The generation of T‐cell memory to protect against tuberculosis

Tuberculosis (TB) kills more individuals each year than any other single pathogen and a more effective vaccine is critical for the global control of the disease. Although there has been recent progress in the clinical testing of candidates, no new vaccine has been licensed for use and correlates of protective immunity in humans have not been defined. Prior Mycobacterium tuberculosis infection does not appear to confer long‐term protective immunity in humans; thus mimicking the natural immune response to infection may not be a suitable approach to develop improved TB vaccines. Data from animal testing are used to progress vaccines through the “vaccine pipeline”, but studies in animals have not been able to predict efficacy in humans. Furthermore, although the generation of conventional CD4+ T‐cell responses are considered necessary to control infection with M. tuberculosis, these do not necessarily correlate with protection induced by candidate vaccines and other immune components may play a role, including donor unrestricted T cells, tissue‐resident memory T cells and anti‐M. tuberculosis antibodies. This review will summarize the current understanding of the protective immune responses following M. tuberculosis infection or vaccination, with a particular focus on vaccines that have recently entered clinical trials.


INTRODUCTION
Tuberculosis has been responsible for more deaths than any other infectious disease in history. 1 Despite the availability of a vaccine (Mycobacterium bovis Bacille Calmette-Gu erin or BCG) and effective antibiotics, globally there are an estimated 1.7 million deaths and 10.4 million new TB cases each year. 2 The limited efficacy of the BCG vaccine, emergence of multidrug-resistant strains of Mycobacterium tuberculosis and HIV coinfection all contribute to the inability of current programs to adequately control TB. The development of a more effective and easily administrable vaccine is necessary for optimal TB control and progress toward ending the global TB epidemic.
There are a number of reasons why an effective new TB vaccine has yet to be developed. Unlike common childhood infections, where pre-exposure vaccination is highly effective, the high proportion of the population latently infected with M. tuberculosis (estimated to be 1.7 billion individuals 3 ) suggests post-exposure vaccines would be necessary for optimal control; modeling of new vaccine efficacy supports this. 4 The choice of antigens and delivery systems is also critical for TB vaccine design. There is no consensus on the most immunodominant antigens of M. tuberculosis and antigen expression differs between active and latent/chronic infection. 5 Furthermore, vaccine delivery should be safe for use in immunocompromised individuals, a potential limitation of live vaccines, while adjuvants for use with subunit vaccines should not induce deleterious inflammation and ideally be capable of mucosal delivery, to target the lung as the primary site of infection. Finally, the immunological mechanisms responsible for protection against TB are not completely defined. While essentially all the current vaccines used in infant/childhood vaccination programs induce protective antibody responses, such responses are not considered essential for protection against M. tuberculosis and the stimulation of long-term memory CD4 + T-cell responses is the goal of anti-TB vaccination strategies. 6 However, the precise phenotype of protective CD4 + T cells is unknown, hindering attempts to rationally design vaccines targeting optimal protective immunity. In this review, we will dissect the adaptive memory response to M. tuberculosis infection and discuss recent developments in the progression and immunological characterization of TB vaccine candidates.

M. TUBERCULOSIS INFECTION AND THE GENERATION OF PROTECTIVE IMMUNITY
A large body of data now exists on T-cell behavior following pathogen exposure. Most of this information, such as the kinetics of T-cell expansion/contraction 7 and the development of T-cell memory 8 stems from the studies of CD8 + T-cell responses against acute viral infection or fast-growing bacteria such as Listeria monocytogenes. This is because of the ease of using such model organisms, as opposed to slow-growing pathogens such as M. tuberculosis, and the availability of tetramers and transgenic T-cell receptor mice to examine antigenspecific CD8 + T-cell immunity. However, a number of studies have identified the critical role for CD4 + T cells in immunity against M. tuberculosis, aided by the recent availability of reagents and models for precise definition of T-cell immunity to M. tuberculosis infection. Delayed priming of M. tuberculosis-specific T cells in the lungs of infected mice has been reported 9 and this presumably contributes to the chronic nature of M. tuberculosis infection. Antigen load is the key determinant in the priming of CD8 + T cells after encounter with mycobacteria 10 and more recent data demonstrate that the abundance and duration of antigen expression defines the function of CD4 + T cells in mice and humans. 11 This further highlights the complexity of TB vaccine design, as the selection of stage-specific antigens would be required to provide protection during active and/or chronic phases of infection.

