Suboptimal Antigen Presentation Contributes to Virulence of Mycobacterium tuberculosis In Vivo Free

This article is featured in In This Issue, p.1

CD4 T cells are essential for control of infection with Mycobacterium tuberculosis. In humans coinfected with M. tuberculosis and HIV, CD4 T cell deficiency leads to a higher frequency of progression to active tuberculosis (TB), a higher frequency of disseminated extrapulmonary disease, and higher mortality (1–3). Similarly, M. tuberculosis infection of mice deficient in CD4 T cells results in higher bacterial burdens in the lung and other tissues and in shortened survival compared with infection of immunocompetent mice (4–6). Although CD4 T cells are essential for control of infection, T cell responses rarely, if ever, eliminate M. tuberculosis from infected humans (7, 8) or animals (9, 10). Consequently, understanding the mechanisms that limit the efficacy of CD4 T cells in TB is essential to guide rational approaches to improving control of TB, including the development of effective vaccines.

Previous studies revealed that M. tuberculosis subverts CD4 T cell–dependent immunity. For example, priming of Ag-specific CD4 T cells occurs much later after M. tuberculosis infection compared with other infections, and this provides time for the bacterial population to markedly expand prior to the appearance of effector T cells in the lungs (11–13). In addition, CD4 effector T cells specific for the immunodominant M. tuberculosis Ag85B are activated poorly at the site of infection in the lungs (14), and regulatory T cells dampen the effector CD4 T cell response during infection (15). Furthermore, mycobacteria were reported to interfere with MHC class II Ag presentation to CD4 T cells in vitro (16–22), although the in vivo significance of this mechanism has not been determined. Because direct recognition of M. tuberculosis–infected cells by CD4 T cells is required for optimal control of intracellular infection in vivo (23), the effectiveness of Ag presentation by infected cells may be an important determinant of the outcome of infection.

M. bovis bacillus Calmette–Guérin (BCG), which has been widely used as a TB vaccine, is less virulent than wild-type (WT) M. bovis and M. tuberculosis. Although the genomic differences between M. bovis BCG (hereafter termed “BCG”) strains and M. tuberculosis are well characterized (24), and the contribution of the loss of the RD-1/Exs-1 locus to attenuation is well established (25–27), the consequences of its attenuation on host–pathogen interactions have not been studied in depth. Similar to control of infection with M. tuberculosis, CD4 T cells are important for control of BCG infection in humans (28, 29) and mice (6, 30–32). However, in contrast to the inability of CD4 T cell responses to eliminate M. tuberculosis, CD4 T cell responses to BCG infection are associated with gradual clearance of the bacteria (31, 33). The differential effects of T cell responses on the course of infection with M. tuberculosis and BCG prompted us to hypothesize that, compared with BCG, M. tuberculosis impedes the generation, activation, or action of CD4 T cells. Because M. tuberculosis resides in professional APCs (34), we further hypothesized that M. tuberculosis impedes CD4 T cell activation by acting on APCs. We found that dendritic cells and macrophages infected with BCG are more capable of activating CD4 T cells in vivo and in vitro than are cells infected with virulent M. tuberculosis H37Rv, and we found evidence that this is attributable to more effective Ag presentation. These results establish that ineffective Ag presentation is associated with virulence in tuberculosis and likely contributes to the ability of M. tuberculosis to evade elimination in immunocompetent hosts.

C57BL/6 mice of WT and TCRβ/δ^−/− genotypes were either bred in the New York University School of Medicine Skirball animal facility or purchased from Taconic Farms for aerosol and intratracheal infection. Mice aged 6–8 wk were used for infection; at various time points following infection, mice were euthanized, and lungs and mediastinal lymph nodes (MDLN) were isolated for CFU enumeration and flow cytometry. Peptide 25–specific TCR-transgenic (P25TCR-Tg) CD4 T cells, specific for Ag85B peptide 25 (aa 240–254 of the mature protein) were isolated from P25TCR-Tg mice on the C57BL/6 background (11, 35). All mouse experiments were performed in accordance with the New York University School of Medicine Institutional Animal Care and Use Committee.

