Evasion and subversion of antigen presentation by <em>Mycobacterium tuberculosis</em>

Background

Through a process of co-evolution, many pathogens have developed strategies to evade or subvert the protective responses directed against them by the immune system of their hosts. In addition to the evasion of innate immune mechanisms (1,2), microbes that invade the bodies of vertebrates often possess mechanisms that enable them to inhibit or alter the outcome of adaptive immunity through the manipulation of antigen presentation pathways (3,4). The mechanisms of inhibition are diverse, and in many cases the microbial genes responsible and their precise modes of action are still unknown. Mycobacterium tuberculosis is one of the most successful of human pathogens, and has acquired the ability to establish primary infections that generally become latent and persist permanently in the body in spite of an intact immune system. The presence of inherited or acquired deficiencies in the immune response allows latent tuberculosis to become active, leading to chronic disease and transmission.

Tuberculosis remains a global public health problem of enormous scope and proportion, with an estimated one third of the world's population harboring latent disease and approximately two million deaths from active disease annually (5). The intersection of the tuberculosis and HIV pandemics in many developing nations has led to enormous morbidity and mortality, as well as increased transmission and emergence of extensively drug resistant strains of M. tuberculosis (6,7). The currently available vaccine for prevention of tuberculosis, the attenuated M. bovis strain known as Bacille Calmette-Guerin (BCG), has proven largely ineffective despite widespread use. This most likely reflects in part the inherent ability of pathogenic mycobacteria to prevent effective host responses. A more complete understanding of how this occurs is crucial to efforts targeted at the design and production of better vaccines. Here we review current knowledge and recent advances in the study of evasion of adaptive immunity by M. tuberculosis, with a particular focus on the subversion or inhibition of antigen presentation to T lymphocytes by this pathogen.

T cell immunity to M. tuberculosis

The central importance of T cells in controlling infection with M. tuberculosis is clearly established in humans and animal models (8,9), and is in significant part due to their unique functions in activating and augmenting the microbicidal effector functions of mononuclear phagocytes. To a great extent this is achieved through the actions of T cell-derived cytokines, most notably interferon-γ (IFNγ) and tumor necrosis factor α (TNFα), which are produced abundantly by activated CD4 ^{T cells that have differentiated into T helper type 1 (Th1) cells under the influence of interleukin-12 (IL-12). These cytokines have multiple important effects including the activation of macrophages to produce nitric oxide and the stimulation of phagosome-lysosome fusion (}10). In addition, CD4 ^{T cells activate macrophages and dendritic cells through surface bound molecules such as CD40 ligand (CD40L), and also mediate cytotoxicity against infected cells through expression of Fas ligand (FasL) and potentially other cytolytic effector molecules (}11-13). MHC class I restricted CD8 ^{T cells also carry out many of these functions including the production of IFNγ, and are believed to make unique contributions to immunity against intracellular bacteria such as} M. tuberculosis through their potent cytolytic activity against infected cells. This cytolytic activity is predominantly mediated by directed secretion of the pore forming protein perforin together with proteases known as granzymes, and a second pathway that induces target cell death by engaging Fas (CD95) with its ligand FasL may also contribute (14,15). Several other T cell subsets are also known to contribute to immunity against M. tuberculosis, including T cells of various phenotypic subsets that recognize lipid antigens presented by CD1 molecules, which are a family of non-MHC encoded class I-like antigen presenting molecules (16). In humans, a major subset of T cells expressing γδ T cell antigen receptors (TCRs) responds to small non-peptide antigens produced by M. tuberculosis and many other bacteria (17). Although the antigen presenting molecules involved in the γδ T cell responses to these unconventional antigens are not known, it is believed that such responses may also contribute to protective immunity (18).

The presentation of specific antigens derived from an intracellular pathogen such as M. tuberculosis is a complex process that involves multiple steps, including antigen uptake, delivery to appropriate cellular compartments, proteolytic cleavage or other enyzmatic processing steps, loading and intracellular trafficking of the antigen presenting molecules. There are many points in this process at which pathogens can potentially interfere to prevent or alter the outcome of peptide antigen recognition. As discussed in the following sections, there is evidence that M. tuberculosis interferes with antigen processing and presentation by the MHC and CD1 systems, and most likely does so in multiple ways.

Antigenic targets of T cell responses to M. tuberculosis

Identification of the specific M. tuberculosis antigens recognized by mice and humans during the course of infection has been vigorously pursued in the interest of obtaining subunit components for vaccine development. A considerable number of specific antigens have been identified and characterized, including not only conventional peptide antigens that bind to classical MHC class I and class II molecules but also a variety of abundant mycobacterial lipids that are presented by CD1 molecules (19,20).

