The Lancet Infectious Diseases 2003; 3:578-590
DOI:10.1016/S1473-3099(03)00741-2
Latent tuberculosis: mechanisms of host and bacillus that contribute to persistent infection
JoAnn M Tufariello a, Prof John Chan a and Prof, Dr JoAnne L Flynn
b 
Summary
Immune response to M tuberculosis
Animal models for study of latent tuberculosis
T cells in latent tuberculosis
Modulation of antigen presentation: results of in-vitro studies
Avoidance of the toxic effects of RNI
Remodelling the phagosome
Cytokines and latent tuberculosis
Interleukin 10?regulation or dysfunction?
Conclusion
Search strategy and selection criteria
References
Summary
Most people infected with Mycobacterium tuberculosis contain the initial infection and develop latent tuberculosis. This state is characterised by evidence of an immune response against the bacterium (a positive tuberculin skin test) but no signs of active infection. It can be maintained for the lifetime of the infected person. However, reactivation of latent infection occurs in about 10% of infected individuals, leading to active and contagious tuberculosis. An estimated 2 billion people worldwide are infected with M tuberculosis-an enormous reservoir of potential tuberculosis cases. The establishment and reactivation of latent infection depend on several factors, related to both host and bacterium. Elucidation of the host immune mechanisms that control the initial infection and prevent reactivation has begun. The bacillus is well adapted to the human host and has a range of evasion mechanisms that contribute to its ability to avoid elimination by the immune system and establish a persistent infection. We discuss here current understanding of both host and bacterial factors that contribute to latent and reactivation tuberculosis.
Back to top
Tuberculosis is a major cause of death worldwide, causing an estimated 1?5?2 million deaths each year and rivalling malaria as the commonest cause of death due to a single infectious agent.1Mycobacterium tuberculosis was first identified as the causative agent by Robert Koch in 1882. Tuberculosis has affected human beings for thousands of years.2 Despite decades of research on chemotherapy for the disease and development of preventive vaccines, it remains a major public?health problem and was declared a global health emergency by WHO in 1993.3 Tuberculosis is preventable and in most cases can be cured, but various demographic and socioeconomic factors make both prevention and treatment difficult. The number of new cases of tuberculosis is projected to reach 11?9 million annually by the year 2006 if current control efforts are not strengthened.4 The recent resurgence in tuberculosis is attributable partly to the AIDS epidemic5 with the emergence of multidrugresistant strains also hindering efforts at control.6,7
M tuberculosis is an extraordinarily effective human pathogen. Surveys with purified protein derivative (PPD) or tuberculin skin tests suggest that one-third of the world's population is infected with the bacillus.8 Primary infection leads to active disease in only a minority (about 10%) of infected individuals,9 in most cases within 2 years. In the remaining 90% of cases the immune system contains the infection, and the individual is non-infectious and symptom-free. This clinical latency can persist throughout the person's lifetime. However, in some circumstances the host immune response is perturbed, and reactivation of latent infection results. This process can occur, for example, through HIV infection, malnutrition, the use of steroids or other immunosuppressive medications, or advanced age.10 Although the estimated lifetime risk of developing reactivation disease is between 2% and 23%,11 the risk for individuals immunosuppressed by HIV infection is estimated to be as high as 10% per year.12 Reactivation of latent infection contributes substantially to the incidence of adult tuberculosis, especially in more developed countries where disease prevalence is fairly low.13,14
Although M tuberculosis is known to have the capacity to persist in human tissues,15 many fundamental questions about the mechanisms of persistence remain unanswered. What regulates the transition from initial growth to persistence, and the transition back to active growth in reactivation of latent disease? What are the bacterial determinants necessary for persistent infection? What is the physiological state of tubercle bacilli during latent infection-are the bacteria in a non-replicating ?spore-like? state or are they at all replicative? To what extent are the existing bacilli metabolically active? What is the microenvironment encountered by M tuberculosis within the granuloma? How does the bacterium counteract or evade host defences and survive in the face of a vigorous host immune response? The answers to these questions are likely to provide insight into both the mechanisms by which M tuberculosis persists within the host and the means of elimination of latent infection, the disease phase that poses the most significant obstacle to the eradication of tuberculosis. Clearly, these are difficult questions to address experimentally, particularly in the human system. In this review we address the immune mechanisms of the host that contribute to control of infection and the ways in which M tuberculosis has evolved to avoid elimination, focusing on the interactions between the bacillus and CD4 T cells, CD8 T cells, and macrophages. Together, these strategies result in the establishment of latent M tuberculosis infection.
Immune response to M tuberculosis
The immune response to M tuberculosis is multifaceted and complex. T cells are an essential component of the protective response, and the interaction of these cells with infected macrophages is crucial for control of infection. The immune response is initiated when M tuberculosis arrives in the alveolar space, where it encounters alveolar macrophages. In response, at least partly through interactions of mycobacterial components with Toll-like receptors (TLR), the macrophages produce inflammatory cytokines and chemokines that serve as signals of infection.16 The bacteria enter the parenchyma and can replicate within the alveolar macrophages or in resident lung macrophages. The signals induced result in migration of monocyte-derived macrophages and resident dendritic cells to the focal site of infection in the lungs. The dendritic cells that engulf bacteria mature17?19 and migrate to the regional lymph node. Once there, CD4 and CD8 T cells are primed against mycobacterial antigens. Primed T cells expand and migrate back to the lungs and then through the lung tissue to the focus of infection, presumably in response to signals such as chemokines produced by or in response to infected cells. The migration of macrophages and T cells (as well as B cells) to the site of infection culminates in formation of a granuloma, a characteristic feature of tuberculosis. In addition to T lymphocytes and macrophages, the granuloma consists of other host cells including B cells, dendritic cells, endothelial cells, fibroblasts, and probably stromal cells (figure 1).20 The relative proportions of the different cell types in the granuloma vary with its age. The granuloma encompasses the bacilli, which reside within macrophages, and serves to wall off the bacteria from the rest of the lung, limiting spread. In addition, the granuloma functions as an immune microenvironment to facilitate interactions between T cells and macrophages and cytokines. However, the granuloma can also provide a home for M tuberculosis for an extended period, because some bacteria can avoid elimination within the granuloma. Although excessive replication of bacteria seems to result in loss of granuloma structure, extensive necrosis, and cavity formation, in most cases the granuloma contains the infection, but a limited number of bacilli survive, which is termed latent tuberculosis. A person who has a delayed-type hypersensitivity response to mycobacterial antigens (PPD) without signs of active tuberculosis is deemed to have latent infection and not to be contagious. Whether the host can ever completely eliminate latent infection is unclear. Some infected individuals do revert to a PPD-negative status; such PPD reversion can occur at a rate of about 5% per year.21 Studies of nursing-home residents also support the occurrence of tuberculin reversion.22,23 However, whether PPD reversion is linked with eradication of persisting tubercle bacilli remains unclear. The bacilli that survive within host tissues are a reservoir of infection that can reactivate to cause active disease many years later, as a result of immune-system compromise. The immune responses necessary for initial containment of infection, as well as long-term maintenance of a latent infection, are beginning to be elucidated.

Click to enlarge image
Figure 1. Immune mechanisms important in the maintenance of latent tuberculosis. The granuloma that forms in response to M tuberculosis consists of macrophages, which can differentiate into epithelioid macrophages or multinucleate giant cells, CD4 and CD8 T cells, and B cells. The T cells produce interferon γ, which activates macrophages. CD8 T cells can lyse infected macrophages or kill intracellular bacteria. Tumour necrosis factor (TNF) is produced by macrophages and T cells. Dendritic cells are also present within the granuloma. A mature granuloma is surrounded by fibroblasts. M tuberculosis is present within the macrophages and also extracellularly if necrosis is present. On depletion of CD4 T cells (eg, during HIV infection), the granuloma does not function as well, production of interferon γ may decrease, and macrophages are less activated. As a result, M tuberculosis begins to multiply and cause reactivation of infection. In the case of TNF neutralisation, the cells within the granuloma are no longer as tightly clustered, perhaps owing to chemokine or adhesion-molecule dysregulation. In addition, the macrophages are not as activated. These defects lead to a disorganised granuloma that is less able to control infection and greater immunopathology.
Back to top
Animal models for study of latent tuberculosis
Understanding of the various model systems available for study of acute and latent tuberculosis is useful. For studies of acute (primary) infection and vaccine testing, mice and guineapigs have been used extensively. Guineapigs are very susceptible to tuberculosis and can succumb to a very lowdose infection. This model has been used extensively to assess vaccine candidates. Mouse strains vary in susceptibility to M tuberculosis disease. An advantage of using mice for studies of infectious diseases is the availability of genetically engineered strains, inbred strains, and immunological reagents.
Establishment of an animal model of latent tuberculosis has been more difficult. First, the features of human latent tuberculosis that are to be modelled must be defined, which is challenging, owing to the limited understanding of this phase of M tuberculosis infection. In human beings, the immune response is crucial in establishing latent infection, since immunocompromised individuals are more susceptible to primary tuberculosis. In addition, there are very small numbers of bacteria in the lungs in latent tuberculosis. Finally, the latent state can be maintained for many years, and spontaneous reactivation is infrequent. However, reactivation can occur in response to declining immune function. The best animal model of latent tuberculosis would encompass all of these features. However, such a model has been difficult to develop, for reasons that are not completely clear. There are, however, a few models used to study latent and reactivation tuberculosis, and each one mimics at least some feature of human latent M tuberculosis infection.
For studies of latent and reactivation tuberculosis, two different mouse models have been used. In the first, mice of a relatively resistant strain (C57BL/6) are infected via aerosol or the intravenous route. The acute infection is controlled by the immune response by 3?4 weeks after infection. However, the immune response does not eliminate the infection but merely restrains it. The bacteria persist in the lungs, spleen, and liver. In the lungs, a high bacterial burden (105−106 colony-forming units) is stably maintained for many months. Although there is slowly progressive lung pathology, the mice do not show overt signs of disease such as weight loss or laboured breathing. There is evidence that the bacteria are in a metabolically quiescent state during this phase of the infection.24 However, this model seems to represent a chronic or persistent infection, rather than the latent infection observed in human beings. Nonetheless, the model does share some features with human latency, in that the chronic infection is controlled by the immune response, granulomas are present in the lungs, and the mice do not seem to be ill.
The second murine model of latency was developed by researchers at Cornell University, USA, in the 1950s.25,26 There have since been several variations on this model. Mice are infected with M tuberculosis and treated with antimycobacterial drugs to reduce the bacterial load to undetectable values. Subsequently, either the mice can spontaneously reactivate the infection, or reactivation can be induced by immunosuppression. However, this model can be difficult to use. Reactivation may be difficult.27 The drug treatment can sterilise the mice under some conditions. However, some researchers have had success with this model in identifying factors involved in maintaining a latent infection.27?29 The model is attractive in that the bacterial numbers are undetectable but can reactivate; the similarity to the human situation is appealing. However, unlike the human situation, antibiotics must be used to set up the latent infection, and the effect of this intervention on the immune response and control of infection is an additional variable.
Other animal models for the study of tuberculosis include rabbits and non-human primates.30 The rabbit model was historically used to study resistance and susceptibility to tuberculosis, and the granulomas that form in the rabbit lungs are remarkably similar to those in human lung.31,32 However, use of this model for study of latent disease has not been fully explored. In terms of immunology, the reagents for detailed studies in the rabbit model are not available. By contrast, the reagents available for immunological studies in non-human primates are excellent, owing to the extensive use of macaque monkeys in studies of simian immunodeficiency virus as a model for HIV. Macaques have been used historically in vaccine and drug-development studies for tuberculosis. Lately, a few groups have used this model to explore questions about pathogenesis and immunology.33?35 We have infected cynomolgus macaques with a low dose (about 25 colonyforming units) of M tuberculosis via bronchoscope.36 After infection, about half of the monkeys developed active chronic infection, characterised by infiltrates visible on chest radiographs, clinical signs, or weight loss. The remainder showed no signs of disease, although they did have delayedtype hypersensitivity responses, as well as positive lymphoproliferative and ELISpot responses to mycobacterial antigens. These monkeys appear to have latent infection. We have observed spontaneous reactivation in a small number of cases, after many months of infection. The macaque model is similar to human beings in terms of range of disease, immunology, and pathology. The granulomas in the lungs are remarkably similar to those in human beings. Thus, this model may be very useful for the study of acute and latent tuberculosis. Studies are under way to test various immunological mechanisms for reactivation of latent infection.