WT M. tuberculosis strain H37Rv and BCG Pasteur were initially acquired from American Type Culture Collection, and the Ag85B deletion mutant (ΔAg85B) H37Rv strain was generated as described previously (11). All bacterial strains were stored at −80°C; bacteria were thawed and cultured to mid-log phase in Middlebrook 7H9 media supplemented with 10% (v/v) ADC enrichment prior to use for aerosol infection of mice or infection of cultured cells. Mice were inoculated with 10² CFU H37Rv or 5 × 10⁴ BCG Pasteur using an Inhalation Exposure Unit (Glas-Col). The dose delivered was verified 1 d following aerosol infection by euthanizing infected mice to isolate and homogenize infected lungs in PBS–Tween-80 (0.5%) for CFU plating on Middlebrook 7H11 medium supplemented with 10% (v/v) ADC enrichment. Infected cells were counted and lysed in PBS–Tween-80 and plated on 7H11 medium to determine multiplicity of infection (MOI) in bone marrow-derived dendritic cells (BMDC) and bone marrow-derived macrophages (BMMΦ).

Single-cell suspensions from infected lungs and lymph nodes were stained using the following fluorescently labeled Abs (BioLegend, BD Pharmingen, or eBioscience): anti-CD3ε allophycocyanin-Cy7 (145-2C11), anti-CD4 (L3T4) Pacific Blue, anti-CD45.1 (A20) PerCP or Alexa Fluor 750, anti-CD69 (H1.2F3) allophycocyanin, anti-CD44 (IM7) PE, and anti–IFN-γ (XMG1.2) PerCP-Cy5.5. MHC class II tetramers I-A^b Ag85B_280–294 PE and I-A^b EsxG_46–61 PE were obtained from the National Institute of Allergy and Infectious Diseases Tetramer Core Facility (Emory University, Atlanta, GA). The epitope recognized by the Ag85B-specific tetramers is also termed Ag85B_240–254, when the amino acid numbering excludes the 40-aa signal peptide. The epitope recognized by the Ag85B tetramers is identical to that of the P25TCR-Tg CD4 T cells used in these studies. Infected BMDC and BMMΦ were stained with the following Abs: anti-CD11c (N418) PerCP, anti-F4/80 (BM8) allophycocyanin, anti–MHC class II IA^b (M5/114.15.2) Pacific Blue, anti-CD86 (PO3) Alexa Fluor 700, anti-CD40 (3/23) PE, and anti–PD-L1 (10F.9G2) PE. Flow cytometry was performed using an LSR II (BD Biosciences) at the New York University Flow Cytometry and Cell Sorting core facility. Flow cytometry data were analyzed using FlowJo software.

Culture filtrates and pellet lysates of mycobacteria were collected from overnight cultures in Saunton’s medium. A total of 0.5 μg total lysate protein and 2.5 μg culture filtrate proteins were loaded and run on 10% Ready Gel Tris-HCl gels (Bio-Rad). Proteins were transferred in Tris-Glycine buffer lacking SDS and blotted for Ag85B (polyclonal rabbit antisera prepared in our laboratory), MPT32 (polyclonal rabbit antisera; BEI; NR-13807), and GroEL2 (BEI; clone IT-70).

Bone marrow was isolated from WT mice aged 6–8 wk. To generate BMDC, bone marrow was cultured for 7 d in complete RPMI 10 media supplemented with 12 ng/ml GM-CSF (PeproTech). The media was replaced every 3 d; on day 7 the floating fraction of cells was collected, magnetically labeled with anti-CD11c (N418) MicroBeads, and sorted using an AutoMACS (Miltenyi Biotec). The CD11c⁺ cells were counted and replated for infection. BMMΦ were differentiated in complete DMEM-10 supplemented with 20% L929-conditioned media. The media was replaced every 3 d for 7 d. On day 7, the adherent fraction of cells was collected by incubation in PBS at 37°C, counted, and replated for infection.

BMDC were infected at an MOI of 2 overnight; the following day, infected cells were treated with amikacin (20 μg/ml) to kill extracellular bacteria. Cells were washed with PBS, collected, and counted. A total of 10⁵ infected BMDC was transferred to WT mice, which received 2 × 10⁶ CFSE-labeled P25TCR-Tg CD4 T cells 1 d prior to the intratracheal transfer of infected BMDC. Mice were euthanized 60 h after BMDC transfer, and lung-draining MDLN were isolated for CFU enumeration and stained for flow cytometric analysis of P25TCR-Tg cell proliferation.