DOES THE IMMUNE RESPONSE "REMEMBER" M. TUBERCULOSIS? IMPLICATIONS FOR TB VACCINE DEVELOPMENT
The rationale for the development of all vaccines currently in use is the "mimicking" of natural immunity induced by the pathogen. This is because of the fact that exposure to common childhood infections, such as measles and chickenpox, results in long-term protection against re-infection. However, recurrent TB is common; for example, studies in Uzbekistan revealed that a third of TB patients successfully treated were subsequently rediagnosed with recurrent TB after an approximate 2-year follow up. 12 In some studies, re-infection was identified as the major cause of recurrence, as opposed to reactivation of latent infection. 13 So how do these findings impact on vaccine development? These studies indicate that primary infection may negatively impact on protective immunity, possibly due to the chronic nature of M. tuberculosis infection. Evidence from the literature supports a model where high level, persisting antigen, such as that occurring during M. tuberculosis infection, may compromise the effective generation of CD4 + T-cell memory responses. For example, persistent viral infection in humans results in CD4 + T cells that display a phenotype and function resembling "effector" type cells, rather than memory cells that were generated in response to short-lived protein antigens. 14 These findings parallel those with CD8 + T cells, in which chronic viral infection appears to favor the generation of effector CD8 + T cells, which fail to acquire the key properties of memory cells. 15 Indeed, more recent analysis of antigen-specific responses revealed that chronic M. tuberculosis infection led to functional exhaustion of CD4 + T cells in mice and humans. 11 Conversely, limited antigen expression correlates with poor T-cell expansion 11,16 and poorly persistent live vaccines display limited protection against M. tuberculosis infection in mice. 17 Thus, a model could be put forward where the immunizing dose and persistence of protective antigens would dictate the quality of the resultant T-cell response ( Figure 1) and new vaccine design should be instructed by these parameters. This is further supported by the recent observation that low antigen dose provides optimal postexposure protection with the H56 fusion protein vaccine candidate in mice, with high antigen dose altering the functional avidity of vaccine-specific CD4 + T cells. 18 Dose escalation studies in humans of H56-containing vaccines demonstrated that low dose of vaccine antigen induced durable antigen-specific CD4 T cells, irrespective of M. tuberculosis infection status. 19