Back to top
T cells in latent tuberculosis
Interferon γ is a key cytokine in the immune response against M tuberculosis. This cytokine activates macrophages to kill the intracellular bacilli. Interferon γ probably has other roles in control of infection, since interferon-γ knockout mice are more susceptible to tuberculosis37,38 than strains deficient in macrophage effector mechanisms, such as inducible nitric oxide synthase (NOS2)39?41 or in T cells.42,43 Certain genetic mutations in human beings lead to deficiencies in interferon-γ signalling, and carriers of the mutations are more susceptible to mycobacterial disease,44 although reported tuberculosis cases in such individuals are rare.
The primary producers of interferon-γ are CD4 and CD8 T cells and natural killer cells. Many studies have implicated T cells in the control of M tuberculosis infection. In particular, CD4 T cells have a central role in protection against tuberculosis. In human beings, infection with HIV leads to loss of CD4 T cells and immunodeficiency. Tuberculosis is a major cause of death in AIDS patients worldwide. The heightened susceptibility of HIV-positive people to active tuberculosis, either from primary infection or by reactivation of latent infection, supports a role for CD4 T cells in defence against this infection.12 In experimental studies, mice deficient in CD4 T cells show impaired ability to control infection and die of tuberculosis.42,45 These mice show deficiency in early production of interferon-γ in the lungs and macrophage activation. Since other cells, including CD8 T cells, also produce interferon-γ in M tuberculosis infection,46,47 the findings in mice deficient in CD4 T cells suggest that early production of interferon-γ by CD4 T cells can determine the outcome of infection, and that CD4 T cells may have other functions, apart from production of interferon-γ, in the protective response against M tuberculosis.
Studies in a model of latent tuberculosis (the chronic infection model described above) provided further evidence of additional roles for CD4 T cells in control of infection. The infection was reactivated in chronically infected mice depleted of CD4 T cells, as shown by the loss of control of the infection, increased pathological features, and death.48 However, analysis of the lung tissue from the mice showed normal concentrations of interferon-γ and NOS2. CD8 T cells increased in number and in expression of interferon-γ to compensate for the lack of CD4 T cells. However, despite this strong response from CD8 T cells, the mice still succumbed to the infection. These findings strongly suggest additional roles for CD4 T cells in control of latent tuberculosis. By contrast, in the Cornell model of latent tuberculosis, depletion of CD4 T cells had little effect on the persistent infection.28 The reasons for these results in the Cornell model are unclear given the apparent requirement for CD4 T cells in control of latent infection in human beings. This discrepancy may be the result of difficulty in using the Cornell model for latent tuberculosis.27
There are several possible functions for CD4 T cells in control of latent tuberculosis.10 These cells could promote apoptosis of infected macrophages in the lungs through Fas/Fas ligand interaction. Apoptosis is thought to be detrimental to intracellular M tuberculosis,49?51 although no studies showing this in vivo have been published. There may be other cytokines produced by mycobacteria-specific CD4 T cells in the granuloma that are important in control of the infection, including interleukin 2 and tumour necrosis factor γ (TNFα).46 CD4 T cells can also induce production of other important cytokines by macrophages or dendritic cells, such as interleukins 12,52 10, and 15.53
CD4 T cells could also activate macrophages through direct contact independently of NOS2. These cells are one of the few types to express CD40 ligand (CD40L), which interacts with CD40 on antigen-presenting cells, such as macrophages and dendritic cells.54 In a murine model of leishmaniasis, CD40?CD40L interaction was essential for macrophage activation and parasite killing.55 However, macrophages deficient in CD40 stimulated with immune CD4 T cells can still kill intracellular M tuberculosis (unpublished, V Lazarevic, JLF), which suggests that this interaction is not required for adequate macrophage activation in tuberculosis.
In some systems, CD4 T cells are necessary for the induction or function of CD8 T cell responses.56?59 In the murine model, CD4 T cells were not required for priming of interferon-γ-producing CD8 T cells specific for M tuberculosis.42,60 However, CD8 T cells in the lungs of mice deficient in CD4 T cells showed impaired ability to kill macrophages infected with M tuberculosis.60 Thus, CD4 T cells were necessary for the cytotoxic but not cytokineproducing function of the CD8 T cells. The underlying mechanism remains unclear but may relate to a reduction in expression of interleukin 15 in the lungs,60 because this cytokine is important in function of CD8 T cells.61?63 This feature may point to an important secondary function of CD4 T cells in control of M tuberculosis infection.
Mycobacteria-specific CD8 T cells could be cytotoxic for infected macrophages, as well as producing cytokines. In human beings, CD8 T cells specific for M tuberculosis can directly kill intracellular bacilli via a granule-associated protein, granulysin.64 CD8 T cells have been reported to be involved in control of both acute and chronic infection,65?68 although in some studies the requirement for CD8 T cells was minor.69,70 The requirement for cytotoxic function by CD8 T cells has not been shown in the mouse model. Although cytotoxic CD8 T cells are present in the lungs of infected mice, mice lacking perforin did not show increased susceptibility to M tuberculosis.71,72 The true contribution of CD8 T cells to protection against tuberculosis cannot be tested in mice, because they do not have a granulysin homologue.
Evidence is emerging for a role of CD8 T cells in control of latent tuberculous infection. People with latent M tuberculosis infection have high frequencies of mycobacteria-specific CD8 T cells.73,74 In the Cornell mouse model of latent tuberculosis, depletion of CD8 T cells caused reactivation of the infection.28 In the chronic tuberculosis murine model, efficient depletion of the activated CD8 T cells from the lungs by use of antibody is difficult (unpublished, JLF), so the possibility of assessing contribution of these cells directly in chronic infection is limited. The cytotoxic potential of CD8 T cells is less in chronic than in acute infection, whereas potential to produce interferon-γ remains high (unpublished, V Lazarevic, JLF). The differential regulation of the function of CD8 T cells in the lungs in tuberculosis is currently under investigation.
Back to top
Modulation of antigen presentation: results of in-vitro studies
There is much evidence that M tuberculosis can interfere with the pathway for processing and presentation of MHC class II antigens. This pathway is crucial for the priming and effector function of CD4 T cells. Since CD4 T cells are an important component of the protective immune response against M tuberculosis, the ability of the tubercle bacillus to affect adversely the efficiency of MHC class II processing and presentation probably contributes to persistence.
In a study designed to characterise M tuberculosis mannose-capped lipoarabinomannan as a potential virulence factor, this complex glycolipid attenuated interferon-γ-induced expression of MHC molecules by a human macrophage-like cell line.75 The ability of M tuberculosis to modulate the expression of class II molecules was later confirmed.76?81 One study on human peripheral-blood-derived monocytes infected with the avirulent vaccine strain BCG did not show decreased expression of HLA class II.82 These discrepant results may be due to the differences in the experimental systems used.
Other mycobacterial components reported to suppress macrophage expression of MHC class II molecules include a 25 kDa glycolipoprotein77 and a 19 kDa lipoprotein.80 The precise mechanisms by which M tuberculosis attenuates expression of MHC class II molecules have not been clearly defined. However, possibilities include intracellular sequestration of immature class II heterodimers,78 the inhibition of interferon-inducible gene expression,75 and downregulation of the expression of class II transactivator.79 Studies involving M tuberculosis mannosecapped lipoarabinomannan,75,82 the 25 kDa glycolipoprotein,77 and the 19 kDa lipoprotein show that mycobacterial components actively contribute to downregulation of MHC class II expression by macrophages infected with M tuberculosis. The ability of mycobacteria to inhibit antigen processing or presentation has been shown in various in-vitro macrophage models.76,80,83 The 19 kDa protein can attenuate the processing of non-mycobacterial soluble antigens and antigen 85B of intact M tuberculosis.80 The inhibitory effects of the 19 kDa protein depend on TLR2. This receptor has an important role in initiating the innate immune response to the tubercle bacillus.84 The TLR2?dependent interference of the MHC class II antigen-presentation pathway occurs only at later times after treatment of macrophages with the 19 kDa lipoprotein.80 The hypothesis is that initial signalling of the 19 kDa lipoprotein through interaction with TLR2 triggers the innate immune response. As the infection progresses, this interaction mediates the inhibition of MHC class II antigen processing and presentation, thus enabling evasion from surveillance by CD4 T cells. Since TLR2 can sample intraphagosomal bacterial components,85 the 19 kDa lipoprotein, if expressed during the chronic phase of infection, could provide a sustained means of downregulating the host's antigen processing and presenting pathway, resulting in a safe haven for the mycobacteria within macrophages.
Phagosomes containing M tuberculosis are competent antigen-presenting organelles.81 Apparently, processing of mycobacterial antigens occurs in these phagosomes to form a complex of peptide and MHC class II antigens.81 Phagosomes containing live bacilli have smaller amounts of the complex than in vacuoles containing dead bacilli. Delay in the maturation of phagosomes containing live but not dead bacilli is also reported. These findings suggest that live mycobacteria can attenuate antigen processing and arrest phagosome maturation. Since these studies involved a time of less than 100 min from initiation of infection to examination of the phagosomes, M tuberculosis must interfere with processing of class II antigens and phagosomal maturation almost immediately after entering the host cells. These observations, together with those showing the delayed kinetics of the 19 kDa lipoprotein-mediated inhibitory effects,80 suggest that M tuberculosis can subvert the MHC class II antigen-presentation pathway in different phases of the tuberculous
Interference with macrophage effector functions Macrophages have a unique role in the host response to mycobacterial infection, since they represent both the primary effector cells for killing M tuberculosis and the primary habitat in which persisting bacilli reside. Here we discuss how M tuberculosis evades two major macrophage antimycobacterial mechanisms-the generation of nitric oxide and related reactive nitrogen intermediates (RNI), and the fusion of phagosomes and lysosomes (figure 2).

Click to enlarge image
Figure 2. Immune evasion strategies of M tuberculosis: interaction with macrophages. The maturation of phagosomes containing M tuberculosis apparently stops at a point close to the acquisition of the GTPase Rab5 (1). This arrest of phagosome biogenesis prevents fusion with lysosomal compartments (3) that have potent antimicrobial activities. Within the phagosome, M tuberculosis is subject to the antimycobacterial effect of reactive nitrogen intermediates (RNI) generated by the macrophage NOS2 (2). In addition, M tuberculosis can inhibit the MHC class II-dependent antigen presentation pathway (4). Thus, M tuberculosis can subvert various antimycobacterial functions of macrophages.
Macrophages activated by T-cell-derived cytokines generate products with antimycobacterial activity, among the best studied of which are the RNI. The antimycobacterial effects of these intermediates were first shown by in-vitro macrophage studies.86,87 Evidence from murine models involving NOS inhibitors or mice with disruption of the nos2 gene further supports a crucial role for RNI in host defence against M tuberculosis, in both acute and chronic persistent infections.39,40,88,89
However, the importance of nitric oxide and other RNI in human defence against M tuberculosis is a matter of substantial controversy,90 partly because the experimental manipulations that have defined the importance of these substances in the mouse model are not applicable to human tuberculosis. Human beings with tuberculosis are not intentionally given inhibitors of NOS, and no genetic disorder that renders individuals unable to produce NOS2 has so far been identified. Genetic defects in components of the cytokine cascade that lead to NOS2 induction, such as interleukin 12 and its receptor and the ligand-binding and signal-transducing portions of the interferon-γ receptor have been associated with increased susceptibility to mycobacterial infection;44 however, these deficiencies in cytokine expression are likely to have more far-ranging effects than merely inhibition of NOS2 production. Treatment with 1,25?dihydroxyvitamin D3 induced NOS2 expression and also limited growth of M tuberculosis in a human promyelocytic cell line,91 and certain polymorphisms in the vitamin-D-receptor gene were under-represented among patients with pulmonary tuberculosis compared with blood-donor controls.92 These findings suggest a link between vitamin D signalling, NOS2 production, and susceptibility to clinical tuberculosis, but again the vitamin-D-receptor mutations could have pleiotropic effects. So the relevance of RNI in controlling M tuberculosis infection in human beings must rely on indirect evidence.
Alveolar macrophages from healthy individuals produce RNI in response to infection with M tuberculosis, and production of the intermediates is associated with constrained intracellular growth of the bacteria.93 Human alveolar macrophages restrict growth of BCG through the induction of NOS2 activity, and this can be prevented by treatment with a NOS2 inhibitor.94 Raised NOS2 expression has been found in human alveolar macrophages from patients with untreated, culture-positive pulmonary tuberculosis, by use of a highly specific antibody to NOS2,95 whereas the amount of nitric oxide in exhaled air was higher in patients with tuberculosis than in controls, and the amount of exhaled nitric oxide decreased after antituberculous chemotherapy.96 Choi and colleagues97 showed by immunostaining that there is greater expression of NOS2 in the inflammatory zone of granulomas, and in areas of pneumonitis, in surgically resected lung tissue of eight patients with tuberculosis than in uninfected controls.
