Portal de Periódicos da CAPES

The many faces of host responses to tuberculosis

2001; Wiley; Volume: 103; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2567.2001.01236.x

1365-2567

Helen Collins, Stefan H. E. Kaufmann,

Immunodeficiency and Autoimmune Disorders

Tuberculosis remains today one of the top three fatal infectious diseases, together with acquired immune deficiency syndrome (AIDS) and malaria. During the last decade, 90 million new infections occurred, resulting in approximately 30 million deaths. Although there is currently effective chemotherapy, consisting of three specific drugs, this regimen must be continued for a period of at least 6 months, which in many cases, results in problems with compliance. Lack of compliance further impacts on the development of multidrug-resistant strains of the bacterium, which consequently raises the cost of treatment, making the expense of curing tuberculosis prohibitive in many developing countries. Despite the enormous numbers of people infected with this organism, it is estimated that only 10% of affected individuals show evidence of clinical symptoms. Many parameters, notably socio-economic factors, co-infection with human immunodeficiency virus (HIV) and genetic predisposition of the host, influence the susceptibility to disease. Much work has been invested to elucidate the biology of the interaction between Mycobacterium tuberculosis and its host, both in experimental animal models and in clinical studies. Here we review some of the latest developments in the understanding of the immune response required to control this pathogen. It is hoped that further progress in this field will lead to a more rational approach towards the development of an effective vaccine and novel chemotherapeutic agents. The establishment of a successful infection by mycobacteria depends on the initial encounter between the pathogen and the host cell, usually the macrophage. Obviously the surface characteristics of both parties will significantly influence the outcome. Although mycobacteria are Gram positive, their wax-rich cell wall confers on them unique features and thus they are classified as acid-fast bacilli. The abundant cell wall glycolipids, including lipoarabinomannan (LAM) and mycolic acids, are responsible for many immunological peculiarities. Mycobacteria have been proposed to bind to a variety of host cell receptors, including Fc receptors (FcR), complement receptors (CR) (both with or without prior opsonization), the macrophage mannose receptor, surfactant protein receptors and CD14 (reviewed in ref. 1), via a variety of its surface molecules (reviewed in ref. 2). It is proposed that the choice of receptor used to enter the macrophage influences the cellular response. For example, internalization of immunoglobulin G (IgG) -opsonized mycobacteria via the FcR induces the production of reactive oxygen intermediates and permits phagosome–lysosome fusion,3 whereas entry of mycobacteria via CR3 prevents the activation of the respiratory burst4 and results in a phagosome that is arrested in its maturation stage at that of an early endosome.5 The diverse array of receptors that could be utilized by mycobacteria to interact with and to enter host cells, makes it unlikely that there is one 'preferred route'. Indeed, the organism appears to have become rather adept at abusing both host cell receptors and other host cell molecules in order to maximize internalization. Thus, in a mechanism specific for pathogenic mycobacteria, the bacteria can associate with C2a to form a C3 convertase, resulting in opsonization of the organism with C3bi and uptake by macrophage CR1/CR3.6 However, the vast redundancy of this system is illustrated by experiments demonstrating that CR3-deficient mice infected with M. tuberculosis show no difference in survival, bacterial burden or granuloma formation as compared to wild-type animals.7 Recently, an essential role for cholesterol in the entry of mycobacteria into macrophages has been described. The depletion of plasma membrane cholesterol specifically inhibited mycobacterial uptake. This observation has important implications for the subsequent intracellular events, as cholesterol mediates the phagosomal association of tryptophane aspartate-containing coat protein (TACO), which prevents the maturation of the phagosome to a phagolysosome8. Much work has focused on the interaction of mycobacteria with the Toll-like receptors (TLR). This receptor family was first identified in Drosophila, where they are important for resistance to microbial pathogens.9 Subsequently a large number of Toll homologues have been identified in mammals, and TLR4 in particular has been demonstrated to be critical in responses to Gram-negative bacteria.9,10 Two TLRs, TLR2 and TLR4, have been implicated in the activation of macrophages by mycobacteria. Viable M. tuberculosis could activate an NF-κB reporter gene in Chinese hamster ovary cells transfected with TLR2 and TLR4 via distinct ligands, independent of CD14 expression. Thus, a soluble, heat-stable factor