Immune cell interactions in tuberculosis

SUMMARY

Despite having been identified as the organism that causes tuberculosis in 1882, Mycobacterium tuberculosis has managed to still evade our understanding of the protective immune response against it, defying the development of an effective vaccine. Technology and novel experimental models have revealed much new knowledge, particularly with respect to the heterogeneity of the bacillus and the host response. This review focuses on certain immunological elements that have recently yielded exciting data and highlights the importance of taking a holistic approach to understanding the interaction of M. tuberculosis with the many host cells that contribute to the development of protective immunity.

INTRODUCTION

Tuberculosis has plagued humans for centuries. The bacillus Mycobacterium tuberculosis (Mtb) is transmitted primarily by aerosols expelled by a person with active tuberculosis. Tuberculosis is primarily considered a pulmonary disease, but the bacillus can infect nearly all organs in the body. Although there are still 10 million new cases of active tuberculosis every year worldwide and an estimated 1.3 million deaths (WHO, 2022), the majority of humans who are infected with Mtb control the infection without intervention. This state of clinically quiescent infection is known as latent tuberculosis (LTBI) and a small percentage of those with LTBI will progress to active tuberculosis years to decades after primary infection. It is now recognized that Mtb infection outcomes encompass a full spectrum from subclinical infection to mild, moderate, or severe active disease. Although controversial, it is possible and even likely that some of those classified with LTBI have cleared the infection (Behr et al., 2019), while others are at extremely low risk of reactivation. It is difficult to classify clinically the state of Mtb infection in humans. This is in part due to the diagnostics used to define Mtb infection or disease, which are based on reported symptoms, the host T cell response to mycobacterial antigens, chest X-rays or CT scans, and/or microscopy or culture of sputum or other patient samples. Thus, the various outcomes of Mtb infection are complex and likely due to a combination of several factors, including host immune responses, bacterial strain, and extent of exposure. Heterogeneity in outcome is prominent not only among infected subjects but also within a single infected person, where disease can regress in one area of the lung and progress simultaneously in another (Lin et al., 2014) (Figure 1).

Figure 1. The heterogeneous M. tuberculosis-host interaction in granulomas determine infection and disease outcomes.

Cellular immunological and scRNA-seq analyses of granulomas, together with imaging and the use of barcoded bacilli, have revealed a high degree of heterogeneity of the granulomatous response among infected non-human primates and humans. In Mtb infection of naive hosts (left panel), the spectrum of heterogeneity spans resolution in restrictive granulomas, i.e., those that kill or limit replication of Mtb, to progression in permissive granulomas with increased bacterial growth and limited killing and dissemination to form new granulomas, representing the extremes on a continuum of disease states. Granulomas harbor a wide variety of immune cells and immunological factors (left and right panels), whose interactions likely govern whether the lesions progress or regress. The significance of the ability of specific cellular and immunological parameters of a granuloma in determining disease outcomes is seen in subjects of immunization-challenge studies (right panel), which can reflect vaccine efficacies. The granulomas developed upon challenge with virulent Mtb after immunization with highly protective vaccines such as i.v. BCG, which affords near sterilizing protection, are absent or smaller relative to unimmunized subjects and contain fewer or no bacilli (right panel). In addition, the lung parenchyma of non-human primates subjected to the i.v. BCG immunization-challenge model harbors CD4 T cells and CD8 T cells, and the airway is populated with both T cells and antibodies, suggesting an important role for these immunological elements in mediating vaccine-engendered protection against Mtb (right panel).

There have been decades of research into the immune responses that dictate protection against tuberculosis, yet we still do not have a firm grasp on the constellation of immune responses necessary to prevent infection or progression to disease. The immune response to Mtb infection is complex, involving many cell types and functions. The field has looked for the “magic bullet” to target for an effective vaccine, without much success. Often tuberculosis research focuses on the role that one cell type, one cytokine, or one pathway plays in control or exacerbation of infection. Instead, we posit that the various aspects of the immune response interact with and influence each other within several different anatomical compartments to lead to the ultimate outcome of the infection—clearance, containment, or progressive disease. That said, there are likely several “combinations” of factors that can result in a successful outcome, just as there are likely other combinations that result in progression to disease (Cadena et al., 2017). A holistic approach to the immunology of Mtb infection and disease is needed to identify mechanisms of protection. This is challenging, as it requires integration of multiple data types across different model systems and in humans. Because the primary battlefield between the host and the bacillus occurs in the tissues (airways, lung, granulomas, and lymph nodes), it can be difficult to extrapolate from samples easily obtained from humans, such as blood and sometimes bronchoalveolar lavage (BAL) fluid. Nonetheless, this is the challenge for modern research in the tuberculosis field—devising methods to interrogate what control of infection represents without a focus on one cell type or function, but instead to identify the collection of responses that can lead to effective elimination of the bacillus either at the time of infection or in persons who are already infected.

In this review, we discuss the various immunologic players in Mtb infection and disease but highlight recent studies that support the notion that the sum of immune responses is more important than any one particular cell type or function. It is also relevant that immune responses in the tissues can evolve over time, highlighting the plasticity of the immune system.

THE GRANULOMA

The granuloma is the pathologic hallmark of Mtb infection. Mtb is transmitted via aerosols from a person with active tuberculosis. Infection most often occurs upon inhalation of individual bacilli into the airways, where the bacillus encounters alveolar macrophages. These macrophages can engulf the bacilli and, in some cases, prevent productive infection. However, when the bacilli (likely in alveolar macrophages or maybe other phagocytes) transit to the lung parenchyma, the host mounts an immune response. Monocytes, macrophages, neutrophils, and dendritic cells are attracted to the site of infection, likely due to chemokine and cytokine signals from the infected cells, initiating granuloma formation. The bacilli are carried to the lung draining lymph nodes where priming of the adaptive immune response is initiated. Murine studies indicate this occurs within 7–14 days of infection (Reiley et al., 2008), although in humans and non-human primates adaptive T cell responses in the periphery are usually not observed until 4–6 weeks post-infection (Capuano et al., 2003; Poulsen, 1950). Primed T cells (and likely B cells) then migrate to the site of infection, resulting in the formation of an organized granuloma, with lymphocytes surrounding a macrophage layer, often with neutrophils surrounding the necrotic center (Figure 1). In non-human primates, the fully formed granuloma can be observed by 3–4 weeks post-infection (Lin et al., 2006). Within granulomas, Mtb can be intracellular, primarily within macrophages or extracellular, primarily within the caseous necrotic center. The immune responses necessary for controlling infection may differ depending on the location of the bacteria within granulomas.

Using DNA barcoded strains of Mtb in macaques, we demonstrated that each granuloma is initiated by a single bacillus; over the next 4–6 weeks that bacillus multiplies to reach ~10⁵ CFU/granuloma, which appears to be the “carrying capacity” of any single granuloma (Lin et al., 2014). At that point, replication can be restrained by the granuloma or the Mtb can escape (disseminate) to form new granulomas in the lung. In most cases, bacterial killing is not robust until ~10 weeks of infection, likely due to a delay in a functioning adaptive immune response in granulomas (Grant et al., 2022; Lin et al., 2014).

Granulomas can form in other tissues as well, most commonly in thoracic lymph nodes. Dissemination of initial infection to the thoracic lymph nodes is common in Mtb infection. Because thoracic lymph nodes are important lymphoid structures, the migration of Mtb to the thoracic lymph nodes is likely critical for the initiation of adaptive immunity. However, lymph nodes can be a prominent site of bacterial persistence and potential source of reactivation (Ganchua et al., 2018, 2020). Humans with Mtb infection often have involved thoracic lymph nodes as seen by imaging and, particularly in children, an infected lymph node can be quite enlarged and cause lung lobe collapse (Cong et al., 2022; Navani et al., 2011). We showed previously that macaque thoracic lymph nodes are relatively poor at killing Mtb, and the data suggest that lymph nodes may be a source of bacterial dissemination (Ganchua et al., 2018).

Granuloma structure and function play a critical role in containment of the infection, yet many of the mechanisms by which this occurs and what leads to failure are still unclear. Holistic studies of granulomas and other pathologies in tuberculosis are required to dissect the complexity of the immune factor interplay in granulomas.

MACROPHAGES

Although the macrophage was discovered a century and a half ago, the biology of this highly complex phagocyte, which plays an important role in defense against microbes and tissue homeostasis, remains incompletely defined (Ginhoux and Guilliams, 2016; Guilliams et al., 2020). Nevertheless, with the help of novel technologies, we now know that there are distinct subsets of tissue-resident macrophages that vary in their ontogeny (bone marrow versus embryonic in origin), homeostasis maintenance, and functions, thus revising the concept of the mononuclear phagocyte system and the M/M2 paradigm of functional polarization (Martinez and Gordon, 2014; Murray, 2017). Importantly, cellular source of macrophage subsets and their tissue-specific niche bear functional consequences (reviewed in Ginhoux and Guilliams, 2016; Guilliams et al., 2020). It is apparent that organ-specific niche can modulate the development and maintenance of tissue-resident macrophages, imprinting upon them distinct attributes and functions, including metabolic activities. Elucidation of the mechanisms by which these processes are regulated will likely provide information that can lead to the design of macrophage-targeting intervention in treating diseases.