Back to top
Avoidance of the toxic effects of RNI
The production of RNI by NOS2 therefore seems to be essential for the containment of M tuberculosis infection. Yet in both mice and human beings, many immunocompetent individuals, though able to contain infection with M tuberculosis, are unable to eliminate the bacterium completely. This feature suggests that M tuberculosis expresses genes that counteract the bactericidal or bacteriostatic effects of RNI; studies by various experimental approaches now support this hypothesis.
A selection strategy was used to identify several candidate resistance genes of M tuberculosis. A DNA library derived from a clinical isolate of M tuberculosis highly resistant to RNI was introduced into a number of RNIsensitive bacterial hosts; the bacteria were then subjected to RNI stress. M tuberculosis genes that permitted survival of the bacterial host strains were identified. This strategy led to the identification of noxR1, conferring enhanced resistance to reactive oxygen intermediates (ROI) and RNI in Escherichia coli and Mycobacterium smegmatis,98 and noxR3, which is able to protect Salmonella typhimurium from ROI and RNI.99 At present, the mechanism by which these genes confer RNI resistance remains to be defined.
Another approach to identify RNI-resistance genes in M tuberculosis involves study of homologues in other bacteria that protect from oxidative stress, an approach made simpler with the publication of the complete genome sequence of M tuberculosis H37Rv, a common laboratory strain.100 This strategy was used to identify alkyl hydroperoxide reductase subunit C (AhpC) as a gene product affording protection against RNI.101 AhpC is a peroxiredoxin initially cloned from E coli and S typhimurium as an oxidative-stress response protein.102?105 Disruption of ahpC rendered S typhimurium highly susceptible to killing by RNI, and this defect was complemented with the ahpC gene of M tuberculosis.101 Mycobacterial ahpC was also able to protect human cells from killing by S-nitrosoglutathione (GSNO) and from apoptosis and cytotoxicity induced by NOS2.101
Subsequent biochemical studies of the wild-type and various site-directed mutants showed that AhpC can metabolise peroxynitrite anion (OONO−) into nitrite, thereby detoxifying this highly reactive species.106 Peroxynitrite is a powerful oxidant formed by the reaction between nitric oxide and superoxide anion (O2−), both of which are produced by activated macrophages, and can exert toxic effects through protein modifications.107 In-vitro studies have shown that although virulent M tuberculosis is very resistant to OONO−, the avirulent M smegmatis and BCG strains are susceptible.108 As for ahpC, a candidate gene approach has also been used to identify additional mycobacterial genes conferring nitric oxide resistance based on their homologues in other organisms. For instance, in studying the specific RNI species that cause antimicrobial effects, St John and colleagues109 elucidated an enzymatic basis for OONO resistance that is distinct from the detoxification of OONO− by AhpC. They focused on the peptide methionine sulphoxide reductase (msrA) gene, shown to have a role in resistance to oxidative stress in yeast and E coli,110,111 and found that E coli lacking MsrA was highly susceptible to killing by nitrite and GSNO (under aerobic conditions, when these species can be converted intracellularly to peroxynitrite), and that this defect was complemented by the msrA gene of M tuberculosis. The results suggest both that oxidation of methionine residues is a target for peroxynitrite, and that MsrA is crucial for repair of peroxynitrite-induced toxicity. The genome of M tuberculosis includes two genes, glbN and glbO, encoding truncated haemoglobin-like proteins of unknown function. A BCG strain in which glbN was disrupted showed reduced capacity to metabolise nitric oxide during stationary phase growth.112 In addition, the glbN mutant strain showed striking inhibition of aerobic respiration in the presence of nitric oxide compared with wild-type bacteria.112 Although M tuberculosis glbN-disrupted strains have yet to be investigated in vivo, these findings point to a role for mycobacterial truncated haemoglobin-like proteins in promoting bacterial resistance to RNI toxicity.
Given the important role of RNI in controlling mycobacterial infection, it is not surprising that tubercle bacilli have developed complementary strategies to combat nitrosative stress and thereby permit persistence within the macrophage, for example by detoxifying peroxynitrite (ahpC) and nitric oxide (glbN) and by repairing peroxynitriteinduced damage (msrA). Together, these mechanisms may contribute significantly to the establishment of persistent infection. Although understandable from the perspective of the bacterium, such redundancy can render the in-vivo demonstration of the significance of these genes challenging. For instance, an ahpC-disrupted strain of M tuberculosis showed increased susceptibility to peroxynitrite (though not to nitric oxide alone) in axenic culture as well as reduced survival in unactivated macrophages, with percentage survival an order of magnitude less than wild-type at 7 days after macrophage infection.113 By contrast, in macrophages stimulated with interferon-γ and lipopolysaccharide no survival difference between wild-type and ahpC-disrupted M tuberculosis strains was apparent, suggesting that activated macrophages have other antimycobacterial effectors, or that bacilli within activated macrophages express other mechanisms that compensate for or mask the absence of AhpC.113 A M tuberculosis ahpC mutant showed no growth defect in lungs and spleens during acute infection of BALB/c mice (up to 56 days after infection).114
Back to top
Remodelling the phagosome
Phagocytosis of pathogenic microorganisms by ?professional? phagocytes such as macrophages and neutrophils is the first step in their eventual degradation, as the phagosome eventually matures into a phagolysosome rich in acid hydrolases with degradative and microbicidal capacity. The maturation of a pathogen-containing phagosome is a highly complex and ordered process that involves the interaction of the phagosome with several endocytic vesicles through fission and fusion events, the molecular details of which are currently being unravelled.115?117
M tuberculosis has adopted a strategy to avoid destruction by lysosomal enzymes, that of interrupting the process of phagosome maturation. The studies of D'Arcy Hart and colleagues118,119 showed that phagosomes containing M tuberculosis do not fuse with lysosomes. This failure of phagolysosomal fusion occurred only with the ingestion of viable bacteria; dead bacteria were effectively delivered to lysosomes. Subsequently, vacuoles containing mycobacteria were shown to be less acidic than lysosomes,120 and a biochemical basis for the acidification defect was provided by studies showing that mycobacterium-containing phagosomes exclude the vesicular proton ATPases.121 Although phagosomes containing live mycobacteria do not fuse with lysosomes, they remain dynamic structures fully capable of fusion with other vesicles, such as early endosomes, and are accessible to internalised plasma membrane components such as the transferrin receptor and to the fluid-phase marker horseradish peroxidase.122?125
Recent advances in defining the biochemical and molecular basis for the trafficking of intracellular vesicles have opened new avenues of exploration into the mechanism of arrested maturation of the mycobacterial phagosome by identifying the basic tethering and fusion machinery involved in vesicular transport.117,126?128 Soluble N−ethylmaleimide?sensitive factor?attachment protein receptors (SNAREs), a family of membrane?anchored proteins, have a key role in mediating membrane fusion. In addition to SNAREs, the machinery required for specific membrane fusion is thought to include small GTPases of the Rab family
Via and co?workers129 showed that BCG?containing phagosomes fail to acquire the late endosomal GTPase Rab7 but retain the early endosomal GTPase Rab5, indicating that the phagosomes retain early endosomal characteristics and remain excluded from the late endosomal compartment. Clemens and colleagues130 reported that HeLa cells overexpressing Rab5c, when infected with M tuberculosis, form phagosomes that stain abundantly and persistently for Rab5. However, these investigators subsequently reported that phagosomes containing M tuberculosis also stained abundantly for the Rab7 GTPase, in a HeLa cell line stably transfected with the human RAB7 gene in a system permitting regulated expression.131 The reasons for the apparently discrepant Rab7 data from these two independent studies are unclear, but possible explanations are the dissimilar experimental systems used, including different mycobacterial species, host cells, and methods of analysis
Evidence has emerged lately supporting the contention that maturation of mycobacterial phagosomes is arrested at a point close to the acquisition of Rab5; EEA1 (early endosome autoantigen), a known Rab5?interacting partner and effector, is excluded from mycobacterium-containing phagosomes.131 The binding of EEA1 to early endosomal membranes depends on the modification of membrane lipids by a phosphatidylinositol 3(OH) kinase enzyme termed hVPS34.133 Both EEA1 and hVPS34 have a role in phagosome biogenesis.132M tuberculosis mannosecapped lipoarabinomannan inhibited activity of phosphatidylinositol 3(OH) kinase,132 and this could be a mechanism by which mycobacteria attenuate EEA1 recruitment. In addition, in comparison with phagosomes containing control beads, phagosomes containing latex beads coated with mannose-capped lipoarabinomannan recruited less syntaxin 6 (a SNARE apparently involved in trans-Golgi network to phagosome trafficking) and accumulated less of the lysosomal enzyme precursor immature cathepsin D. These findings point to a role for mannose-capped lipoarabinomannan in interfering with phagosomal acquisition of components from the trans-Golgi network, including lysosomal hydrolases.134 Therefore, blocking of recruitment of syntaxin 6 by mannose-capped lipoarabinomannan may account for the diminished accumulation of lysosomal markers in mycobacteriumcontaining phagosomes. As further evidence that mycobacterial lipids exert multiple and far-ranging effects in the infected macrophage, fluorescently labelled surface glycolipids of BCG are released from the mycobacterial cell wall and move throughout the endocytic network, ultimately entering many intracellular compartments.135
The phagosome biogenesis pathway, and its modification by M tuberculosis gene products, could differ depending on the activation state of the macrophage. Phagosomes containing M avium (MAC), like those containing M tuberculosis or BCG, also show limited fusion with lysosomes;136 however, when macrophages are costimulated with interferon-γ and lipopolysaccharide, the MAC-containing phagosomes can acidify, accumulate proton ATPases, acquire other characteristics of phagosome maturation, and ultimately exert microbicidal effects.137 This apparent ability of the host immune response to affect the fate of M tuberculosis is reminiscent of the observation that antibody-coated bacilli are defective in the prevention of phagolysosomal fusion and yet maintain the ability to survive in macrophages.138 The mechanism underlying this latter observation is not known at present. Clearly, a complex interplay between the host immune response and bacterial gene products determines the fate of M tuberculosis within the phagosome.
Another mechanism by which mycobacteria could interfere with phagolysosomal fusion is by retention on mycobacterial phagosomes of the host protein TACO (tryptophan aspartate rich coat protein).139 TACO, or mouse coronin 1, was originally described in the free-living soil amoeba Dictyostelium discoideum; coronin-deficient mutants show abnormalities in phagocytosis and cytokinesis.140 TACO remains associated with phagosomes containing viable, but not heat-killed, BCG, and transfection of a non-phagocytic melanoma cell line with cDNA encoding TACO blocked the transfer of mycobacteria to late endosomal/lysosomal organelles and resulted in slightly greater intracellular survival of mycobacteria.139 However, the importance of the association of TACO with mycobacterial phagosomes in preventing fusion with lysosomes remains controversial. Schuller and colleagues141 reported lately that coronin 1 was retained by phagosomes containing clumps of ten to 20 bacteria, but not by phagosomes containing individual bacteria, where coronin 1 was recruited by the early phagosomes but then lost within 24 h. This finding challenges the significance of TACO retention in the maturation arrest of mycobacterial phagosomes and raises the possibility that clumping of bacteria may represent a virulence factor for M tuberculosis.142
D discoideum has been used as a model system for studying mycobacterial?host interactions, with Mycobacterium marinum as the bacterial pathogen.143 In human beings M marinum causes cutaneous disease as a result of skin trauma in contaminated water (swimming pool granulomas, fish-handler's nodules), and in ectothermic animals such as fish and amphibians the organism can cause a disseminated systemic infection characterised by chronic granulomatous lesions and persistence of organisms within tissues.144 Solomon and coworkers143 found that, in contrast to expectations, M marinum showed improved intracellular growth in a coronin-deficient mutant of D discoideum compared with wild-type. They speculated that the uptake of mycobacteria may proceed along a unique pathway, distinct from the ?default pathway? of phagocytosis, in which D discoideum coronin mutants were previously shown to have defects,145 and that in the absence of competition from this alternative pathway the intracellular growth is improved. The investigators also considered that their findings may differ from previous work showing phagocytosis defects in coronin null mutants because they used amoebic monolayers for the M marinum infections rather than suspension cultures. In any case, the results clearly show that coronin is dispensable for M marinum uptake and growth within D discoideum; another possibility is that M tuberculosis (and its close counterpart M bovis BCG) have different requirements from M marinum for establishment of intracellular infection, or that the essential presence of a coronin isoform may differ for host cells as diverse as amoebic cells and mammalian macrophages.
