Cutting Edge: Bacillus Calmette–Guérin–Induced T Cells Shape Mycobacterium tuberculosis Infection before Reducing the Bacterial Burden

Bacillus Calmette–Guérin (BCG), the current tuberculosis (TB) vaccine, is effective at preventing disseminated disease in infants and young children (1). However, in most settings it provides little or no protection against adult pulmonary TB, the transmissible form of disease (2). Thus, despite widespread BCG immunization for nearly a century, Mycobacterium tuberculosis kills over 1.5 million people every year, more than any other single infectious agent (3). A better TB vaccine is urgently needed but attaining this goal has been surprisingly difficult (4). Furthermore, because BCG reduces childhood mortality, a new vaccine will likely be added to a regimen that includes BCG, rather than replace it (5). To develop a strategy that builds upon BCG-mediated protection, we must first understand how BCG shapes immunity to M. tuberculosis, especially during early stages of infection when protective immunity is established.

In mice, pulmonary M. tuberculosis burdens are equivalent between BCG-immunized and control mice until 2 weeks postinfection (6). The failure of BCG to impact the M. tuberculosis burden during the first 2 weeks of infection has been attributed to the delayed arrival of T cells in the lung (7). However, BCG-specific T cells have been shown to be present in the lungs (8) of immunized mice even prior to M. tuberculosis challenge, indicating that impaired T cell recruitment cannot fully account for the inability of BCG to induce early protection.

In this study, we used the mouse model to investigate the impact of BCG on the early immune response to M. tuberculosis infection. Our findings reveal unexpected roles for CD4 T cells in the following areas: 1) accelerating the translocation of M. tuberculosis–infected alveolar macrophages (AM) into the lung interstitium; 2) recruiting monocyte-derived macrophages (MDM); and 3) promoting the early transfer of M. tuberculosis from AM to other phagocytes. Despite these effects, a vaccine-induced reduction in the lung bacterial burden does not occur until colocalization of CD4 T cells with infected macrophages, which is delayed until 2 weeks postinfection, even in vaccinated animals.

ESAT-6 TCRtg (C7) mice were provided by Eric Pamer (9) and have been described previously. OTII mice were provided by Michael Gerner (University of Washington). C57BL/6 and MHC class II (MHCII)^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All mice were housed in specific pathogen–free conditions at Seattle Children’s Research Institute. Experiments were performed in compliance with the Seattle Children’s Research Institute Animal Care and Use Committee. Both male and female mice between the ages of 8–12 wk were used.

BCG-Pasteur was cultured in Middlebrook 7H9 broth at 37°C to an OD of 0.2–0.5. Bacteria was diluted in PBS and 1 × 10⁶ CFU in 200 μl was injected s.c. After immunization, mice were rested for 8 wk prior to M. tuberculosis infection.

Infections were performed with wild-type (WT) H37Rv M. tuberculosis or H37Rv transformed with an mCherry reporter plasmid (10). Mice were enclosed in a Glas-Col Inhalation Exposure System, and 50–100 CFU were deposited directly into the lungs.

For intratracheal labeling, 30 min prior to sacrifice, mice were anesthetized with 20% isoflurane in propylene glycol (Thermo Fisher Scientific) and 0.25 μg of CD45.2 PE-Cy7 in 50 μl of PBS and was pipetted into the airway. For i.v. labeling, mice were anesthetized as above and infused with CD45.2 PE 10 min prior to sacrifice.

Mouse lungs were homogenized in HEPES buffer with Liberase Blendzyme 3 (70 μg/ml; Roche) and DNaseI (30 μg/ml; Sigma-Aldrich) using a gentleMacs dissociator (Miltenyi Biotec). Lungs were incubated at 37°C for 30 min and then further homogenized with the gentleMacs. Cells were filtered through a 70-μm cell strainer and resuspended in RBC lysis buffer (Thermo Fisher Scientific) prior to a PBS wash. Cells were next incubated with 50 μl of Zombie Aqua viability dye (BioLegend) for 10 min at room temperature (RT). Viability dye was quenched with 100 μl of Ab mixture in 50% FACS buffer (PBS containing 2.5% FBS and 0.1% NaN₃)/50% Fc-block buffer. Staining was performed for 20 min at 4°C. Cells were washed with FACS buffer and fixed with 2% paraformaldehyde for 1 h prior to analysis on an LSR II flow cytometer (BD Biosciences). When stain sets contained tetramers, staining was performed for 1 h at RT. Ag85B (I-A(b) 280-294) and TB10.4 (K(b) 4-11) tetramers were obtained from the National Institutes of Health Tetramer Core Facility. Intracellular phospho-S6 (Ser235/236 2F9; Cell Signaling Technology) staining was performed for 1 h at RT with the FoxP3/Transcription Factor Staining Kit (Invitrogen).