Immunodominant protein antigens of M. tuberculosis

The immunodominant T cell antigens recognized by M. tuberculosis infected humans or animals are mostly secreted proteins, including such well studied examples as ESAT-6, CFP-10, the Ag85 family of mycolyl transferases, Mtb32a, Mtb9.8, Mtb8.4, TB10.4 and lipoproteins such as the 19-kDa and 38-kDa antigens (21-24). Many of these antigens are currently being studied as components of new candidate vaccines (25-28), and there is clear evidence for their presentation to T cells by the MHC class I and class II pathways. Infection with M. tuberculosis can generate strong immune responses against a relatively limited number of epitopes of these immunodominant secreted protein antigens early in the course of infection. This has been observed for antigens such as ESAT-6 and Ag85B, which are secreted early after infection and tend to dominate as targets of the T cell responses in experimentally infected animals or humans with naturally acquired tuberculosis. On the other hand, T cell responses to cytosolic non-secreted proteins have generally been found to be absent or weak in infected animals, although cellular and humoral responses have occasionally been reported to these (29-32). Recent experiments comparing M. tuberculosis strains that were engineered to either secrete or retain within the bacterial cell the antigen CFP-10 confirmed that CD4 ^{and CD8 T cell responses} in vivo to this normally immunodominant antigen are dependent on its secretion (33). The finding that only secreted bacterial proteins seem to generate strong responses in animals that are naturally or experimentally infected with M. tuberculosis suggests there may be an active inhibition of priming against protein antigens that remain associated with the bacterial cell.

Mycobacterial lipids and other nonpeptide antigens of M. tuberculosis

Mycobacteria produce many unique lipids and glycolipids, and some of these have been found to be specific T cell antigens that are presented by MHC class I-like CD1 molecules. In humans, there are five forms of CD1, and three of these (CD1a, CD1b and CD1c – the so-called group 1 CD1 proteins) have been shown to present mycobacterial lipids and glycolipids. The ligands presented by CD1b include free and monoglycosolated mycolic acids, phosphatidylinositols (LAM, PIM₂ and PIM₆) and diacylated sulfoglycolipids (34,35). CD1c binds and presents β1-mannosylated phospholipids that are products of a polyketide synthesis pathway (mycoketides), and CD1a presents an acylated non-ribosomally synthesized peptide siderophore known as didehydroxymycobactin (34). Mice only express one form of CD1 which is most homologous to human CD1d (classified as group 2 CD1), and this seems to be less involved in presentation of mycobacterial lipid antigens than the human group 1 CD1 proteins. However, mouse CD1d has been reported to bind mycobacterial tetramannosylated phosphatidylinositols (PIM₄), although it is unclear whether responses to this glycolipid contribute to the immune response in M. tuberculosis infected mice (36). A major subset of human γδ T cells is well documented to be responsive to low molecular weight non-peptidic antigens produced by M. tuberculosis and many other bacteria. The specific antigens that have been identified are primarily intermediates produced by pathways of isoprene biosynthesis, and include isopentenyl pyrophosphate (IPP) and 4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP) (37,38). The mechanism by which such small alkyl phosphate compounds are presented to γδ T cells remains unknown, although it is believed that the known MHC and CD1 antigen presenting molecules are not involved (17,37,38).

Strategies for evasion or subversion of T cell responses by M. tuberculosis

It has been well recognized for more than a decade that successful intracellular pathogens such as M. tuberculosis frequently possess multiple highly evolved mechanisms to disrupt antigen processing and presentation (39,40). In recent years, detailed mechanistic studies have begun to delineate the specific strategies of immune evasion used by M. tuberculosis, and the mechanisms by which different pathways of antigen presentation are inhibited or manipulated by this pathogen (41,42).

Infection of antigen presenting cells by M. tuberculosis

The fact that M. tuberculosis directly and efficiently infects professional antigen presenting cells is likely to have a major impact on how this pathogen subverts and evades antigen specific T cell responses. Macrophages are the major cell type infected by M. tuberculosis in vivo, and provide the most important site for intracellular replication. Macrophages also have the capability of presenting peptides to effector T cells on both MHC class I and class II molecules, and depend on activation by T cells to exert their microbicidal functions against mycobacteria such as nitric oxide production (43). While antigen presentation by macrophages is likely to be critical to expression of the antimycobacterial functions of previously primed effector or memory T cells, it is unlikely that macrophages are effective at priming the initial T cell responses of naïve T cells against mycobacterial antigens. This function is provided by various subsets of dendritic cells, which are highly specialized to process and present all types of antigens to naïve CD4 ^{and CD8 T cells. Recent studies have shown that dendritic cells are also infected at high levels during} M. tuberculosis infection in mice, suggesting that direct manipulation of dendritic cell function by intracellular mycobacteria is likely to be a prominent feature of infection (44).

The use of M. tuberculosis engineered to express green fluorescent protein (GFP) has allowed the accurate localization of the bacteria in vivo after infection. Using this technique it has been shown that lung myeloid dendritic cells (CD11c^{CD11b are among the major cell types that become infected after introduction of aerosolized} M. tuberculosis into the lungs, along with alveolar macrophages (CD11c^{CD11b), recruited interstitial macrophages (CD11cCD11b), monocytes and neutrophils (}44). Dendritic cells infected in the lung are then able to transport mycobacterial antigens to the mediastinal lymph nodes to initiate the adaptive immune response. However, macrophages and dendritic cells that have not been sufficiently activated do not seem capable of killing the intracellular bacteria and serve as a reservoir of M. tuberculosis which could be responsible for the dissemination and delay of the adaptive immune response. These infected phagocytes may serve the role of “Trojan horse” by carrying the bacteria away from the primary focus of infection to widely seed many other tissues. Because of the marked migratory potential of tissue dendritic cells, these cells may be particularly likely to play a significant role in bacterial dissemination (45,46).

