Selecting antigenic targets for vaccines
For decades, it was believed that only living vaccines (like BCG) could generate the long-lived response necessary to combat M. tuberculosis infection and this had a major influence on the search for immunologically relevant TB antigens (138). However, in 1994, Andersen and colleagues, and subsequently other labs, reported the protective effect of vaccination with culture-filtrate proteins (CFPs) prepared from log-phase M. tuberculosis cultures in mice and guinea pigs, and demonstrated that the protection was transferable by CD4+ T cells (138). The demonstration that non-living vaccines based on secreted proteins could effectively protect against subsequent M. tuberculosis infection in animal models led to the initiation of extensive antigen discovery programs that aimed to identify crucial antigenic molecules. The initial antigens were isolated from filtrates of cultures of actively growing bacteria, which led to the hypothesis that proteins secreted by living bacilli in the phagosome might be the first antigens to be presented to the immune system in the early phase of infection, and consequently an immune response toward these proteins might be more effective at stimulating a protective immune response (138, 139). Antigens from culture filtrates such as ESAT-6, Ag85A/B and TB10.4 have demonstrated good protective efficacy when used as vaccines against an acute infection with M. tuberculosis, and these antigens are presently in clinical trials where the aim is to boost BCG-induced immunity (140–143). However, as noted above, the ability of M. tuberculosis to develop a latent infection allows it to outlast an immune response generated by vaccination early in life. Moreover, the vaccines in clinical development so far were all derived from actively replicating bacteria, and have all been assessed as prophylactic vaccines (140–143). The primary measure of their efficiency has been their ability to restrict early bacterial growth and dissemination. Preliminary studies suggest that they may have limited activity against dormant bacilli. This is not particularly surprising, as M. tuberculosis is able to establish latency and survive in an intracellular habitat for many years by making major changes in gene expression and, therefore, presumably in the antigenic repertoire presented to the immune system. More recent vaccine development strategies are therefore testing the assumption that this change in the antigenic repertoire should be reflected in the vaccines administered to individuals harboring a latent infection. The obvious conclusion is that such vaccines should contain antigens specifically expressed by the dormant bacteria, and this has spurred detailed studies of the gene expression pattern in these bacteria.
How does the dormant M. tuberculosis bacteria differ from the actively growing bacteria?
An effective vaccine against M. tuberculosis needs to consider the complexity of M. tuberculosis' lifestyle. Exposure to M. tuberculosis often results in lifelong infection due to the large range of evasion mechanisms deployed by the bacterium. The acute phase of M. tuberculosis infection is characterized by rapid bacterial growth and the development of an initial immune response dominated by recognition of secreted bacterial antigens (138, 139, 144, 145). Macrophages and lymphocytes migrate to the site of infection, resulting in the formation of granulomas in the lungs. In the majority of cases, the infection is brought under control by the immune system – even if the pathogen is not eliminated. However, the bacterium responds to the hostile environment of the host and enters a stage (often referred to as dormancy or latency) characterized by a drastically altered metabolism and a significant change in gene expression (146–149). It is unclear at present whether the bacteria in this stage are truly dormant: it is more likely that they persist through limited but continuous replication, or perhaps as a continuum of active and less-active forms (150). The outcome is a latent stage of infection without clinical symptoms that may last for many years or even decades. Latency is a dynamic process in which bacterial outgrowth is controlled by the immune response and, as described above, the bacteria attempt to subvert that immune response. This is a delicate balance that can change at any point (e.g., immunosuppression by HIV), leading to rapid bacterial replication and clinical reactivation of TB (3, 108, 151, 152). Considering the phenotypic change of the bacterium during the different stages of M. tuberculosis infection, it is most likely that a successful vaccine against TB may need to induce immune recognition of a broad spectrum of bacterial antigens.