mediates TLR2-dependent activation, while a heat-sensitive cell wall-associated factor induced activation via TLR4. The cell wall component, LAM, isolated from fast-growing mycobacteria, stimulated TLR2- but not TLR4-mediated activation, whereas LAM isolated from M. tuberculosis or M. bovis bacillus Calmette–Guèrin (BCG) failed to activate via either receptor.11 The 19 000 MW lipoprotein of M. tuberculosis induced the production of the T helper type 1 (Th1) cell promoting the cytokine interleukin-12 (IL-12) from primary human monocytes in a TLR2-dependent mechanism and this receptor was sufficient to mediate this effect. The ability of this lipoprotein to activate the promotor for inducible nitric oxide synthase (iNOS), was abolished by the presence of a transfected dominant negative mutant TLR2 receptor.12 This enzyme is responsible for the catalysis of reactive nitrogen intermediates (NO) from l-arginine, which forms one of the primary microbicidal mechanisms of macrophages (see below). TLR2 has also been shown to be recruited specifically to macrophage phagosomes containing yeast, and can trigger tumour necrosis factor (TNF) production in response to Gram-positive, but not Gram-negative, bacteria.13 To date it is unclear whether the TLRs are recruited to the mycobacteria phagosome. However, taken together these data indicate that a variety of bacterial components, including a diverse array from mycobacteria, can activate signalling cascades within the host cell which direct the subsequent development of an immune response (Fig. 1). Stages of the immune response to Mycobacterium tuberculosis. Having gained entry into the macrophage, M. tuberculosis faces the problem of establishing residence inside a primary host effector cell. To this end, mycobacteria have evolved mechanisms to exploit the macrophage as an intracellular niche. One of the major problems is acquisition of essential nutrients in the intracellular environment. The macrophage requires iron as a cofactor in the induction of microbicidal effector mechanisms, while the mycobacteria themselves have an obligate requirement for iron for their intracellular survival. Thus, there is competition between the host and M. tuberculosis for the acquisition of this essential molecule. The host cell acquires iron via the transferrin receptor (TfR), which internalizes extracellular iron bound to transferrin and lactoferrin. This complex is then trafficked to an early endosomal recycling compartment, where the mildly acidic conditions facilitate the release of the iron from the receptor. There are several strategies that M. tuberculosis employs to ensure that its iron supply is not restricted. Primarily, the mycobacterial phagosome is restricted in its maturation state to that of an early endosome which resides in the recycling endosomal pathway. This results in free access to the transferrin receptor with its iron bound to transferrin.14 Additionally mycobacteria have developed specialized iron-binding molecules, siderophores, which have a high affinity for intracellular iron and transfer iron from host proteins to specialized mycobactin molecules in the mycobacterial cell wall.15In vitro studies using gallium, which accumulates intracellularly and disrupts iron acquisition by M. tuberculosis, revealed that treatment of infected macrophages with this compound resulted in the killing of M. tuberculosis both extracellularly and intracellularly.16 In line with this, a mutant strain of M. tuberculosis that was deficient in the synthesis of a subset of siderophores was impaired in its intracellular growth.17 In the clinical situation, there is an increased incidence of tuberculosis among people suffering from dietary iron overload, which is characterized by iron deposition in macrophages and parenchymal cells18. This well illustrates the delicate balance that must exist inside the host cell: too much iron down-regulates microbicidal effector mechanisms and favours the growth of the pathogen, whereas too little iron is inhibitory to the induction of antimicrobial processes. Possible mechanisms for manipulating intracellular iron concentration during M. tuberculosis infection have recently become a focus in our laboratory, and preliminary experiments in mice suggest that the local administration of iron-chelating compounds alters the course of pulmonary tuberculosis (Collins et al. unpublished observations). Recently evidence has been provided that in chronically infected lung tissues, i.e. at late stages of infection, M. tuberculosis obtains carbon from fatty acids. One pathway that is required for this acquisition is the glyoxylate shunt. One enzyme of this pathway, isocitrate lyase, is up-regulated by M. tuberculosis organisms when they are inside macrophages.19 A mutant M. tuberculosis strain with a disruption in the gene encoding isocitrate lyase was attenuated in its ability to sustain a persistent infection in mice, but was dispensable during the acute phase of growth. Furthermore, in vitro infection of macrophages revealed that the expression of isocitrate