Mice were infected with H37Rv M. tuberculosis–mCherry and sacrificed at day 10 and day 14. Lungs were excised and submerged in BD Cytofix fixative solution diluted 1:3 with PBS for 24 h at 4°C. Lungs were washed two times in PBS and dehydrated in 30% sucrose for 24 h prior to OCT embedding and rapid freezing in a methylbutane–dry ice slurry. Twenty-micrometer sections were stained overnight at RT (with the following Abs: CD11c BV480; HL3 [BD Biosciences], CD11b BV510; M1/70 [BD Biosciences], CD45.2 Alexa Fluor 700; 104 [BioLegend], MHCII allophycocyanin-Fire750; M5/114.15.2 [BioLegend], Siglec-F BV421; E50-2440 [BD Biosciences], p-S6 Alexa Fluor 488; 2F9 [Cell Signaling Technology], CD3 CF633; 17A2, CD4 CF660C; RM4-5) and coverslipped with Fluoromount G mounting media (Southern Biotech). Images were acquired on a Leica SP8 confocal microscope, compensated for fluorophore spillover using LAS X (Leica) and analyzed with Imaris (Bitplane) and FlowJo (11). More details can be found in Supplemental Fig. 3.

To deplete T cells, mice were i.p. injected with 400 μg of anti-CD4 GK1.5 or anti-CD8 2.43 (Bio X Cell) in PBS at day 1, day 4, and day 10 relative to infection. To block T cell egress, mice were i.p. injected with 1 mg/kg FTY720 (Sigma-Aldrich) in water daily, starting 2 d prior to harvest.

WT CD45.1/2 F1 mice were irradiated with 1000 rad and reconstituted with a 1:1 mixture of CD3-depleted (Miltenyi Biotec) CD45.1 B6.SJL:CD45.2 MHCII^−/− bone marrow. At day 56 after reconstitution, mice were immunized with BCG.

CD4 T cells from ESAT-6–specific (C7) CD90.1⁺ and OVA-specific (OTII) CD45.1⁺ TCR transgenic mice were negatively enriched from spleens using EasySep magnetic microbeads (STEMCELL Technologies). T cells were Th1 polarized as follows: 1.6 × 10⁶ transgenic T cells were cultured with 8.3 × 10⁶ irradiated CD3⁻ splenocytes. Five micrograms per milliliter of ESAT-6 or OVA peptide, 10 ng/ml IL-12, and 10 μg/ml of anti–IL-4 Ab (R&D Systems) were added at day 0. At day 3, cells were split 1:2, and 10 ng/ml IL-12 was added (R&D Systems). On day 5, Th1 cells were i.v. injected into mice infected with M. tuberculosis 10 d prior (Fig. 3C) or 35 d prior (Supplemental Fig. 2A).

To understand the effects of BCG immunization on early M. tuberculosis infection, we examined the pulmonary M. tuberculosis burdens in BCG-immunized and control mice. Consistent with prior reports (6, 7), lung burdens rose similarly in both groups through 2 wk (Fig. 1A). At day 15, the M. tuberculosis burden in the immunized group began to diverge and was reduced by one log by day 21. These findings are consistent with the idea that BCG-induced immunity is not initiated until the third week of M. tuberculosis infection.

We recently found that M. tuberculosis first infects AM before disseminating to other cells, including polymorphonuclear neutrophils (PMN) and MDM (12). As tissue-resident and recruited phagocytes have been shown to differ in their capacity to curb M. tuberculosis replication (13, 14), we next asked whether immunization alters the proportions of cell types that harbor infection. Consistent with the similar M. tuberculosis burdens at day 14, the numbers of cells harboring fluorescent M. tuberculosis (M. tuberculosis–mCherry) were also similar in each group (Fig. 1B). Surprisingly, even at this early phase, we observed a dramatic shift in the composition of infected cells. At day 14, by gating on mCherry⁺ cells, we found that the proportion of M. tuberculosis–infected AM was significantly reduced in immunized animals compared with controls, with a corresponding increase in infected PMN and MDM (Fig. 1C, 1D). There was no effect on the infection of dendritic cells (Supplemental Fig. 1A). We confirmed these findings using confocal microscopy and quantitative histocytometry (11), wherein most M. tuberculosis was within Siglec-F⁺ AM at day 14 in controls but within CD11b⁺ Siglec-F⁻ cells (primarily PMN and MDM) in immunized mice (Fig. 1E, 1F). As M. tuberculosis dissemination to PMN and MDM requires translocation of infected AM to the lung interstitium (12), we next assessed whether this translocation was accelerated in immunized mice. Indeed, intratracheal Ab administration, which specifically labels alveolar-localized cells (12), revealed significantly increased interstitial localization (label-negative) of infected AM in immunized mice at day 14 (Fig. 1G). Importantly, the changes in infected cell types was independent of changes in AM or PMN cellularity, as we found no differences in the numbers of these cells following immunization or infection (Supplemental Fig. 1B). Finally, immunization significantly enhanced MDM recruitment to the lung at day 14 (Fig. 1H), which was not observed at earlier time points or prior to infection (Supplemental Fig. 1B), suggesting that the accelerated recruitment of MDM in immunized mice begins between day 10 and day 14. Thus, although BCG does not impact the pulmonary M. tuberculosis burden in the first 2 wk of infection, it accelerates the translocation of infected AM from alveoli to the lung interstitium, MDM recruitment, and M. tuberculosis dissemination to PMN and MDM.