Back to top
Cytokines and latent tuberculosis
Although a protective role for TNF against M tuberculosis is well established in mice,27,146?149 there was no evidence until recently that this cytokine is important for the control of tuberculosis in human beings. In fact, TNF was thought to contribute to the immunopathology of tuberculosis in both mice and human beings.150?153 Administration of thalidomide (which suppresses TNF production) to individuals with M tuberculosis infection was reported to result in improvement of overall clinical status, including weight gain.150 Indeed, the pathogenetic effects of TNF have been well documented in human diseases other than tuberculosis.154 For example, cachexia observed in cancer patients has been ascribed to the immunopathological effect of TNF.155 Clearly, TNF has both adverse and beneficial effects in the human immune response to pathogens, tumours, and tissue injury.
Attenuation of the biological activity of TNF has lately become an important therapeutic intervention in the management of a wide variety of chronic inflammatory diseases.156,157 TNF blockade effectively ameliorates inflammation in rheumatoid arthritis and Crohn's disease. These two disorders are among a rapidly expanding list of inflammatory diseases for which the efficacy of TNF blockade is being investigated.157 Currently two major anti-TNF protocols are being used for the treatment of various chronic inflammatory diseases.158,159 Infliximab is an anti-TNF neutralising antibody, and etanercept is a recombinant molecule made up of two identical chains of the human 75 kDa TNF receptor 2 fused to the Fc portion of human IgG1.
A major side-effect of anti-TNF therapy is the increased risk of developing tuberculosis.157?160 This is the most convincing evidence that TNF has a role in the control of human tuberculosis. Up to November 30, 2001, 110 cases of tuberculosis in patients receiving anti-TNF treatment had been reported to the US Food and Drug Administration.160 This increased risk has mandated screening for, and treatment of, individuals at risk of developing tuberculosis or with active infection before the initiation of anti-TNF therapy.161 The need for this prudent practice is emphasised by the non-typical presentation of tuberculosis associated with TNF blockade, which can delay diagnosis.162 A retrospective study of 70 reported cases of tuberculosis associated with anti-TNF therapy before disease manifestation showed that 56% of the study participants had extrapulmonary infection and 22% developed dissemination.162 This disease manifestation is highly nontypical, since extrapulmonary disease and disseminated infection occur in the general population at rates of 18% and 2%, respectively. Data gathered by review of the clinical history and the demographic characteristics of these patients, including age, the temporal relation between anti-TNF therapy and disease manifestation, prevalence of tuberculosis in the geographical location, and the history of exposure to M tuberculosis, suggested that most of the tuberculosis cases associated with TNF blockade represent reactivation of a chronic infection.162
The mechanisms underlying the TNF-blockade-induced reactivation of latent tuberculosis have not been clearly defined. Histological examination of tissues procured by open lung biopsy of an index case, in whom cultureconfirmed tuberculosis developed 7 weeks after a single dose of infliximab, revealed copious infiltration of inflammatory cells with no evidence of acid-fast bacilli in the lesion.162 In addition, pulmonary granulomas, a prominent feature in tuberculous tissues, were not seen in the lung specimens of this patient.162 Thus, TNF seems to have a role in maintaining the organisation of the granuloma, the integrity of which contributes to the containment of M tuberculosis.
Certain histological features observed in the pulmonary lesions of human beings with infliximab-related tuberculosis162 are very similar to the lung lesions of persistently infected mice undergoing TNF-blockadeinduced reactivation.147 In the chronic infection model, TNF neutralisation by monoclonal antibody resulted in reactivation disease characterised by an increase in tissue bacterial burden and mortality, and a greatly enhanced inflammatory reaction in the granulomatous lesions in the lungs. Augmentation of the inflammatory response was reflected by pronounced infiltration of immune cells associated with granuloma disorganisation. Significantly, the highest bacterial burden in the lungs of TNF-neutralised mice, attained 3 weeks after initiation of infection with virulent M tuberculosis, was 107 colony-forming units.147 This bacterial load remained constant until all mice neutralised for TNF died from the infection. Within the time frame of these studies, this bacterial burden is not generally fatal. The dissociation between bacterial load and mortality has been confirmed in another murine tuberculosis model-SPCTNFRIIFc transgenic mice with lung-specific expression of a soluble TNF inhibitor.163
Analysis of pulmonary gene expression in TNFneutralised mice of the low-dose model showed adequate amounts of interleukin 12 and interferon-γ, compared with controls. Expression of NOS2 was low but not absent. Together, these findings suggest that the severe inflammatory response associated with TNF blockade has a significant role in causing the increased mortality observed in these mice. A corollary of this notion is that in the chronic phase of tuberculous infection, TNF plays a part in regulating the inflammatory response and maintaining the integrity of the granuloma. The protective role of TNF in persistent tuberculosis is further underlined by studies in the Cornell model of latent tuberculosis.27 TNF neutralisation after establishment of apparent sterility resulted in disease reactivation. Thus, animal studies designed to assess the function of TNF in persistent infection have provided results that predicted the risks of reactivation tuberculosis during TNF blockade therapy.
Although TNF blockade is well established as an effective treatment for various chronic inflammatory diseases, the mechanisms by which it attenuates inflammation have not been defined. In patients with Crohn's disease, TNF blockade induced T-cell or macrophage apoptosis164,165 and led to downregulation of interferon-γ expression in human T cells.166 TNF has a role in apoptosis of macrophages infected with M tuberculosis also.50,167 TNF, with its ability to regulate the expression of chemokines, chemokine receptors, and adhesion molecules, is a potent modulator of cell migration.147,168 This attribute, together with the evidence that TNF-blockade-associated tuberculosis shows granuloma disorganisation, suggests that this cytokine has an important role in the maintenance of the granulomatous response in latent tuberculosis. A disorganised granuloma may be deficient in restricting inflammation to a localised environment. This deficiency could lead to extension of the inflammatory response into otherwise unaffected lung tissues, resulting in the aberrant pathology observed in tuberculous patients treated with TNF blockade.147,168 Characterisation of the mechanisms by which TNF neutralisation ameliorates chronic inflammation will probably shed light on how this means of immunomodulation leads to reactivation of latent tuberculosis, and how the host controls latent M tuberculosis infection.
Back to top
Interleukin 10?regulation or dysfunction?
Interleukin 10 is an immunoregulatory cytokine produced by T cells, macrophages, and dendritic cells. It can deactivate macrophages and can dampen the immune response to prevent or limit pathology from an over-exuberant inflammatory response to a pathogen. In transgenic mice expressing interleukin 10 compared with control mice, BCG was present in slightly higher numbers in spleen and liver;169 this feature suggests that increased interleukin 10 could be detrimental to bacterial control. However, interleukin-10−/− mice were similar to wild-type mice in the ability to control virulent M tuberculosis aerosol infection, indicating that absence of interleukin-10 production did not result in improved bacterial control, even after 8 months of infection.170,171 Interleukin 10 does affect the immune response in infected mice; interleukin-10−/− mice had somewhat higher production of interferon-γ in the lungs after M tuberculosis infection (unpublished data). However, in resistant mice such as the C57BL/6 strain, the immune response seems to be sufficient, and increasing interferon-γ production does not have a major effect on control of infection.
In studies of reactivation induced by TNF neutralisation in mice, interleukin-10 mRNA expression in the lungs increased,147 leading to the hypothesis that increased interleukin 10, in conjunction with decreased TNF, might be driving the reactivation. To test this possibility, we treated chronically infected interleukin-10−/− or wild type mice with anti-TNF monoclonal antibody and compared the reactivation rate (H Scott, JLF, JC, unpublished). Although all animals showed reactivation after TNF neutralisation, the interleukin-10−/− mice showed delayed reactivation, suggesting that the cytokine contributes to reactivation. Histology suggested that the lung macrophages were more activated in the interleukin-10−/− mice than in control mice.
Transgenic mice, in which interleukin 10 is expressed under the control of the interleukin 2 promoter and thus is increased when activated T cells are at the site of infection, were also more prone to reactivation than wild-type (C57BL/6) mice.172 In that study, interleukin 10 was increased in the lungs of the transgenic mice throughout infection, but bacterial numbers were higher in the transgenic mice only in the chronic phase (>70 days after infection). Interleukin 12 and M tuberculosis specific interferon-γ production in the lungs was decreased in the interleukin-10 transgenic mice compared with wild-type. In general, the phenotype of these mice was similar to a mouse strain CBA/J that is ?reactivation-prone?, or less able to control chronic infection than the resistant C57BL/6 strain.172,173 These findings suggest that increased interleukin− 10 production at the site of infection (ie, the lungs) increases susceptibility to reactivation. In the human setting, such an increase could conceivably result from a secondary infection or an immunopathological process in the lungs.
Interleukin 10 also seems to inhibit or modulate initiation of the immune response against M tuberculosis. Dendritic cells from interleukin-10−/− mice produced more interleukin 12 than wild-type dendritic cells on infection with BCG or CD40 stimulation in vitro.174 When infected dendritic cells were transferred to mice, interleukin-10−/− dendritic cells were more efficient than wild-type cells at priming an interferon-γ-producing T-cell response against mycobacterial antigens. Increased priming by interleukin-10−/− dendritic cells was associated with increased migration of these cells to the draining lymph nodes as well as increased interleukin-12 production by the resident dendritic cells within those lymph nodes. Thus, interleukin 10 can affect migration of dendritic cells, perhaps by modulation of the chemokine receptors on the dendritic cells,175 and interleukin-12 production at the priming or effector site. In the human system, production of interleukin 12 by dendritic cells in response to M tuberculosis infection is also regulated by interleukin 10, which could affect T-cell priming in vitro.176 In that study, interleukin-12 production by human macrophages was completely inhibited by endogenous interleukin-10 production, although this effect was overcome by the presence of interferon-γ.
Interleukin 10 is thought to be important in limiting immunopathology caused by an over-exuberant immune response. This cytokine may have a role in the chronic phase of infection, since downregulation of a type 1 immune response is likely to be beneficial to the host, at least in terms of lung pathology. Although interleukin 10 is produced at low levels in the lungs of C57BL/6 mice infected with M tuberculosis, in situations of substantial pathology or uncontrolled infection, interleukin-10 expression in the lungs is greatly increased (unpublished, V Lazarevic, JLF). The macrophages appear to be the primary source of interleukin? 10 expression under these circumstances. This may be an attempt by the host to dampen an immune response that increases substantially in response to high bacterial load.
Back to top
Conclusion
Control of the initial phase of M tuberculosis infection is likely to be important to the downstream events of the disease. The immunological mechanisms involved in maintaining a latent infection are complex, but are clearly necessary to prevent reactivation. Although the host response is essential to control of infection, the tubercle bacillus participates in the establishment of latency by using various strategies to evade elimination by the host. The bacteria can wait for the immune response to falter and then can emerge to cause active tuberculosis. The constant battle between the host and the microbe occurs at the level of the granuloma. The interplay of cytokines, including interferon-γ, interleukins 12 and 10, and TNF, is likely to be crucial to the ultimate fate of a person with M tuberculosis infection. In addition, the contribution of T cells and macrophages, as well as dendritic cells, to the successful immune response is undisputed. Tipping the balance of the response in one direction or the other, particularly in the context of the local immune environment of the granuloma, may determine whether an infection will remain latent or reactivate to cause active tuberculosis. Further studies, particularly at the level of the granuloma, will elucidate this complex interaction and perhaps provide surrogate markers for disease progression in human beings.
Back to top
Search strategy and selection criteria
Data for this review were identified though searches of Medline and PubMed, through references cited in relevant articles, and through searches of the authors' files. Search terms used were ?mycobacterium?, ?tuberculosis?, ?latency?, ?persistence?, ?animal models?, ?T cells?, ?macrophage?, ?nitric oxide?, ?phagosome?, ?antigen presentation?, ?tumour necrosis factor?, and ?IL-10?. Only papers published in English were reviewed.
Back to top
Conflicts of interest
None declared.
Acknowledgments
Our work is supported by grants AI49157, HL68526, HL71241, and AI50732 (JC, JLF), AI37859 and AI47485 (JLF), and AI49375 (JMT) from the National Institutes of Health, and grant CI-016?N from the American Lung Association (JLF). We thank the members of the Flynn and Chan laboratories for helpful discussions and unpublished data.