lyase was prolonged in activated macrophages in comparison to resting cells.19,20 This suggests that the manipulation of the nutritional requirements of M. tuberculosis, coupled with the immune status of the host, dramatically alters the course of infection and could open up potential avenues for therapeutic intervention. As critical as it is for M. tuberculosis to reside inside the macrophage to perpetuate infection, it is more important from the host's perspective to eliminate the pathogen via the activation of microbicidal mechanisms, such as the induction of reactive nitrogen and oxygen intermediates. Experimental M. tuberculosis infection of iNOS knockout mice, reveals a greatly increased susceptibility of these mice, as measured by an increase in bacterial loads, as well as a decreased survival time.21,22 Recently a role for superoxide has also been demonstrated by using mice lacking the cytosolic p47 (phox) gene which is essential for NADPH-dependent production of superoxide radicals. Phox–/– mice showed an increase in bacterial loads during the early infection period with M. tuberculosis, however, once interferon-γ (IFN-γ) -secreting antigen-specific T lymphocytes appeared in the lung, the mutant mice were once again able to control the infection, and bacterial loads stabilized.23 This suggests that early IFN-γ production and the induction of iNOS are not sufficient to control initial M. tuberculosis replication. This is similar to the situation observed in experimental infections with Salmonella typhimurium, where NADPH phagocyte oxidase was required early in infection, whereas iNOS-dependent mechanisms were important later for bacterial clearance.24 Despite experimental models implicating NO production as a critical molecule in the anti-mycobacterial response, in humans its role has been more controversial. However, in vitro experiments have demonstrated the induction of NO in human monocytes,25 as well as in human alveolar macrophages, and the presence of NO correlated with the growth inhibition of M. tuberculosis.26 What is apparent from all of these observations, is that the production of IFN-γ is critical in the control of M. tuberculosis infection, whether produced early in infection as a by-product of the activation of immune defence mechanisms, or by antigen-specific T cells following the induction of specific immunity. This has been demonstrated in a variety of systems. Experimentally, mice deficient in either IFN-γ itself, or IL-12, a critical cytokine in the induction of IFN-γ, were highly susceptible to challenge with M. tuberculosis.27,28 Clinically, there exists a group of patients with inherited complete or partial IFN-γ receptor deficiency who are highly susceptible to infections with intracellular bacteria, including environmental mycobacteria such as M. avium and M. smegmatis, and suffer from disseminated M. bovis BCG infection. Furthermore, as in the experimental system, patients with inherited deficiencies in IL-12p40 or the IL-12 receptor are similarly susceptible (reviewed in ref. 29). However, surprisingly, these patients did not suffer from detectable M. tuberculosis infection. The reason for this is unclear, perhaps the relatively short life span of these individuals, as well as their other recurrent bacterial infections, makes it less likely that they are exposed to M. tuberculosis. In light of the experimental and clinical data, recent efforts have focused on determining the exact requirements for IFN-γ in controlling M. tuberculosis infection. The examination of tuberculous granulomas has revealed contradictory results – lung tissues from patients suffering from tuberculosis were examined for the presence of a variety of cytokines. Whereas transforming growth factor-β (TGF-β) could be detected in some samples, IFN-γ was absent from all specimens tested.30 In apparent contrast to these findings, a study of five established granulomas revealed positive staining for IFN-γ in all five lesions by in situ hybridization.31 The differences may be due to the stage of granuloma examined, which is obviously difficult to control in a clinical situation. Experimentally, two recent approaches have been undertaken to determine the local effects of IFN-γ production in the lung. Kaplan and colleagues infected IFN-γ knockout mice with recombinant BCG-secreting murine IFN-γ. In comparison to mice who received BCG alone, the mice that were infected with IFN-γ-BCG had lower bacterial loads, more differentiated granulomas and increased expression of iNOS.32 In studies from our own laboratory we constructed a transgenic mouse that expressed IFN-γ under the control of a lung-specific promotor on an IFN-γ knockout background, creating a mouse that only expressed IFN-γ in the lung. These mice were then challenged by aerosol with M. tuberculosis, and the bacterial loads and survival were measured over the course of infection. In comparison to the IFN-γ-deficient animals, the transgenic mice had lower bacterial loads over the early infection period (up to 30 days post-infection), and delayed