As the primary host cell for Mtb, macrophages have been widely used to study the Mtb-host interaction. Based on our current understanding of tissue-resident macrophage biology (Ginhoux and Guilliams, 2016; Guilliams et al., 2020), the variability of results generated by in vitro studies, which use cell lines and primary macrophages derived from different tissue compartments, is not unexpected. Relevant to Mtb, a pulmonary pathogen, lung-resident macrophages are made up of distinct subsets—broadly designated as alveolar macrophages (AM) and interstitium macrophages (IM) (reviewed in Ginhoux and Guilliams, 2016; Guilliams et al., 2020), which are located in the alveolus and the interstitium of the lungs, respectively (Figure 2). In mice, AM are of embryonic origin and thus bone marrow independent, and they are self-replenishing (Figure 2). Under certain experimental conditions, however, including irradiation and infections that lead to depletion of AM, monocytes can participate in replenishing this population (Ginhoux and Guilliams, 2016; Guilliams et al., 2020). Emerging evidence supports the notion that subsets of AM exist (Pisu et al., 2021). Similarly, IM are heterogeneous and of both embryonic and monocytic origin (Gibbings et al., 2017; Sabatel et al., 2017; Tan and Krasnow, 2016; Ural et al., 2020) with subsets that are functionally different from one another and from AM (Gibbings et al., 2017; Schyns et al., 2019; Ural et al., 2020). How do these spatially and functionally distinct macrophage subsets modulate infection and disease outcome in tuberculosis?

Figure 2. The interaction between lung macrophages and M. tuberculosis shapes the host response to the tubercle bacillus and influences the development of drug tolerance.

In a mammalian host, alveolar macrophages (AM) of embryonic origin populate the air space prior to birth. (1) The AM population is self-renewing and represents the host cell that Mtb apparently targets primarily upon entering into the lung alveolar space. (2) Mtb-infected AM, through IL-1 receptor (IL-1R) signaling-dependent mechanisms (2), translocate from the alveolar space into the lung interstitium (3), serving as the vehicle that carries bacilli into the parenchyma. The translocation of infected AM is dependent on IL-1R, MyD88, and ASC in a macrophage-extrinsic fashion through non-hematopoietic cells. Translocated Mtb-infected AM multiplies in the interstitium, forming cellular aggregates (4), followed by recruitment from the vasculature (5) of neutrophils and monocyte-derived macrophages (6), which are then infected. The parenchyma of the lungs contains interstitial macrophages (IM), which are of bone marrow origin (7) and can be infected with Mtb (8). IM are a highly heterogeneous population, consisting of multiple subsets that interact with Mtb differentially, raising the possibility that distinct macrophage sub-populations may contribute to granuloma heterogeneity (Figure 1). One functional characteristic that reflects subset heterogeneity of AM and IM macrophage populations is the levels of stress these phagocytes imposed on Mtb, as assessed by fitness reporter mycobacterial strains (9). Mouse studies have shown that the AM and IM lineages display distinct metabolism (possibly pre-programed as a result of their different origins [1 and 7] and niche [2 and 8)] that differentially regulates replication of intracellular bacilli.) These cellular parameters translate to AM being more permissive for Mtb growth compared with IM. The levels of stress encountered by Mtb intracellularly can promote drug tolerance (10), an intermediate state conducive to the development of drug resistance. Permissive macrophages are depicted by higher number of paler bacilli in the phagosomes, while restrictive macrophages are those with lower number and darker bacilli in the vacuoles. The darker color of the bacilli denotes higher levels of macrophage-imposed intracellular stress.

Using Mtb fluorescent reporter strains that enable the assessment of mycobacterial fitness via determination of the replication status of intracellular bacilli and of the levels of stress they are subjected to, in conjunction with in vivo modeling and transcriptome analysis, Huang et al. (2018) reported that interaction of Mtb with AM and IM in mice leads to remarkably divergent infection outcomes. Compared with IM, AM exert less stress on intracellular Mtb and are relatively permissive for bacterial replication. In line with this observation, depletion of AM leads to reduced lung bacterial burden, while the opposite is observed upon ablation of blood monocytes and IM (Huang et al., 2018). Transcriptional data showed that metabolically, the permissive AM, which harbor relatively higher number of Mtb, exhibit a preference for fatty acid oxidation, while the relatively restrictive IM that carry a lower bacterial burden displays a propensity for the glycolytic pathway. This metabolism/Mtb control observation is congruent with the known preferred lipid diet of intracellular bacilli (Laval et al., 2021) and is in line with the results of studies involving inhibition of specific metabolic pathways in macrophages and in mice. The significance of these observations is buttressed by the results of a dual RNA sequencing (RNA-seq) study that enabled simultaneous characterization of the intracellular bacterial and host cell transcriptomes using sorted AM and IM from infected mice (Pisu et al., 2020). This latter study extends the previously observed differential metabolisms displayed by AM and IM (Huang et al., 2018) to include a range of distinctive transcriptome signatures, including varied levels of stress signals imposed by reactive nitrogen intermediates and restriction of iron availability (Pisu et al., 2020). The dual RNA-seq platform constitutes a versatile approach that should support detailed analysis of the host and bacillary in vivo responses during Mtb infection. Indeed, using the dual RNA-seq platform, together with the fitness reporter Mtb strains and detection of specific cell surface marker expression to study in vivo host-cell-bacterial interaction at the single-cell level, a recent study revealed that infected AM and IM in the lungs of mice3 weeks after an intranasal infection with a relatively high inoculum are composed of heterogeneous subpopulations(four each for AM and IM), with AM and IM containing both mycobacterial growth-restrictive and -permissive subsets; this functional diversity is further supported by their transcriptome profiles (Pisu et al., 2021).While the precise characteristics of these pulmonary macrophage subsets remain to be determined, the fact that individual subsets harbor bacilli with varied fitness phenotypes suggests that interaction of Mtb with distinct subpopulations may differentially modulate infection and disease outcome. The alveolar macrophage subsets identified in this mouse study can be detected in BAL fluid obtained from healthy uninfected humans, suggesting their relevance in human tuberculosis (Pisu et al., 2021). It is noteworthy that transposase-accessible chromatin sequencing (ATAC-seq) data suggest that preexisting epigenetic imprinting contributes to how each lung macrophage subset responds to Mtb infection (Pisu et al., 2021). Importantly, a recent study involving single-cell RNA-seq (scRNA-seq) analysis has revealed diverse macrophages subsets in the lungs of Mtb-infected macaques (Gideon et al., 2022). The results of these highly informative studies underscore the complexity of the lung macrophage landscape, characterized by distinct subsets that display differential interaction with Mtb (Huang et al., 2018; Pisu et al., 2021), raising the possibility that (1) the resulting differential interactions between subsets of these phagocytes with Mtb contribute, at least in part, to the heterogeneity in disease outcome in tuberculosis and the variability of granulomas observed in the lungs of a single infected individual (Cadena et al., 2017) and (2) targeting specific macrophage subsets and their local niche may enable manipulation of infection and disease outcomes and should pave the way for future experiments designed to unravel the relative contributions of the various lung-resident macrophage subpopulations in regulating the host response to the pathogen (Huang et al., 2018; Pisu et al., 2021).

TRAFFICKING MTB—THE ALVEOLAR SACINTERSTITIUM RELOCATION

The early sequence of events that occur after Mtb enters the alveolar space has been poorly understood until recently (Figure 2). Using a fluorescent Mtb strain to track lung host cell dynamics in acute Mtb infection in mice, a recent report revealed that AM constitute the primary cell type that Mtb parasitizes as early as 48 h post-inoculation (Cohen et al., 2018). Infected AM gain access to the lung interstitium during the innate phase of infection, where these phagocytes proliferate to form aggregates, followed by dissemination of bacilli to varied lung myeloid cell populations, mostly involving recruited neutrophils and monocyte-derived macrophages (Cohen et al., 2018). Pursuing the observation that infected AM upregulate the expression of pro-IL-1β, the effect of elements involved in IL-1 production and the IL-1R signaling pathway (Cohen et al., 2018) on AM translocation was examined, targeting IL-1R, MyD88, and ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain), factors that have been reported to play significant roles in mediating innate immunity in tuberculosis. Using specific gene-disrupted mice and mixed bone marrow chimeras, the studies revealed that AM translocation is dependent on IL-1R, MyD88, and ASC in a macrophage-extrinsic fashion through non-hematopoietic cells (Cohen et al., 2018). Further, in agreement with the observation that IL-1 expression in tuberculosis is dependent on the mycobacterial ESX-1 secretion system, AM translocation was markedly attenuated in cells infected with an RD1 deletion Mtb mutant (Cohen et al., 2018 and references therein). Together, the results suggest that IL-1β mediates a crosstalk between infected AM and pulmonic non-hematopoietic cells (possibly lung epithelial cells) to enable the entry of infected macrophages into the interstitium from the airway. It will be of interest to determine the cellular source of IL-1 in this early phase of infection and to identify the non-hematopoietic cells that participate in the ESX-1-dependent relocation process. Importantly, the IL-1R signaling pathway is relevant clinically. A recent report noted that varied transmission capacity of clinical Mtb isolates is related to their discrepant proficiency to promote the migration of infected AM from the air space into the interstitium in an IL1-R signaling-dependent manner (Lovey et al., 2022). The high transmission (HT) Mtb strain, relative to the low transmission (LT) counterpart, triggers a more rapid relocation of infected AM into the interstitium that is associated with earlier dissemination of bacilli to the draining lymph nodes, higher levels of Th1 priming, and enhanced ability to bacterial control. Mice with chronic HT infection develop necrotic lung lesions that in humans, may develop into cavities to promote transmission (Hunter, 2011). The highly complex and heterogeneous granulomas modulate both mycobacterial containment and immune-mediated inflammatory damage in tuberculosis (Bhattacharya et al., 2021; Cadena et al., 2017; Krug et al., 2021; McCaffrey et al., 2022; Roca and Ramakrishnan, 2013; Tobin et al., 2012; Wells et al., 2021). Understanding the mechanisms underlying the regulation of the exacerbated granulomatous response in the HT strain could illuminate how human tissue-damaging immunopathology develops (Hunter, 2011). Worthy of note, in the above-described macrophage studies, the results discussed have been generated from models involving Mtb infection of naive mice. Given that a significant proportion of humans have been vaccinated with BCG and/or exposed to Mtb, it would be of interest and important to study the early events of Mtb-macrophage interaction using animals that are not mycobacteria naive.

