Bacterial immunostat : Mycobacterium tuberculosis lipids and their role in the host immune response

The lipid-rich cell wall of Mycobacterium tuberculosis is a dynamic structure that is involved in the regulation of the transport of nutrients, toxic host-cell effector molecules, and anti-tuberculosis drugs. It is therefore postulated to contribute to the longterm bacterial survival in an infected human host. Accumulating evidence suggests that M. tuberculosis remodels the lipid composition of the cell wall as an adaptive mechanism against host-imposed stress. Some of these lipid species (trehalose dimycolate, diacylated sulphoglycolipid, and mannan-based lipoglycans) trigger an immunopathologic response, whereas others (phthiocerol dimycocerosate, mycolic acids, sulpholipid-1, and di-and polyacyltrehalose) appear to dampen the immune responses. These lipids appear to be coordinately expressed in the cell wall of M. tuberculosis during different phases of infection, ultimately determining the clinical fate of the infection. This review summarizes the current state of knowledge on the metabolism, transport, and homeostatic or immunostatic regulation of the cell wall lipids, and their orchestrated interaction with host immune responses that results in bacterial clearance, persistence, or tuberculosis.


INTRODUCTION
Tuberculosis remains an important underlying cause of death from lung disease worldwide.Exposure to Mycobacterium tuberculosis, a causative agent of tuberculosis, can result in immediate clearance, establishment of latent tuberculosis infection, a rapidly progressive disease, or reactivation disease 1 .More than 90% of otherwise healthy individuals who become newly infected with M. tuberculosis do not develop an active disease 1 .However, about 60% fail to eliminate the bacteria and they may remain infected for life 1 .It is currently not known why only a subset of those with latent tuberculosis infection progresses to develop active disease.Progression from latent infection to active disease mostly occurs in individuals without any obvious conditions related to the host.Furthermore, a subset of those who develop tuberculosis and complete a course of treatment will relapse.In human, the rate of relapse varies between 8-13%, depending on the prevalence of multidrugresistant tuberculosis (MDRTB) in that population 2 .Various M. tuberculosis factors, as well as host factors, have been proposed to affect these clinical outcomes, but in this review, we discuss the possible role of M. tuberculosis cell wall lipids in determining these outcomes.
During a pulmonary infection, aerosolized M. tuberculosis enters the lungs and encounters the alveolar air space.The alveolar air space is shaped by type I and II pneumocytes.Type I pneumocytes cover about 96% of the alveolar surface area, whereas type II cells cover about 4% 3 .Traditionally, it was thought that M. tuberculosis is engulfed by alveolar macrophages after entering the lungs, but the cellular architecture of the alveolar space suggests that M. tuberculosis will most likely interact with these pneumocytes first 3 .Later, macrophages and lymphocytes migrate to the site of infection and form a granuloma (reviewed in) 4 .
One unique feature of M. tuberculosis is its lipid-rich envelope (Figure 1) 5,6 .About 40% of its cell-wall dry weight is comprised of lipids and, as such, a large portion of the coding capacity of the bacterial genome is devoted to lipid biosynthesis and degradation 7 .Several cell wall lipids are associated with a distinct host innate immune response during the early phase of infection.Differential expression of these lipids determines whether or not M. tuberculosis will enter a chronic phase of infection.
The ability of M. tuberculosis to persist within a host appears to rely on its beta-oxidation and glyoxylate metabolic cycles.In mouse, the bacterium utilizes fatty acids as an exclusive carbon source 8 .During the beta-oxidation cycle, even chain fatty acids are catabolized into acetyl coenzyme A (acetyl-CoA), and odd chain fatty acids into acetyl-CoA and propionyl-CoA 8,9 .The bacillus prevents toxic accumulation of propionyl-CoA by synthesizing methylmalonyl-CoA-based lipids [phthiocerol dimycocerosate (PDIM), sulpholipid-1 (SL-1), and di-and polyacyltrehalose], as suggested by Muñoz-Elías et al. 10 .Glyoxylate cycle converts acetyl-CoA molecules derived from fatty acid beta-oxidation into tricarboxylic acid cycle intermediates 11 .In addition, acetyl-CoA is a substrate for mycolic acid (MA), a malonyl-CoA-based lipid 11,12 .M. tuberculosis adjusts its lipid metabolism as it encounters limiting nutrient conditions imposed by the host.By alternatively activating the two pathways, the bacteria synthesize both methylmalonyl-and malonyl-based lipids, which allow M. tuberculosis to establish persistence.