CD4 T cells from the lymph nodes of P25TCR-Tg mice were magnetically labeled with CD4 (L3T4) MicroBeads and sorted using an AutoMACS. CD4⁺ P25TCR-Tg cells were cocultured with irradiated splenocytes from C57BL/6 mice in the presence of Th1-skewing cytokines, as previously described (14). Frozen stocks of Th1-differentiated P25TCR-Tg cells were stored in liquid nitrogen after differentiation and thawed for coculture with infected BMDC and BMMΦ.

BMDC and BMMΦ were infected overnight at an MOI of 2. The following day, infected cells were treated with amikacin (20 μg/ml) to kill extracellular bacteria and washed extensively with PBS. The infected cells were collected, counted, and assessed for viability. Infected cells were replated at 2 × 10³, 2 × 10⁴, and 2 × 10⁵ cells/well in a 96-well plate, and 2 × 10⁴ Th1 P25TCR-Tg cells were added to each well. Infected cells were also lysed in PBS–Tween-80 and plated at various dilutions to determine MOI. Supernatants from these cocultures were collected and filtered 6 h later for mouse IFN-γ ELISA (BD Biosciences).

To test the hypothesis that differential T cell activation or effector action contributes to the distinct outcomes of infection with BCG or M. tuberculosis, we first confirmed that T cells are required for clearance of BCG from mice after aerosol infection. Initial studies with aerosol inocula of 10²–10³ BCG/mouse resulted in clearance of BCG from lungs of WT mice within 25 d of infection, with minimal induction of T cell responses, indicating that innate mechanisms are sufficient to clear BCG after these low-dose inocula (data not shown). Increasing the aerosol inoculum to 5 × 10⁴ BCG/mouse established infection in the lungs where the bacterial burden increased, albeit with a growth curve dissimilar to that of M. tuberculosis, in WT or T cell–deficient (TCRβ/δ^−/−) C57BL/6 mice during the first 28 d. After 28 d postinfection, BCG continued to expand in the lungs of T cell–deficient mice, whereas the bacteria began to be cleared from the lungs of WT mice (Fig. 1A). These data indicate that aerosol BCG infection with a higher inoculum establishes infection in the lungs that requires T cells for control and clearance. In contrast, the presence of T cells in WT mice prevented progressive growth of M. tuberculosis after day 28, but the bacteria were not cleared from the lungs (Fig. 1B). Therefore, T cells contribute to resolution of BCG infection, but they are unable to resolve M. tuberculosis infection. These findings support the hypothesis that M. tuberculosis interferes with activation and/or effector action of T cells in vivo and, thereby, avoids being cleared.

Because CD4 T cells are essential for control of M. tuberculosis in mice (4) and humans (36), and because in vitro studies indicate that M. tuberculosis can inhibit Ag presentation to CD4 T cells (37), we focused our subsequent studies on CD4 T cells. To test the hypothesis that CD4 T cell interactions with APCs differ during BCG or M. tuberculosis infection, we first characterized the earliest encounter of Ag-specific CD4 T cells and APCs postinfection: priming of T cells in the local lymph node. For these analyses, we used P25TCR-Tg CD4 T cells, which recognize peptide 25 (aa 240–254 of the mature protein) of Ag85B (11, 35, 38), expressed by both BCG and M. tuberculosis (24). We transferred CFSE-labeled, P25TCR-Tg CD4 T cells to WT mice 1 d prior to aerosol infection with either BCG or M. tuberculosis and assessed in vivo proliferation (as dilution of CFSE) of P25TCR-Tg CD4 T cells from the lung-draining MDLN at multiple time points. P25TCR-Tg CD4 T cells proliferated as early as 4 d after aerosol infection with BCG (Fig. 2A, 2C), whereas proliferation was not observed before 11 d postinfection with M. tuberculosis (Fig. 2B, 2C). Consistent with the earlier onset of CFSE dilution as an indication of proliferation, more CFSE^dilute P25TCR-Tg cells were present in the MDLN between 4 and 11 d following BCG infection than in lymph nodes of M. tuberculosis–infected mice (Fig. 2C). The earlier priming of P25TCR-Tg cells during BCG infection resulted in the earlier arrival of CFSE^dilute P25TCR-Tg cells in lung (Fig. 2D). Consistent with earlier activation of P25TCR-Tg cells in the MDLN, live BCG were transported to the MDLN significantly earlier than M. tuberculosis (Fig. 2E). We considered the possibility that the earlier P25TCR-Tg activation and proliferation observed in the MDLN could be due to the higher aerosol inoculum of BCG and the early presence of a large number of Ag-producing BCG in the MDLN. However, this was not the full explanation, because we also observed that fewer BCG CFU initiated proliferation in the MDLN; ∼60 BCG CFU were present in the MDLN at the onset of P25TCR-Tg cell proliferation on day 4, whereas initial proliferation in response to M. tuberculosis required >1000 bacteria in the MDLN on day 11 (Fig. 2E). These results indicate that earlier priming of Ag-specific CD4 T cells after BCG infection may be due to two factors: the earlier appearance of bacteria in the lymph node and priming of Ag-specific CD4 T cells in the presence of fewer bacteria than are required after M. tuberculosis infection.