This unexpected impact of BCG on the early dynamics of infection led us to next investigate how immunization affects the kinetics of T cell recruitment to the lung. Before infection, Ag-specific CD4 (Ag85B) and CD8 (TB10.4) T cells could be identified in lung cell suspensions of immunized mice (Fig. 2A–C). Although ∼25% of the Ag85B-specific CD4 T cells were located in the lung parenchyma (as evidenced by their failure to stain with i.v. CD45 Ab), virtually all of the TB10.4-specific CD8 T cells resided in the vasculature (Fig. 2D). This difference is consistent with a recent report that lung CD4 T resident memory cells may be maintained for longer than CD8 T cells (15). Following infection, immunized mice had significantly more TB10.4-specific cells, as well as Ag85B-specific CD4 T cells, in the lung parenchyma than controls as early as day 10; by day 14 they contained >5-fold more (Fig. 2B, 2C). Thus, BCG induces a small population of lung-resident M. tuberculosis–specific CD4 T cells prior to infection. Postinfection, BCG accelerates the pulmonary accumulation of both CD4 and CD8 M. tuberculosis–specific T cells even before impacting the M. tuberculosis burden.

Given the presence of lung-resident M. tuberculosis–specific T cells in immunized mice prior to infection, we next determined whether T cells play a role in the accelerated transfer of M. tuberculosis from AM to other myeloid cells. CD4 or CD8 T cells were depleted from immunized mice beginning 1 d prior to M. tuberculosis–mCherry infection, and lung cells were assessed at day 14 (Supplemental Fig. 1C, 1D). In the absence of CD4 T cells, the accelerated transfer of M. tuberculosis from AM to PMN and MDM was partially reversed, whereas CD8 T cell depletion had no effect (Fig. 3A). Interestingly, the accelerated MDM recruitment (Fig. 1H) was also abolished by CD4 depletion (Fig. 3B). To assess whether antigenic recognition by CD4 T cells even in the absence of vaccination was sufficient to transfer M. tuberculosis infection from AM to other phagocytes, we adoptively transferred transgenic M. tuberculosis ESAT-6–specific CD4 T cells (C7), or OVA-specific CD4 T cells (OTII) as a control, into unimmunized mice that were infected with M. tuberculosis 10 d prior. Four days later (day 14) we observed that ESAT-6–specific but not OVA-specific T cells were sufficient to induce the transfer of infection out of AM (Fig. 3C), suggesting that Ag-specific CD4 T cell activation is needed to mediate this effect. Finally, we investigated whether direct recognition of M. tuberculosis–infected cells by CD4 T cells was required for the early dissemination out of the AM niche. WT (CD45.1):MHCII^−/− (CD45.2) mixed bone marrow chimeras were generated, BCG immunized, and infected with M. tuberculosis–mCherry to test whether MHCII^−/− AM, which cannot present Ag to CD4 T cells, would retain M. tuberculosis longer than WT AM. At day 14, BCG induced the accelerated transfer of M. tuberculosis from AM to other myeloid cells irrespective of intrinsic MHCII expression (Fig. 3D). Taken together, the results indicate that BCG-induced CD4 T cells promote the early transfer of M. tuberculosis from AM to other myeloid cells in a process that seems to require antigenic recognition but does not require direct cognate interactions between T cells and M. tuberculosis–infected AM. Our finding that CD4 T cells promote MDM recruitment to the lung, thereby providing new bacterial targets, may help explain the increased proportion of infected MDM in immunized animals. This recruitment likely relates to T cell production of cytokines, such as IFN-γ and TNF, which are known to trigger the release of chemokines that act on MDM (e.g., CCL2 and CXCL10) (16).