Here, we will first review the effect of other M. tuberculosis cell wall lipids [trehalose dimycolate (TDM) and mannanbased lipoglycans] during the early, innate immunity phase of interaction with the host.Then, we will review the interaction between the products of lipid beta-oxidation and the host that becomes important during the transition into a chronic infection phase when host adaptive immunity sets in.These lipid-induced immune responses are summarized in Table 1.

EARLY PHASE OF INFECTION AND MYCOBACTERIUM TUBERCULOSIS LIPID METABOLISM
The innate immune system comprises the first line of host defense against M. tuberculosis.As part of this initial response, pattern recognition receptors Toll-like receptor (TLR)-2 and TLR-4 on alveolar macrophages and dendritic cells (and probably also on pneumocytes) engage in recognizing TDM (6,6'-dimycoloyl-α-d-trehalose) and mannosylated lipoarabinomannan of M. tuberculosis 13,14 .

Trehalose dimycolate
Trehalose dimycolate, formerly referred to as cord factor, is a glycolipid where a disaccharide is linked to two mycolyl chains.Its precursor, trehalose monomycolate (TMM), has only one mycolate chain.
The cord factor has long been described as a virulence factor of M. tuberculosis.Division of the H37 strain into highly virulent (H37Rv) and avirulent (H37Rva) variants according to specific culture characteristics led Middlebrook et al. 15 to link bacterial morphology in liquid and solid media to strain virulence.A parenteral injection of the cord factor was toxic to mice and inhibited leukocyte migration 16 .The toxic compound was later identified as TDM 17 .Injection of TDM into mouse lungs elicited granulomatous response 18 .
Cell wall MAs are normally covalently linked to other molecules, such as TDM and arabinogalactan.Recently, Cantrell et al. 19 and Ojha et al. 20 reported that M. tuberculosis cell wall contains unattached MA (free MA or FM) (Figure 1).One potential source of FM is the hydrolysis of TDM.Ojha et al. 21identified a gene Msmeg_1529 that encodes a cutinase-like serine carboxyesterase responsible for TDM hydrolysis in Mycobacterium smegmatis.The enzymatic activity of Msmeg_1529 was specific for purified TDM from M. smegmatis and M. tuberculosis 21 .Among seven cutinaselike proteins with esterase activities previously described in M. tuberculosis, the authors suggested that Rv3452 is a possible TDM hydrolase in M. tuberculosis because of its closest relatedness to Msmeg_1529 21 .Indeed, it was shown later that Rv3452, and also Rv3451, share > 50% amino acid similarity with Msmeg_1529 22 .Experiments revealed that Rv3451 is the primary TDM hydrolase in M. tuberculosis 22 .Thus, TDM may be one source of FM.In fact, Yang et al. 22 estimated that about 20% of FM in the cell wall may be derived from the hydrolysis of TDM.
Ishikawa et al. 23 showed that macrophage inducible C-type lectin (Mincle) is a receptor for mycobacterial TDM.The activation of Mincle leads macrophages to produce nitric oxide and induce the production of tumor necrosis factor (TNF) and MIP-2 (also called CXCL2), and a subsequent granuloma formation 23 .
An alternative model for macrophage activation involves a myeloid differentiation factor 88 (MyD88)-dependent response to TDM 14,24,25 .Bone marrow-derived macrophages (BMMs) from MyD88-knocked out were not able to produce proinflammatory cytokines in response to TDM-coated polystyrene microspheres 24 .Bowdish et al. 25 proposed a model, where macrophage receptor with collagenous structure (MARCO), a class A scavenger receptor, cooperates with TLR-2 and cluster of differentiation 14 (CD14) in TDM recognition and signaling to mount an efficient inflammatory response to M. tuberculosis.Using deoxyribonucleic acid (DNA) microarrays, Sakamoto et al. 14  coated microspheres and showed that matrix metalloproteinases (MMPs) 8, 9, 12, 13, 14, and 19, were up-regulated in a MyD88-dependent manner after stimulation with TDM.Matrix metalloproteinases are a family of enzymes implicated in tissue remodeling and chronic inflammation conditions, and are involved in granuloma formation, cavitation, and fibrous granuloma capsule breakdown 26 .