Secreted proteins as immunological decoys

The highly coordinated expression of one or a small set of antigens that dominates the immune response may represent the use of immunological “decoys” by M. tuberculosis to distract the immune response and prevent it from responding against target antigens that are more relevant to host protection. In theory, these decoy proteins would contain peptides with high affinity for MHC class I and II molecules that out-compete other subdominant but potentially more important epitopes for the formation of protective effector and memory T cells. In some cases, such as with the immunodominant secreted Ag85B antigen, the protein is not essential for bacterial survival and its expression is shut off after infection is established (47). In this way, a subversion of the T cell response can occur in which strong responses are primed that subsequently fail to recognize the pathogen. Effector T cells primed against such decoy antigens would thus survey the body for antigenic targets that are no longer present, and would be unable to recognize infected macrophages containing bacteria that do not express the decoy epitope but instead express other antigens to which the immune system has not been primed. Once the initial targets of the primary immune response have been established, it is possible that M. tuberculosis may actively inhibit the priming of new responses against subdominant epitopes or those which are expressed in a delayed manner, thus preventing truly protective responses from developing (48). Regulation of antigen expression in this way has been suggested to result in part from feedback loops that allow the bacterium to regulate its protein secretion systems through a process related to quorum sensing (49).

Nonsecreted or cytoplasmic proteins of mycobacteria have generally not been found to be major antigenic targets of T cell responses in M. tuberculosis infected animals, although humoral responses have sometimes been found against such antigens (29,31,32). However, the fact that nonsecreted mycobacterial proteins do not strongly stimulate T cell responses during infection with live mycobacteria does not necessarily mean that these antigens cannot be the targets for protective T cell responses. In fact, a few studies have shown that vaccination of animals to specifically stimulate immunity against nonsecreted protein antigens can lead to a significant level of protection against subsequent mycobacterial challenge (27,50). This suggests that the failure of nonsecreted proteins to effectively prime MHC class I or class II restricted T cell responses during infection with M. tuberculosis could be the result of active immune evasion mechanisms. On the other hand, the observation that killed mycobacterial vaccines are less efficient than live attenuated vaccines for eliciting protective immunity argues that protein secretion may be needed to generate effective immunity (51). This is a critical point for ongoing efforts to produce more effective vaccines, particularly since most of the target antigens that are currently being used in vaccine development are immunodominant secreted antigens that could potentially represent ineffective and relatively non-protective decoy antigens (25,52). More analysis of the impact of immunizing against secreted as opposed to nonsecreted target antigens will be needed to resolve this important point.

Delayed priming of adaptive immunity in M. tuberculosis infection

A key feature of the adaptive immune response to M. tuberculosis in the lung is that it is remarkably delayed and slow to develop, which in mouse models allows the bacteria to grow uncontrolled for approximately 21 days (8). The recent development of transgenic mice expressing TCRs specific for MHC class II presented peptides of M. tuberculosis Ag85B and ESAT-6 has significantly enhanced the study of the early events of the adaptive immune response in the lungs after a low dose aerosol infection (53-55). Ag85B and ESAT6 are both secreted antigens that are highly expressed during the initial stages of infection, and then decline after approximately three weeks. Studies using adoptive transfer of T cells from these TCR transgenic mice have emphasized the slow pace of the evolution of the T cell response even against strongly expressed secreted antigens during the initial phase of pulmonary tuberculosis (53-55). These studies also show that T cell responses in mice following aerosol infection with M. tuberculosis are initiated in the draining mediastinal lymph nodes, and that this requires the transport of mycobacteria into the lymph nodes by migrating lung dendritic cells. The delay in T cell priming can be accounted for by the finding that migration of infected dendritic cells from the lung to lymph nodes is also delayed, requiring 11-12 days to first become detectable (44,55).

These findings suggest that one way in which M. tuberculosis avoids elimination by the host immune system involves delaying the onset of adaptive responses until the bacteria have established a protected niche in which they can persist permanently. Thus, while M. tuberculosis produces many secreted antigens that are highly immunogenic during the early phase of infection, the host immune system is not able to mount a rapid protective response against these to control bacterial growth. It is noteworthy that T cell responses following an aerosol M. tuberculosis infection seem to be significantly more delayed than those which develop after intravenous infection (48), suggesting that the delay in T cell priming could be largely due to inhibitory effects of the pathogen that are exerted specifically on antigen presenting cells of the lungs. This may reflect a specialized adaptation of this bacterium to its natural route of infection. The slow development of adaptive immunity following M. tuberculosis infection is in marked contrast to the kinetics typically observed for acute viral infections of the lung, or with other intracellular bacteria such as Listeria monocytogenes (56). In the sections that follow, known or postulated mechanisms are reviewed that subvert, evade or alter the kinetics of the antimycobacterial immune response by direct inhibition of key steps in antigen presentation.

Manipulation of MHC class II antigen processing and presentation pathways by M. tuberculosis

MHC class II restricted CD4 ^{T cells are well known to be critical for control of} M. tuberculosis infection in experimental animals and humans (8), and it is therefore extremely probable that the MHC class II pathway for antigen presentation is targeted by the bacterium to enhance its intracellular survival and persistence. In fact, multiple mechanisms have been described by which M. tuberculosis can potentially affect antigen presentation by MHC class II molecules (Table 1 and Figure 1).