DOI:10.1016/S1473-3099(03)00741-2
Latent tuberculosis: mechanisms of host and bacillus that contribute to persistent infection
JoAnn M Tufariello a, Prof John Chan a and Prof, Dr JoAnne L Flynn


Summary
Immune response to M tuberculosis
Animal models for study of latent tuberculosis
T cells in latent tuberculosis
Modulation of antigen presentation: results of in-vitro studies
Avoidance of the toxic effects of RNI
Remodelling the phagosome
Cytokines and latent tuberculosis
Interleukin 10?regulation or dysfunction?
Conclusion
Search strategy and selection criteria
References
Summary
Most people infected with Mycobacterium tuberculosis contain the initial infection and develop latent tuberculosis. This state is characterised by evidence of an immune response against the bacterium (a positive tuberculin skin test) but no signs of active infection. It can be maintained for the lifetime of the infected person. However, reactivation of latent infection occurs in about 10% of infected individuals, leading to active and contagious tuberculosis. An estimated 2 billion people worldwide are infected with M tuberculosis-an enormous reservoir of potential tuberculosis cases. The establishment and reactivation of latent infection depend on several factors, related to both host and bacterium. Elucidation of the host immune mechanisms that control the initial infection and prevent reactivation has begun. The bacillus is well adapted to the human host and has a range of evasion mechanisms that contribute to its ability to avoid elimination by the immune system and establish a persistent infection. We discuss here current understanding of both host and bacterial factors that contribute to latent and reactivation tuberculosis.
Back to top
Tuberculosis is a major cause of death worldwide, causing an estimated 1?5?2 million deaths each year and rivalling malaria as the commonest cause of death due to a single infectious agent.1Mycobacterium tuberculosis was first identified as the causative agent by Robert Koch in 1882. Tuberculosis has affected human beings for thousands of years.2 Despite decades of research on chemotherapy for the disease and development of preventive vaccines, it remains a major public?health problem and was declared a global health emergency by WHO in 1993.3 Tuberculosis is preventable and in most cases can be cured, but various demographic and socioeconomic factors make both prevention and treatment difficult. The number of new cases of tuberculosis is projected to reach 11?9 million annually by the year 2006 if current control efforts are not strengthened.4 The recent resurgence in tuberculosis is attributable partly to the AIDS epidemic5 with the emergence of multidrugresistant strains also hindering efforts at control.6,7
M tuberculosis is an extraordinarily effective human pathogen. Surveys with purified protein derivative (PPD) or tuberculin skin tests suggest that one-third of the world's population is infected with the bacillus.8 Primary infection leads to active disease in only a minority (about 10%) of infected individuals,9 in most cases within 2 years. In the remaining 90% of cases the immune system contains the infection, and the individual is non-infectious and symptom-free. This clinical latency can persist throughout the person's lifetime. However, in some circumstances the host immune response is perturbed, and reactivation of latent infection results. This process can occur, for example, through HIV infection, malnutrition, the use of steroids or other immunosuppressive medications, or advanced age.10 Although the estimated lifetime risk of developing reactivation disease is between 2% and 23%,11 the risk for individuals immunosuppressed by HIV infection is estimated to be as high as 10% per year.12 Reactivation of latent infection contributes substantially to the incidence of adult tuberculosis, especially in more developed countries where disease prevalence is fairly low.13,14
Although M tuberculosis is known to have the capacity to persist in human tissues,15 many fundamental questions about the mechanisms of persistence remain unanswered. What regulates the transition from initial growth to persistence, and the transition back to active growth in reactivation of latent disease? What are the bacterial determinants necessary for persistent infection? What is the physiological state of tubercle bacilli during latent infection-are the bacteria in a non-replicating ?spore-like? state or are they at all replicative? To what extent are the existing bacilli metabolically active? What is the microenvironment encountered by M tuberculosis within the granuloma? How does the bacterium counteract or evade host defences and survive in the face of a vigorous host immune response? The answers to these questions are likely to provide insight into both the mechanisms by which M tuberculosis persists within the host and the means of elimination of latent infection, the disease phase that poses the most significant obstacle to the eradication of tuberculosis. Clearly, these are difficult questions to address experimentally, particularly in the human system. In this review we address the immune mechanisms of the host that contribute to control of infection and the ways in which M tuberculosis has evolved to avoid elimination, focusing on the interactions between the bacillus and CD4 T cells, CD8 T cells, and macrophages. Together, these strategies result in the establishment of latent M tuberculosis infection.
Immune response to M tuberculosis
The immune response to M tuberculosis is multifaceted and complex. T cells are an essential component of the protective response, and the interaction of these cells with infected macrophages is crucial for control of infection. The immune response is initiated when M tuberculosis arrives in the alveolar space, where it encounters alveolar macrophages. In response, at least partly through interactions of mycobacterial components with Toll-like receptors (TLR), the macrophages produce inflammatory cytokines and chemokines that serve as signals of infection.16 The bacteria enter the parenchyma and can replicate within the alveolar macrophages or in resident lung macrophages. The signals induced result in migration of monocyte-derived macrophages and resident dendritic cells to the focal site of infection in the lungs. The dendritic cells that engulf bacteria mature17?19 and migrate to the regional lymph node. Once there, CD4 and CD8 T cells are primed against mycobacterial antigens. Primed T cells expand and migrate back to the lungs and then through the lung tissue to the focus of infection, presumably in response to signals such as chemokines produced by or in response to infected cells. The migration of macrophages and T cells (as well as B cells) to the site of infection culminates in formation of a granuloma, a characteristic feature of tuberculosis. In addition to T lymphocytes and macrophages, the granuloma consists of other host cells including B cells, dendritic cells, endothelial cells, fibroblasts, and probably stromal cells (figure 1).20 The relative proportions of the different cell types in the granuloma vary with its age. The granuloma encompasses the bacilli, which reside within macrophages, and serves to wall off the bacteria from the rest of the lung, limiting spread. In addition, the granuloma functions as an immune microenvironment to facilitate interactions between T cells and macrophages and cytokines. However, the granuloma can also provide a home for M tuberculosis for an extended period, because some bacteria can avoid elimination within the granuloma. Although excessive replication of bacteria seems to result in loss of granuloma structure, extensive necrosis, and cavity formation, in most cases the granuloma contains the infection, but a limited number of bacilli survive, which is termed latent tuberculosis. A person who has a delayed-type hypersensitivity response to mycobacterial antigens (PPD) without signs of active tuberculosis is deemed to have latent infection and not to be contagious. Whether the host can ever completely eliminate latent infection is unclear. Some infected individuals do revert to a PPD-negative status; such PPD reversion can occur at a rate of about 5% per year.21 Studies of nursing-home residents also support the occurrence of tuberculin reversion.22,23 However, whether PPD reversion is linked with eradication of persisting tubercle bacilli remains unclear. The bacilli that survive within host tissues are a reservoir of infection that can reactivate to cause active disease many years later, as a result of immune-system compromise. The immune responses necessary for initial containment of infection, as well as long-term maintenance of a latent infection, are beginning to be elucidated.

Click to enlarge image
Figure 1. Immune mechanisms important in the maintenance of latent tuberculosis. The granuloma that forms in response to M tuberculosis consists of macrophages, which can differentiate into epithelioid macrophages or multinucleate giant cells, CD4 and CD8 T cells, and B cells. The T cells produce interferon γ, which activates macrophages. CD8 T cells can lyse infected macrophages or kill intracellular bacteria. Tumour necrosis factor (TNF) is produced by macrophages and T cells. Dendritic cells are also present within the granuloma. A mature granuloma is surrounded by fibroblasts. M tuberculosis is present within the macrophages and also extracellularly if necrosis is present. On depletion of CD4 T cells (eg, during HIV infection), the granuloma does not function as well, production of interferon γ may decrease, and macrophages are less activated. As a result, M tuberculosis begins to multiply and cause reactivation of infection. In the case of TNF neutralisation, the cells within the granuloma are no longer as tightly clustered, perhaps owing to chemokine or adhesion-molecule dysregulation. In addition, the macrophages are not as activated. These defects lead to a disorganised granuloma that is less able to control infection and greater immunopathology.
Back to top
Animal models for study of latent tuberculosis
Understanding of the various model systems available for study of acute and latent tuberculosis is useful. For studies of acute (primary) infection and vaccine testing, mice and guineapigs have been used extensively. Guineapigs are very susceptible to tuberculosis and can succumb to a very lowdose infection. This model has been used extensively to assess vaccine candidates. Mouse strains vary in susceptibility to M tuberculosis disease. An advantage of using mice for studies of infectious diseases is the availability of genetically engineered strains, inbred strains, and immunological reagents.
Establishment of an animal model of latent tuberculosis has been more difficult. First, the features of human latent tuberculosis that are to be modelled must be defined, which is challenging, owing to the limited understanding of this phase of M tuberculosis infection. In human beings, the immune response is crucial in establishing latent infection, since immunocompromised individuals are more susceptible to primary tuberculosis. In addition, there are very small numbers of bacteria in the lungs in latent tuberculosis. Finally, the latent state can be maintained for many years, and spontaneous reactivation is infrequent. However, reactivation can occur in response to declining immune function. The best animal model of latent tuberculosis would encompass all of these features. However, such a model has been difficult to develop, for reasons that are not completely clear. There are, however, a few models used to study latent and reactivation tuberculosis, and each one mimics at least some feature of human latent M tuberculosis infection.
For studies of latent and reactivation tuberculosis, two different mouse models have been used. In the first, mice of a relatively resistant strain (C57BL/6) are infected via aerosol or the intravenous route. The acute infection is controlled by the immune response by 3?4 weeks after infection. However, the immune response does not eliminate the infection but merely restrains it. The bacteria persist in the lungs, spleen, and liver. In the lungs, a high bacterial burden (105−106 colony-forming units) is stably maintained for many months. Although there is slowly progressive lung pathology, the mice do not show overt signs of disease such as weight loss or laboured breathing. There is evidence that the bacteria are in a metabolically quiescent state during this phase of the infection.24 However, this model seems to represent a chronic or persistent infection, rather than the latent infection observed in human beings. Nonetheless, the model does share some features with human latency, in that the chronic infection is controlled by the immune response, granulomas are present in the lungs, and the mice do not seem to be ill.
The second murine model of latency was developed by researchers at Cornell University, USA, in the 1950s.25,26 There have since been several variations on this model. Mice are infected with M tuberculosis and treated with antimycobacterial drugs to reduce the bacterial load to undetectable values. Subsequently, either the mice can spontaneously reactivate the infection, or reactivation can be induced by immunosuppression. However, this model can be difficult to use. Reactivation may be difficult.27 The drug treatment can sterilise the mice under some conditions. However, some researchers have had success with this model in identifying factors involved in maintaining a latent infection.27?29 The model is attractive in that the bacterial numbers are undetectable but can reactivate; the similarity to the human situation is appealing. However, unlike the human situation, antibiotics must be used to set up the latent infection, and the effect of this intervention on the immune response and control of infection is an additional variable.
Other animal models for the study of tuberculosis include rabbits and non-human primates.30 The rabbit model was historically used to study resistance and susceptibility to tuberculosis, and the granulomas that form in the rabbit lungs are remarkably similar to those in human lung.31,32 However, use of this model for study of latent disease has not been fully explored. In terms of immunology, the reagents for detailed studies in the rabbit model are not available. By contrast, the reagents available for immunological studies in non-human primates are excellent, owing to the extensive use of macaque monkeys in studies of simian immunodeficiency virus as a model for HIV. Macaques have been used historically in vaccine and drug-development studies for tuberculosis. Lately, a few groups have used this model to explore questions about pathogenesis and immunology.33?35 We have infected cynomolgus macaques with a low dose (about 25 colonyforming units) of M tuberculosis via bronchoscope.36 After infection, about half of the monkeys developed active chronic infection, characterised by infiltrates visible on chest radiographs, clinical signs, or weight loss. The remainder showed no signs of disease, although they did have delayedtype hypersensitivity responses, as well as positive lymphoproliferative and ELISpot responses to mycobacterial antigens. These monkeys appear to have latent infection. We have observed spontaneous reactivation in a small number of cases, after many months of infection. The macaque model is similar to human beings in terms of range of disease, immunology, and pathology. The granulomas in the lungs are remarkably similar to those in human beings. Thus, this model may be very useful for the study of acute and latent tuberculosis. Studies are under way to test various immunological mechanisms for reactivation of latent infection.
