Cutting Edge: Control of Mycobacterium tuberculosis Infection by a Subset of Lung Parenchyma–Homing CD4 T Cells

Mycobacterium tuberculosis is a major contributor to global human morbidity and mortality, with 8.6 million new cases of disease and 1.3 million deaths annually (1). The only available vaccine, Bacillus Calmette-Guérin, protects against disseminated tuberculosis (TB) in children, but it confers little or no protection against pulmonary disease in adults. Indeed, the development of novel vaccines for M. tuberculosis has proven very difficult (2).

CD4 T cell–deficient HIV-infected individuals, mice, and nonhuman primates depleted of CD4 T cells (3–5), humans with inborn errors in genes involving IFN-γ signaling, and mice deficient in IFN-γ signaling or T-bet are all extremely susceptible to M. tuberculosis infection (6–8), indicating that Th1-polarized effector responses play a central role in host resistance to TB. However, there is also evidence that CD4 T cells can mediate control of M. tuberculosis in an IFN-γ–independent manner (9, 10), and IFN-γ responses do not predict host resistance to M. tuberculosis infection (11). In fact, there are no validated correlates of protection against TB, and there is a great need for a better understanding of the properties of a protective host response against M. tuberculosis infection.

Recently, it has become clear that, following the resolution of acute viral infection, peripheral nonlymphoid tissues harbor a subset of nonrecirculating, tissue-resident CD4 and CD8 T cells that are distinct from memory T cells in secondary lymphoid tissue (12–14). However, it is not clear whether recirculating and tissue-localizing subsets of effector CD4 T cells exist in the context of chronic M. tuberculosis infection. In this study, we show that two types of Th1 cells with different phenotypic, migratory, and host-protective capacities populate the lung parenchyma and vasculature. These results identify a subpopulation of the M. tuberculosis–specific CD4 T cell response with enhanced lung-homing ability as a key cell type in control of pulmonary M. tuberculosis infection.

C57BL/6, B6.SJL (CD45.1), and TCRα^−/− mice were obtained from Taconic Farms (Germantown, NY). All animals were housed at the American Association for the Accreditation of Laboratory Animal Care–approved facility at the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health, according to the National Research Council Guide for the Care and Use of Laboratory Animals. Mice were used according to an animal study proposal approved by the NIAID Animal Care and Use Committee. Mice were aerosol exposed to ∼100 CFU the H37Rv strain of M. tuberculosis (Glas-Col, Terre Haute, IN).

Mice were injected i.v. with 2.5 μg fluorochrome-labeled anti-CD45.2 or anti-CD45 Ab, and after 3 min, PBLs, bronchoalveolar lavage fluid, and lungs were harvested. For direct ex vivo intracellular cytokine staining (ICS), lungs were processed in the presence of brefeldin A (eBioscience, San Diego, CA). For T cell stimulations, cells were incubated with 5 μg/ml ESAT-6_1–20 peptide. I-A^bESAT-6_4–17 and I-A^bEsxG_46–61 MHC tetramers were produced by the NIAID Tetramer Core Facility (Emory University, Atlanta, GA).

M. tuberculosis–infected CD45.1 congenic mice were injected i.v. with anti-CD45–PE on day 30 postinfection (p.i.), and live CD45⁺ or CD45⁻ CD4 T cells were sorted to >98% purity with a FACSAria II (BD Biosciences). For migration experiments, ∼5 × 10⁵ cells of each population were transferred into infection-matched CD45.2 congenic recipient mice. For the protection experiments, TCRα^−/− mice that had been infected with M. tuberculosis 7 d before adoptive transfer were used as recipients.

To determine the distribution of CD4 T cells among the airways, tissue parenchyma, and blood vasculature within the lungs during M. tuberculosis infection, we used a well-established intravascular (iv) staining technique (15, 16) (Fig. 1A). As expected, CD4 T cells in the PBLs all stained with the injected Ab, and cells in the bronchoalveolar lavage fluid were uniformly negative (Fig. 1B). In contrast, CD4 T cells in the lung were clearly divided into iv stain–negative (iv⁻) and iv stain–positive (iv⁺) populations, corresponding to cells in the lung parenchyma and vasculature, respectively. The anatomical localization of CD4 T cells was confirmed by immunofluorescence microscopy, which showed that cells staining with the injected anti-CD45 Ab were exclusively located within the CD31⁺ blood vessels (Fig. 1C).