Mycolic acids
Originally, MAs have not been described as lipids that play a role in the early phase of immune response to infection.However, it has been since suggested that some FMs in the bacterial cell wall are derived from the hydrolysis of TDM 22 and may interact with host cells to induce some type of innate immune response.
In fact, MAs are the major lipids of a protective layer of cell wall, serving as structural units of the cell wall and envelope (Figure 1).They exist in homologous series of fatty acids differing by 28 atomic mass units (a two-carbon unit).In M. tuberculosis, these are extremely hydrophobic C 54 -C 63 fatty acids with C 22 -C 24 α side chains 27 .Three classes of M. tuberculosis MAs exist, alpha, keto, and methoxy-MA 27,28 ; they differ in the functional groups attached to the meromycolate portion of the molecule.
Mycolic acid induces the accumulation of cholesterol inside peritoneal and alveolar macrophages.Macrophages containing cholesterol and lipid droplets resemble foamy macrophage derivatives observed in tuberculous granulomas 29 .Inflammatory response to intratracheal instillation of different doses of MA or TDM was compared; TDM was less effective in inducing the production of interleukin (IL)-6 and IL-12.However, TDM was very effective in inducing tumor necrosis factor alpha (TNF-α), a unique property of this glycolipid.Experiments with TLR-2-and TLR-4-deficient mice yielded similar results, excluding these receptors as potential pattern recognition targets of MA 29 .Recently, Sequeira et al. 30 showed that FM can inhibit TLR-2mediated pro-inflammatory response in RAW 264.7 cells and in human lung epithelial cells (A549).
Indeed, MA may constitute a scaffold for the stimulation of CD1-restricted T cells by mycobacterial lipid antigens 31 .When the antigenicity of glycerol monomycolate (GroMM) was evaluated, the hydroxyl group of glycerol and MA chain length were shown to be critical for triggering T cell responses 31 .It was previously demonstrated that FM stimulates CD1b-restricted T cells, indicating that this structure might form stimulatory complexes with CD1b 32 .
Minor structural alterations of trehalose-attached MA have a profound effect on murine pro-inflammatory response.MA cyclopropane ring is required for virulence and long-term persistence of pathogenic mycobacteria in mice 33,34 .Also, the relative differences in the amounts of keto-and methoxy-MA influence the intra-macrophage growth rate of M. tuberculosis 35 and the absence of keto-and methoxy-MA is associated with bacterial attenuation in mice 36 .
The abundance of MA in M. tuberculosis cell wall appears to affect the course of tuberculosis in infected people.The putative mycobacterial transporter mce1 might play a role in the regulation of cell wall MA abundance in M. tuberculosis.M. tuberculosis has four homologs of mce operon (mce1-4) encoding ATP-binding cassette transporters (ABC transporters) possibly involved in lipid import 7,37,38 .Cantrell et al. 19 and Forrellad et al. 39 reported that disruption of mce1 operon results in a M. tuberculosis mutant that accumulates several-fold more FM in its cell wall than wild type bacteria.Both groups suggested that the operon might encode an importer system involved in recycling of MA.Also, disruption of the mce1operon renders the cells hypervirulent in mice 40,41 .The mutant is unable to induce a strong T-helper 1 (Th1) type T cell immune response and organized granuloma formation in lungs 41 .Taken together, these observations suggest that the mce1 operon is involved in the remodeling of M. tuberculosis cell wall, dampening proinflammatory response in mouse that is associated with rapid progression to death 40,41 .The operon is not expressed in the wild type strain during the first 4-8 weeks of infection in mouse 42 , suggesting that the M. tuberculosis cell wall lipid profile can be altered according to the course and/or time of infection.