Back to top
T cells in latent tuberculosis
Interferon γ is a key cytokine in the immune response against M tuberculosis. This cytokine activates macrophages to kill the intracellular bacilli. Interferon γ probably has other roles in control of infection, since interferon-γ knockout mice are more susceptible to tuberculosis37,38 than strains deficient in macrophage effector mechanisms, such as inducible nitric oxide synthase (NOS2)39?41 or in T cells.42,43 Certain genetic mutations in human beings lead to deficiencies in interferon-γ signalling, and carriers of the mutations are more susceptible to mycobacterial disease,44 although reported tuberculosis cases in such individuals are rare.
The primary producers of interferon-γ are CD4 and CD8 T cells and natural killer cells. Many studies have implicated T cells in the control of M tuberculosis infection. In particular, CD4 T cells have a central role in protection against tuberculosis. In human beings, infection with HIV leads to loss of CD4 T cells and immunodeficiency. Tuberculosis is a major cause of death in AIDS patients worldwide. The heightened susceptibility of HIV-positive people to active tuberculosis, either from primary infection or by reactivation of latent infection, supports a role for CD4 T cells in defence against this infection.12 In experimental studies, mice deficient in CD4 T cells show impaired ability to control infection and die of tuberculosis.42,45 These mice show deficiency in early production of interferon-γ in the lungs and macrophage activation. Since other cells, including CD8 T cells, also produce interferon-γ in M tuberculosis infection,46,47 the findings in mice deficient in CD4 T cells suggest that early production of interferon-γ by CD4 T cells can determine the outcome of infection, and that CD4 T cells may have other functions, apart from production of interferon-γ, in the protective response against M tuberculosis.
Studies in a model of latent tuberculosis (the chronic infection model described above) provided further evidence of additional roles for CD4 T cells in control of infection. The infection was reactivated in chronically infected mice depleted of CD4 T cells, as shown by the loss of control of the infection, increased pathological features, and death.48 However, analysis of the lung tissue from the mice showed normal concentrations of interferon-γ and NOS2. CD8 T cells increased in number and in expression of interferon-γ to compensate for the lack of CD4 T cells. However, despite this strong response from CD8 T cells, the mice still succumbed to the infection. These findings strongly suggest additional roles for CD4 T cells in control of latent tuberculosis. By contrast, in the Cornell model of latent tuberculosis, depletion of CD4 T cells had little effect on the persistent infection.28 The reasons for these results in the Cornell model are unclear given the apparent requirement for CD4 T cells in control of latent infection in human beings. This discrepancy may be the result of difficulty in using the Cornell model for latent tuberculosis.27
There are several possible functions for CD4 T cells in control of latent tuberculosis.10 These cells could promote apoptosis of infected macrophages in the lungs through Fas/Fas ligand interaction. Apoptosis is thought to be detrimental to intracellular M tuberculosis,49?51 although no studies showing this in vivo have been published. There may be other cytokines produced by mycobacteria-specific CD4 T cells in the granuloma that are important in control of the infection, including interleukin 2 and tumour necrosis factor γ (TNFα).46 CD4 T cells can also induce production of other important cytokines by macrophages or dendritic cells, such as interleukins 12,52 10, and 15.53
CD4 T cells could also activate macrophages through direct contact independently of NOS2. These cells are one of the few types to express CD40 ligand (CD40L), which interacts with CD40 on antigen-presenting cells, such as macrophages and dendritic cells.54 In a murine model of leishmaniasis, CD40?CD40L interaction was essential for macrophage activation and parasite killing.55 However, macrophages deficient in CD40 stimulated with immune CD4 T cells can still kill intracellular M tuberculosis (unpublished, V Lazarevic, JLF), which suggests that this interaction is not required for adequate macrophage activation in tuberculosis.
In some systems, CD4 T cells are necessary for the induction or function of CD8 T cell responses.56?59 In the murine model, CD4 T cells were not required for priming of interferon-γ-producing CD8 T cells specific for M tuberculosis.42,60 However, CD8 T cells in the lungs of mice deficient in CD4 T cells showed impaired ability to kill macrophages infected with M tuberculosis.60 Thus, CD4 T cells were necessary for the cytotoxic but not cytokineproducing function of the CD8 T cells. The underlying mechanism remains unclear but may relate to a reduction in expression of interleukin 15 in the lungs,60 because this cytokine is important in function of CD8 T cells.61?63 This feature may point to an important secondary function of CD4 T cells in control of M tuberculosis infection.
Mycobacteria-specific CD8 T cells could be cytotoxic for infected macrophages, as well as producing cytokines. In human beings, CD8 T cells specific for M tuberculosis can directly kill intracellular bacilli via a granule-associated protein, granulysin.64 CD8 T cells have been reported to be involved in control of both acute and chronic infection,65?68 although in some studies the requirement for CD8 T cells was minor.69,70 The requirement for cytotoxic function by CD8 T cells has not been shown in the mouse model. Although cytotoxic CD8 T cells are present in the lungs of infected mice, mice lacking perforin did not show increased susceptibility to M tuberculosis.71,72 The true contribution of CD8 T cells to protection against tuberculosis cannot be tested in mice, because they do not have a granulysin homologue.
Evidence is emerging for a role of CD8 T cells in control of latent tuberculous infection. People with latent M tuberculosis infection have high frequencies of mycobacteria-specific CD8 T cells.73,74 In the Cornell mouse model of latent tuberculosis, depletion of CD8 T cells caused reactivation of the infection.28 In the chronic tuberculosis murine model, efficient depletion of the activated CD8 T cells from the lungs by use of antibody is difficult (unpublished, JLF), so the possibility of assessing contribution of these cells directly in chronic infection is limited. The cytotoxic potential of CD8 T cells is less in chronic than in acute infection, whereas potential to produce interferon-γ remains high (unpublished, V Lazarevic, JLF). The differential regulation of the function of CD8 T cells in the lungs in tuberculosis is currently under investigation.
Back to top
Modulation of antigen presentation: results of in-vitro studies
There is much evidence that M tuberculosis can interfere with the pathway for processing and presentation of MHC class II antigens. This pathway is crucial for the priming and effector function of CD4 T cells. Since CD4 T cells are an important component of the protective immune response against M tuberculosis, the ability of the tubercle bacillus to affect adversely the efficiency of MHC class II processing and presentation probably contributes to persistence.
In a study designed to characterise M tuberculosis mannose-capped lipoarabinomannan as a potential virulence factor, this complex glycolipid attenuated interferon-γ-induced expression of MHC molecules by a human macrophage-like cell line.75 The ability of M tuberculosis to modulate the expression of class II molecules was later confirmed.76?81 One study on human peripheral-blood-derived monocytes infected with the avirulent vaccine strain BCG did not show decreased expression of HLA class II.82 These discrepant results may be due to the differences in the experimental systems used.
Other mycobacterial components reported to suppress macrophage expression of MHC class II molecules include a 25 kDa glycolipoprotein77 and a 19 kDa lipoprotein.80 The precise mechanisms by which M tuberculosis attenuates expression of MHC class II molecules have not been clearly defined. However, possibilities include intracellular sequestration of immature class II heterodimers,78 the inhibition of interferon-inducible gene expression,75 and downregulation of the expression of class II transactivator.79 Studies involving M tuberculosis mannosecapped lipoarabinomannan,75,82 the 25 kDa glycolipoprotein,77 and the 19 kDa lipoprotein show that mycobacterial components actively contribute to downregulation of MHC class II expression by macrophages infected with M tuberculosis. The ability of mycobacteria to inhibit antigen processing or presentation has been shown in various in-vitro macrophage models.76,80,83 The 19 kDa protein can attenuate the processing of non-mycobacterial soluble antigens and antigen 85B of intact M tuberculosis.80 The inhibitory effects of the 19 kDa protein depend on TLR2. This receptor has an important role in initiating the innate immune response to the tubercle bacillus.84 The TLR2?dependent interference of the MHC class II antigen-presentation pathway occurs only at later times after treatment of macrophages with the 19 kDa lipoprotein.80 The hypothesis is that initial signalling of the 19 kDa lipoprotein through interaction with TLR2 triggers the innate immune response. As the infection progresses, this interaction mediates the inhibition of MHC class II antigen processing and presentation, thus enabling evasion from surveillance by CD4 T cells. Since TLR2 can sample intraphagosomal bacterial components,85 the 19 kDa lipoprotein, if expressed during the chronic phase of infection, could provide a sustained means of downregulating the host's antigen processing and presenting pathway, resulting in a safe haven for the mycobacteria within macrophages.
Phagosomes containing M tuberculosis are competent antigen-presenting organelles.81 Apparently, processing of mycobacterial antigens occurs in these phagosomes to form a complex of peptide and MHC class II antigens.81 Phagosomes containing live bacilli have smaller amounts of the complex than in vacuoles containing dead bacilli. Delay in the maturation of phagosomes containing live but not dead bacilli is also reported. These findings suggest that live mycobacteria can attenuate antigen processing and arrest phagosome maturation. Since these studies involved a time of less than 100 min from initiation of infection to examination of the phagosomes, M tuberculosis must interfere with processing of class II antigens and phagosomal maturation almost immediately after entering the host cells. These observations, together with those showing the delayed kinetics of the 19 kDa lipoprotein-mediated inhibitory effects,80 suggest that M tuberculosis can subvert the MHC class II antigen-presentation pathway in different phases of the tuberculous
Interference with macrophage effector functions Macrophages have a unique role in the host response to mycobacterial infection, since they represent both the primary effector cells for killing M tuberculosis and the primary habitat in which persisting bacilli reside. Here we discuss how M tuberculosis evades two major macrophage antimycobacterial mechanisms-the generation of nitric oxide and related reactive nitrogen intermediates (RNI), and the fusion of phagosomes and lysosomes (figure 2).

Click to enlarge image
Figure 2. Immune evasion strategies of M tuberculosis: interaction with macrophages. The maturation of phagosomes containing M tuberculosis apparently stops at a point close to the acquisition of the GTPase Rab5 (1). This arrest of phagosome biogenesis prevents fusion with lysosomal compartments (3) that have potent antimicrobial activities. Within the phagosome, M tuberculosis is subject to the antimycobacterial effect of reactive nitrogen intermediates (RNI) generated by the macrophage NOS2 (2). In addition, M tuberculosis can inhibit the MHC class II-dependent antigen presentation pathway (4). Thus, M tuberculosis can subvert various antimycobacterial functions of macrophages.
Macrophages activated by T-cell-derived cytokines generate products with antimycobacterial activity, among the best studied of which are the RNI. The antimycobacterial effects of these intermediates were first shown by in-vitro macrophage studies.86,87 Evidence from murine models involving NOS inhibitors or mice with disruption of the nos2 gene further supports a crucial role for RNI in host defence against M tuberculosis, in both acute and chronic persistent infections.39,40,88,89
However, the importance of nitric oxide and other RNI in human defence against M tuberculosis is a matter of substantial controversy,90 partly because the experimental manipulations that have defined the importance of these substances in the mouse model are not applicable to human tuberculosis. Human beings with tuberculosis are not intentionally given inhibitors of NOS, and no genetic disorder that renders individuals unable to produce NOS2 has so far been identified. Genetic defects in components of the cytokine cascade that lead to NOS2 induction, such as interleukin 12 and its receptor and the ligand-binding and signal-transducing portions of the interferon-γ receptor have been associated with increased susceptibility to mycobacterial infection;44 however, these deficiencies in cytokine expression are likely to have more far-ranging effects than merely inhibition of NOS2 production. Treatment with 1,25?dihydroxyvitamin D3 induced NOS2 expression and also limited growth of M tuberculosis in a human promyelocytic cell line,91 and certain polymorphisms in the vitamin-D-receptor gene were under-represented among patients with pulmonary tuberculosis compared with blood-donor controls.92 These findings suggest a link between vitamin D signalling, NOS2 production, and susceptibility to clinical tuberculosis, but again the vitamin-D-receptor mutations could have pleiotropic effects. So the relevance of RNI in controlling M tuberculosis infection in human beings must rely on indirect evidence.
Alveolar macrophages from healthy individuals produce RNI in response to infection with M tuberculosis, and production of the intermediates is associated with constrained intracellular growth of the bacteria.93 Human alveolar macrophages restrict growth of BCG through the induction of NOS2 activity, and this can be prevented by treatment with a NOS2 inhibitor.94 Raised NOS2 expression has been found in human alveolar macrophages from patients with untreated, culture-positive pulmonary tuberculosis, by use of a highly specific antibody to NOS2,95 whereas the amount of nitric oxide in exhaled air was higher in patients with tuberculosis than in controls, and the amount of exhaled nitric oxide decreased after antituberculous chemotherapy.96 Choi and colleagues97 showed by immunostaining that there is greater expression of NOS2 in the inflammatory zone of granulomas, and in areas of pneumonitis, in surgically resected lung tissue of eight patients with tuberculosis than in uninfected controls.
