 187: 3607–3619. , and (
- 1996) An in vitro model for sequential study of shiftdown of Mycobacterium tuberculosis through two stages of nonreplicating persistence. Infect Immun 64: 2062–2069. , and (
- 1997) Regulation of acetate metabolism in Corynebacterium glutamicum: transcriptional control of the isocitrate lyase and malate synthase genes. Arch Microbiol 168: 262–269. , , , , , and (
- 1994) Metabolism of Mycobacterium tuberculosis. In Tuberculosis: Pathogenesis, Protection, and Control. Bloom, B.R.E. (ed.). Washington, DC: American Society for Microbiology Press, pp. 353–385. , and (
- 1999) Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc Natl Acad Sci USA 96: 12833–12838. , , , , , , and (
- 2004) Persistent and dormant tubercle bacilli and latent tuberculosis. Front Biosci 9: 1136–1156. (
- Top of page
- Results and discussion
- Experimental procedures
- Supporting Information
|MMI_7399_sm_TableS1-8.pdf||1488K||Supporting info item|
|MMI_7399_sm_TableS4.xls||113K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.