Figure 1

Disruption of MHC Class II presentation by Mycobacterium tuberculosis

The various steps in the MHC class II processing and presentation pathway that are known or postulated to be influenced by M. tuberculosis infection are illustrated. A) New synthesis of MHC class II molecules is blocked by TLR2 signaling due to mycobacterial products such as the 19-kDa lipoprotein. B) Intracellular trafficking of MHC class II is disrupted by the suppression of cathepsin S, which is due to induction of IL-10 by mycobacterial infection. C) Generation of peptide antigens for loading onto MHC class II in relevant endocytic compartments (MIIC) is also inhibited by several effects of mycobacterial infection, including inhibition of phagosome-lysosome fusion, neutralization of phagosomal pH by urease (UreC) and by blockade of recruitment of the vaculolar proton ATPase. D) A proposed inhibition of autophagy and autophagic vacuole formation also eliminates a potential source of antigenic peptides that can load MHC class II molecules. E) The reduction of peptide antigen availability and incomplete cleavage of MHC class II associated Ii resulting from cathepsin S suppression result in a reduced transport of stable peptide loaded MHC class II molecules to the APC surface.

Transcriptional downregulation of MHC class II molecules by M. tuberculosis

Studies focusing on the 19-kDa lipoprotein antigen (also known as LpqH), a major triacylated exported cell wall protein of M. tuberculosis, demonstrated the ability of this ligand for mammalian toll-like receptor-2 (TLR2) to globally reduce the surface levels of MHC class II molecules expressed by M. tuberculosis infected macrophages (57-59). This effect appears to be linked to excessive or prolonged TLR2 signaling, which reduces activity of the MHC class II transcriptional transactivator (CIITA) and also interferes with IFNγ signaling. Detailed in vitro studies of this phenomenon revealed that isoforms of the transcription factor C/EBP that suppress transcription of CIITA are induced following TLR2-dependent stimulation of macrophages by 19-kDa lipoprotein. These C/EBP isoforms are further modified by a mitogen activated protein kinase (MAPK) that is also activated by TLR2 signaling, and this increases their binding affinity to promoter regions I and IV of the CIITA gene to inhibit its expression (60-62).

The reduction of MHC class II expression induced by 19-kDa lipoprotein has been shown clearly in experiments in vitro with cultured cells, but the in vivo relevance of this mechanism during infection remains unclear. Experiments in vivo showed indistinguishable levels of MHC class II on cells either expressing or lacking TLR2 in mixed bone marrow chimeras that were reconstituted with a mixture of wild type and TLR2-deficient bone marrow (63). This result indicates that TLR2 expression is not required on macrophages for inhibition of their MHC class II synthesis, and that other mechanisms that may be independent of TLR2 could account for this effect during M. tuberculosis infection in vivo (64,65). Other recent in vivo data show that the decrease in MHC class II expression on lung APCs during infection is restricted mainly or exclusively to cells that are directly infected with mycobacteria (66). This raises the question of how the 19-kDa lipoprotein-induced downregulation of MHC class II is restricted to infected but not uninfected cells, given that this lipoprotein is secreted by the bacterium. It is possible that the lipoprotein is unstable and short lived so that only secretion in an autocrine manner (i.e., by cells that are directly infected) is sufficient to give TLR2 activation, or that secretion directly within the phagosomal lumen is required.

In addition, experiments have not yet fully clarified what the impact of the 19-KDa lipoprotein-induced MHC class II downmodulation is on the presentation of mycobacterial protein antigens. Although inhibition of the presentation of exogenously added ovalbumin has been shown for various APCs purified from the lungs of M. tuberculosis infected mice (66), it remains to be determined whether this effect is directly related to activities of the 19-kDa lipoprotein. Likewise, data are not yet available for the effect of 19-KDa lipoprotein on presentation of actual bacterial proteins produced in the infected cell. It should be noted that 19-kDa lipoprotein is only one of multiple known TLR2 ligands in M. tuberculosis, others of which include the lipoproteins LprA, LprG and 38-kDa lipoprotein (also known as PstS-1, phoS or phoS1) as well as the small secreted protein ESAT-6 and complex glycolipids in the lipoarabinomannan (LAM) family (67-70). All of these M. tuberculosis-derived ligands for TLR2 could potentially contribute to the modulation of MHC class II levels, and may synergize with numerous other microbial products to influence the outcome of the adaptive immune response to M. tuberculosis.