We next examined the kinetics of CD4 T cell recruitment into the lungs after M. tuberculosis infection. Very low numbers of CD4 T cells were recovered from the lungs of naive mice, and <5% were iv⁻ (Fig. 1D). At the peak of CD4 T cell numbers at day 30, ∼45% of lung CD4 T cells were in the parenchyma. The percentage of CD4 T cells that was iv⁻ continued to increase until reaching a plateau of ∼65% at day 60, which remained stable for ≥6 mo p.i. Interestingly, the frequency of I-A^bESAT-6_4–17– and I-A^bEsxG_46–61–specific CD4 T cells was highest in the vasculature (Fig. 1E). Moreover, the frequency of Ag-specific CD4 T cells in the lung vasculature was >5-fold higher than in the circulating blood, indicating that iv⁺ T cells are not simply “blood contamination.” We also found that M. tuberculosis–specific CD8 T cells were most enriched in the lung vasculature (Supplemental Fig. 1). Together, these data indicate that significant populations of M. tuberculosis–specific CD4 T cells are located in both the lung parenchyma and vasculature throughout the course of M. tuberculosis infection.

We next compared the function of lung iv⁻ and iv⁺ M. tuberculosis–specific CD4 T cells. Similar to the increased frequencies of tetramer⁺ cells (Fig. 1E), a higher percentage of iv⁺ cells produced IFN-γ compared with iv⁻ cells after in vitro stimulation with ESAT-6_1–20 peptide (Fig. 2A). To compare cytokine production by iv⁻ and iv⁺ Ag-specific cells in vivo more directly, lung lymphocytes were isolated in the presence of brefeldin A, and IFN-γ production by CD4 T cells was assessed by direct ex vivo ICS. Surprisingly, we observed ∼2-fold more IFNγ⁺ I-A^bESAT-6_4–17–specific CD4 T cells in the lung vasculature compared with the parenchyma by direct ICS (Fig. 2B), and iv⁺ M. tuberculosis–specific CD4 T cells also expressed higher levels of T-bet (Fig. 2C). When accounting for the increased overall number of CD4 T cells in the vasculature and their higher cytokine-producing activity, ∼75% of the total direct ex vivo IFNγ⁺ CD4 T cells in the lung were in the vasculature at 4 wk p.i. (data not shown). These data indicate that, at the peak of T cell clonal expansion, the majority of the active Th1 response against M. tuberculosis infection (as measured by IFN-γ production) occurs in the blood vasculature and not in the parenchyma.

We next compared the phenotype of I-A^bESAT-6_4–17–specific CD4 T cells in both lung compartments. We found that iv⁻ M. tuberculosis–specific T cells exhibited much higher levels of activation markers, including CD69 and PD-1 (Fig. 2D, Supplemental Fig. 2), whereas iv⁺ cells expressed a high level of KLRG1, which has been associated with terminal effector T cells (17). Among the chemokine receptors analyzed (Supplemental Fig. 2), strikingly, the majority of iv⁻ CD4 T cells showed a high level of CXCR3 expression, whereas most of the iv⁺ cells expressed CX3CR1 (Fig. 2E). I-A^bESAT-6_4–17–specific CD4 T cells in the iv⁻ and iv⁺ compartments displayed the same preferential expression of CXCR3 and CX3CR1, respectively (Fig. 2E). Collectively, these data demonstrate a phenotypic dichotomy in M. tuberculosis–specific CD4 T cells whereby CXCR3 marks highly activated, parenchyma-localized cells, and CX3CR1 and KLRG1 identify cells capable of the highest IFN-γ production that are enriched within the lung blood vessels.