Free mycolic acid (FM), a major compound involved in biofilm formation 20,21 , is released from TDM at some point in late infection; the latter is recognized for its ability to activate the innate immune response 14,23 .At that infection stage, (1) the capacity of M. tuberculosis to induce pro-inflammatory response is diminished; (2) the bacterium becomes drug tolerant 22 ; and (3) nutrient uptake increases as the host imposes bacterial starvation 22 .
Queiroz et al. 43 used high-resolution quantitative time-offlight liquid chromatography-mass spectrometry and identified 26 acyl forms of lipid species whose levels were either decreased or increased in mce1 mutant compared with wild type.The mutant strain showed lower levels of different acyl forms of diacyltrehalose (DAT), diacylated sulphoglycolipids (Ac 2 SGL), PDIM, and phosphatidylethanolamine; and increased levels of acyl forms of alpha-, keto-, and methoxy-MA species and hydroxyphthioceranic acid 43 .These cell wall changes appeared to profoundly affect the host response, perhaps ultimately determining the clinical outcome in an infected host.

Mannan-based lipoglycans
Cell wall-associated lipids, such as phosphatidyl-myoinositol mannosides (PIM), and the glycolipids lipomannan (LM) and lipoarabinomannan (LAM), play a key role in modulating the host response during infection by interacting with different receptors on macrophages and dendritic cells 44 .The structure and biosynthesis of PIM, LM, and LAM have already been reviewed 45,46,47 .
Much of the literature focuses on the role of these molecules in maintaining mycobacterial cell wall integrity 48 and growth 49 , but PIMs, LM, LAM, and mannose-capped lipoarabinomannan (Man-LAM) (Figure 1) also play an important role in host defense, and hence, induction of host responses.Interestingly, LM, LAM, Man-LAM, and PIMs distinctly modulate the host immune response.Both phospho-myo-inositol-dimannoside (PIM 2 ) and phospho-myo-inositol-hexamannoside (PIM 6 ) have been described as agonists of TLR-2 50,51 ; Dao et al. 52 showed that LM but not PIM 2 strongly induces IL-12 production and apoptosis.On the other hand, LAM inhibits IL-12 production by human dendritic cells and modulates M. tuberculosis-induced macrophage apoptosis 53 .Man-LAM mimics a mammalian cell mannose receptor, which enables M. tuberculosis to evade host defense mechanisms, inhibiting phagosome maturation by suppressing the production of IL-12 and TNF and increasing IL-10 production by dendritic cells or monocytic cells 54 .
Dao et al. 52 postulated that the LM: LAM ratio in the mycobacterial cell wall can directly affect the virulence of M. tuberculosis and the outcome of an infection.LM activates the inflammatory response of the host.LM, but not LAM, activates the macrophages by eliciting the expression of co-stimulatory molecules CD40 and CD86 in a TLR-2-and MyD88-dependent manner.The signaling pathway leading to cytokine production appears to be independent of TLR-4 and TLR-6 55 .Similarly, Dao et al. 52 demonstrated that LM induces cell signaling specifically through TLR-2 ligation.Based on a structure-function evaluation, these authors suggested that the addition of arabinan to LAM masks the activity of the LM mannan core, preventing LAM from inducing apoptosis and IL-12 expression.Interestingly, although PIM 2 is the lipid anchor of LM, it does not play a role in this induction.Since PIM 2 contains only two mannose residues, Dao et al. 52 suggested that increased mannose content is required for the stimulation of IL-12 expression 52 .
Krishnan et al. 56 evaluated the response of BMMs, dendritic cells, and mice to whole lipid extracts from the six major genetic lineages of M. tuberculosis and concluded that the distinct cell envelope lipid profiles affect mycobacterial virulence and immune innate response.These data reinforce the role of M. tuberculosis cell wall lipid components as a determining factor in the outcome of early phase M. tuberculosis infection.