MHC class II intracellular trafficking in M. tuberculosis infected cells

Mycobacterium tuberculosis can affect the intracellular trafficking of MHC class II molecules by interfering with cathepsins and by alkalinization of intracellular compartments (71). Cathepsins are cysteine proteases that are essential for processing of the MHC class II associated invariant chain (Ii), which is a necessary step in peptide loading and subsequent cell surface expression of MHC class II molecules. Studies of cathepsin S (CatS) have shown this particular member of the cathepsin family to be the most important or possibly the only protease that is capable of mediating the late steps of Ii cleavage needed to generate MHC class II molecules that can be efficiently loaded with peptide antigens (72,73). In vitro experiments showed that mycobacterial infection of human macrophages caused reduction of CatS activity and gene expression through the induction of the inhibitory cytokine IL-10 (74). This was associated with an intracellular retention of MHC class II molecules, and a reduced expression of peptide loaded MHC class II complexes at the cell surface. This effect was reversed by addition of anti-IL-10 antibodies, which restored expression of active CatS and export of mature MHC class II molecules to the surface of infected cells. Other experiments showed that a recombinant BCG strain that was engineered to secrete CatS could also restore cell surface levels of MHC class II molecules (75). In related studies, it was reported that intraphagosomal production of urease encoded by the mycobacteria ureC gene, which hydrolyses urea into carbon dioxide and ammonia, is also involved in disruption of MHC class II trafficking leading to intracellular retention. This is believed to be due to the effect of urease on preventing the acidification of MHC class II processing and loading compartments, which may also prevent the activity of proteases such as CatS (76). Other reports have also implicated inhibition of expression or activity of other cathepsin proteases, such as cathepsins L and D, in the impaired processing and presentation of antigens by MHC class II in mycobacteria infected cells (77,78).

Disruption of dendritic cell maturation by M. tuberculosis

A number of studies have shown that M. tuberculosis infection of dendritic cells can alter the normal process of maturation of these cells, which is crucial to their ability to efficiently prime antigen specific T cells. At least two studies using cultured human monocyte-derived dendritic cells have found that these cells fail to show normal maturation following infection with M. tuberculosis, as indicated by a lack of rapid mobilization of MHC class II molecules to the cell surface that is normally associated with the maturation process (79,80). Other data indicate that immune evasion may occur not through a blockade of DC maturation, but rather by stimulating a poorly coordinated maturation that causes effective antigen processing to cease before M. tuberculosis antigen production begins (81). According to this model, rapid maturation of DCs infected by M. tuberculosis results in the movement of the great majority of MHC class II molecules to the cell surface, coupled with the cessation of new MHC class II molecule synthesis. This suggests that by the time peptide antigens secreted by M. tuberculosis become available for processing and loading onto MHC class II molecules in endocytic compartments, the majority of MHC class II molecules already may have moved to the cell surface to limit the pool of these molecules available for peptide loading. This “kinetic model” of immune evasion may explain how M. tuberculosis can severely restrict the ability of directly infected DCs to present bacterial antigens. However, such a mechanism would not clearly apply to cross presentation of secreted antigens by uninfected DCs, which may offer a potential explanation for the apparent bias toward recognition of secreted protein antigens by T cells in M. tuberculosis infected animals.

Inhibition of phagosome-lysosome fusion and phagosome acidification

One of the earliest virulence traits described for M. tuberculosis was its ability to block the fusion of phagosomes with lysosomes in infected macrophages (78,82-84). This mechanism not only allows the organism to avoid exposure to lysosomal hydrolases, but probably also contributes to keeping its antigens from being delivered to an intracellular environment in which proteolytic processing and loading of the resulting peptides onto nascent MHC class II molecules can occur. Studies using murine macrophages in vitro have found that MHC class II complexes loaded with mycobacteria-derived peptides originate in the phagosomes of these cells, but are not detected in lysosomes or late endosomes that do not contain the bacteria (85). This suggests that by maintaining an intraphagosomal environment that lacks the proteases and low pH required for efficient MHC class II/peptide complex formation, M. tuberculosis may delay or prevent effective antigen presentation to CD4 ^{T cells.}

The mechanisms by which M. tuberculosis inhibits phagosome maturation and phagosome-lysosome fusion has been an area of active investigation, but remains relatively poorly understood. Initial molecular and cell biological studies showed that mycobacterial phagosomes failed to incorporate vacuolar ATPases, which could account for their inability to acidify (78,86). Other elegant studies showed that mycobacterial phagosomes retained a protein known as coronin-1 or TACO on the cytoplasmic face of their limiting membranes, whereas phagosomes that undergo fusion with lysosomes tended to rapidly lose the association with this protein (87,88). Such studies provided confirmation and molecular detail about the inhibition of the normal process of phagosome-lysosome fusion in M. tuberculosis infected macrophages, but did not offer direct insight into the mechanism by which the pathogen controls this process. More recently, studies have identified a serine/threonine kinase encoded by the M. tuberculosis pknG gene as a potential effector of the inhibition of phagosome-lysosome fusion. Expression of the product of this gene (M. tuberculosis protein kinase G) in the nonpathogenic species M. smegmatis was sufficient to cause this inhibition, and deletion of pknG from M. tuberculosis caused the bacteria to localize to lysosome-like structures (89). Surprisingly, a follow up study failed to reveal an impact of pknG on MHC class II restricted antigen presentation as assessed by CD4 ^{T cell activation (}90). Although further investigation will be needed to resolve this point, a possible explanation for this paradox is that the T cell responses studied were against secreted mycobacterial antigens for which phagosome-lysosome fusion may not be essential. Thus, additional studies to assess presentation of nonsecreted or more tightly bacterial cell-associated antigens could potentially reveal an impact of pknG on the MHC class II presentation system.