The finding that priming of P25TCR-Tg CD4 T cells after aerosol infection happens in the presence of fewer BCG than M. tuberculosis in the local lymph node suggested that BCG-infected cells may be superior to M. tuberculosis–infected cells in activating CD4 T cells. To compare P25TCR-Tg CD4 T cell responses to equal numbers of BCG and M. tuberculosis in vivo, we used intratracheal transfer of infected BMDC (39, 40). We transferred 10⁵ BCG- or M. tuberculosis–infected BMDC intratracheally to mice that received CFSE-labeled P25TCR-Tg CD4 T cells 1 d earlier. P25TCR-Tg cell proliferation in the lung-draining lymph node was quantitated ∼60 h after transfer of the infected BMDC. We found that when equivalent numbers (∼2000) of BCG or M. tuberculosis were present in the lung-draining lymph node after transfer of infected BMDC (Fig. 3A), BCG was more effective in priming P25TCR-Tg CD4 T cells (Fig. 3B). Although ∼70% of P25TCR-Tg cells had proliferated in response to BCG, only ∼40% of P25TCR-Tg cells had proliferated in response to M. tuberculosis at the same time point (Fig. 3C). Consistent with this finding, the total number of CFSE^dilute P25TCR-Tg cells present in BCG-infected lymph nodes was 15-fold greater than in M. tuberculosis–infected lymph nodes (Fig. 3D). As an additional indication of Ag-specific CD4 T cell priming, the frequency of P25TCR-Tg cells expressing CD69 was higher in response to BCG than M. tuberculosis (Fig. 4E). These results support the hypothesis that cells infected with BCG are more efficient than those infected with M. tuberculosis at activating P25TCR-Tg CD4 T cells in vivo.

We next examined the dynamics and activation of CD4 T cell populations in the lungs following infection with either BCG or M. tuberculosis. The frequency (Fig. 4A) and number (Fig. 4B) of CD4 T cells increased markedly in the lungs between 14 and 28 d postinfection. Notably, both the frequency and number of CD4 T cells were higher in lungs of mice infected with M. tuberculosis, indicating that the inability of T cells to clear M. tuberculosis is not due to deficient trafficking of CD4 T cells to the lungs. Despite differences in the total number of CD4 T cells, the overall frequencies of lung CD4 T cells expressing CD69 were similar in M. tuberculosis–infected mice and BCG-infected mice (Fig. 4C), suggesting that the general phenotype and state of activation of CD4 T cells in the lungs do not explain the inability of T cells to clear M. tuberculosis infection. When we used MHC class II tetramers (I-A^b:Ag85B_280–294 and I-A^b:EsxG_46–61) to characterize activation of Ag-specific CD4 T cells in the lungs 21 d after aerosol infection, we found similar numbers of Ag85B- and EsxG-specific CD4 T cells in mice infected with M. tuberculosis or BCG (Fig. 4D). We also found similar frequencies of IFN-γ⁺ Ag85B- and EsxG-specific CD4 T cells in both groups of mice (Fig. 4E, 4F), despite lower lung bacterial burdens of BCG compared with M. tuberculosis (Fig. 4H). Notably, significantly greater proportions of Ag85B- and EsxG-specific CD4 T cells expressed CD69 in lungs of BCG-infected mice (Fig. 4G). Because these results suggest that CD4 T cell responses are more efficacious in the context of BCG infection than virulent M. tuberculosis infection, we hypothesized that any difference in T cell efficacy is related to differences in interactions of CD4 T cells with infected APCs, such as dendritic cells and macrophages.