MYCOBACTERIAL LIPIDS DURING TRANSITION TO THE CHRONIC PHASE OF INFECTION
Once M. tuberculosis survives the early assaults from antimicrobial host effector cells and molecules, it enters a chronic infection phase.This transition involves yet another set of mycobacterial lipids.The host's adaptive immune response at this stage of infection comprises the formation of T cell-dependent granulomas designed to contain the bacilli to prevent their spread and facilitate their elimination 57 .The control of M. tuberculosis depends on a joint effort of T cells, macrophages, and dendritic cells 58 , and M. tuberculosis survives within macrophages inside the granulomas.Major histocompatibility complex (MHC) class II molecules are loaded with mycobacterial peptides that are presented to CD4 T cells.CD8 T cell stimulation requires the loading of MHC I molecules by mycobacterial peptides in the cytosol, by egression of mycobacterial antigens into the cytosol 59 .
The MHC class I-like CD1 antigen-presenting molecules recognize the abundant and diverse set of lipid-containing antigens and present these amphipathic antigens for recognition to T cell receptors (reviewed by Barral and Brenner 60 ).Many lipid species expressed by M. tuberculosis can be presented by CD1a, CD1b, and CD1c, including MA, lipoglycans (such as LAM) and PIMs, lipopeptides, mannosyl-β-1phosphomycoketides, and sulpholipids 60 .Other cell wall lipids that have been described as activating/modulating the immune response include PDIM, DAT, and polyacyltrehalose (PAT) 61 .
Accumulating evidence suggests that the beta-oxidation pathway is a source of lipids that protect M. tuberculosis against host-imposed stress during the chronic phase of infection 62 .As described below, these lipids play a role in dampening the host immune response or protecting the tubercle bacilli against bactericidal intermediates.These beta-oxidation products include DAT, PAT, sulpholipid-1 (SL-1), and PDIM.

Di-and polyacyltrehalose
Mycobacterium tuberculosis glycolipids (DAT, PAT, and triacyltrehaloses) are located in the outer layer of the cell wall (Figure 1) and contain di-and tri-methyl branched fatty acids, only found in pathogenic species of mycobacteria.Hatzios et al. 63 suggested that acyl transferase PapA3 first acylates the 2-position of one of the trehalose glucose residues with a palmitoyl group, to form trehalose 2-palmitate (T2P).A mycolipenoyl group, synthesized by the polyketide synthase Pks3/4, a product of the mas-like gene (msl3) 64,65 , is then transferred to the 3-position of T2P by PapA3 to generate DAT 66 .DAT may either be transported to the cell surface or serve as a biosynthetic intermediate that is further elaborated with mycolipenic acids to yield PAT. Gene mmpL10, which encodes a putative lipid transporter and clusters with papA3 and pks3/4 genes, may encode a transporter of PAT or its precursor (Figure 1).DAT may be transported to the cell surface without further modifications 63 .Thus, the extracellular synthesis of PAT from DAT might constitute yet another mechanism by which M. tuberculosis protects itself against the bactericidal activities of the host during infection.
DAT extracted from M. tuberculosis H37Rv and M. fortuitum ATCC 6841 inhibits the proliferation of mouse T cells in a dose-dependent manner in vitro, which suggests that this glycolipid could play a role in T cell hyporesponsiveness and immunosuppression associated with M. tuberculosis infections 67 .The disruption of the M. tuberculosis msl3 gene provided additional information about the role of polyacylated trehaloses in mycobacterial virulence.A msl3 mutant does not produce mycolipanoic and mycolipenic (phthienoic) acids required for the synthesis of DAT and PAT.In the absence of glycolipids to anchor the trehalose at the cell surface, this mutant forms bead-like aggregates with no discernable decrease in growth rate.Capsule attachment is probably defective in the mutant, and, consequently, the hydrophobic core of the wall might be exposed 64,68 .The msl3 mutant infects both activated and resting macrophages derived from BALB/c mice two-to fourfold more effectively than H37Rv, and is more efficiently taken up by the macrophages 68 .However, the ability of the msl3 mutant to replicate and to persist within macrophages is not altered, suggesting that DAT and PAT are unlikely involved in the survival of M. tuberculosis within phagocytic cells 68 .