Prevention of autophagy-related antigen processing by M. tuberculosis

Autophagy is a homeostatic and inducible process by which the cells can eliminate damaged organelles or other unwanted structures within the cell. The process is initiated by the formation of a double membrane-enclosed structure that forms an autophagic vacuole that engulfs organelles or other inclusions in the cytosol. The material engulfed by the vacuole can then be digested following fusion with lysosomes, and becomes a potential source of peptides for antigen presentation (91). Moreover, it has been shown that many pathogens can become localized to autophagic vacuoles within host cells (92). A number of recent studies have provided strong connections between the process of autophagy and MHC class II presentation of peptides from intracellular proteins. In general, these studies show that induction of autophagy in antigen presenting cells increases the presentation of peptides from intracellular and lysosomal sources, and augments the MHC class II processing and presentation machinery of the cell (93-96). For example, using the autophagy-related LC3 protein as a marker of autophagosomes, it has been shown that these colocalize with MHC class II related compartments in dendritic cells (94,95,97).

In the case of M. tuberculosis infected cells, it has been shown that the bacteria can be cleared by autophagy (92,97,98). Nutrient starvation or IFNγ treatment can induce autophagy in M. tuberculosis infected cells, and phagosomes containing the bacilli acquire lysosomal markers and become acidified during this process. Another recent study showed that M. tuberculosis localization to autophagosomes in infected macrophages was markedly increased by treatment of the cells with LPS, defining a toll-like receptor-4 (TLR4) mediated pathway for induction of autophagy (99). These effects are likely to be involved in decreasing the survival of M. tuberculosis in cells induced to undergo autophagy, and, it is reasonable to speculate that M. tuberculosis may target this pathway to increase its intracellular survival and inhibit the presentation of antigens by MHC class II. Consistent with such a proposed mechanism of immune evasion, a recent study showed that autophagy enhances the efficacy of BCG vaccination by increasing mycobacterial peptide presentation in mouse dendritic cells (100).

Effects of M. tuberculosis on presentation by MHC class I molecules

Peptides presented by MHC class I molecules are the principle target structures of CD8 ^{cytolytic T cells, which are important for immunity against viruses and many other intracellular pathogens. Earlier dogmas held that peptides presented by MHC class I molecules were derived only from protein antigens that were produced in the cytosol of target cells or that somehow gained access to the cytosolic compartment. This seemed to rule out a role for MHC class I presentation of antigens from} M. tuberculosis, since that organism and its antigens were believed to be confined to either endosomal compartments or to the extracellular space. This idea is now known to be incorrect for several reasons. First of all, it is now becoming accepted that M. tuberculosis does gain access directly to the cytosol of infected cells, as shown originally in a mouse macrophage cell line and human monocyte-derived macrophages (101) and more recently confirmed in elegant studies of infected human monocyte-derived dendritic cells (102). Secondly, mechanisms by which endocytosed antigens can be processed and presented by the MHC class I system, a process generally referred to as cross presentation, are now well documented and are strongly believed to be relevant to M. tuberculosis infection (Table 1 and Figure 2) (103,104).

Figure 2

MHC class I presentation pathways in Mycobacterium tuberculosis infection

The large cell on the left of the figure represents a macrophage infected with M. tuberculosis. Newly synthesized MHC class I molecules in the endoplasmic reticulum (ER) are loaded with peptides that are produce by the cytosolic proteosome complex and transported into the ER lumen by TAP. Additional trimming of the cytosol derived peptides can occur as a result of aminopeptidase activity in the ER lumen. Escape of mycobacterial proteins from the phagosome into the cytosol can lead to peptide presentation by this classical MHC class I pathway (A). Mechanisms for loading of peptides onto MHC class I molecules in the lumen of the phagosome are also likely to exist. This vacuolar pathway for cross presentation (B) may involve transfer of ER membrane components (e.g., newly synthesized MHC class I complexes and TAP) to the phagosome membrane, enabling the loading of peptides generated in the cytosol. Alternatively, peptides may be generated by proteases in the phagosome lumen, and these may be loaded by a process of peptide exchange onto MHC class I molecules recycling from the plasma membrane. The so-called “detour pathway” (C) is a third way that peptides from a vacuolar intracellular pathogen such as M. tuberculosis can be cross presented by MHC class I. In this case, an infected cell must first die by apoptosis, and the released apoptotic vesicles carry the mycobacterial antigens into uninfected dendritic cells. Current evidence suggests that all of these pathways are likely to be actively inhibited or effectively bypassed during M. tuberculosis infection (see text and Table 1 for details).

One potential pathway for cross presentation by MHC class I molecules of mycobacterial proteins originating in the lumen of a phagosome or phagolysosome involves their translocation to the cytosol (105-109), which is proposed to involve a transporter in the phagosomal membrane referred to as the dislocon. Mycobacterial proteins transported into the cytosol in this way would be expected to undergo cytosolic processing, which involves polyubiquitination, proteolysis by the proteosome and ultimately transport into the lumen of the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP) where binding to nascent MHC class I molecules occurs with high efficiency (Figure 2)(110). The identification of the dislocon remains controversial, although it is postulated to be related or possibly identical to protein transporting complexes found on the ER membrane, such as Sec61 which translocates proteins into the ER lumen or the AAA-ATPase p97 which dislocates proteins out of the ER as part of the ER associated degradation (ERAD) pathway (111). In support of this, proteomic analysis of isolated phagosomes has detected the presence of Sec61 and the AAA-ATPase p97 (112), although these findings are of uncertain significance due to possible contamination of phagosome preparations with ER membranes (113).