THE T CELL RESPONSE

It must be recognized that currently, there are no true correlates of protection against tuberculosis, and although many immune cell types have been implicated as being important in prevention of tuberculosis disease in animal models, the mechanisms by which the immune response can control or eliminate infection remain poorly understood. T cells have long been recognized as necessary for control of Mtb infection and progression to disease. However, as discussed below, recent data support that there is a wide range of T cell types and functions that can contribute to protection or even exacerbate disease. Critical to effective control of Mtb infection and disease is the balance of T cell responses—neither too little nor too much—so that the infection is controlled but inflammation is restrained (Barber et al., 2019; Gideon et al., 2015; Kauffman et al., 2021). Inflammation is a major factor in tuberculosis disease, so a delicate balance of immune responses in granulomas, lung tissue, and lymph nodes is necessary to prevent excessive inflammation.

Much of the attention over the past several decades has been on CD4 T cells and their importance in control of Mtb infection and disease. There is no doubt that CD4 T cells are a key factor in the immune response to this pathogen. However, emerging evidence indicates that they are not the only important cell type and which functions of CD4 T cells are most relevant is controversial. Mice deficient in CD4 T cells, either genetic knockouts or through antibody depletion, have higher Mtb bacterial burdens compared with control animals (Caruso et al., 1999; Saunders et al., 2002; Scanga et al., 2000). Th1 CD4 T cells (expressing cytokines IFN-γ, TNF, and/or IL-2) are generally considered to be most important against Mtb, since Th1 cells can activate macrophages to induce anti-bacterial activity. In addition, murine knockouts of IFN-γ, TNF, or IL-12 genes (reviewed in Cooper, 2009), neutralization of TNF in NHPs or in humans (as a treatment for inflammatory diseases) (Keane et al., 2001; Lin et al., 2016, 2010), and genetic studies of mycobacterial susceptibility in humans have demonstrated the importance of these cytokines in resistance to tuberculosis. However, potential vaccine candidates that induce CD4 T cells producing IFN-γ or other Th1 responses have not necessarily resulted in enhanced protection in animal models or in human clinical trials (Tameris et al., 2014, 2013). In murine models, there is evidence that IFN-γ produced by CD4 T cells contributes to protection (Green et al., 2013) but also evidence that there are IFN-γ-independent mechanisms of CD4 T cells that are important (Sakai et al., 2016). Depletion of CD4 T cells in acute infection in macaques leads to exacerbated disease in most animals (Lin et al., 2012) HIV infection in humans is a major risk factor for development of active tuberculosis, and the risk is increased even before CD4 T cells are substantially depleted (Gupta et al., 2013; Lawn et al., 2009). In macaques with LTBI, SIV infection, as a model for HIV, caused reactivation in a majority of animals, while CD4 depletion caused reactivation tuberculosis in only a subset of animals (Bucsxan et al., 2019; Diedrich et al., 2010, 2020). This suggests that SIV (and likely HIV) has additional effects on the immune system beyond CD4 T cells, leading to increased susceptibility to tuberculosis.

In addition to Th1 cytokines, CD4 T cell production of IL-17 (Th17 cells) has been shown in murine models to enhance chemokines in lungs that induce recruitment of IFN-γ producing CD4 T cells to the lungs during Mtb infection (Khader et al., 2007) and in response to BCG vaccination ( Gopal et al., 2012). Neutralization of IL-17 during the early phase of infection in mice did not exacerbate tuberculosis (Segueni et al., 2016b). IL-17 was implicated in control of chronic Mtb infection in murine models although the production was primarily from γδ T cells whereas Th17 cells specific for Mtb were rare. During acute infection, the primary source of IL-17 was also shown to be from γδ T cells (Lockhart et al., 2006). There is also evidence that IL-17 can contribute to expansion of γδ T cells in non-human primates (Shen et al., 2015). In macaque models and in humans, CD4 T cells producing IL-17 are at relatively low levels compared with Th1 cells in blood and in tissues (Gideon et al., 2015; Nikitina et al., 2018). Nonetheless, in a model of mucosal BCG vaccination, Th17 cells were implicated as a correlate of protection (Dijkman et al., 2019). In intravenous (i.v.) BCG-vaccinated animals, which show robust protection against infection and disease, Th17 cells were only a small population in blood or tissues (Darrah et al., 2020). In a recent study where scRNA-seq was applied to granulomas, a subset of T cells (termed T1/T17) that included both CD4 and CD8 T cells expressed transcription factors that are associated with Th17 or Th1 cells but that did not express IL-17 was shown to be significantly associated with reduced bacterial burden (Gideon et al., 2022). An intriguing study using resected human lung tissue suggested that tissue-resident IL-17 expressing cells could participate in restricting Mtb growth (Ogongo et al., 2021). However, another study of the tuberculin skin test (TST) response in humans demonstrated that an IL-17 signaling pathway was upregulated in the TST of persons with active tuberculosis compared with those with LTBI, although the source of IL-17 was not reported (Pollara et al., 2021). In that study, IL-17 was hypothesized to contribute to tuberculosis pathology. In a study combining human, non-human primate and murine tuberculosis data, innate lymphocytes of the ILC3 class which produce IL-17 and IL-22 were implicated in the formation of bronchial associated lymphoid tissue (iBALT) in lungs, and ILC3-deficient mice had a modest increase in lung bacterial burden (Ardain et al., 2019). In human lungs, tissue-resident CD4 T cells expressing IL-17 were expanded compared with blood although at a lower frequency than CD4 T cells expressing canonical Th1 cytokines and IL-17 in lungs was inversely correlated with inflammation (Ogongo et al., 2021). In this same study, adding IL-17 or IL-2 to a 3D in vitro granuloma model reduced Mtb bacterial growth. In summary, Th17 cells and other producers of IL-17 may have different effects depending on the stage of infection or in the context of vaccination. They may well play a supporting role through interactions with other cells important for protection.

Immune responses do not exist in a vacuum and CD4 T cells can also exert other functions, including help for inducing B cell and CD8 T cell responses that may play a role in prevention of tuberculosis disease and induction of migration signals for other cells. TNF is a cytokine known to be important in protection against infection and reactivation tuberculosis in humans, nonhuman primates, and mice (Flynn et al., 1995; Keane et al., 2001; Lin et al., 2010). CD4 T cells are one source of TNF in Mtb infections, although macrophages are likely a larger contributor (Allie et al., 2013). The receptor for TNF on myeloid cells is critical to control of Mtb infections in mice (Segueni et al., 2016a), reinforcing that TNF (produced by T cells or myeloid cells) is an activator of macrophages. Data from mice suggest that CD4 T cells are important for inducing appropriate CD8 T cell responses, possibly through production of IL-2 or by activation of antigen presenting cells (Serbina et al., 2001).

Although CD4 T cells are clearly important in control of Mtb infection and disease, the role of adaptive CD8 T cells is less well studied (reviewed in Jasenosky et al., 2015; Lin and Flynn, 2015). In mice, there are conflicting data regarding the importance of these cells in control of tuberculosis. MHC-class-I-and TAP-deficient mice were both more susceptible to tuberculosis disease (Behar et al., 1999; Flynn et al., 1992; Schaible et al., 2002), but depletion of CD8 T cells was reported to only increase disease in the chronic phase of infection (van Pinxteren et al., 2000). The Behar lab has demonstrated enhanced protection with Mtb-specific CD8 T cells in mice, although not all Mtb-specific CD8 T cells recognized infected macrophages (Woodworth et al., 2008; Yang et al., 2018). In one study, human CD8 T cell lines were shown to primarily recognize heavily infected macrophages (Lewinsohn et al., 2003) and recognition of Mtb-infected macrophages by T cells is an area of active research. In addition, human CD8 T cells induced by vaccination with an adenovirus expressing Mtb antigens often did not recognize Mtb-infected macrophages (Nyendak et al., 2016). Antigen cross-presentation on MHC class I could underlie this phenomenon, and it will be important to understand the role of CD8 T cells that recognize infected cells as well as those that recognized uninfected cells presenting Mtb antigens in the context of granuloma level control of infection. Understanding the mechanisms by which CD8 T cells do or fail to recognize Mtb-infected macrophages will be critical in designing CD8-focused vaccine strategies.