Back to top
Avoidance of the toxic effects of RNI
The production of RNI by NOS2 therefore seems to be essential for the containment of M tuberculosis infection. Yet in both mice and human beings, many immunocompetent individuals, though able to contain infection with M tuberculosis, are unable to eliminate the bacterium completely. This feature suggests that M tuberculosis expresses genes that counteract the bactericidal or bacteriostatic effects of RNI; studies by various experimental approaches now support this hypothesis.
A selection strategy was used to identify several candidate resistance genes of M tuberculosis. A DNA library derived from a clinical isolate of M tuberculosis highly resistant to RNI was introduced into a number of RNIsensitive bacterial hosts; the bacteria were then subjected to RNI stress. M tuberculosis genes that permitted survival of the bacterial host strains were identified. This strategy led to the identification of noxR1, conferring enhanced resistance to reactive oxygen intermediates (ROI) and RNI in Escherichia coli and Mycobacterium smegmatis,98 and noxR3, which is able to protect Salmonella typhimurium from ROI and RNI.99 At present, the mechanism by which these genes confer RNI resistance remains to be defined.
Another approach to identify RNI-resistance genes in M tuberculosis involves study of homologues in other bacteria that protect from oxidative stress, an approach made simpler with the publication of the complete genome sequence of M tuberculosis H37Rv, a common laboratory strain.100 This strategy was used to identify alkyl hydroperoxide reductase subunit C (AhpC) as a gene product affording protection against RNI.101 AhpC is a peroxiredoxin initially cloned from E coli and S typhimurium as an oxidative-stress response protein.102?105 Disruption of ahpC rendered S typhimurium highly susceptible to killing by RNI, and this defect was complemented with the ahpC gene of M tuberculosis.101 Mycobacterial ahpC was also able to protect human cells from killing by S-nitrosoglutathione (GSNO) and from apoptosis and cytotoxicity induced by NOS2.101
Subsequent biochemical studies of the wild-type and various site-directed mutants showed that AhpC can metabolise peroxynitrite anion (OONO−) into nitrite, thereby detoxifying this highly reactive species.106 Peroxynitrite is a powerful oxidant formed by the reaction between nitric oxide and superoxide anion (O2−), both of which are produced by activated macrophages, and can exert toxic effects through protein modifications.107 In-vitro studies have shown that although virulent M tuberculosis is very resistant to OONO−, the avirulent M smegmatis and BCG strains are susceptible.108 As for ahpC, a candidate gene approach has also been used to identify additional mycobacterial genes conferring nitric oxide resistance based on their homologues in other organisms. For instance, in studying the specific RNI species that cause antimicrobial effects, St John and colleagues109 elucidated an enzymatic basis for OONO resistance that is distinct from the detoxification of OONO− by AhpC. They focused on the peptide methionine sulphoxide reductase (msrA) gene, shown to have a role in resistance to oxidative stress in yeast and E coli,110,111 and found that E coli lacking MsrA was highly susceptible to killing by nitrite and GSNO (under aerobic conditions, when these species can be converted intracellularly to peroxynitrite), and that this defect was complemented by the msrA gene of M tuberculosis. The results suggest both that oxidation of methionine residues is a target for peroxynitrite, and that MsrA is crucial for repair of peroxynitrite-induced toxicity. The genome of M tuberculosis includes two genes, glbN and glbO, encoding truncated haemoglobin-like proteins of unknown function. A BCG strain in which glbN was disrupted showed reduced capacity to metabolise nitric oxide during stationary phase growth.112 In addition, the glbN mutant strain showed striking inhibition of aerobic respiration in the presence of nitric oxide compared with wild-type bacteria.112 Although M tuberculosis glbN-disrupted strains have yet to be investigated in vivo, these findings point to a role for mycobacterial truncated haemoglobin-like proteins in promoting bacterial resistance to RNI toxicity.
Given the important role of RNI in controlling mycobacterial infection, it is not surprising that tubercle bacilli have developed complementary strategies to combat nitrosative stress and thereby permit persistence within the macrophage, for example by detoxifying peroxynitrite (ahpC) and nitric oxide (glbN) and by repairing peroxynitriteinduced damage (msrA). Together, these mechanisms may contribute significantly to the establishment of persistent infection. Although understandable from the perspective of the bacterium, such redundancy can render the in-vivo demonstration of the significance of these genes challenging. For instance, an ahpC-disrupted strain of M tuberculosis showed increased susceptibility to peroxynitrite (though not to nitric oxide alone) in axenic culture as well as reduced survival in unactivated macrophages, with percentage survival an order of magnitude less than wild-type at 7 days after macrophage infection.113 By contrast, in macrophages stimulated with interferon-γ and lipopolysaccharide no survival difference between wild-type and ahpC-disrupted M tuberculosis strains was apparent, suggesting that activated macrophages have other antimycobacterial effectors, or that bacilli within activated macrophages express other mechanisms that compensate for or mask the absence of AhpC.113 A M tuberculosis ahpC mutant showed no growth defect in lungs and spleens during acute infection of BALB/c mice (up to 56 days after infection).114
Back to top
Remodelling the phagosome
Phagocytosis of pathogenic microorganisms by ?professional? phagocytes such as macrophages and neutrophils is the first step in their eventual degradation, as the phagosome eventually matures into a phagolysosome rich in acid hydrolases with degradative and microbicidal capacity. The maturation of a pathogen-containing phagosome is a highly complex and ordered process that involves the interaction of the phagosome with several endocytic vesicles through fission and fusion events, the molecular details of which are currently being unravelled.115?117
M tuberculosis has adopted a strategy to avoid destruction by lysosomal enzymes, that of interrupting the process of phagosome maturation. The studies of D'Arcy Hart and colleagues118,119 showed that phagosomes containing M tuberculosis do not fuse with lysosomes. This failure of phagolysosomal fusion occurred only with the ingestion of viable bacteria; dead bacteria were effectively delivered to lysosomes. Subsequently, vacuoles containing mycobacteria were shown to be less acidic than lysosomes,120 and a biochemical basis for the acidification defect was provided by studies showing that mycobacterium-containing phagosomes exclude the vesicular proton ATPases.121 Although phagosomes containing live mycobacteria do not fuse with lysosomes, they remain dynamic structures fully capable of fusion with other vesicles, such as early endosomes, and are accessible to internalised plasma membrane components such as the transferrin receptor and to the fluid-phase marker horseradish peroxidase.122?125
Recent advances in defining the biochemical and molecular basis for the trafficking of intracellular vesicles have opened new avenues of exploration into the mechanism of arrested maturation of the mycobacterial phagosome by identifying the basic tethering and fusion machinery involved in vesicular transport.117,126?128 Soluble N−ethylmaleimide?sensitive factor?attachment protein receptors (SNAREs), a family of membrane?anchored proteins, have a key role in mediating membrane fusion. In addition to SNAREs, the machinery required for specific membrane fusion is thought to include small GTPases of the Rab family
Via and co?workers129 showed that BCG?containing phagosomes fail to acquire the late endosomal GTPase Rab7 but retain the early endosomal GTPase Rab5, indicating that the phagosomes retain early endosomal characteristics and remain excluded from the late endosomal compartment. Clemens and colleagues130 reported that HeLa cells overexpressing Rab5c, when infected with M tuberculosis, form phagosomes that stain abundantly and persistently for Rab5. However, these investigators subsequently reported that phagosomes containing M tuberculosis also stained abundantly for the Rab7 GTPase, in a HeLa cell line stably transfected with the human RAB7 gene in a system permitting regulated expression.131 The reasons for the apparently discrepant Rab7 data from these two independent studies are unclear, but possible explanations are the dissimilar experimental systems used, including different mycobacterial species, host cells, and methods of analysis
Evidence has emerged lately supporting the contention that maturation of mycobacterial phagosomes is arrested at a point close to the acquisition of Rab5; EEA1 (early endosome autoantigen), a known Rab5?interacting partner and effector, is excluded from mycobacterium-containing phagosomes.131 The binding of EEA1 to early endosomal membranes depends on the modification of membrane lipids by a phosphatidylinositol 3(OH) kinase enzyme termed hVPS34.133 Both EEA1 and hVPS34 have a role in phagosome biogenesis.132M tuberculosis mannosecapped lipoarabinomannan inhibited activity of phosphatidylinositol 3(OH) kinase,132 and this could be a mechanism by which mycobacteria attenuate EEA1 recruitment. In addition, in comparison with phagosomes containing control beads, phagosomes containing latex beads coated with mannose-capped lipoarabinomannan recruited less syntaxin 6 (a SNARE apparently involved in trans-Golgi network to phagosome trafficking) and accumulated less of the lysosomal enzyme precursor immature cathepsin D. These findings point to a role for mannose-capped lipoarabinomannan in interfering with phagosomal acquisition of components from the trans-Golgi network, including lysosomal hydrolases.134 Therefore, blocking of recruitment of syntaxin 6 by mannose-capped lipoarabinomannan may account for the diminished accumulation of lysosomal markers in mycobacteriumcontaining phagosomes. As further evidence that mycobacterial lipids exert multiple and far-ranging effects in the infected macrophage, fluorescently labelled surface glycolipids of BCG are released from the mycobacterial cell wall and move throughout the endocytic network, ultimately entering many intracellular compartments.135
The phagosome biogenesis pathway, and its modification by M tuberculosis gene products, could differ depending on the activation state of the macrophage. Phagosomes containing M avium (MAC), like those containing M tuberculosis or BCG, also show limited fusion with lysosomes;136 however, when macrophages are costimulated with interferon-γ and lipopolysaccharide, the MAC-containing phagosomes can acidify, accumulate proton ATPases, acquire other characteristics of phagosome maturation, and ultimately exert microbicidal effects.137 This apparent ability of the host immune response to affect the fate of M tuberculosis is reminiscent of the observation that antibody-coated bacilli are defective in the prevention of phagolysosomal fusion and yet maintain the ability to survive in macrophages.138 The mechanism underlying this latter observation is not known at present. Clearly, a complex interplay between the host immune response and bacterial gene products determines the fate of M tuberculosis within the phagosome.
Another mechanism by which mycobacteria could interfere with phagolysosomal fusion is by retention on mycobacterial phagosomes of the host protein TACO (tryptophan aspartate rich coat protein).139 TACO, or mouse coronin 1, was originally described in the free-living soil amoeba Dictyostelium discoideum; coronin-deficient mutants show abnormalities in phagocytosis and cytokinesis.140 TACO remains associated with phagosomes containing viable, but not heat-killed, BCG, and transfection of a non-phagocytic melanoma cell line with cDNA encoding TACO blocked the transfer of mycobacteria to late endosomal/lysosomal organelles and resulted in slightly greater intracellular survival of mycobacteria.139 However, the importance of the association of TACO with mycobacterial phagosomes in preventing fusion with lysosomes remains controversial. Schuller and colleagues141 reported lately that coronin 1 was retained by phagosomes containing clumps of ten to 20 bacteria, but not by phagosomes containing individual bacteria, where coronin 1 was recruited by the early phagosomes but then lost within 24 h. This finding challenges the significance of TACO retention in the maturation arrest of mycobacterial phagosomes and raises the possibility that clumping of bacteria may represent a virulence factor for M tuberculosis.142
D discoideum has been used as a model system for studying mycobacterial?host interactions, with Mycobacterium marinum as the bacterial pathogen.143 In human beings M marinum causes cutaneous disease as a result of skin trauma in contaminated water (swimming pool granulomas, fish-handler's nodules), and in ectothermic animals such as fish and amphibians the organism can cause a disseminated systemic infection characterised by chronic granulomatous lesions and persistence of organisms within tissues.144 Solomon and coworkers143 found that, in contrast to expectations, M marinum showed improved intracellular growth in a coronin-deficient mutant of D discoideum compared with wild-type. They speculated that the uptake of mycobacteria may proceed along a unique pathway, distinct from the ?default pathway? of phagocytosis, in which D discoideum coronin mutants were previously shown to have defects,145 and that in the absence of competition from this alternative pathway the intracellular growth is improved. The investigators also considered that their findings may differ from previous work showing phagocytosis defects in coronin null mutants because they used amoebic monolayers for the M marinum infections rather than suspension cultures. In any case, the results clearly show that coronin is dispensable for M marinum uptake and growth within D discoideum; another possibility is that M tuberculosis (and its close counterpart M bovis BCG) have different requirements from M marinum for establishment of intracellular infection, or that the essential presence of a coronin isoform may differ for host cells as diverse as amoebic cells and mammalian macrophages.