Sulpholipid-1
Since the discovery of sulpholipids by Middlebrook et al. 69 in the early 1950s many in vivo and in vitro studies have suggested that these complex lipids play a role in the virulence of the tubercle bacillus.SL-1 comprises a trehalose-2-sulfate (T2S) core decorated with four acyl groups: a straight chain fatty acid (palmitate or stearate) and three multiply methyl-branched (hydroxy) phthioceranoic acids.The sulphotransferase Stf0 initiates SL-1 biosynthesis by sulfating the abundant disaccharide trehalose to form T2S 70 .The acyltransferase PapA2 then catalyzes the esterification of T2S at the 2-position to generate a monoacylated intermediate monoacylated-SGL 71 .The polyketide synthase Pks2 synthesizes methyl-branched (hydroxy) phthioceranoyl 65 and PapA1 transfers the product of Pks2 to the 3-position of monoacylated-SGL, yielding Ac 2 SGL 71 .MmpL8 may flip Ac 2 SGL across the plane of the cell membrane, allowing it to be further modified to mature SL-1 before delivery to the cell wall 72 (Figure 1).Acylation of Ac 2 SGL with a phthioceranic acid group and another hydroxyphthioceranic acid group completed the biosynthesis of SL-1 70 .
The role of Ac 2 SGL and SL-1 in immunopathogenesis of M. tuberculosis has been described 73,74 .SL-1 is the most abundant lipid on the M. tuberculosis surface and absent in nonpathogenic Mycobacterium species.SL-1 and its precursor, Ac 2 SGL, are thus important targets in biomarkers studies.
Goren et al. 73 showed that SL-1 purified from M. tuberculosis strain H37Rv is a potent inhibitor of phagolysosome formation in cultured mouse peritoneal macrophages.The authors suggested that the sulfatide impairs either the phagosomal or lysosomal membranes.SL-1 also directly inhibits macrophage priming, which results in reduced superoxide anion (O 2 -) release, reduced phagocytosis, and reduced release of IL-1 75 , and affects neutrophil signal transduction pathways 76 .Nevertheless, the role of SL-1 in the virulence of M. tuberculosis remains questionable 77 .
Kato and Goren 78 showed that SL-1 damages the mitochondrial structure in synergy with TDM, but no toxic effect was noted in mice injected intraperitoneally with SL-1 alone.Later 72,79 , the construction of mmpL8 and pks2 mutants clarified the role of SL-1 and its precursors in the pathogenesis of M. tuberculosis.MmpL8 is a member of a family of 13 predicted lipid transporters and pks2 has high homology to the mycocerosic acid synthase (mas) gene, whose product is involved in the synthesis of a class of multiple methyl-branched fatty acids 65 .M. tuberculosis mmpL8 mutant is unable to produce SL-1 and accumulates Ac 2 SGL in the cytoplasm.The pks2 mutant that synthesizes neither SL-1 nor Ac 2 SGL grows at rates indistinguishable from those of the wild type in the spleen, liver, and lung, indicating that SL-1 is not required for the growth of M. tuberculosis in vivo.In contrast, the mmpL8 mutant was attenuated in all these organs 72 .These results suggest that Ac 2 SGL is the immunogenic metabolite of the SL-1 pathway.
Gilleron et al. 74 confirmed these findings by showing that Ac 2 SGL binds to the MHC-like lipid receptor CD1b and stimulates the expression of interferon (IFN)-γ and IL-2 in CD8 + T cells obtained from tuberculin skin test-positive donors 74 .Thus, since the biosynthesis of SL-1 is completed by acylation of Ac 2 SGL with a phthioceranic acid group and another hydroxyphthioceranic acid group, it is possible that M. tuberculosis uses these two fatty acids in a balanced fashion to modulate the host immune response to bacterial advantage 74 .

Phthiocerol dimycocerosate
By using signature-tagged mutagenesis (STM), Cox et al. 80 isolated three M. tuberculosis transposon mutants unable to replicate within the mouse lung.These transposons disrupted the mas, fadD28 and mmpL7 genes which encode enzymes and proteins implicated in the synthesis, assembly and, subcellular localization of PDIM 80 .Also, by using a cell-free system, Trivedi et al. 81 delineated the enzymology of PDIM assembly and the precise roles of several proteins involved in the biosynthesis of this lipid.Some genes described as involved in PDIM synthesis are located within 50kbp pps locus 82 .