NEW TB VACCINE CANDIDATES AND T-CELL MEMORY
The major limitation in TB vaccine development is the lack of defined immune correlates of vaccine-induced protection in humans. The cytokine IFN-c is essential for protection against M. tuberculosis infection in mice and IFN-c deficiency in humans leads to disseminated mycobacterial infection. 20,21 However, numerous studies in mice and humans have revealed that IFN-c is not a reliable correlate of protection and indeed high levels of IFN-c may be a more reliable marker of immunopathology and bacterial load (reviewed in Reference 22). CD4 + T cells expressing multiple cytokines (e.g. IFN-c, TNF and/or IL-2), were found to strongly correlate with protection in a murine model of Leishmania major 23 and similar findings have been described in mice for M. tuberculosis infection 24,25 . However, despite CD4 + IFN-c + TNF + IL-2 + T cells being induced after boosting of BCG vaccines with the MVA85A vaccine in humans, no improved protection over BCG was observed. 26 Thus, other properties of CD4 + T-cell responses may be required for protective responses. Many studies have correlated with the level of antigen-specific central memory T cells (T CM ) and effector memory T cells with the ability to contain chronic infections, including TB. In humans infected with M. tuberculosis CD4 + T cells appear to be more of an effector memory T cells as opposed to T CM phenotype, suggesting chronic infection impairs the generation of protective T cells. 27,28 Lindenstrøm et al. demonstrated that BCG protection wanes in chronically infected mice in parallel with the loss of T CM of a defined phenotype (CD4 + IL-2 + KLRG1 À ) which could be counteracted by boosting BCG with the H1/IC31 vaccine. 25 Importantly, a similar T CM subset were identified in humans vaccinated with H1/IC31 and it would be of interest to examine if these cells correlate with protection in efficacy trials. 29 Finally, enhanced protection in mice conferred by vaccination with a recombinant BCG strain (VMP1002) correlated with high numbers of T CM and adoptive transfer of these cells confirmed their critical role in protection. 30 Thus, vaccines with the capacity to preferentially induce long-lasting T CM subsets may provide optimal protection against M. tuberculosis infection in humans.
IL-17-secreting CD4 + T cells (Th17 cells) play a critical role in immunity against a number of bacterial and fungal pathogens (reviewed in Reference 31). Th17 cells have been reported to either play a protective role 32,33 or confer pathological effects 34 during mycobacterial infection. In particular, it has been reported that Th17 cells are important in enhancing the recruitment of neutrophils and Th1 CD4 + T cells by the secretion of various chemokines, which correlates with the control of M. tuberculosis infection. 32 Similar to that observed with IFN-c, the production of IL-17 alone does not seem to be sufficient to mediate protection 32 but Th17 responses do correlate with improved protection by the VPM1002 vaccine in mice. 35 Thus, vaccine-induced production of IL-17 may contribute to protection by inducing the recruitment of neutrophils and circulating CD4 + T cells to the site of infection, as well as inducing early maturation of the granuloma. 36 A role for CD8 + T cells?
While CD8 + T cells are critical for the control of viral and some bacterial infections, their role in immunity  Reference 37). In mice, BCG is a poor stimulator of CD8 + T-cell responses compared with M. tuberculosis 10 and this is the rationale for the development of vaccines targeting CD8 + T-cell expansion. 38 However, viral and subunit vaccines that lead to strong CD8 + T-cell responses in mice did not improve protection against M. tuberculosis infection. 39,40 Both these studies, however, used the immunodominant M. tuberculosis antigen, TB10.4, and TB10.4-specific CD8 + T cells appear unable to recognize M. tuberculosisinfected macrophages, 41 which suggests that M. tuberculosis may subvert the CD8 + T-cell responses as a virulence strategy. Despite this, a number of TB vaccines in clinical trials are designed to elicit a strong CD8 + T-cell response ( Table 1). The VPM1002 vaccine induces a defined subset of CD8 + T cells expressing IL-17 in humans 42 ; however, the potential pathological role of these cells in inflammation indicates that their expansion may need to be tightly regulated. 43 A recombinant human cytomegalovirus vaccine that stimulates CD8 + T cells results in protective immunity against M. tuberculosis infection in nonhuman primates, and it would be of interest to observe if a similar finding is observed in human trials. 44