To better understand the difference in activation of Ag-specific CD4 T cells by BCG compared with M. tuberculosis, we investigated the activation of Th1-differentiated P25TCR-Tg CD4 effector T cells in vitro. At BMDC/T cell ratios of 1:10 or 1:1, P25TCR-Tg Th1 effector T cells consistently secreted significantly more IFN-γ in response to BCG-infected BMDC than to M. tuberculosis–infected BMDC (Fig. 5A). Similarly, P25TCR-Tg effector cells secreted more IFN-γ in response to BCG-infected BMMΦ compared with M. tuberculosis–infected macrophages at all APC/T cell ratios examined (Fig. 5B). BCG-infected cells were consistently superior to M. tuberculosis–infected cells as activators of P25TCR-Tg T cells when the MOI was titrated (data not shown). Thus, consistent with the in vivo proliferation results, BCG-infected BMDC and macrophages are superior to M. tuberculosis–infected cells for activating Th1 P25TCR-Tg CD4 effector cells in vitro.

One potential explanation for the difference in P25TCR-Tg CD4 T cell responses to BCG and M. tuberculosis is that BCG might express greater quantities of Ag85B. To assess this possibility, we compared Ag85B protein expression and secretion by BCG and M. tuberculosis by immunoblotting bacterial lysates and culture filtrates. This revealed that, contrary to expectation, BCG expresses less Ag85B than does M. tuberculosis H37Rv in both bacterial lysates and culture filtrates (Fig. 5C). Therefore, despite expressing lower quantities of Ag85B, BCG is able to activate P25TCR-Tg CD4 T cells more effectively than is M. tuberculosis. These data indicate that the same Ag is presented more efficiently by BCG-infected cells than by M. tuberculosis–infected cells.

Because differences in efficiency or effectiveness of T cell activation can be a function of differential Ag presentation and/or differential costimulatory or inhibitory signaling, we investigated the possibility that BCG-infected cells express higher levels of MHC class II and/or costimulatory molecules to explain their superior T cell–stimulating activity. We found that BCG and M. tuberculosis–infected BMDC expressed similar levels of MHC class II on the cell surface (Fig. 6A) and that the costimulatory molecules CD86 and CD40 and the inhibitory molecule PD-L1 were expressed at higher levels on M. tuberculosis–infected cells than on BCG-infected cells (Fig. 6A). Moreover, the difference in activation of P25TCR-Tg CD4 effector cells was abrogated when we pulsed BCG-infected or M. tuberculosis–infected BMDC (Fig. 6B) or BMMΦ (Fig. 6C) with a saturating concentration of peptide 25 (the epitope recognized by P25TCR-Tg cells). Together, these results indicate that differential costimulatory or inhibitory molecule expression on BCG- and M. tuberculosis–infected cells does not account for the differences in T cell activation observed. Instead, these results indicate that less Ag is presented by M. tuberculosis–infected cells than by BCG-infected cells.