Many studies on CD8 T cells in tuberculosis have measured cytokines such as IFN-γ and TNF rather than the more canonical function of CD8 T cells, which is cytotoxicity. There are multiple cytotoxic effectors described for CD8 T cells (and also innate cytotoxic cells), including granzymes (A, B, H, K, and M), granulysin, and perforin. Perforin forms a pore that is used to deliver cytotoxic molecules to a target cell. A potential role for CD8 T cells in control of Mtb is recognizing and destroying infected macrophages or directly killing Mtb inside the cells. There is also growing evidence that granzymes can be secreted and have extracellular functions including modulation of other cells, which could be important in the context of the granuloma. The exact functions of the different granzymes are still being investigated in other disease systems; however, Robert Modlin’s group reported that granulysin could directly kill Mtb in vitro (Stenger et al., 1998). Recently, a small fragment of granulysin has been shown to have antimicrobial activity against extracellular Mtb, internalized by human macrophages in vitro and restricted replication of intracellular Mtb (Noschka et al., 2021). Unlike humans and non-human primates, mice do not express granulysin. “Poly-cytotoxic” CD8 T cells (expressing granzyme B or CCL5, granulysin, and perforin) in human blood were associated with improved control of Mtb infection (Stegelmann et al., 2005). Zheng Chen’s group implicated CD8 T cells in the modest protection provided by intradermal BCG vaccination in macaques (Chen et al., 2009). Several groups have reported a series of Mtb antigens that are recognized by human CD8 T cells. There is wide variability among humans in terms of antigen recognition, and there does not appear to be a single or a few immunodominant antigens (Lin and Flynn, 2015; Lindestam Arlehamn et al., 2014). There is growing interest in the potential roles of cytotoxic CD8 T cells in control of tuberculosis, although much remains to be understood. Recent studies on macaque granulomas using scRNA-seq demonstrate the diversity of CD8 T cell responses at the site of infection, including several cytotoxic subsets that are associated with local control or killing of Mtb (Gideon et al., 2022). Identifying strategies to enhance CD8 T cells and harness their cytotoxic potential in the context of vaccines could be a game changer. Such strategies might include viral vectors, including the CMV-based tuberculosis vaccine discussed below, but also adjuvants such as IL-15 superagonists (N803) that enhance CD8 T cell responses in other systems (Berrien-Elliott et al., 2022; Miller et al., 2022).

Non-classical CD8 T cells recognize glycolipid antigens instead of protein antigens in the context of the MHC class Ib molecules CD1a–c. Humans, non-human primates, and guinea pigs express MHC CD1a–c, but mice do not. Several mycobacterial lipid antigens have been identified for CD1-restricted T cells, including glucose monomycolate, mycoketides, diacyltrehalose, and sulfoglycolipids (James et al., 2018; Reijneveld et al., 2021a, 2021b; Van Rhijn et al., 2013). The TCRs that recognize mycolyl lipid in the context of CD1b are germline encoded, often referred to as GEM TCRs. CD1-restricted T cells are expanded in Mtb-infected humans and in macaques vaccinated with BCG via the i.v. route (Layton et al., 2021). The role of these cells in protection is less clear. Immunization of guinea pigs with certain lipids showed modest improvements in tuberculosis-related pathology (Dascher et al., 2003).

In addition to adaptive CD8 T cell responses, there are also several innate-like lymphocytes that can express cytokine and/or cytotoxic effectors, including NK T cells, γδ T cells, and mucosal associated invariant T cells (MAITs) (collectively referred to as donor unrestricted T cells), and NK and other innate-like lymphocytes (ILCs). NK cells in blood have been associated with LTBI in humans (Chowdhury et al., 2018). Innate-like lymphocytes in airways or lung tissue may be the first line of defense against Mtb infection by modulating the AMs to restrict or kill Mtb bacilli. Several groups have postulated a protective role for MAIT cells, which have a restricted TCR usage and recognize non-peptide antigens presented by the class Ib molecule MR1 (reviewed in Gold et al., 2015). MAITS originally were demonstrated to recognize riboflavin intermediates produced by bacteria but more recently have been shown to respond to stimulation with mycobacterial antigens and even lysates from bacteria that do not produce riboflavin (reviewed in Lepore et al., 2021). MAITs are present in airways of mice, humans, and non-human primates and can produce effector molecules including cytokines and cytotoxic molecules. Although they are considered to be innate-like T cells, they can also have adaptive qualities and may act as a bridge between the innate and immune response to pathogens (Napier et al., 2015). MAITs may be among the first cells to interact with pathogens at mucosal surfaces, such as the airways. It was recently shown that these cells are clonally expanded in the airways of humans with pulmonary tuberculosis (Wong et al., 2019). In contrast, the frequency of MAITs in human and non-human primate granulomas is quite low (Gideon et al., 2022; Ogongo et al., 2020). Increasing the frequency of MAITs in mice with a small molecule led to improved control of Mtb (Sakai et al., 2021a). However, a similar strategy in macaques was complicated by an unexpected blunting of the MAIT response with no effect on bacterial burden (Sakai et al., 2021b). Thus, the roles or importance of MAITS in control of Mtb infection remains unclear.

γδ T cells have received a fair amount of attention over the years in the context of Mtb infection. These T cells recognize phosphoantigens such as isopentenyl pyrophosphate and (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP) and vγ9/vδ2 T cells, which are the majority of γδ T cells in primates, are expanded during Mtb infection or BCG vaccination in humans and non-human primates (Hoft et al., 1998). HMBPP is produced by mycobacterial species and may account for the increase in γδ T cells following mycobacterial exposure although mycobacterial glycolipids were also shown to activate vγ9/vδ2 T cells (Xia et al., 2016). Although generally considered as innate-like T cells, data in non-human primates indicate that they have adaptive T cell qualities and memory responses (Qaqish et al., 2017; Shen et al., 2019). Chen et al. expanded vγ9/vδ2 T cells ex vivo and adoptively transferred them into macaques followed by Mtb infection, which reduced overall lung bacterial burden ~3–4-fold compared with naive macaques. In a subsequent study, this same group immunized macaques with a Listeria monocytogenes strain expressing HMBPP and found similar reductions in bacterial burden following Mtb challenge providing evidence that vγ9/vδ2 T cells may play a role in limiting tuberculosis disease (Shen et al., 2019). Recent data from Mtb-infected humans showed that the γδ T cell repertoire was clonally expanded in lung tissue and dominated by Vδ1 (Ogongo et al., 2020). γδ T cells are also present in macaque granulomas and express cytotoxic molecules (Gideon et al., 2022). γδ T cells can activate dendritic cells, leading to enhanced T cell responses. There are conflicting results in mice regarding the role of γδ T cells in protection against Mtb. In mice lacking γδ T cells due to a knockout in the delta gene, some studies showed modest increases in bacterial burden early following low-dose infection (Ladel et al., 1995) but not following high-dose infection or during chronic infection (Turner et al., 2001). Another study showed no effect on Mtb bacterial burden but suggested a role for these cells in limiting pathology in the lungs (D’Souza et al., 1997). At this time, it is not possible to selectively deplete γδ T cells in non-human primates, and there are some dissimilarities between the murine and primate γδ T cell responses to Mtb. Although more studies are needed, these cells, which are also found in granulomas (Gideon et al., 2022; Napier et al., 2015), may be playing a beneficial role in control of Mtb infection as well as influencing other immune cells.

One of the conundrums in Mtb infection is the relatively slow adaptive T cell response to the pathogen in animal models and in humans. The TST, which relies on a delayed-type hypersensitivity response and therefore measures adaptive T cell responses, takes up to 6 weeks after infection to become positive in humans and in macaque models (Capuano et al., 2003; Poulsen, 1950). Mice also have a relatively slow adaptive T cell response. In mice, various studies have suggested this is due to inhibition of antigen presentation of Mtb antigens, delayed migration of the bacilli to the lung draining lymph nodes, and direct interference with T cell responses (Chackerian et al., 2002; Myers et al., 2013; Reiley et al., 2008). This delay in induction of T cell responses is likely due, at least in part, to the relatively small numbers of infecting bacilli and the slow growth of the organism, so that there is insufficient antigen detected by the immune system early after infection. A recent study from our group showed that in macaques adaptive T cells and innate lymphocytes are present in granulomas, but at early time points (4 weeks) innate lymphocytes are expressing the T-bet transcription factor that regulates cytokine and cytotoxic genes while the adaptive T cells are T-bet negative (Grant et al., 2022). At this early time point, a subset of CD4 T cells express the more global transcription factor RORα. By 12 weeks, the CD8 T cells in granulomas have a significant induction of Tbet, but it takes even longer for a reliably measurable frequency of T-bet+ CD4 T cells to appear. While the T-bet+ CD4 T cell population is enriched for cytokine production, the T-bet+ CD8 T cells produce minimal cytokines but express granzyme B. Thus, while innate lymphocytes likely contribute to early containment of Mtb infection in granulomas, there is a slow evolution of the functional adaptive T cell response. Even in mice, Mtb in lungs grows to a high level with minimal bacterial killing until after 4 weeks of infection. Mtb killing in macaque granulomas is not readily observed until ~10 weeks post-infection (Lin et al., 2014). This delay in functional adaptive T cell responses likely allows the Mtb bacillus to establish a foothold in the lungs resulting in either a persistent or progressive infection. Notable was that a majority of T cells in granulomas express none of the five transcription factors analyzed (Tbet, RORα, RORγ, GATA3, or Foxp3) (Grant et al., 2022). It has also been shown that a majority of T cells in granulomas do not express cytokines (Gideon et al., 2015; Grant et al., 2022). These findings suggest that many of the T cells are not stimulated in granulomas perhaps due to granuloma structure or that many are not Mtb antigen specific and perhaps traffic to the granuloma in response to inflammation (Millar et al., 2021). We previously investigated whether T cells in granulomas were functionally exhausted but did not find evidence for this (Wong et al., 2018); this is supported by the lack of exhaustion markers on T cells in human granulomas based on spatial imaging via MIBI-TOF (McCaffrey et al., 2022). An effective vaccine likely needs to either prevent infection or induce Mtb-specific T resident memory cells that can quickly limit or extinguish infection when the bacilli enter the lungs.