The observation that a PDIM-deficient M. tuberculosis H37Rv mutant replicates less well and elicits fewer lung surface tubercles than a PDIM-producing H37Rv strain in a guinea pig model suggested for the first time that PDIMs may play a role in M. tuberculosis pathogenesis 83 .In their study, Cox et al. 80 also showed that M. tuberculosis mutant strains unable to synthetize PDIM were attenuated in mice.These strains were defective in cell wall biosynthesis, displayed strikingly altered colony morphology on solid medium, compared with the wild type, and their growth in mouse lungs was impaired 80 .
Phthiocerol dimycocerosate may be also involved in cell wall permeability.Insertion mutants unable to synthesize or subcellular translocation of PDIM have higher cell wall permeability and are more sensitive to detergents than the wild-type strain.Chenodeoxycholate is a negatively-charged hydrophobic probe used for the evaluation of the fluidity of mycobacterial cell wall lipids 84 .Its uptake by M. tuberculosis strain devoid of PDIM was significantly higher than in the wild type 85 .
Compared with mice infected with wild-type M. tuberculosis, bacterial loads in the lungs and spleen of mice over a 4-month period were reduced in animals intranasally infected with a PDIM-deficient mutant 86 .The differences in bacterial loads during the acute phase of infection are probably caused by a delay in granuloma formation in mice infected with PDIMdeficient mutant.M. tuberculosis PDIMs do not contribute to the inhibition of phagolysosome fusion within resting macrophages but contribute to the protection against bactericidal activity of reactive nitrogen intermediates 86 .Finally, the PDIM-deficient mutant induced the secretion of higher amounts of TNF-α and IL-6 by macrophages and dendritic cells, 48 and 72h after infection, than the wild type 86 .Also, Astarie-Dequeker et al. 87 showed that PDIM interferes with phagocytosis and blocks phagosomal acidification in a cholesterol-dependent manner.These data suggest that M. tuberculosis uses PDIM for the modulation of the early host immune response and to dampen the bactericidal activity of macrophages 86 .Thus, PDIM appears to affect the infection outcome at both early and chronic phases of disease.

LIPID SYNTHESIS CONTROL AND A BACTERIAL IMMUNOSTAT
Mycobacterium tuberculosis manages its cell wall lipid content in order to modulate the host immune response over the course of infection.Analysis of the lipid-induced immune response generated from the products of beta-oxidation pathway, suggests that DAT/PAT, SL-1, and PDIM collectively contribute to diminishing the immunopathologic response induced by M. tuberculosis.Jain et al. 9 showed that PDIM and SL-1 syntheses are coupled via metabolic flux of a common precursor, methylmalonyl-CoA.The overexpression of methylmalonyl-CoA mutase, which catalyzes the conversion of the citric acid cycle intermediate succinyl-CoA to methylmalonyl-CoA, led to a decrease in the total abundance of PDIM and SL-1 in vivo; this suggested that during chronic infection, M. tuberculosis preferentially uses the beta-oxidation pathway and not the citric acid cycle as a source of methylmalonyl-CoA 9 .The beta-oxidation of even chain fatty acids exclusively yields acetyl-CoA units, while beta-oxidation of odd chain fatty acids yields both acetyl-CoA and propionyl-CoA 10 .Thus, during the later phase of infection and after the activation of host immune response, M. tuberculosis uses the beta-oxidation pathway to metabolize intracellular odd chain lipids to synthetize DAT/ PAT, SL-1, and PDIM.During this phase of infection, local immunosuppression induced by these lipids may contribute to the establishment of a long-term M. tuberculosis infection.This notion is supported by a recent study by Mendum et al. 62 , who showed that M. tuberculosis survival in dendritic cells depends on multi-gene loci associated with specifi c functional groups, such as the synthesis of PDIM, phenolic glycolipids, and SL-1, and cholesterol metabolism.