NEW CORRELATES OF VACCINE-INDUCED PROTECTION AGAINST TB
More recent analysis of pathogen immunity has identified T-cell populations other than T CM or effector memory T cells that may play a role in vaccine efficacy, in particular tissue-resident memory T cells. 45 Resident memory T cells reside in the mucosal tissues such as the lung and do not recirculate through the blood or the lymphatics. Lymphocytes resident in the lung are sufficient for BCGinduced protection of mice against M. tuberculosis infection, as blocking egress of cells from the secondary lymphoid organs did not alter the protective effect of the vaccine. 46 Mucosal delivery of BCG in multiple animal models has shown an improvement in protective efficacy compared with parenteral delivery 47,48 although intriguingly earlier studies showed no difference between vaccination routes. 49 The reason for this difference is unclear but may relate to use of antibiotics to clear bacterial load after immunization in the Palendira study, as it was subsequently identified that vaccine persistence and BCG antigen load impacts T-cell immunity. 10,17 Investigating other BCG delivery routes may be an important consideration, as intravenous (i.v.) delivery of BCG in nonhuman primates improves protection against aerosol M. tuberculosis challenge and i.v. administration of attenuated Plasmodium falciparum sporozoites promotes sustained protection against malaria challenge in humans. 50  Pulmonary delivery of fusion proteins with Bacillus subtilis spores 51 or recombinant Influenza A expressing M. tuberculosis antigens also induced lung-resident memory T cells, which could confer protection in the absence of circulating T CM. 52 More recently, a new subset termed stem cell memory T cells have been identified, that represent the earliest and longest lasting developmental stage of memory T cells and have a greater self-renewing capacity and proliferative potential compared with T CM and effector memory T cells . 53 In M. tuberculosis-infected individuals and BCG vaccines, antigen-specific CD4 + stem cell memory T cells are detected and display effector expression including Th1 cytokine release and expression of cytotoxic molecules. 54 Thus, a major challenge in TB vaccine development is to determine the relative importance of each of these memory T cells and define if effective vaccines should induce all or some of these subsets for maximal protection.
In addition to the "conventional" T cells described above, that recognize peptide antigens presented in the context of classical MHC class I or MHC class II molecules, unconventional T cells with invariant or semivariant TCRs may also play a role in anti-TB immunity. Donor unrestricted T cells are a diverse set of T-cell groups including Natural Killer T cells, CD1-restricted T cells and mucosa-associated invariant T cells, the latter recognizing the evolutionarily conserved major histocompatibility complex-like molecule MR1. The antigens recognized by these cells encompass a wide range of natural occurring molecules such as lipids and metabolites, many of which are derived from microbes including M. tuberculosis. Humans CD1-restricted T cells recognizing M. tuberculosis cell wall products have been identified, 55 while mucosaassociated invariant T cells have been shown to detect intracellular infection with M. tuberculosis. 56 Donor unrestricted T cells can produce effector cytokines, display cytolytic activity and are thought to represent an early line of defence against M. tuberculosis. 57 Therefore, targeting unconventional T-cell responses in humans may represent a feasible strategy for the early control of mycobacterial infection, particularly with the use of live vaccines; indeed, mucosa-associated invariant T cells recognize macrophages infected with BCG 58 and it would be of particular interest to determine whether other live vaccines in human trials (e.g. VPM1002 and MTBVAC, see Table 1) activate Donor unrestricted T cells.
A role for antibodies in TB vaccines?
The possible role of antibodies in the control of TB is covered elsewhere in this special feature. In terms of vaccine development, most focus has been on targeting T-cell responses and only a small number of vaccines have been developed to specifically target humoral responses, or examine antibody generation after vaccination. A conjugate vaccine candidate composed of Ag85B fused with Arabinomannan, a mycobacterial target of the humoral response, could improve survival of M. tuberculosis-infected mice and guinea pigs. 59 Vaccination of mice with the ID93/GLA-SE vaccine, currently in clinical trials, revealed vaccine-induced memory T-cell responses are impaired in B cell-deficient mice. 60 Antibody responses have been examined in humans after protein/adjuvant vaccines; however, a correlation with protective immunity is yet to be examined (Table 1). Post hoc analysis of the MV85A vaccine trial demonstrated that the production of IgG antibodies specific for the Ag85A was associated with reduced TB disease risk, 61 and as such the potential targeting of humoral immunity to improve anti-TB immunity warrants further exploration.

CONCLUSION
Despite many years of concerted research and development, a new TB vaccine to replace the existing BCG has yet to emerge. This is partly because of the fact that defined correlates of protection have not been identified. Identifying such correlates would allow the early demonstration of new vaccine efficacy, thus permitting prioritization of candidate TB vaccines for human efficacy testing and minimize the time and cost associated with late-stage efficacy trials. The importance of "conventional" T-cell subsets (CD4 + T cells and CD8 + T cells) in response to infection with M. tuberculosis is well established, and while particular cells subsets, such as multifunctional Th1 CD4 + T cells, appear essential for protection, their presence is not a reliable correlate of vaccine-mediated protection against TB disease. 62 While animal models have been instrumental in the selection of vaccine candidates for clinical progression, clinical trial endpoints will be required for defining correlates of protective immunity. The recent demonstration of the protective effects of BCG revaccination against M. tuberculosis infection in adolescents 63 and the M72/ ASO1 E protein/adjuvant against pulmonary TB in adults with latent TB infection 64 provides opportunities to define the parameters that correlate with protection against TB and instruct the development of improved vaccines. Such analysis would need to include the multitude of other immune parameters associated with M. tuberculosis infection, including tissue resident and/or stem cell memory T cells, Th17 cells, B cells or unconventional T cells. 57 The exact contribution of this spectrum of immune responses to protective immunity in humans and their role as correlates of protection are to be resolved, highlighting the complexity of vaccine development against pathogens highly evolved to combat immune detection and clearance.