THE B CELL AND ANTIBODY RESPONSE

Antibodies (Ab) exert their antimicrobial activities via the antigen-binding fragment (Fab) and the interaction of the crystallizable fragment (Fc) with Fc receptors (FcRs) on effector cells (Bournazos et al., 2015; Chan et al., 2014; Lu et al., 2018), regulating neutralization, opsonization, opsonophagocytosis, complement fixation, and a broad range of effector functions of innate immune cells, including antigen presentation and cytokine production (Figure 3). The effect of the Fc-FcγR interaction is particularly far reaching, as FcγRs are expressed on a wide variety of immune cells (Bournazos et al., 2015; Chan et al., 2014; Lu et al., 2018). Ample evidence indicates that Ab and B cells (via antigen presentation and cytokine and Ab production) can protect against microbes (including intracellular pathogens) by influencing T cells responses, such as the development of memory, as well as by modulating vaccine efficacy (Bournazos et al., 2015; Chan et al., 2014; Lu et al., 2018; Lund, 2008; Lund and Randall, 2010). B cell and Ab can thus play a role in inducing protection and improving vaccine efficacy against infection for which cell-mediated immune responses are pivotal in controlling the pathogens (Figure 3).

Figure 3. The B cell and humoral immune response to Mtb.

The host response to Mtb begins with the presentation of Mtb antigens (Ag) by antigen-presenting cells to cognate CD4⁺ and CD8⁺ T cells, initiating the development of specific T cell lineages (1). In parallel, naive B cells interact with Mtb component (2), which results in the development of activated B cells. The initial encounters of T cells and B cells with Mtb Ag are an essential first step that leads to the formation in the germinal center of T follicular helper cells (Tfh) and germinal center B cells (3), which interaction is critical for the generation of memory B cells (4), and long-lived antibody (Ab)-producing plasma cells (5). Memory B cells can react rapidly in response to Mtb antigenic challenge, and the plasma cells (5), which produce high quality Ab (6), are critical for the generation and maintenance of serological memory. These latter two components are pivotal to the development of a robust and effective humoral immune response to microbes and Ab-based vaccines. In addition, Mtb Ag-activated B cells (7) can produce soluble factors such as pro-inflammatory and anti-inflammatory cytokines to modulate the development of anti-tuberculosis immunity (8). Additionally, the multifunctional B cells can contribute to the regulation of the immune response to Mtb via other effector functions such as the presentation of Ag to T cells (9). Together, the capacity of B cells to present Ag and to produce cytokines and Mtb-specific Ab can contribute significantly to the regulation of the granulomatous response during Mtb infection. These properties of B cells, in turn, play a role in mediating interaction between B cell and humoral immunity to the broad gamut of immune cells present in the tuberculous granuloma (Figure 1) to determine infection and disease outcomes, granuloma heterogeneity, and the levels of immunopathology (10). Of the major effector functions of B cell and humoral immunity (6, 8, and 9), the role of Ab in modulating anti-tuberculosis immune responses is perhaps the best studied. Specific humoral responses are associated with distinct infection and disease outcomes in human subjects. These are likely related to diverse Ab-mediated effector functions (11) as regulated by differential glycosylation of the Fc fragment of immunoglobulins (Ig) and the distinct functionality of the range of Ab isotypes. The interaction between the Fc of Ab and Fc receptors (FcR) can significantly regulate Ab-mediated effector functions of Ig to (1) modulate opsonophagocytosis, (2) block invasion of the pathogen, (3) enhance dendritic cell (DC) function via engagement of immune complex (IC) by DC to augment T cell responses, (4) mediate Ab-dependent cell-mediated cytotoxicity (ADCC) by targeting NK cells, and (5) Ab-dependent cell-mediated phagocytosis (ADCP) by targeting macrophages and neutrophils. These Ab-mediated effector functions, in turn, modulate infection and disease outcomes, granuloma heterogeneity, and the levels of immunopathology (12).

The notion that B cells and Ab can protect against Mtb is still a topic of considerable debate (Chan et al., 2014), not least because of the apparent low risk for developing tuberculosis in individuals receiving rituximab, a B-cell-depleting mAb (Cantini et al., 2014). Since rituximab is not efficacious in depleting plasma cells, anti-tuberculous serological memory imparted by plasma cells could still be operative in persons at risks for developing tuberculosis, thereby averting overt disease manifestation. Nevertheless, passive immunization studies using Mtb antigen-specific monoclonal Ab (mAb) support a protective role for Ab in tuberculosis (Chan et al., 2014). Existing evidence supports that B cells and Ab modulate immune responses to Mtb (Chan et al., 2014) and are required for the development of optimal anti-tuberculosis immunity, regulating cytokine production, neutrophilic response and granulomatous inflammation, and modulating vaccine-elicited immunity in mice. A role for B cells and Ab in regulating immune responses to tuberculosis has also been observed in non-human primate tuberculosis models (Phuah et al., 2016, 2012). These studies prompted efforts in characterizing the anti-tuberculosis humoral immunity in persons manifesting a spectrum of infection and disease outcomes (Li et al., 2017; Lu et al., 2016, 2019; Chowdhury et al., 2018; Zimmermann et al., 2016) (Figure 3). Thus, IgG in sera from persons with LTBI is superior to that from patients with active disease in mediating NK cell and macrophage anti-tuberculosis effector functions, with a predilection for FcγRIII binding. These Abs exhibit enhanced capacity to activate NK cells that is associated with augmented Ab-dependent cell-mediated cytotoxicity (ADCC), and with increased ability to promote Ab-dependent cell-mediated phagocytosis (ADCP) and restrict Mtb growth in macrophages. These functional differences are associated with distinct patterns of Ab glycosylation (Lu et al., 2016, 2019), which is known to modulate IgG functions (Alter et al., 2018; Wang and Ravetch, 2019). The NK cell/ADCC observation is supported by an independent study examining three separate cohorts of varied demographics (Chowdhury et al., 2018). NK cells kill pathogen-infected cells by ADCC (Bournazos et al., 2015), a process initiated by binding of NK FcγRIII with the Fc portion of Ab that reacts with pathogen antigens expressed on the surface of infected host cells. NK cells (Heron et al., 2012) and IgG (Reynolds et al., 1991) (the key mediator of ADCC) are present in the alveolus, raising the possibility that ADCC may play a role in ablating infected macrophages at the early alveolar phase of infection (Bournazos et al., 2015). To leverage ADCC for Mtb control, identification of the mycobacterial antigens expressed on the surface of infected cells, particularly macrophages, the primary host cells for Mtb, is critical.

Further supporting a role for IgG in the control of tuberculosis, it has been shown that polyclonal IgG from persons with LTBI and the resister phenotype (see below) (Behr et al., 2021; Kroon et al., 2020; Simmons et al., 2018), but not from subjects with active tuberculosis, protects against Mtb challenge in mice (Chen et al., 2020; Li et al., 2017). Intriguingly, IgA mAb cloned from persons with active tuberculosis or LTBI inhibit Mtb invasion of epithelial cells and macrophages, while that of IgG1 subtype promote entry (Zimmermann et al., 2016), underscoring the importance of isotype-dependent Ab-mediated effector functions in controlling Mtb. Indeed, mice deficient in the inhibitory FcγRIIB exhibit enhanced Mtb containment and attenuated lung inflammation that was associated with an augmented pulmonary Th1 response (Maglione et al., 2008). Conversely, Mtb-infected mice deficient in the γ-chain shared by activating FcγRs exhibit enhanced susceptibility and exacerbated immunopathology concomitant with increased production of the immunosuppressive cytokine IL-10. Thus, engagement of distinct FcγR can divergently affect cytokine production and susceptibility during Mtb infection. Elucidation of the Fc-FcγR reaction and the downstream events triggered by this interaction will likely shed light on mechanisms that regulate the development of anti-tuberculosis immunity. Cloning and molecular characterization of mAb from infected hosts (Ishida et al., 2021; Watson et al., 2021; Zimmermann et al., 2016), particularly in full length, will advance our understanding of how Abs regulate anti-tuberculosis immunity.