In the case of M. tuberculosis infected cells, it is possible that the bacterium itself may cause leakiness of the phagosomal membrane, and this could allow bacterial proteins to enter the cytosolic processing pathway. For example, the secreted M. tuberculosis protein ESAT-6 and its chaperone partner CFP-10 have been found to have membrane disrupting properties (114), suggesting that production of these proteins in the phagosome lumen could be responsible for leakage into the cytosol. In support of such a mechanism, it was shown that M. tuberculosis infection of cells facilitated the MHC class I presentation of endosomally delivered ovalbumin through a process that required the presence of live bacteria (109). In addition, the previously documented escape of M. tuberculosis bacilli into the cytoplasm of human cultured DCs was shown to require an intact RD1 locus, which encodes the ESAT-6 and CFP-10 proteins and their secretion apparatus (102). However, whether DCs remain competent to process and present antigens after M. tuberculosis has undergone escape into the cytosol is unclear, since these cells are likely to rapidly undergo necrotic cell death that is triggered by the bacteria (115,116). Presumably, this pathway would be less effective for non-secreted bacterial protein antigens which would remain within the phagosome, and would thus be predicted to favor the presentation of secreted bacterial proteins.

An alternative mechanism that does not depend on phagosomal escape of antigens into the cytosol is the vacuolar model for cross presentation. This proposes that peptides derived from bacterial proteins are generated within the phagosomal lumen, and loaded directly on MHC class I molecules that recycle through the endocytic system (Figure 2) (104,117). This process requires the presence of proteases within the phagosome that can cleave the mycobacterial proteins to generate the peptides required for loading MHC class I molecules, and could potentially be relevant for both secreted and nonsecreted bacterial proteins if sufficient degradation of the bacterial cells occurs within the phagosome lumen. Although it is unclear to what extent this vacuolar pathway may account for cross presentation of antigens by MHC class I in M. tuberculosis infected cells, there is some evidence that peptides from Ag85B can be loaded and presented in this way (78,85). Whether M. tuberculosis actively interferes with MHC class I presentation mediated by either the cytosolic or vacuolar cross presentation pathways is unknown. However, one study has reported that the M.tuberculosis 19-KDa lipoprotein can reduce processing of protein antigens by the vacuolar pathway for MHC class I presentation (118). In addition it would be anticipated that the previously discussed inhibition of phagosome-lysosome fusion, prevention of phagosome acidification and down modulation of cathepsins associated with M. tuberculosis infection would all interfere with a vacuolar MHC class I cross presentation pathway.

Another pathway for cross presentation of antigens by MHC class I molecules exists which has been more strongly implicated in CD8 ^{T cell responses to} M. tuberculosis. This is known as the “detour pathway”, and depends upon the uptake by uninfected antigen presenting cells of bacterial antigens originating from infected cells that have undergone death by apoptosis (Figure 2) (119-121). Although the mechanisms involved in this antigen transfer are not completely resolved, it has been shown that M. tuberculosis infected macrophages that are induced to undergo apoptosis by serum starvation shed subcellular vesicles that are thought to be derived from membrane blebs of the dying cells. These contain bacterial proteins and lipids, and can be efficiently taken up by healthy, uninfected DCs (120). Antigens taken up in this way are delivered to a TAP dependent pathway for MHC class I presentation, and can therefore prime CD8 ^{T cell responses. Interestingly, the detour pathway also appears to be effective for uptake of lipid antigens for CD1 presentation, but does not seem to enhance MHC class II restricted antigen presentation (}120,121).

Arguing in favor of the relevance of the detour pathway to immune responses against M. tuberculosis is the well documented observation that pathogenic mycobacteria actively block the ability of infected macrophages to undergo apoptosis (122-125), and most likely escape eventually from infected cells by causing a form of necrotic cell death (116,126,127). M. tuberculosis and other pathogenic mycobacteria appear to have evolved powerful mechanisms for controlling the process of host cell death, and this is likely to be a key factor for the survival and persistence of these bacteria in immunocompetent hosts. Recently, two genes of M. tuberculosis that are directly involved in blocking apoptosis of infected macrophages have been identified. One of these is secA2 (Rv1821), which encodes a component of a virulence related secretion system and is known to be involved in the transport of the enzyme superoxide dismutase (SodA) out of the bacterial cell. Given that SodA has also been implicated in the inhibition of apoptosis (128), it seems likely that secretion of this enzyme by intracellular mycobacteria is important for controlling the levels of reactive oxygen intermediates that may act as a trigger for apoptosis if present at sufficiently high levels (129). The other identified M. tuberculosis antiapoptosis gene is nuoG (Rv3151), which encodes a subunit of the large, proton extruding type I NADH dehydrogenase complex in the mycobacterial membrane (130). In support of the relevance of antiapoptosis mechanisms to the ability of M. tuberculosis to evade presentation by the detour pathway, it has been shown that a mutant strain of M. tuberculosis with a deletion of the secA2 gene not only fails to block macrophage apoptosis but also concurrently stimulates markedly enhanced CD8 ^{T cell priming (}131). Along similar lines, it has been reported that a recombinant strain of BCG that induces increased levels of apoptosis of infected macrophages produces greater protective immunity against M. tuberculosis challenge when used as a vaccine in mice (132). Thus, the blockade of apoptosis and induction of necrosis may be one of the main strategies by which M. tuberculosis evades or delays antigen presentation by MHC class I molecules.