Back to top
Cytokines and latent tuberculosis
Although a protective role for TNF against M tuberculosis is well established in mice,27,146?149 there was no evidence until recently that this cytokine is important for the control of tuberculosis in human beings. In fact, TNF was thought to contribute to the immunopathology of tuberculosis in both mice and human beings.150?153 Administration of thalidomide (which suppresses TNF production) to individuals with M tuberculosis infection was reported to result in improvement of overall clinical status, including weight gain.150 Indeed, the pathogenetic effects of TNF have been well documented in human diseases other than tuberculosis.154 For example, cachexia observed in cancer patients has been ascribed to the immunopathological effect of TNF.155 Clearly, TNF has both adverse and beneficial effects in the human immune response to pathogens, tumours, and tissue injury.
Attenuation of the biological activity of TNF has lately become an important therapeutic intervention in the management of a wide variety of chronic inflammatory diseases.156,157 TNF blockade effectively ameliorates inflammation in rheumatoid arthritis and Crohn's disease. These two disorders are among a rapidly expanding list of inflammatory diseases for which the efficacy of TNF blockade is being investigated.157 Currently two major anti-TNF protocols are being used for the treatment of various chronic inflammatory diseases.158,159 Infliximab is an anti-TNF neutralising antibody, and etanercept is a recombinant molecule made up of two identical chains of the human 75 kDa TNF receptor 2 fused to the Fc portion of human IgG1.
A major side-effect of anti-TNF therapy is the increased risk of developing tuberculosis.157?160 This is the most convincing evidence that TNF has a role in the control of human tuberculosis. Up to November 30, 2001, 110 cases of tuberculosis in patients receiving anti-TNF treatment had been reported to the US Food and Drug Administration.160 This increased risk has mandated screening for, and treatment of, individuals at risk of developing tuberculosis or with active infection before the initiation of anti-TNF therapy.161 The need for this prudent practice is emphasised by the non-typical presentation of tuberculosis associated with TNF blockade, which can delay diagnosis.162 A retrospective study of 70 reported cases of tuberculosis associated with anti-TNF therapy before disease manifestation showed that 56% of the study participants had extrapulmonary infection and 22% developed dissemination.162 This disease manifestation is highly nontypical, since extrapulmonary disease and disseminated infection occur in the general population at rates of 18% and 2%, respectively. Data gathered by review of the clinical history and the demographic characteristics of these patients, including age, the temporal relation between anti-TNF therapy and disease manifestation, prevalence of tuberculosis in the geographical location, and the history of exposure to M tuberculosis, suggested that most of the tuberculosis cases associated with TNF blockade represent reactivation of a chronic infection.162
The mechanisms underlying the TNF-blockade-induced reactivation of latent tuberculosis have not been clearly defined. Histological examination of tissues procured by open lung biopsy of an index case, in whom cultureconfirmed tuberculosis developed 7 weeks after a single dose of infliximab, revealed copious infiltration of inflammatory cells with no evidence of acid-fast bacilli in the lesion.162 In addition, pulmonary granulomas, a prominent feature in tuberculous tissues, were not seen in the lung specimens of this patient.162 Thus, TNF seems to have a role in maintaining the organisation of the granuloma, the integrity of which contributes to the containment of M tuberculosis.
Certain histological features observed in the pulmonary lesions of human beings with infliximab-related tuberculosis162 are very similar to the lung lesions of persistently infected mice undergoing TNF-blockadeinduced reactivation.147 In the chronic infection model, TNF neutralisation by monoclonal antibody resulted in reactivation disease characterised by an increase in tissue bacterial burden and mortality, and a greatly enhanced inflammatory reaction in the granulomatous lesions in the lungs. Augmentation of the inflammatory response was reflected by pronounced infiltration of immune cells associated with granuloma disorganisation. Significantly, the highest bacterial burden in the lungs of TNF-neutralised mice, attained 3 weeks after initiation of infection with virulent M tuberculosis, was 107 colony-forming units.147 This bacterial load remained constant until all mice neutralised for TNF died from the infection. Within the time frame of these studies, this bacterial burden is not generally fatal. The dissociation between bacterial load and mortality has been confirmed in another murine tuberculosis model-SPCTNFRIIFc transgenic mice with lung-specific expression of a soluble TNF inhibitor.163
Analysis of pulmonary gene expression in TNFneutralised mice of the low-dose model showed adequate amounts of interleukin 12 and interferon-γ, compared with controls. Expression of NOS2 was low but not absent. Together, these findings suggest that the severe inflammatory response associated with TNF blockade has a significant role in causing the increased mortality observed in these mice. A corollary of this notion is that in the chronic phase of tuberculous infection, TNF plays a part in regulating the inflammatory response and maintaining the integrity of the granuloma. The protective role of TNF in persistent tuberculosis is further underlined by studies in the Cornell model of latent tuberculosis.27 TNF neutralisation after establishment of apparent sterility resulted in disease reactivation. Thus, animal studies designed to assess the function of TNF in persistent infection have provided results that predicted the risks of reactivation tuberculosis during TNF blockade therapy.
Although TNF blockade is well established as an effective treatment for various chronic inflammatory diseases, the mechanisms by which it attenuates inflammation have not been defined. In patients with Crohn's disease, TNF blockade induced T-cell or macrophage apoptosis164,165 and led to downregulation of interferon-γ expression in human T cells.166 TNF has a role in apoptosis of macrophages infected with M tuberculosis also.50,167 TNF, with its ability to regulate the expression of chemokines, chemokine receptors, and adhesion molecules, is a potent modulator of cell migration.147,168 This attribute, together with the evidence that TNF-blockade-associated tuberculosis shows granuloma disorganisation, suggests that this cytokine has an important role in the maintenance of the granulomatous response in latent tuberculosis. A disorganised granuloma may be deficient in restricting inflammation to a localised environment. This deficiency could lead to extension of the inflammatory response into otherwise unaffected lung tissues, resulting in the aberrant pathology observed in tuberculous patients treated with TNF blockade.147,168 Characterisation of the mechanisms by which TNF neutralisation ameliorates chronic inflammation will probably shed light on how this means of immunomodulation leads to reactivation of latent tuberculosis, and how the host controls latent M tuberculosis infection.
Back to top
Interleukin 10?regulation or dysfunction?
Interleukin 10 is an immunoregulatory cytokine produced by T cells, macrophages, and dendritic cells. It can deactivate macrophages and can dampen the immune response to prevent or limit pathology from an over-exuberant inflammatory response to a pathogen. In transgenic mice expressing interleukin 10 compared with control mice, BCG was present in slightly higher numbers in spleen and liver;169 this feature suggests that increased interleukin 10 could be detrimental to bacterial control. However, interleukin-10−/− mice were similar to wild-type mice in the ability to control virulent M tuberculosis aerosol infection, indicating that absence of interleukin-10 production did not result in improved bacterial control, even after 8 months of infection.170,171 Interleukin 10 does affect the immune response in infected mice; interleukin-10−/− mice had somewhat higher production of interferon-γ in the lungs after M tuberculosis infection (unpublished data). However, in resistant mice such as the C57BL/6 strain, the immune response seems to be sufficient, and increasing interferon-γ production does not have a major effect on control of infection.
In studies of reactivation induced by TNF neutralisation in mice, interleukin-10 mRNA expression in the lungs increased,147 leading to the hypothesis that increased interleukin 10, in conjunction with decreased TNF, might be driving the reactivation. To test this possibility, we treated chronically infected interleukin-10−/− or wild type mice with anti-TNF monoclonal antibody and compared the reactivation rate (H Scott, JLF, JC, unpublished). Although all animals showed reactivation after TNF neutralisation, the interleukin-10−/− mice showed delayed reactivation, suggesting that the cytokine contributes to reactivation. Histology suggested that the lung macrophages were more activated in the interleukin-10−/− mice than in control mice.
Transgenic mice, in which interleukin 10 is expressed under the control of the interleukin 2 promoter and thus is increased when activated T cells are at the site of infection, were also more prone to reactivation than wild-type (C57BL/6) mice.172 In that study, interleukin 10 was increased in the lungs of the transgenic mice throughout infection, but bacterial numbers were higher in the transgenic mice only in the chronic phase (>70 days after infection). Interleukin 12 and M tuberculosis specific interferon-γ production in the lungs was decreased in the interleukin-10 transgenic mice compared with wild-type. In general, the phenotype of these mice was similar to a mouse strain CBA/J that is ?reactivation-prone?, or less able to control chronic infection than the resistant C57BL/6 strain.172,173 These findings suggest that increased interleukin− 10 production at the site of infection (ie, the lungs) increases susceptibility to reactivation. In the human setting, such an increase could conceivably result from a secondary infection or an immunopathological process in the lungs.
Interleukin 10 also seems to inhibit or modulate initiation of the immune response against M tuberculosis. Dendritic cells from interleukin-10−/− mice produced more interleukin 12 than wild-type dendritic cells on infection with BCG or CD40 stimulation in vitro.174 When infected dendritic cells were transferred to mice, interleukin-10−/− dendritic cells were more efficient than wild-type cells at priming an interferon-γ-producing T-cell response against mycobacterial antigens. Increased priming by interleukin-10−/− dendritic cells was associated with increased migration of these cells to the draining lymph nodes as well as increased interleukin-12 production by the resident dendritic cells within those lymph nodes. Thus, interleukin 10 can affect migration of dendritic cells, perhaps by modulation of the chemokine receptors on the dendritic cells,175 and interleukin-12 production at the priming or effector site. In the human system, production of interleukin 12 by dendritic cells in response to M tuberculosis infection is also regulated by interleukin 10, which could affect T-cell priming in vitro.176 In that study, interleukin-12 production by human macrophages was completely inhibited by endogenous interleukin-10 production, although this effect was overcome by the presence of interferon-γ.
Interleukin 10 is thought to be important in limiting immunopathology caused by an over-exuberant immune response. This cytokine may have a role in the chronic phase of infection, since downregulation of a type 1 immune response is likely to be beneficial to the host, at least in terms of lung pathology. Although interleukin 10 is produced at low levels in the lungs of C57BL/6 mice infected with M tuberculosis, in situations of substantial pathology or uncontrolled infection, interleukin-10 expression in the lungs is greatly increased (unpublished, V Lazarevic, JLF). The macrophages appear to be the primary source of interleukin? 10 expression under these circumstances. This may be an attempt by the host to dampen an immune response that increases substantially in response to high bacterial load.
Back to top
Conclusion
Control of the initial phase of M tuberculosis infection is likely to be important to the downstream events of the disease. The immunological mechanisms involved in maintaining a latent infection are complex, but are clearly necessary to prevent reactivation. Although the host response is essential to control of infection, the tubercle bacillus participates in the establishment of latency by using various strategies to evade elimination by the host. The bacteria can wait for the immune response to falter and then can emerge to cause active tuberculosis. The constant battle between the host and the microbe occurs at the level of the granuloma. The interplay of cytokines, including interferon-γ, interleukins 12 and 10, and TNF, is likely to be crucial to the ultimate fate of a person with M tuberculosis infection. In addition, the contribution of T cells and macrophages, as well as dendritic cells, to the successful immune response is undisputed. Tipping the balance of the response in one direction or the other, particularly in the context of the local immune environment of the granuloma, may determine whether an infection will remain latent or reactivate to cause active tuberculosis. Further studies, particularly at the level of the granuloma, will elucidate this complex interaction and perhaps provide surrogate markers for disease progression in human beings.
Back to top
Search strategy and selection criteria
Data for this review were identified though searches of Medline and PubMed, through references cited in relevant articles, and through searches of the authors' files. Search terms used were ?mycobacterium?, ?tuberculosis?, ?latency?, ?persistence?, ?animal models?, ?T cells?, ?macrophage?, ?nitric oxide?, ?phagosome?, ?antigen presentation?, ?tumour necrosis factor?, and ?IL-10?. Only papers published in English were reviewed.
Back to top
Conflicts of interest
None declared.
Acknowledgments
Our work is supported by grants AI49157, HL68526, HL71241, and AI50732 (JC, JLF), AI37859 and AI47485 (JLF), and AI49375 (JMT) from the National Institutes of Health, and grant CI-016?N from the American Lung Association (JLF). We thank the members of the Flynn and Chan laboratories for helpful discussions and unpublished data.
Comment