Until recently, little was known about the conditions that induce dormancy and the bacterial response to those conditions. It has been known that control of bacterial replication in animal models requires the production of IFN-γ, TNF-α and nitric oxide (76, 87, 88, 103, 107, 108, 110, 151) and that exposure of the bacteria or bacterially infected cells to these agents in vitro or to conditions thought to reflect the conditions inside the granuloma such as limited access to iron, oxygen or nutrients leads to a drastic down-regulation of genes that are highly recognized by TB patients in the early phase of infection (146, 147). Mimicking these conditions and inducing bacterial dormancy in vitro has been the subject of intensive research in recent years. O2 depletion has been the most comprehensively studied and provides a link between the avascular environment of the encapsulated granuloma and the capacity of M. tuberculosis to adapt to hypoxic conditions. Wayne and colleagues demonstrated, in a series of important studies, that a gradual depletion of O2 changes bacterial respiration toward nitrate reduction and induces significant metabolic, chromosomal and structural changes in the bacteria consistent with dormancy (153–155). Recent work using whole genome microarrays has identified >200 genes whose expressions are rapidly altered by defined hypoxic conditions and has identified the dosR regulon that consists of 48 genes (156, 157). The dosR regulon is up-regulated by bacterial sensing of low, non-toxic concentrations of NO and appears to prepare M. tuberculosis for dormancy (158). Similarly, other conditions thought to reflect in vivo infection, such as growth in activated macrophages or within artificial granulomas, has been demonstrated to up-regulate the dosR genes, and an analogous switch in gene expression during chronic infection of mice has been seen (159). Hypoxia-driven dormancy seems to be reversible, as provision of O2, even after long periods of hypoxia-induced bacteriostasis, results in resuscitation and bacterial replication. Recent data suggest that synchronous resuscitation of the surviving dormant bacteria may be promoted by pheromone-like substances (the so-called resuscitation-promoting factors) secreted from slowly replicating bacteria and expressed in M. tuberculosis-infected patients (160, 161). Some of these substances may also promote bacterial spreading and transmission by dissolving the macrophage cell wall through lysozyme-like activity (162).
Nutrient starvation is another factor expected to be encountered by the bacteria in vivo and therefore has been used in vitro by Duncan and colleagues to induce a state of non-replicating persistence with decreased respiration. Proteome and microarray analysis demonstrated that a large number of transcriptional changes occurred, but interestingly, although some of the DosR genes were also up-regulated by starvation, the overall pattern differed significantly from that induced by hypoxia, which would suggest the involvement of a regulon different from DosR (147). Many of these changes appeared to involve lipid metabolism, consistent with earlier findings that long-term survival in the murine lung requires that M. tuberculosis express isocitrate lyase, an enzyme essential for the metabolism of fatty acids and for virulence in vivo (163). Importantly, this gene was necessary for replication of the bacteria in the late stage of infection in normal mice, whereas bacteria with a disruption of the gene still multiplied in IFN-γ knockout mice. This suggests that the metabolism of M. tuberculosis in vivo is profoundly influenced by the host response to infection. It is possible that activated macrophages are more easily able to deprive the bacteria of nutrients [perhaps by resisting changes to phagosome trafficking – (55, 65, 117)] and that the bacteria switch their metabolism to fatty acid degradation in response to this. This hypothesis is supported by the examination of the transcription profile of M. tuberculosis grown in activated murine macrophages or in the lungs of infected mice, which indicates that M. tuberculosis adapts to immune activation by expressing fatty acid-degrading enzymes and secreting siderophores to facilitate the acquisition of iron (157). This finding underscores the complexity of the bacterial transcriptional response to the multiple environmental signals encountered during its intracellular lifestyle and recent work (discussed in the last section of this chapter) is focusing on how to design vaccines that target the bacteria in its dormant phase.
While the antigens used in vaccines are crucial, it is important to stress that any vaccine against infection with M. tuberculosis should induce the correct response against the antigens used. This is particularly important, because, as discussed above, it appears that M. tuberculosis has developed the ability to divert immune responses away from those that confer optimal protection and to change its protein expression according to the immune pressure that it is under – including the expression of proteins to directly interfere with the host's immune response and so-called decoy proteins such as the 27 kDa antigen (69, 70).