dissemination to liver and spleen. However, the local IFN-γ secretion could not compensate fully, as bacterial loads were lower and survival times were longer in wild-type controls. A possible explanation for this is that the constitutive IFN-γ in the transgenic animals is produced by the epithelial clara cells, and that any cells recruited to the site of infection, such as T cells, will be unable to secrete this cytokine (Reuter et al. unpublished results). Although IFN-γ is a major cytokine involved in the control of M. tuberculosis infection, obviously there are others involved. As mentioned above, IL-12 is critical in the development of a Th1 response and the production of IFN-γ, but additionally TNF can synergize with IFN-γ to activate macrophages. However, a more important role for TNF is probably in granuloma formation in tuberculosis. Thus, mice deficient in TNF exhibited poorly formed granulomas with areas of extensive necrosis. The failure to form organized granulomas resulted in widespread dissemination of M. tuberculosis and the rapid death of the animals. Notably, a comparison of macrophages within the lungs of mutant vs. wild-type animals revealed equivalent levels of major histocompatibility complex (MHC) class II and iNOS, suggesting that the critical role of this cytokine is in granuloma organization, rather than in the activation of T cells and macrophages.33 However, as with many other infections, the production of TNF must be finely balanced, as its overproduction leads to increased cellular accumulation, compromising lung function and exacerbating tissue damage. The preferred habitat of mycobacteria is the phagosome of the macrophage. From this cellular compartment mycobacterial antigens are readily accessible to the MHC class II processing machinery, resulting in the activation of specific CD4+ T cells. Consequently the importance of this T-cell subset in controlling acute mycobacterial infections has long been proposed, and confirmed in a variety of experimental models using antibody depletion and knockout mouse strains deficient in either CD4 or MHC class II.34,35 The dramatic increase in susceptibility to tuberculosis of patients infected with HIV further underlines the critical role of this cell subset.36 The main function of CD4+ T cells in tuberculosis is thought to be the production of cytokines, specifically IFN-γ, which – as already mentioned – is critical for macrophage activation and the subsequent induction of microbicidal mechanisms. IFN-γ production by CD4+ T cells seems to be required early in infection, as 2 weeks after M. tuberculosis infection both CD4–/– and MHC class II–/– mice exhibited 50% less IFN-γ in the lung but by 4 weeks post-infection the levels of this cytokine were equivalent in both mutant and wild-type mice.35 In the mutant mice, the IFN-γ production was compensated by CD8+ T cells.34 Furthermore, in a murine model of latent tuberculosis, the antibody-mediated depletion of CD4+ T cells resulted in the rapid reactivation of a persistent M. tuberculosis infection, despite the continuous presence of IFN-γ and the unimpaired production of iNOS.35 These results suggest that CD4+ T cells have an additional role, independent of IFN-γ production, in preventing the reactivation of tuberculosis. In humans, some experimental evidence suggests that CD4+ T cells perform an additional cytolytic function, particularly in the local immune response in the lung.37 One possible mechanism is that this is mediated by Fas-ligand-induced apoptosis of infected macrophages, which subsequently reduces the viability of the intracellular mycobacteria,38 although the involvement of apotosis in the control of mycobacterial infections has been questioned by others.39 Despite the intraphagosomal location of mycobacteria, it is now accepted that CD8+ T cells participate in a successful immune response against the organism. Experimentally, mice deficient in β2-microglobulin (β2m),40 the transporter associated with antigen processing (TAP),41 CD8α and perforin42,43 were all more susceptible to infection with M. tuberculosis than were wild-type mice, although to different extents. In a direct comparison of these mouse strains, Sousa and colleagues determined that both in terms of survival and bacterial loads, the β2m-deficient animals were the most susceptible, followed by the TAP-deficient animals, suggesting that MHC class I processing and subsequent activation of CD8+ T cells is required in the control of M. tuberculosis infection.43 Studies from our laboratory have extended this comparison to include mutant mice that are totally devoid of MHC class Ia molecules in the H-2b haplotype.44 These mice exhibit an increased susceptibility as compared to wild-type mice to both intravenous infection and to low-dose aerosol challenge with M. tuberculosis. However, similar to the findings for CD8α-deficient mice,43 they were less susceptible over the course of infection than β2m knockout mice, suggesting that the lack of classical MHC class I-restricted CD8+ T cells was not the complete explanation (Rolph et al. in press). Possible explanations for the increased susceptibility of the β2m knockout mice include the following: β2m stabilizes the surface expression of non-classical MHC class Ib and the MHC class I like CD1 molecules, which are both non-polymorphic. Thus, T cells restricted by these components could contribute to protective immunity (see below). β2m also interacts with Hfe, an atypical MHC class I protein that is involved in the regulation of iron absorption. As a consequence of this, these mutant mice exhibit tissue iron overload.45 As discussed earlier, this may impair host defence mechanisms as well as providing the necessary iron for the intracellular requirements of mycobacteria. Despite the experimental findings underlining the importance of CD8+ T cells, it still remains puzzling how phagosomally derived antigens interact with the MHC class I processing machinery, which primarily presents endogenously derived antigens, or those derived from pathogens which reside in the cytoplasm. From other intracellular bacterial infections, and also from model systems using exogenous antigens such as ovalbumin, it appears that there is an 'alternative' MHC class I processing pathway permitting the processing of phagosomal antigens.46 This pathway seems to be preferentially utilized by particulate antigens, including bacteria. An attractive theory is that as many intracellular bacteria, including M. tuberculosis, induce apoptosis of host cells and the formation of apoptotic blebs, these blebs could be engulfed by macrophages or dendritic cells and presented in a TAP-dependent fashion to CD8+ T cells, as has previously been described for Salmonella infections.47 Recently, large numbers of apoptotic cells have been demonstrated in the caseous foci of granulomas of patients suffering from tuberculosis, but were rarely seen in productive granulomas without detectable necrosis.48 Interestingly, from in vitro experiments with infected monocytes/macrophages, the induction of apoptosis is a very early event following infection, but does not result in a significant decrease in mycobacterial viability. The CD1 molecules are divided into two groups – Group I (CD1a,b,c) and Group II (CD1d). Humans possess both groups, whereas in other mammalian species, such as rabbits and mice, only group II CD1d is present. Group I CD1a,b and c molecules have been shown to present a variety of mycobacterial cell wall glycolipids, including phosphoinositol mannosides, LAM, mycolic acids and hexosyl-1-phosphoisoprenoids, to T cells which express the αβ T-cell receptor (TCR) and are either CD4– CD8– or CD8+ (reviewed in refs 49–51). Recent studies investigating the intracellular localization of mycobacteria and CD1a,b and c in human dendritic cells revealed that CD1a is almost exclusively expressed on the cell surface, similar to conventional MHC class I molecules. CD1b molecules were primarily found in mature phagolysosomes whereas CD1c was expressed both on the cell surface and in early endosomes.52,53 As mycobacterial phagosomes are restricted in their maturation state to that of early endosomes, it is conceivable that lipid antigens could interact with CD1c in this compartment. However, upon activation with IFN-γ, phagolysosome fusion is promoted54 which may lead to the co-localization of mycobacterial antigens with CD1b. Furthermore, lipids shed from mycobacteria inside cells can be taken up by uninfected cells in their vicinity, suggesting that there is transfer of mycobacterial antigens between cells, possibly from macrophages, the major habitat of mycobacteria to dendritic cells, the major CD1-expressing, antigen-presenting cell.52,55 Recently, our laboratory has provided evidence that lipid-containing vesicles are transferred from infected macrophages to uninfected cells, and that this transfer can be blocked by soluble CD14 or soluble Integrin binding motif (RGD) (U. Schaible, personal communication). Murine group II CD1 molecules (CD1d) control the development of natural killer (NK) T cells which have the ability to produce cytokines rapidly upon activation. Although to date there has been no concrete identification of a bacterial ligand for CD1d, it has been shown that both murine and human T cells react specifically with α-galactosyl ceramide derived from a marine sponge.56 Despite the attractive theory that this cell population could account for the increased susceptibility of the β2m–/– mice, experiments with mice deficient in CD1d have found that there is no exacerbation of M. tuberculosis infection as compared to controls.41 Of course this does not rule out the possibility that, as mice do not have the group II CD1 molecules, CD1-restricted T cells play a more substantial role in human infections, or that these T cells may function as a 'second line of defence' once the major T-cell subsets are compromised or absent. CD8+ T cells restricted by the non-classical MHC class Ib molecule (H-2M3), which recognize N-formylated peptides, have been shown to participate in the immune response to both Listeria