Influence of M. tuberculosis on lipid antigen presentation by CD1

Mycobacterium tuberculosis contains abundant and diverse lipids in its cell wall, and the presentation of some of these by CD1 molecules may account for a significant component of T cell mediated immunity in humans (133). Consistent with a proposed role in host resistance to mycobacteria, there have been suggestive data that M. tuberculosis may interfere with lipid antigen presentation by CD1 molecules (Table 1). Since macrophages in general do not express the group 1 CD1 molecules (CD1a, CD1b and CD1c), the transfer of lipids from infected cells to uninfected DCs may be critical for mycobacterial lipid presentation by CD1. Interestingly, one study has shown that this lipid transfer may occur by the detour pathway described above for MHC class I cross presentation, which depends on the ability of an infected macrophage to undergo death by apoptosis (121). Thus, the ability of M. tuberculosis to block apoptosis of infected cells may also have important implications for evasion of lipid antigen presentation by CD1. It has also been suggested that apolipoprotein E produced by macrophages could function as a carrier to shuttle lipids from infected cells to DCs (134), although whether this occurs in M. tuberculosis infection or is in any way blocked by the bacterium remain to be determined.

Direct evidence for immune evasion mechanisms targeting CD1 presentation was provided by one provocative study showing a dramatic decrease of group 1 CD1 surface expression following infection of human monocyte-derived DCs in vitro with virulent M. tuberculosis. In this case, the reduced CD1a, CD1b and CD1c expression was determined to be associated with reduced mRNA transcripts, suggesting a mechanism by which the bacterium represses the transcription of CD1 genes (135). In addition, a few studies have shown that M. tuberculosis infection of monocytes may impair their upregulation of CD1 expression during subsequent differentiation into DCs (136,137). Although some of these findings appear quite dramatic and biologically significant, the mechanisms responsible have not been explored and the potential significance has not yet been extended to in vivo models. Given that CD1 molecules are generally loaded with their antigens within endocytic compartments (16), the ability of M. tuberculosis to arrest phagosome maturation and prevent acidification of endosomal compartments in which it resides could potentially interfere with CD1-dependent antigen presentation.

Natural Killer T cells (NKT cells) recognize lipid antigens presented by the CD1d molecule (also known as group 2 CD1). These cells can produce large amounts of IFNγ when activated, and have been proposed to serve as part of the first line of defense against many pathogens. NKT cells appear to be activated during mycobacterial infection in mice, and when activated early in M. tuberculosis infection by the glycolipid ligand α-galactosylceramide they can slow the progression of disease (138). It has also been shown that the transfer of additional NKT cells into wild type mice can increase their resistance to M. tuberculosis (139). Surprisingly, mice that are NKT cell deficient due to deletion of genes for CD1d (CD1d ^{mice) or for the Jα segment of the NKT cell TCR (Jα18 mice) have generally not been found to have altered susceptibility to} M. tuberculosis infection (140-142). This raises the possibility that CD1d presentation or other factors involved in the activation of NKT cells may be down modulated by M. tuberculosis infection, although this has not yet been directly demonstrated.

Relevance to vaccine development and disease control

Mycobacterium tuberculosis is an extraordinarily successful pathogen, and its ability to infect and persist in mammals that have intact immune systems indicates that it has acquired powerful strategies for avoiding host immunity. Although the mechanisms are only beginning to be clearly defined, the material reviewed here provides strong support for the general view that M. tuberculosis has multiple and highly evolved mechanisms that allow it to maintain the upper hand in the host-pathogen interaction. Interfering with or modulating antigen presentation to T cells is probably among the highest priorities for M. tuberculosis, given the central role of T cells in resistance and control of infection with this organism. As described in the foregoing sections, we now have good reasons to believe that all of the known major pathways for antigen presentation are affected by M. tuberculosis. In some cases, this involves direct inhibitory effects on antigen presenting molecules or the specific pathways in which they operate (Table 1). However, other effects may be far more subtle, such as alterations in the kinetics of highly coordinated processes such as DC maturation, or the favored presentation of immunological decoys that subvert the immune response by distracting it from recognition of more relevant targets. The range and complexity of these effects point to an ancient and highly evolved relationship between pathogen and host.

Evasion and subversion of antigen presentation by M. tuberculosis is likely to represent one of the key issues in the ongoing effort to develop a truly effective tuberculosis vaccine. While the attenuated M. bovis strain BCG continues to be widely used, it has not proven in clinical studies to be a highly effective vaccine for prevention of infection or disease caused by M. tuberculosis (143). In fact, in some populations studied, the widespread use of BCG appears to have no detectable influence on the prevalence of pulmonary tuberculosis. Considering that BCG is derived from a pathogenic species of mycobacterium that is closely related to M. tuberculosis, it is likely that the vaccine strain has retained many immune evasion properties that prevent it from inducing highly protective and durable adaptive immunity. Thus, by further developing our understanding of the mechanisms by which mycobacteria interfere with antigen presentation and subvert or evade adaptive immunity, the road may eventually be cleared to development of more effective vaccines for prevention of tuberculosis.

Acknowledgments