New targets for vaccine development
Improved understanding of antigen expression patterns has led to a new phase in the intense research on subunit vaccines for TB. Subunit vaccines offer several significant advantages over BCG: first and foremost is the ability to produce a defined product, including antigens expressed by the bacteria in different phases of the infection (discussed in detail below), second is the ability to choose a delivery system that stimulates specifically the kind of immune response – a Th1 dominated response – needed and finally, because they need not be restricted in their growth (or are designed not to require growth in the host) by prior immunity to mycobacteria, their activity in individuals sensitized by environmental mycobacteria or BCG should not be impacted. In a highly cited study, six different atypical mycobacteria strains isolated from soil and sputum samples from Karonga district in Northern Malawi (a region in which BCG vaccination has no effect against pulmonary TB) were investigated in the mouse model. Two of these strains from the Mycobacterium avium complex were found to block BCG activity completely. Importantly, the efficacy of a subunit vaccine (in this case, the Ag85B-ESAT-6 fusion discussed below) was completely unaffected by prior sensitization (17). This makes subunit vaccines highly attractive for the boosting strategy. In addition, most subunit vaccines under development use either replication-deficient vectors, or are non-living, meaning that they pose no threat even in HIV-positive individuals. This makes them suitable for vaccination programs in TB-endemic regions, where the TB and HIV epidemics are ever more closely intertwined.
The vaccines being developed fall into two categories. The first is vaccines aimed at replacing BCG, conferring longer and/or more effective protection. At present, it is unlikely that a subunit vaccine can replace BCG in the near future, due to the latter's low cost, safety record and extensive use worldwide, and this ‘BCG replacement’ vaccine strategy is therefore mostly focused on recombinant BCG or attenuated M. tuberculosis vaccines.
The second strategy involves vaccines designed to be administered to already BCG-vaccinated individuals to further boost (and hopefully prolong) the BCG-induced immunity. Compared with recombinant mycobacterial vaccines, where it is unclear whether such an attenuated vaccine is virulent enough to overcome the existing anti-mycobacterial immunity due to earlier exposure to environmental mycobacteria or a prior BCG vaccination, subunit vaccines do not appear to be affected by – and may even benefit from – existing anti-mycobaterial immunity. Therefore, the obvious choice is to use the mycobacterial vaccines for priming, and subunit vaccines as boosters, allowing designers of boosting vaccines to take advantage of the prevalence of BCG vaccination and the likelihood that this will persist at least for the foreseeable future. However, because a vaccine administered as a booster to adolescents or older children may also be given to individuals who did not receive the BCG vaccine, or who received an ineffective BCG vaccination (incorrectly administered, or with a vaccine that was too old or incorrectly stored), a booster vaccine should also be able to prime an effective immune response. As a result, all of the vaccines currently in clinical trials were initially screened in animal models for the ability to prime a protective immune response at least as efficacious as BCG (141, 143). Because booster vaccines by definition will be administered later in life, the assumption that two billion people are latently infected with M. tuberculosis means that any booster vaccine will also of necessity be administered to large numbers of latently infected individuals. This raises the question of safety and any such vaccine will need to be rigorously screened for safety in M. tuberculosis-infected individuals. However, it also raises the following question – can we design a vaccine that can help people who are already infected, either because they did not receive a primary vaccination or because it did not prevent a latent infection (not an unlikely scenario in the case of BCG-vaccinated individuals)? Mathematical modeling suggests that a post-exposure vaccine effective at preventing disease in latently infected individuals would cause a significant decrease in the number of new cases in the short term, but that over time, a combination of pre- and post-exposure vaccine would have a larger effect (164). The ideal approach would therefore be a single vaccine that is effective against both acute and latent infection, i.e. a vaccine that can counteract M. tuberculosis in different stages of the infection. However, no such ‘multistage’ vaccine currently exists (165, 166).