The major finding of this study is that, compared with an attenuated mycobacterial strain (BCG), a virulent strain of M. tuberculosis (H37Rv) activates Ag-specific CD4 T cells poorly in vitro and in vivo, and this differential T cell activation is associated with clearance of BCG and persistence of M. tuberculosis. We found that BCG-infected cells are more efficient and more efficacious than M. tuberculosis–infected cells for activating Ag85B-specific CD4 T cells, despite the production of larger quantities of Ag85B by M. tuberculosis. The superior ability of BCG-infected cells to activate Ag85B-specific CD4 T cells was manifest by a lower number of BCG-infected cells needed to initiate proliferation of CD4 T cells after aerosol infection, indicating that fewer Ag-producing BCG are needed for sufficient Ag presentation. When the number of bacteria and the number of infected dendritic cells were adjusted to be equivalent and transferred intratracheally to naive mice, BCG-infected cells activated a larger proportion of adoptively transferred CD4 T cells than when M. tuberculosis–infected cells were transferred. The experimental evidence supports the conclusion that superior CD4 T cell activation is due to superior Ag presentation, because the difference in T cell activation could not be explained by differential expression of costimulatory or inhibitory molecules on the infected cells, and epitope peptide complementation abrogated the difference between BCG- and M. tuberculosis–infected cells. Importantly, superior Ag presentation and CD4 T cell activation are tied to clearance of BCG from the lungs. These findings support previously published evidence that M. tuberculosis interferes with MHC class II Ag presentation, as well as provide novel evidence that this mechanism operates in vivo and is associated with persistent infection with M. tuberculosis.

Studies of diverse pathogens, including Salmonella (41), HIV (42), and Leishmania (43), revealed that strains with higher virulence can inhibit or evade Ag processing and presentation. These examples underscore the importance of Ag presentation in the overall pathology of infectious disease, where attenuated pathogens are associated with superior CD4 T cell activation and superior control of infection. Similarly, mycobacteria were reported to be capable of limiting global MHC class II Ag processing or presentation to CD4 T cells (16, 44–46). Several studies showed that various mycobacteria can impact the presentation of exogenous Ag by macrophages (18, 19, 47). In addition, live mycobacteria were found to limit the presentation of endogenous bacterial Ags to CD4 T cells (16, 20, 48). Mycobacterial strains with different virulence (H37Rv, H37Ra, Erdman, and BCG) were used in various studies to demonstrate that members of the M. tuberculosis complex are capable of limiting Ag presentation. One study directly compared mycobacteria with different virulence to determine their relative ability to present MHC class II–restricted mycobacterial Ag (20); in this study, BCG and M. tuberculosis showed limited T cell activation compared with avirulent H37Ra, suggesting that the ability to present Ag to CD4 T cells can further stratify the virulence of mycobacteria. Our findings with BCG and M. tuberculosis reinforce the evidence that virulence in mycobacteria is associated with poor activation of Ag-specific CD4 T cells, as well as extend the findings of previous studies by revealing that poor activation of CD4 T cells by virulent M. tuberculosis also occurs in vivo. The importance of the efficacy of Ag presentation by M. tuberculosis–infected cells is emphasized by our earlier finding that direct recognition of infected cells by CD4 T cells is critical for control of intracellular M. tuberculosis in vivo (23). The in vitro observations reported in this article and those of earlier in vivo work (14) demonstrate that providing exogenous epitope peptide can enhance CD4 P25TCR-Tg cell activation, indicating that suboptimal abundance of peptide:MHC complexes, a result of limited Ag presentation, is an important determinant of CD4 T cell activation in the context of TB.

A recent study evaluated P25TCR-Tg cell activation after intradermal injection of BCG in mice. Consistent with our findings, that study found that BCG activation in the draining lymph node occurs early following intradermal infection (49). Based on our results, we propose that poor Ag presentation by M. tuberculosis–infected cells is a significant factor that limits the efficacy of CD4 T cell responses and allows persistence of the bacteria. If M. tuberculosis–infected cells present other mycobacterial Ags as poorly as they do Ag85B, it is unclear that vaccine strategies that only induce larger numbers of Ag-specific T cells will have greater efficacy than the vaccines developed to date. Our findings imply that further studies of a wide spectrum of TB Ags, as well as studies to understand the virulence mechanisms that limit Ag presentation, will be necessary to improve immunity to M. tuberculosis and contribute to overcoming the global problem of TB.

We thank the New York University Flow Cytometry and Sorting core facility for expertise and technical assistance. We thank Smita Srivastava, Ph.D. for assisting with intratracheal infection of mice and thoughtful discussion of these data. We acknowledge the National Institutes of Health Tetramer Core Facility (contract HHSN272201300006C) for provision of MHC class II tetramers. We also thank members of the Ernst laboratory for thoughts and suggestions for this work.

The authors have no financial conflicts of interest.