A role for the less-studied IgM in protection against Mtb is supported by a recent i.v. BCG immunization study in NHP showing that this vaccination strategy elicits a superior Ab response in blood and BAL fluid, relative to that engendered via conventional intradermal vaccination, with IgM being the most prominent isotype associated with bacterial burden control (Irvine et al., 2021). Both natural and immune IgM can protect against microbes, including intracellular pathogens (Baumgarth, 2011; Racine and Winslow, 2009). Innate-like B-1 B cells, which can play a role in infection (Smith and Baumgarth, 2019), are the primary producer of natural IgM (Baumgarth, 2011; Racine and Winslow, 2009). While evidence exists that B-1 B cells contribute to immune responses to Mycobacterium (Acosta et al., 2021), their role in defense against Mtb remains to be examined. Ab, including IgM, are present in airways (Reynolds et al., 1991; Torrelles and Schlesinger, 2017) and thus can potentially mediate antimicrobial effects in the early innate phase of Mtb infection. Vaccines protocols that can engender rapid appearance of Mtb-specific Ab in the airway upon Mtb entry may prove protective.

Recent advances in our understanding of tuberculosis epidemiology and of the highly complex granulomas in animals and humans (Cadena et al., 2017; Cohen et al., 2022; Gideon et al., 2022; McCaffrey et al., 2022; Wells et al., 2021) have led to reconsideration of the definitions of the various Mtb infection and disease states. Much debate revolves around the immunological characteristics of resisters (Behr et al., 2021; Simmons et al., 2018). The level of heterogeneity of subjects categorized into distinct states likely parallels that of lung granulomas (Cadena et al., 2017; Gideon et al., 2022) (Figure 1); this may account, at least in part, for the variability of results in human tuberculosis studies. Precise definitions for the various infection and disease states will improve study design and can impact tuberculosis public health policy (Behr et al., 2021; Simmons et al., 2018). One such state that has recently become a focus of investigation is that of the resisters, defined as individuals who remain TST- or IGRA (interferon-γ release assay)-negative despite extensive Mtb exposure) (Behr et al., 2021; Simmons et al., 2018). However, the immunological and microbiological definition of this entity is unclear (Behr et al., 2021; Simmons et al., 2018). A recent study supports the notion that resisters are or were in fact infected and developed adaptive immunity to Mtb, as the subjects exhibit a non-IFN-γ T cell response to the diagnostic Mtb-specific antigens ESAT6 and CFP10, and their Mtb-specific Ab display class switching and relatively high binding avidity (Lu et al., 2019). Thus, resisters most likely represent a heretofore uncharacterized state of Mtb infection. The Ab response of resisters exhibits features distinct from that of LTBI subjects, with a preponderance of IgG1 that is associated with unique glycosylation patterns. The non-IFN-γ T cell response is associated with augmented CD40L expression. CD40L is expressed by a wide range of cells, which engages with its receptor CD40 to regulate innate and adaptive immunity, including T-cell-dependent B cell activation (Elgueta et al., 2009; Parker, 1993). It is thus possible that the CD40L-high CD4 T cells of resisters are poised to drive B cell development. More studies are required to assess the significance of the resister CD40L phenotype and the non-IFN-γ T cell response. It would be of interest to determine whether resisters remain infected long-term and whether their anti-tuberculosis immunity is superior to that of LTBI subjects. A protective role for resister IgG is reinforced by the observation that passive transfer of IgG from an independent “resister” cohort protects against Mtb in mice (Li et al., 2017).

The germinal center (GC) reaction is critical for the development of effective B cell and humoral immunity (Victora and Nussenzweig, 2022). Within the GC, developing B cells receive help from T follicular helper cells (Tfhs) to attain a durable, high-affinity Ab response, with the development of memory B cells and long-lived plasma cells (Victora and Nussenzweig, 2022). GC-like structures are present in the lungs of Mtb-infected hosts, including humans (Chan et al., 2014; Tsai et al., 2006; Ulrichs et al., 2004). There is evidence that these ectopic GCs (Pitzalis et al., 2014) are associated with favorable Mtb infection outcome (Foreman et al., 2016; Khader et al., 2009; Slight et al., 2013), but their significance in regulating immune responses to Mtb, particularly locally in the lungs, has not been formally evaluated. Despite the critical role of Tfh in promoting the development of humoral immunity and the recent recognition of B cells and Ab as important factors in defense against Mtb in humans (Chan et al., 2014; Li et al., 2017; Lu et al., 2016, 2019; Chowdhury et al., 2018; Zimmermann et al., 2016), the precise role of Tfh in tuberculosis remains incompletely defined. It is possible that subjects manifesting variable infection and disease outcomes with varied Ab profiles may harbor distinct Tfh subsets that can differentially regulate humoral responses in tuberculosis, thereby influencing the susceptibility to infection or disease.

Emerging evidence also suggests that B cells have the capacity to modulate immune responses to Mtb (Chan et al., 2014). B cell subsets with distinct phenotypic markers are associated with varied infection and disease states of tuberculosis (Joosten et al., 2016; Loxton and van Rensburg, 2021). The roles of these subsets in Mtb infection remain to be defined. B cells can regulate immune responses through the production of a broad range of cytokines (Chan et al., 2014; Fillatreau, 2018; Lund, 2008). For example, B cells generate IFN-β, a type I IFN, upon exposure to Mtb or its components (Bénard et al., 2018). Importantly, B cells in human tuberculous pleural fluid express high levels of type I IFN mRNA. Type I IFN promotes the development of anti-inflammatory human and mouse macrophages (Bénard et al., 2018), and Mtb can induce type I IFN expression to decrease resistance to host defense (Manca et al., 2001; Wiens and Ernst, 2016) by attenuating IL-1α/β production (Mayer-Barber et al., 2011; Novikov et al., 2011). It is thus possible that B cell type I IFN plays a role in diminishing anti-tuberculosis immunity. It will be of interest to determine the relative contribution of B cell type I IFN compared with that produced by other cells at the site of infection. In a tumor model, B cells generate the neurotransmitter GABA (γ-aminobutyric acid) to drive the development of anti-inflammatory IL-10-producing macrophages, resulting in attenuation of CD8 T cell cytotoxicity (Zhang et al., 2021). In a low-dose virulent Erdman Mtb infection in mice, B cell deficiency results in exacerbated granulomatous inflammation characterized by IL-17-driven neutrophilia (Kozakiewicz et al., 2013; Maglione et al., 2007). Conversely, in chronic tuberculosis induced by the Mtb CD1551 strain, the granulomatous response in B-cell-deficient mice is attenuated relative to wild type (Bosio et al., 2000). While these divergent results can be due to the use of dissimilar Mb strains, it is possible that B cells in acute and chronic tuberculosis are functionally different as a result of distinct Ab repertoires against antigens differentially expressed in the two phases of infection (Lavollay et al., 2008), varied cytokine profiles (Chan et al., 2014; Fillatreau, 2018; Lund, 2008) and/or capacity to present antigens (Lanzavecchia, 1990; Vascotto et al., 2007). In chronic tuberculosis, excessive inflammation can cause undesirable lung damage in the form of cavitary lesions which promote Mtb transmission (Bhattacharya et al., 2021; Hunter, 2020), a hindrance to tuberculosis control (Hunter, 2020; Modlin and Bloom, 2013). It is thus possible that B cells can be targeted to curb transmission.

B cells and Abs are present in abundance amid a wide variety of immune cells in the lungs of an infected host (Phuah et al., 2012, 2016; Tsai et al., 2006) and play a role in regulating anti-tuberculosis immunity (Carpenter and Lu, 2022; Chan et al., 2014). The local environment of airway, granuloma, or lymph node is likely to promote the interaction of the multifunctional B cells and Ab with other immunological factors (Figure 3). It is these intricate interactions between the diverse granuloma immune cell types and factors and cross-modulation that will ultimately shape the development of innate and adaptive immune responses to Mtb to influence infection and disease outcomes of tuberculosis.

MYCOBACTERIAL HETEROGENEITY AND ANTIBIOTIC TOLERANCE: FROM THE PERSPECTIVE OF MACROPHAGE-MTB INTERACTION

Microbes, including Mtb, exhibit antibiotic persistence, an innate phenotypic heterogeneity characterized by the survival of a small subpopulation (<1%) of drug-sensitive bacteria during in vitro treatment with a bactericidal antibiotic without being genetically resistant (Balaban et al., 2019; Bigger, 1944; Dhar and McKinney, 2007; Lewis, 2010). Bigger designated the surviving bacteria “persisters” and posited that this phenomenon is clinically relevant (Bigger, 1944). Studying the persisters in Mtb-infected mice treated with anti-tuberculosis drugs (McCune et al., 1956, 1966a, 1966b; McCune and Tompsett, 1956), McDermott espoused the significant relevance of this phenomenon in the treatment of infections, including that caused by Mtb. Accumulating evidence support that stresses exerted upon Mtb by infected macrophages augment bacterial heterogeneity, a state that supports survival and the development of antibiotic tolerance (Adams et al., 2011; Chung et al., 2022; Hicks et al., 2018; Liu et al., 2016; Manina et al., 2015; Pisu et al., 2021). Drug tolerance represents an intermediate step toward attaining drug resistance (Hicks et al., 2018; Levin-Reisman et al., 2017), which poses a major obstacle to effective tuberculosis control (WHO, 2022). Manina et al. showed that the levels of Mtb heterogeneity in standard in vitro growth condition can be augmented by a number of stresses, including those imposed by host immunity stemming from infected macrophages and in mice (Manina et al., 2015). A non-growing yet metabolically active Mtb subpopulation was only present in bacteria procured from the lungs of wild type but not immunocompromised IFN-γ-deficient mice, supporting a role for host immune pressure in the development of Mtb heterogeneity. This study also suggested the existence of mechanisms of drug persistence that go beyond the acquisition of a slow growth or non-replicative states, cellular parameters generally thought to be associated with this phenomenon (Balaban et al., 2013). Indeed, it has been shown that macrophage-induced expression of an efflux pump by actively growing Mtb can mediate drug tolerance (Adams et al., 2011).