We next compared the ability of the parenchymal and iv subsets to migrate into M. tuberculosis–infected lungs by transferring FACS-purified iv⁻ or iv⁺ CD4 T cells into infection matched congenic recipient mice. We found that ∼60% of the I-A^bESAT-6_4–17–specific CD4 T cells derived from the iv⁻ donors efficiently migrated back into the lung parenchyma, whereas only ∼5% of the iv⁺ donors migrated into the lung tissue in the same time period (Fig. 3A). A similar difference was observed between iv⁻ and iv⁺ total effector (CD44^hiFoxp3⁻) donor CD4 T cells (Fig. 3B, 3C). It was shown that naive T cells can migrate into M. tuberculosis–infected lungs (18); thus, we also analyzed the ability of naive (CD44^loFoxp3⁻) CD4 T cells to migrate into the lungs of M. tuberculosis–infected recipients. Naive T cells derived from either the parenchyma or vasculature migrated equally into the lung parenchyma after adoptive transfer (Fig. 3B, 3C). Interestingly, naive T cells migrated more efficiently into the lung than did the iv⁺ effector T cells and were even comparable to the parenchymal donor effector T cells. Together, these data indicate that the iv effector CD4 T cells are unexpectedly poor at migrating into the lung parenchyma.

Although the vast majority of the iv⁺ effector CD4 T cells were KLRG1⁺, we routinely observed that a small subpopulation of KLRG1⁻ iv⁺ effector cells expressed CXCR3 at levels similar to the iv⁻ effectors (Fig. 2D, 2E). Thus, we next asked whether the iv KLRG1⁺ and KLRG1⁻ effector CD4 T cells differed in their ability to migrate into the parenchyma by gating on the subsets after recovery from the recipients. We found that the KLRG1⁻ iv⁺ donor cells migrated efficiently into the lung parenchyma upon adoptive transfer (Fig. 3B, 3C). In contrast, the KLRG1⁺ iv⁺ donor cells were extremely poor at migrating into the tissue. Collectively, these data show that the anatomical localization of pulmonary effector T cells largely reflects the cell’s migratory ability.

We compared the ability of lung iv⁻ and iv⁺ CD4 T cells to control bacterial replication upon adoptive transfer into infected recipients. Because it is often difficult to observe protection of M. tuberculosis–infected mice by transfer of CD4 T cells into normal recipients, we chose to use T cell–deficient mice as hosts. Although the use of lymphopenic mice is a caveat of this experiment, it allowed us to observe large effects of transferred CD4 T cells on bacterial control. FACS-purified iv⁻ or iv⁺ CD4 T cells isolated from the lungs at 30 d p.i. were injected into TCRα^−/− mice that had been infected with M. tuberculosis 7 d previously, and the recipients were euthanized on day 28 p.i. A total of 5 × 10⁵ CD4 T cells was transferred, which corresponded to 3.2 × 10⁵ iv⁻ and 3.3 × 10⁵ iv⁺ CD44^hiFoxp3⁻ effector cells and 2.2 × 10⁴ iv⁻ and 2.1 × 10⁴ iv⁺ I-A^bESAT-6_4–17–specific T cells. We found that donor CD4 T cells initially isolated from either the lung parenchyma or vasculature of the donor mice gave rise to both iv⁻ and iv⁺ cells in the recipient lungs (Fig. 4A). The total number of donor CD44^hiFoxp3⁻ effector and I-A^bESAT-6_4–17–specific CD4 T cells in the parenchyma was ∼2-fold higher in the lungs of mice reconstituted with iv⁻ cells compared with iv⁺ donors (Fig. 4A). Although iv⁻ donor cells repopulated the parenchyma more efficiently, iv⁺ donors produced higher levels of IFN-γ in the direct ICS assay (Fig. 4B). This increased cytokine secretion by iv⁺ donor cells was observed when comparing either the expanded iv⁻ or iv⁺ cells.

This experimental design allowed us to compare bacterial loads in the setting of higher parenchymal effector cell numbers versus higher in vivo IFN-γ production. Both groups of recipient mice showed reduced bacterial loads in the lungs compared with the unreconstituted mice (Fig. 4C). However, TCRα^−/− mice injected with iv⁺ donor cells showed an ∼4-fold reduction in bacterial loads, whereas the recipients of iv⁻ donor cells showed an ∼18-fold decrease in CFU compared with the unreconstituted controls (Fig. 4C). Therefore, CD4 T cell–mediated protection against M. tuberculosis infection was associated with a higher number of CD4 T cells in the lung parenchyma rather than a higher frequency of IFNγ⁺ cells in the vasculature.