monocytogenes and Salmonella typhimurium.57 Recently, it was found that immunization of mice with H-2M3-binding N-formylated peptides elicited immune reactivity against M. tuberculosis.58 Studies from our own department have demonstrated CD8+ T cells from M. tuberculosis-infected mice that react with the N-formylated peptides f-MIVIL and f-MIGWII. Taken together, these data suggest that these T cells also contribute to the immune response to M. tuberculosis (A. Sponaas, personal communication). The effector mechanisms of CD8+ T cells, regardless of their MHC restriction, are primarily cytokine production and cytolysis. Following antigen recognition, CD8+ T cells can produce IFN-γ and TNF which, as described above, are critical cytokines in the activation of the microbicidal mechanisms of macrophages. In tuberculosis patients, IFN-γ-secreting CD8+ T cells could be detected by ELISPOT from the peripheral blood, and these T cells recognized peptides defined from ESAT-6 in a classical MHC class I-restricted manner. Furthermore, in healthy household contacts, as well as in patients with inactive, self-healed pulmonary tuberculosis, 1 : 2500 peripheral blood lymphocytes were specific for a single M. tuberculosis ESAT-6 epitope, suggesting that these cells play a long-term protective role in humans.59 CD8+ T cells also perform cytolytic functions directed against M. tuberculosis-infected macrophages. To date, the contribution of these cytotoxic functions in tuberculosis has been controversial. Mice deficient in either perforin or granzyme, molecules critical for the major lytic pathway of CD8+ T cells, do not appear to be more susceptible to aerosol challenge with M. tuberculosis, nor do there appear to be major histological differences in granuloma formation in the early phase of infection (up to 60 days)42,60 and only a modest increase in bacterial burden at later stages of infection.43 However, recent studies have demonstrated the presence of MHC class I restricted, β2m-dependent, M. tuberculosis-specific CD8+ cytotoxic T cells in the lungs and draining lymph nodes of M. tuberculosis-infected mice. These cells were capable of lysing infected macrophages in a perforin-dependent manner.61 Furthermore, the group I CD1-restricted T cells expressing CD8 are also capable of lysing M. tuberculosis-infected macrophages through a mechanism involving the introduction of cytotoxic granules into the cell via perforin-mediated pores. Notably, this mechanism resulted in a reduction of the viability of the intracellular mycobacteria which was directly attributed to the actions of granulysin, which was found to kill extracellular mycobacteria by altering their membrane integrity.62 This ability to reduce intracellular mycobacterial viability was exclusive only to those CD8+ T cells that effected cytolysis via perforin and granulysin, as CD4– CD8–, CD1-restricted CD8+ T cells that lysed target cells via the Fas–FasL pathway did not affect the viability of the pathogen.39 Thus, with an obvious mechanism affecting bacterial viability that is essentially dependent on the coexistence of perforin and granulysin, it is surprising that the perforin gene knockout animals were not more susceptible to M. tuberculosis infection. A possible explanation for this is that a compensatory mechanism exisits in perforin-deficient animals. Indeed, the numbers of activated CD8+ T cells that secreted IFN-γ were approximately four times higher in the lungs of perforin knockout mice following M. tuberculosis infection,61 a phenomenon also described for infections with lymphochoriomeningitis virus. However, in the viral system, perforin was proposed to be involved in down-regulating T-cell responses to reduce immune-mediated damage.63 As no increase in immune-mediated pathology was observed in perforin-deficient animals following M. tuberculosis infection, how critical this molecule is in controlling tuberculosis requires further elucidation. In mice and humans, T cells expressing the γδ TCR make up 1–5% of the peripheral blood and lymphoid organs. By adulthood, 50% of all human γδ T cells express Vγ2δ2. This subpopulation of T cells has been implicated in the response to mycobacterial infections, mainly from the observations that they accumulate early, both in the lesions of leprosy patients64 and in experimental M. tuberculosis infection of mice,65 and that mycobacterial phospholigands stimulate all Vγ2δ2-bearing cells. The exact manner of this activation is not fully understood but it results in a high frequency of mycobacteria-specific γδ T cells, at least in the same order of magnitude as the clonal expansion of αβ T cells.66 However, using gene knockout mice, once again the results have been more controversial, and appear to vary according to the dose and route of infection. Thus, δ TCR gene-disrupted mice infected with M. tuberculosis in a low-dose aerosol challenge showed no increase in susceptibility as compared to wild-type animals.67 In contrast to these findings, intravenous infections of TCR γδ knockout mice with a low dose of M. tuberculosis