The significance of host immunity in driving Mtb heterogeneity has been underscored in a recent study, showing that enhanced mutation rates are detected in isolates collected from HIV-negative but not HIV-positive subjects with active tuberculosis prior to the initiation of anti-tuberculous therapy (Liu et al., 2020). Using fitness reporter Mtb strain to probe the relationship of host immune pressure and the development of drug tolerance in vivo (Liu et al., 2016), it was shown that levels of Mtb tolerance to INH and rifampin derived from lung myeloid cells of Mtb-infected mice treated with these two agents are proportional to the degree of activation of the host cells (Liu et al., 2016). In addition, in vitro experiments demonstrated that activated macrophages promote the capacity of Mtb to develop tolerance to anti-tuberculosis drugs (Liu et al., 2016). Thus, it appears that the host macrophage environment contributes significantly to the development of drug tolerance, the latter an obstacle to tuberculosis treatment.

THE UNDERSTUDIED NON-MACROPHAGE MYELOID CELLS

Ample evidence support the notion that host-Mtb interaction involves a range of myeloid cells in the lungs of a tuberculous host, including macrophages, dendritic cells, monocytes, neutrophils, and even mast cells (Gideon et al., 2022; McCaffrey et al., 2022; Srivastava et al., 2014; Wolf et al., 2007), all of which are present in granulomatous tissues in mice (Srivastava et al., 2014; Tsai et al., 2006; Wolf et al., 2007), NHP (Gideon et al., 2022), and humans (McCaffrey et al., 2022). Thus, infected myeloid cell subsets can contribute to shaping the immune responses to Mtb and modulating the overall biology of Mtb, thereby influencing the outcome of the infection (Srivastava et al., 2014). Cohen et al. (2018) showed that in the acute phase of murine tuberculosis, the infection foci established in the interstitium by aggregates of translocated Mtb-infected AM mediate dissemination of bacilli into recruited neutrophils and monocyte-derived macrophages. What remains to be defined is how this initial bacterial dissemination into recruited myeloid cells occurs and the precise downstream events following this spread. In a zebra fish model of Mycobacterium marinum infection, resident macrophages in the hindbrain of the infected Danio are the first immune cells that the invading pathogen enters and therefore, could be considered the AM equivalent of Mtb lung infection (Cambier et al., 2017, 2014). M. marinum phenolic glycolipids triggered the production of CXCL2 by infected hindbrain macrophages, leading to the recruitment of the more growth permissive monocytes, to which bacilli within macrophages are transferred to cause disease dissemination (Cambier et al., 2014, 2017). These observations suggest that the neutrophils and monocytes contribute to the immune response to Mtb and to disease pathogenesis, but relative to macrophages, these versatile myeloid cells (Borregaard, 2010; Galli et al., 2011; Swirski et al., 2014; Xie et al., 2020) are understudied (Pisu et al., 2021; Srivastava et al., 2014; Wolf et al., 2007). For example, the precise role in tuberculosis of neutrophils, perhaps the first immune cells that arrive at the site of Mtb infection (Seiler et al., 2003; Tsai et al., 2006), remains to be defined. Evidence exists that neutrophilia may reflect failed Th1 immunity in tuberculosis (Nandi and Behar, 2011) and that neutrophils can modulate vaccine efficacy (Kozakiewicz et al., 2013). The interaction of Mtb with neutrophils can augment dendritic cell migration to lymph nodes to promote the initiation of adaptive immunity (Blomgran and Ernst, 2011). A role for neutrophils to protect against Mtb remains controversial and is likely contextual (see references in Blomgran and Ernst, 2011; Kozakiewicz et al., 2013). Neutrophilia has been associated with exuberant lung pathology and poor Mtb control in susceptible mice (Eruslanov et al., 2005; Keller et al., 2006). Neutrophils, recruited to infected lung via an IL-1b- and 12/15-lipoxygenase-dependent pathway, have been reported to provide an environment with nutrients that promote Mtb growth, and this phenomenon has clinical implications (Mishra et al., 2017). More recently, it has been shown that loss of the autophagy factor ATG5 in neutrophils enhances Mtb susceptibility and ATG5 can attenuate immunopathology caused by neutrophils (Kimmey et al., 2015), and specific apoptotic pathways in neutrophils can mediate protection against Mtb (Stutz et al., 2021). It is well established that a neutrophil and type 1 IFN signature is observed in blood from humans with active tuberculosis (Berry et al., 2010) and is recapitulated in mice (Moreira-Teixeira et al., 2020). A role for myeloid cells in shaping the immune response to Mtb is further underscored by a recent study designed to map specific microenvironments in tissues procured from humans with active tuberculosis using multiplexed ion beam imaging by time of flight (MIBI-TOF) to visualize 37 proteins (McCaffrey et al., 2022). Of note, one of the microenvironments, made up of myeloid cells comprising macrophages, neutrophils, monocytes, and dendritic cells, which displays expression of the immunoregulatory molecules indolamine 2,3-dioxygenase 1 and PD-L1, is characterized by the absence of IFN-g expression and enhanced levels of TGF-b and regulatory T cells. The immunoregulatory features of this myeloid microenvironment is apparent in the blood of tuberculosis patients as assessed by transcriptomic analysis. Importantly, PD-L1 expression is associated with progression to active tuberculosis. It is possible that the immunoregulatory cells in the myeloid microenvironment reported by McCaffrey et al. (2022) may be related to myeloid-derived suppressor cells described in tuberculosis (Dorhoi et al., 2020). The role of this myeloid niche in tuberculous granulomatous tissues in modulating the immune response to Mtb warrants further investigation.

TYPE 2 IMMUNE RESPONSES

Although historically, tuberculosis has been considered to be dominated by type 1 immune responses, including Th1 T cells, recent studies support that type 2 immune cells and responses may play a role in Mtb infection outcome. Eosinophils are found in human and macaque granulomas, although usually at low levels, and at least in mice appear to participate in control of Mtb (Bohrer et al., 2021). Mtb does not appear to infect eosinophils in mice, macaques, or human granulomas but instead these cells influence the microenvironment of the granuloma and are early to arrive at the site of infection. Mast cells are another key player in type 2 immune responses. From scRNA-seq data on macaque granulomas at 10 weeks post-infection, mast cells and plasma cells are found together in granulomas that arise early in infection and have high bacterial burdens (Gideon et al., 2022). Although there was no evidence for IgE production by the plasma cells, the mast cells demonstrated expression of IL-13 (Gideon et al., 2022). A fascinating study by Cronan et al. demonstrated in a zebrafish model using M. marinum that signaling through the IL-4 receptor (by IL-4 or IL-13) and Stat6 was critical to early granuloma formation, specifically by differentiation of macrophages into epithelioid macrophages (Cronan et al., 2021). Without Stat6 signaling, granulomas in the fish did not form properly and had poor control of M. marinum. Interaction analyses in the scRNA-seq granuloma study suggested that mast and plasma cells drive an early self-reinforcing type 2 immune environment (Gideon et al., 2022). The presence of fibroblasts at higher levels in conjunction with mast cells and plasma cells in high burden granulomas could indicate that this constellation of cells reflects a wound-healing environment, possibly trying to initiate and wall off the granuloma when adaptive responses have not yet begun to kill the bacilli. A recent study showed co-localization of Mtb antigens with mast cells in human lung tissue as well as production of cytokines including IL-17 that may influence migration of cells to granulomas and proteases that may contribute to fibrosis (Garcia-Rodriguez et al., 2021). Thus, type 2 immune responses may participate in control of infection through early granuloma formation and function and balancing of type 1 and other inflammatory responses but also may exacerbate disease depending on timing and location of the cytokines and cells during infection.

VACCINES AND CLINICAL CONUNDRUMS

To truly stem the tide of tuberculosis, we need new and effective vaccines. For the first time in decades, there have been positive developments in the vaccine space. In clinical trials, a subunit vaccine, M72, composed of two Mtb proteins with AS01E adjuvant was shown to reduce the incidence of active tuberculosis in people with clinically latent infection by 49.7% (Tait et al., 2019). Presumably, this subunit vaccine boosted CD4 T cell responses in LTBI subjects that reduced the chance of progressing to active tuberculosis. The same vaccine was tested in rhesus macaques, but as a prevention of infection or disease in BCG-vaccinated macaques. Naive (i.e., without prior Mtb infection) macaques were vaccinated intradermally with BCG and then boosted with M72/AS01 or M72 proteins expressed in an Ad5 viral vector (Darrah et al., 2019). In this scenario, M72 did not provide protection against tuberculosis disease beyond the small amount of protection provided by BCG. It remains to be seen in clinical trials whether M72/AS01E will provide protection against Mtb infection or disease in humans without LTBI.