In this study, we demonstrate that the pulmonary CD4 T cell response against M. tuberculosis is composed of two major subsets that either enter the lung parenchyma or reside within the vasculature. The parenchymal effectors express CXCR3 and are PD-1^hi/CD69^hi, which likely reflects their access to Ag within the tissue. In contrast, most of the iv CD4 T cells express CX3CR1 and have a more terminally differentiated phenotype (KLRG1^hi/T-bet^hi). A recent study (19) noted that, in M. tuberculosis–infected mice, PD-1^hi CD4 T cells are highly proliferative, whereas KLRG1^hi cells produce more IFN-γ. Our results indicate that these differences that were reported previously between PD-1^hi and KLRG1^hi CD4 T cells reflect the phenotype of T cells within the parenchyma and blood vessels, respectively, highlighting the importance of this iv-staining technique in the study of cellular immune responses in M. tuberculosis–infected lungs. Indeed, many previous studies of cellular immunity to M. tuberculosis may warrant re-evaluation with this technique.

We also found that parenchymal CD4 T cells migrate rapidly back into the parenchyma, whereas only a small population of iv cells does so in the same amount of time. Although KLRG1^hi iv T cells are the most abundant subset in the entire lung at the peak of clonal expansion, they are poor at entering the lung parenchyma (even compared with naive T cells). In contrast, a small subset of iv T cells negative for KLRG1 efficiently migrates into the lung. This indicates that parenchymal precursor cells within blood vessels may be a small population of effector T cells with enhanced lung-migratory activity that are distinct from the majority of iv cells that have very poor ability to enter the lung. However, further investigation is needed to fully characterize the relationship between the two subsets.

Although the lung-homing subset of CD4 T cells mediates the most efficient control of M. tuberculosis infection, the majority of active IFN-γ secretion comes from CD4 T cells in the lung vasculature. It is unclear to what extent Ag recognition and innate cytokine signals contribute to IFN-γ production by iv CD4 T cells, but it raises the interesting possibility that iv T cells could have some unrecognized contribution to bacterial control or regulation of other cell types during M. tuberculosis infection. It is also not clear why the parenchymal CD4 T cells make less IFN-γ compared with the T cells in the blood vessels. The parenchymal effector cells could be actively inhibited by signals in the tissue, such as PD-1, or the parenchymal and iv T cells may represent distinct lineages of Th1 cells with different effector programs. These two scenarios are not mutually exclusive. Regardless, despite their reduced IFN-γ production in comparison with iv cells, the parenchymal effector CD4 T cells did produce some IFN-γ, and perhaps this amount was sufficient to mediate the IFN-γ–dependent effects on bacterial control. A recent study (20) showed that the major role of CD4 T cells in control of M. tuberculosis infection is dependent on direct contact between the CD4 T cells and infected APCs. Therefore, it is likely that the enhanced protective capacity of the parenchymal-homing subset is principally due to its ability to gain access to the infected cells in the granuloma; conversely, the relative ineffectiveness of the CX3CR1⁺ subset, despite its enhanced IFN-γ production, is simply due to its inability to enter the lung tissue. Interestingly, it is also possible that the parenchyma-homing subset mediates the still-unidentified IFN-γ–independent effector functions that are able to induce control of M. tuberculosis infection (9, 10).

Lastly, the findings reported in this article might have implications for understanding the correlates of protection against M. tuberculosis infection. Our observations that the CD4 T cells producing the highest level of IFN-γ are not the same as the cells that efficiently enter the lung and control infection may help to explain why IFN-γ responses notoriously do not correlate with resistance to M. tuberculosis. These data illustrate how a better understanding of the heterogeneity of M. tuberculosis–specific CD4 T cells could provide rationale for the development of novel therapies and vaccine regimens for TB by identifying the properties of protective CD4 T cell responses.

We thank Bishop Hague and Kevin Holmes of the NIAID Flow Cytometry Core Facility for FACS sorting.

The authors have no financial conflicts of interest.