Given that CD4 T cells seem to induce the transfer of infection independent of direct interactions with infected cells, we next sought to monitor T cell activation in the lung using phospho-S6 (p-S6), a ribosomal protein that is rapidly phosphorylated after TCR engagement and can be detected both by flow cytometry and confocal microscopy. Although previous work has shown that p-S6 specifically marks T cells that have recently engaged their TCR (peaking at 4 h and resolving within 24 h) under homeostatic conditions (17), we first confirmed this specificity in the context of M. tuberculosis–infected lungs by showing robust p-S6 expression by adoptively transferred TCR transgenic M. tuberculosis–specific (ESAT-6; C7) CD4 T cells compared with irrelevant TCR transgenic T cells (OVA-specific) (Supplemental Fig. 2A). Next, we examined p-S6 in the endogenous, polyclonal CD4 T cell population to monitor the kinetics and location of TCR signaling during early M. tuberculosis infection. Although there were very few p-S6⁺ CD4 T cells in the lungs of either control or immunized mice prior to M. tuberculosis challenge, as measured by flow cytometry (Supplemental Fig. 2B), p-S6⁺ CD4 T cells were readily identified in the vasculature of both unimmunized and immunized mice by day 10 and were present in higher numbers in the lung parenchyma of immunized mice compared with controls (Fig. 4A, Supplemental Fig. 2C). To assess the intrapulmonary location of these p-S6⁺ cells, we performed confocal imaging and quantitative histocytometry (Supplemental Fig. 3). Consistent with our flow cytometry data, there were significantly more p-S6⁺ cells in lung sections from immunized mice at day 10 (Fig. 4B, 4C), but, surprisingly, few of these cells were located near infected cells (Fig. 4B, 4G). Thus, although BCG induces early T cell recruitment and activation, at day 10 this occurs primarily in uninfected areas of the lung. This may be due to M. tuberculosis Ag export from infected to uninfected APCs (18) but could represent T cells that have recently trafficked from the lymph node (LN). To address this, we treated mice with FTY720, which blocks lymphocyte egress from lymphatic tissue, for 48 h prior to analysis (Supplemental Fig. 2D). Although FTY720 treatment almost completely eliminated p-S6⁺ CD4 T cells in the lung vasculature at day 10, the number of p-S6⁺ T cells in the lung parenchyma (i.v.⁻) was only partially reduced (Fig. 4D). These results suggest that most of the recently activated T cells in the vasculature and a subset of those in the lung parenchyma had recently egressed from the LN. However, some of the p-S6⁺ cells in the lung parenchyma at day 10 were likely activated in the lungs. Taken together, these results indicate that the activation of BCG-induced CD4 T cells, which occurs distal to sites of infection, shapes immunity to M. tuberculosis challenge earlier than previously appreciated by facilitating the pulmonary recruitment of MDM and accelerating the transfer of M. tuberculosis from AM to other myeloid cells. This transfer likely influences the ability of the BCG-immunized host to control M. tuberculosis, as prior studies have shown that tissue-resident versus recruited macrophages differ profoundly in their capacity to control M. tuberculosis replication (13, 14). Future studies are needed to elucidate the overall impact on protection because the settings in which distinct macrophage types mediate enhanced immunity remain unclear.

Interestingly, BCG-induced CD4 T cells only begin to curb M. tuberculosis replication at day 14, when they finally colocalize with cells harboring M. tuberculosis, as evidenced by the identification of many p-S6⁺ cells in the lung parenchyma and at sites of infection compared with controls (Fig. 4E–G). This is consistent with the finding that optimal immunity against M. tuberculosis requires direct interactions between Ag-specific CD4 T cells and infected cells (19). Importantly, in contrast to our findings at day 10, at day 14, FTY720 had no effect on the number of p-S6⁺ CD4 T cells recovered from the lung parenchyma of BCG-immunized mice (Fig. 4H), suggesting that almost all of these T cells were activated in the lungs rather than the LN. Why do T cells and M. tuberculosis–infected cells not colocalize earlier? The AM is the first cell type to become infected and remains the primary infected cell type for at least a week (12). During this time, the immune system appears largely unaware of the looming threat, as few MDM or PMN are recruited to the lung. The recent finding that AM infection is noninflammatory and poorly induces chemokines may help explain the covert nature of early infection (21). Identifying vaccination approaches that enable T cells to colocalize with M. tuberculosis–infected AM may promote earlier M. tuberculosis control. Together, these results demonstrate that BCG immunization shapes early T cell and myeloid cell responses in the lung and new vaccine strategies should consider the dynamics of both these compartments.

We thank Peter Andersen, Joshua Woodworth, and Rasmus Mortensen for critical feedback of the manuscript as well as Bridget Alexander, the flow cytometry core, and vivarium staff for technical assistance.

The authors have no financial conflicts of interest.