revealed slightly increased bacterial growth in the mutant animals between days 15 and 30, although at later time-points both control and knockout animals were able to control the infection. Intravenous challenge with higher doses however, proved fatal for the γδ T-cell-deficient mice, while control animals survived.68 One difference that was apparent, however, even under experimental conditions showing no increased susceptibility of δ TCR–/– mice, was the difference in the type of granuloma formed. The δ TCR knockout mice developed pyogenic granulomas with a substantial infiltrate of neutrophils in contrast to the more lymphocytic granulomas of the wild-type mice.67 This suggests a role for γδ T cells in directing cellular infiltration and granuloma formation which occurs early in infection. Similar to experimental murine infections, clinical studies in humans have also revealed seemingly contradictory results. Thus, Vγ2δ2-positive cells are reduced in the blood and broncheoalveolar lavage fluid of patients with active pulmonary tuberculosis compared to patients with other granulomatous disease,69 whereas a later study revealed that γδ T cells of tuberculosis patients were highly activated in comparison to those from patients without disease.70 Furthermore, a subset of Vγ2δ2 T cells has been shown to kill macrophages infected with M. tuberculosis in a granule-dependent pathway that results in the loss of mycobacterial viability, similar to that described for the CD1-restricted CD8+ T cells.71 From all of this experimental and clinical evidence, one possible mechanism is that γδ T cells are recruited early into the site of mycobacterial infection where they direct the formation of the granuloma. However, upon interaction with mycobacterial antigens, and ligation of the TCR, the cells become activated, resulting in an up-regulation of FasL. This would then induce apoptosis of Fas-bearing cells and may explain the regulatory role of γδ T cells in the granuloma as well as the disappearance of γδ T cells from the blood of tuberculosis patients.72,73 From what has been discussed, the intertwined liaison between M. tuberculosis and the immune system should be obvious. The long-term habitation of the organism within macrophages has resulted in the development of numerous strategies for competition with one of the major host effector cells. A similar complexity concerns the relationship with T lymphocytes as major mediators of acquired resistance against tuberculosis. Experiments in the viral system and with model antigens have shown us that antigen-processing pathways segregate into MHC I, for newly synthesized antigens, such as those of viral origin, and MHC II for antigens originating in the endosomal compartment. Although M. tuberculosis has chosen the endosomal compartment as its preferred habitat, both MHC I- and MHC II-restricted T cells participate in protection. Moreover, the so-called unconventional T cells, notably the CD1-restricted T cells specific for mycobacterial glycolipids, apparently contribute to an optimal response against M. tuberculosis. This suggests that the long standing confrontation between mycobacteria and the host has selected unique cell subsets which are focused on the complex cell wall of mycobacteria and which are not required for anti-viral defence. A better understanding of the relationship between the tubercle bacillus and its host will provide important guidelines for the rational combat against this disease, which is still of major importance, even today.74 After all, less than 10% of the 3 billion individuals infected with M. tuberculosis have developed the disease, and for many of these, disease development is promoted by immunodeficiency, such as concurrent HIV infection. This strongly supports the argument that the immunocompetent host is well equipped to keep the pathogen in check. However, the recent demonstration of exogenous re-infection in tuberculosis patients who have undergone curative treatment,75 indicates that the immune response in at least a proportion of tuberculosis sufferers fails to control the pathogen satisfactorily. For these individuals, a vaccine that outperforms M. tuberculosis by inducing a more potent immune response than natural infection is required. Hence, understanding the immune response during natural infection will help us optimize the immune response against the pathogen, doubtless, an ambitious goal. The availability of the genetic blue-print of M. tuberculosis76, deeper insights into the immune response against this pathogen, as well as the increasing knowledge about optimization of immunization strategies, provide good reasons to hope that this goal can be achieved The authors gratefully acknowledge support from Bundesministerium für Bildung und Forschung (BMBF) joint project 'Mykobakterielle Infektionen'; EU project 'TB vaccine cluster'; Chiron Behring TB vaccine research co-operation; WHO GPV-VRD 'Rational design of anti-TB vaccines' and Deutsche Forschungsgemeinschaft (DFG) priority programme 'Neue Vakzinierungsstrategien'.