Another recent clinical trial involved revaccination with BCG (Nemes et al., 2018). Humans previously vaccinated at birth with BCG were revaccinated intradermally with the standard human BCG dose or with an adjuvanted subunit vaccine H4. Although there was no difference in the rate of Mtb infection in any group, as assessed by the conversion of the IGRA, which is a blood-based assay that measures T cell IFN-γ production to Mtb-specific proteins, BCG revaccination showed a significant decrease in sustained IGRA conversion compared with those who were not revaccinated (46.3% versus 24.5%). In other words, a higher percentage of BCG revaccinated people who had converted their IGRA to positive then reverted to negative. H4 vaccination also increased the IGRA reversion rate, although not significantly different than the control group. As noted, the control group (BCG at birth but no subsequent vaccination) had a relatively high rate of IGRA reversion (25%).

What IGRA reversion signifies clinically is not clear. It could mean clearance of initial Mtb infection or lower Mtb bacterial burden, or it could point out the inadequacies of IGRA as a diagnostic measure. In fact, in human cohorts with known Mtb exposures who remain IGRA negative, known as “resisters,” it was demonstrated that in fact resisters do have T cell and antibody responses to Mtb proteins, but their T cells do not produce IFN-γ in response to ESAT-6 and CFP10 (Lu et al., 2019). Resisters may have a lower risk of progression to active tuberculosis (Stein et al., 2019). Recent data indicate that IGRA reverters do have T cell responses to Mtb proteins as measured by flow cytometry indicating that they were productively infected, and their responses are consistent with well-controlled infection (Mpande et al., 2021). It is possible that “resisters” and “reverters” are similar phenotypes, depending on when the IGRA assays were done post-exposure. More studies are needed to determine what IGRA reversion and persistent IGRA-negative results in Mtb-exposed persons signify in terms of risk of active tuberculosis. Since diagnosis of possible Mtb infection relies on measuring a small fraction of the immune responses to the pathogen rather than the actual presence of the pathogen, the interpretation of a positive or negative IGRA in terms of risk of disease remains challenging. This highlights the failings of our common diagnostic tools for detecting Mtb infection and disease, especially in developing countries. Recent studies have demonstrated the potential utility of transcriptional signatures of risk of disease in humans with LTBI, some that can identify development of active tuberculosis 12–18 months prior to symptoms (Penn-Nicholson et al., 2020; Suliman et al., 2018; Zak et al., 2016).

In the last several years, there have been studies in non-human primates demonstrating that robust protection against tuberculosis, even sterilizing immunity against infection, is possible, providing hope for a path to effective vaccines. Rhesus macaques are quite susceptible to tuberculosis, with nearly all animals succumbing to active tuberculosis within a few months post-infection with a virulent Mtb strain. A rhesus CMV vectored vaccine, expressing 6–9 Mtb proteins, provided nearly 70% protection against Mtb disease in rhesus macaques, and induced strong T cell responses (Hansen et al., 2018). An attenuated Mtb strain resulted in reduced disease following challenge with a less virulent Mtb strain in rhesus macaques (Kaushal et al., 2015). The only approved tuberculosis vaccine for humans, BCG, is administered intradermally to newborns. However, changing the route to i.v. and increasing the dose of BCG in rhesus macaques resulted in remarkable protection (90%), including sterilizing immunity, against Mtb infection or disease (Darrah et al., 2020). In contrast, intradermal BCG at the standard dose or high-dose aerosolized BCG did not provide protection against disease in macaques; high-dose intradermal BCG was not significantly better than low-dose BCG ID, but there was a subset of animals that showed some protection against disease. It is possible that higher doses of intradermal BCG in humans, possibly even in a revaccination trial, could improve protection, although additional studies would be needed. In contrast, delivery of BCG via bronchoscope directly in the airways of rhesus macaques resulted in protection against disease and perhaps even against infection in a subset of animals (Dijkman et al., 2019).

The rhesus macaque model sets a very high bar for protection as these animals are much more susceptible than humans to infection and disease. Thus, these recent successes offer both hope and opportunities to define what is necessary for an effective vaccine in humans. BCG delivered either by the i.v. route or via bronchoscope directly into the lungs resulted in much higher levels of mycobacteria-specific T resident memory cells in lungs compared with BCG delivered via other routes (Darrah et al., 2020). Although this has not been reported for the CMV-tuberculosis vaccine, one can speculate that the persistence of the CMV vector and the attenuated Mtb strain may also induce T resident memory cells or other protective factors in tissues. Unlike studies in mice, concurrent infection with Mtb in macaques provides robust protection against reinfection, providing yet one more model of protection that can be useful for delineating mechanisms of protection (Cadena et al., 2018). Ongoing and future studies including depletion of T cell subsets or B cells, neutralization of cytokines, and in-depth analyses of local immune responses in these vaccination and protection models can point to the critical protective mechanisms. These studies will provide crucial information about how best to construct an effective vaccine for humans.

It is also important to identify correlates of protection, which are not necessarily mechanisms of protection. Correlates of protection are critical for clinical vaccine trials and should be measurable in an easily accessible sample, such as blood (Nemes et al., 2022). Correlates may be a combination of innate responses, T cells, antibodies, or even gene expression or metabolic signatures. Such studies are being conducted in the clinical trials that have provided some measure of success. Correlates can also be assessed in the animal models of vaccine-induced protection, which will be important as more vaccines move into clinical trials.

It must be acknowledged that much of our understanding of events in the tissues, including formation, establishment, progression, and regression of granulomas, comes from animal models—mice, non-human primates, zebrafish, and others. Such studies are next to impossible to perform in humans. However, most studies in animal models are in naive (with respect to mycobacteria) animals, while in most countries humans are vaccinated at birth with BCG. In addition, in some parts of the world, exposure to non-tuberculous or environmental mycobacteria may influence subsequent exposure to Mtb. These present a limitation to translation of animal studies to humans. This is particularly important for vaccine studies, since vaccines must perform in humans who have been previously vaccinated with BCG. There are ongoing studies in macaques to vaccinate infants with BCG for future novel vaccine studies, but these are limited and expensive. Nonetheless, it is important to consider these limitations in interpreting the results from animal studies.

Due to the nature of Mtb infection and the lack of validated correlates of protection, clinical trials of vaccines are long and expensive, requiring large numbers of participants. Several infectious disease fields have benefited from human challenge models for assessing vaccines, including influenza and malaria (Minhinnick et al., 2016). Human challenges provide a rapid assessment of vaccine efficacy in a controlled setting with relatively few volunteers. However, a human challenge model of Mtb infection poses questions of safety and feasibility, including the difficulty in assessing infection or bacterial burden in vivo, the potential for latent infection that could reactivate later, and the long treatment regimen using drugs with known side effects. Still, there are efforts in this arena, mostly focusing on BCG challenge, either in skin (Minhinnick et al., 2016) or in lungs. PPD pulmonary challenge in persons with LTBI has been performed for several years to identify the recall response in lungs (Silver et al., 2003; Walrath et al., 2005). BCG skin challenge in the setting of vaccines has been done in humans with some modest success (Harris et al., 2014). However, whether the skin immune responses reflect the critical lung immune responses in terms of protection is unclear. BCG lung challenges have also been performed in proof-of-concept studies (Davids et al., 2020), although not yet in the setting of vaccines. Although difficult to develop, safe human challenge models for tuberculosis would be game changer in the vaccine world.

CONCLUDING REMARKS

The immune response to M. tuberculosis is complex and multilayered, and there are still many unanswered questions about protective or exacerbating immune responses. It is impossible to be truly comprehensive in a single review on the immunology of tuberculosis, and thus, there are many important studies that were not included in this short review. Our opinion is that the various host factors do not work in an independent fashion, either in airways, lungs, granulomas, or lymph nodes, but instead interact with each other. The sum total of immune interactions ultimately determines the outcome of infection, either by allowing or preventing initial infection, priming of the immune responses in lymph nodes, and influencing the outcome of each individual granuloma. There are multiple examples of cells and secreted factors, including chemokines, cytokines, and cytotoxic effector molecules, influencing other cells, not only in Mtb infection but in many other biological systems. We hypothesize that there are several interaction pathways that can lead to either success or failure in Mtb infection outcomes, involving multiple cell types and effectors. In other words, not all cell types or effectors are required in all situations, but that the long evolution of the human host with Mtb has resulted in multiple players that can contain or resolve infection. Similarly, Mtb survives and multiplies in humans and therefore there are pathogenic features of Mtb that resist elimination by the host, resulting in a host-pathogen standoff. The explosion of new data in the realm of tuberculosis immunology highlights various pathways to successful control or elimination of the bacillus through interventions or vaccines. It will likely require engagement of various arms of the immune system, as well as multiple cell types and effectors, with further research and translation of the findings to humans, to stem the tide of the tuberculosis epidemic.