The manner in which M. tuberculosis regulates the cell wall lipid content during the course of infection and how the temporal lipid profi le determines the fate of infection remain to be clarifi ed.Perez et al. 88 described the possible contribution of PhoP-PhoR, a bacterial two-component signal transduction system involved in an adaptive response to a variety of stimuli.Disruption of this system markedly affects the ability of the tubercle bacillus to replicate in cellular and animal models 88 .PhoP positively regulates the synthesis SL-1, DAT, and PAT 89 .Also, Galagan et al. 90 proposed a regulatory network model that operates during hypoxia where PhoP regulates whiB3, with both PhoP and WhiB3 regulating PAT/DAT and SL-1.WhiB3 is involved in the maintenance of redox homeostasis by regulating fatty acid metabolism, and modulates the biosynthesis of PAT, DAT, SL-1, PDIM, and triacylglycerol 91 .
Another suggestion concerning M. tuberculosis cell wall homeostasis comes from studies of the putative lipid transporter system encoded by mce operons.It has been long recognized that the thickness of the cell wall of M. tuberculosis cells changes during stationary phase of growth 92 .Cantrell et al. 19 and Forrellad et al. 39 independently reported increased abundance of FM in the cell wall of M. tuberculosis mce1 mutants, suggesting that the increased thickness of the cell wall may be attributed to such lipids.Such cells may become impermeable to the host's effector molecules, albeit they may also experience the stress of nutrient starvation.Uchida et al. 42 observed that during the fi rst 8 weeks of infection in mouse, the mce1 operon in wildtype M. tuberculosis is repressed by a negative transcriptional regulator Mce1R, which is similar to the constitutive lack of expression in the mce1 mutant.During this nutrient-limited state, M. tuberculosis may down-regulate the propionyl-CoA/ methylmalonyl-CoA pathway involved in the synthesis of SL-1, DAT, and PAT, while increasing the synthesis of MA, which has even lower pro-infl ammatory activity 8,30 .Dunphy et al. 93 proposed that the mce1 operon is involved in MA import.A prolonged starvation of the cells with thickened cell wall may serve as a signal to de-repress the mce1 operon to allow the importation of FM, a carbon source.The starvation signal may also trigger the synthesis of MA via the acetyl-CoA/malonyl-CoA pathway 27 .Queiroz et al. 43 showed that the beta-oxidation cycle might prefer the acetyl-CoA over the propionyl-CoA pathway in the mce1 operon mutant, as evidenced by decreased levels of SL-1, DAT, and PDIM precursors, with a concomitant increase in MA levels.The mce1 operon thus appears to play an important role in regulating MA abundance and utilization, allowing bacterial persistence (Figure 2).
In conclusion, M. tuberculosis appears to have developed an elaborate system for a homeostatic regulation of its cell wall lipid composition.It comprises a dynamic adaptive mechanism for the establishment of persistence in the host.Factors that disturb this homeostasis lead to disease on one side or bacterial clearance on the other.Collectively, the cell wall lipids serve as a type of immunologic thermostat (immunostat model, as depicted in Figure 2) in an infected host.Given this model, the big questions that can be asked are, 1) Which changes in the Mycobacterium tuberculosis cell wall lipid content that occur during progression to active tuberculosis can be targeted by drugs to prevent reactivation TB in those who are latently infected?2) What host immunological signals are recognized by Mycobacterium tuberculosis that induce the bacterium to modulate its cell wall lipid contents for its persistence?And 3) Can we identify a vaccine or an immunomodulator that can shift the equilibrium state towards elimination of the bacteria in a host?Better understanding of the interaction between M. tuberculosis lipids and the immune response could lead to the development of such immunomodulators distinct from the traditional protein antigen-based vaccines, or the development of drugs against novel targets in M. tuberculosis to tip the balance in favor of the host.

Lipids Features of the biosynthetic pathway Related immune responses Induced immune activity
evaluated a transcriptional response of BMMs to TDM-

TABLE 1
Mycobacterium tuberculosis lipid-induced immune response and main metabolic pathways involved in lipid synthesis.