Antigen-Specific CD8⁺ T Cells and the Development of Central Memory during Mycobacterium tuberculosis Infection

Abstract

Whether true memory T cells develop in the face of chronic infection such as tuberculosis remains controversial. To address this question, we studied CD8⁺ T cells specific for the Mycobacterium tuberculosis ESAT6-related Ags TB10.3 and TB10.4. The shared epitope TB10.3/10.4_20–28 is presented by H-2 K^d, and 20–30% of the CD8⁺ T cells in the lungs of chronically infected mice are specific for this Ag following respiratory infection with M. tuberculosis. These TB10.3/10.4_20–28-specific CD8⁺ T cells produce IFN-γ and TNF and express CD107 on their cell surface, which indicates their likely role as CTL in vivo. Nearly all of the Ag-specific CD8⁺ T cells in the lungs of chronically infected mice had a T effector cell phenotype based on their low expression of CD62L and CD45RB. In contrast, a population of TB10.3/10.4_20–28-specific CD8⁺ T cells was identified in the lymphoid organs that express high levels of CD62L and CD45RB. Antibiotic treatment to resolve the infection led to a contraction of the Ag-specific CD8⁺ T cell population and was accompanied by an increase in the proportion of CD8⁺ T cells with a central memory phenotype. Finally, challenge of memory-immune mice with M. tuberculosis was accompanied by significant expansion of TB10.3/10.4_20–28-specific CD8⁺ T cells, which suggests that these cells are in fact functional memory T cells.

Our ability to contain, control, and in some cases eliminate human disease caused by pathogenic bacteria and viruses has been augmented by the capacity to elicit protective immunity by vaccinating susceptible hosts with attenuated or inactivated microbes. Unfortunately, this approach has not been completely successful against Mycobacterium tuberculosis. Although bacillus Calmette-Guerin (BCG),⁴ which is an attenuated form of the closely related species Mycobacterium bovis, has been used as a vaccine for nearly a century, there is little evidence that this strategy has had a significant impact on the control of human tuberculosis. Consequently, expensive multidrug treatment regimens are required for the 8–10 million cases of tuberculosis each year. Many innovative strategies are being used to develop an effective vaccine for M. tuberculosis. An important adjunct to this effort is gaining a better understanding of natural immunity to tuberculosis. After all, whereas the efficacy of M. bovis BCG for the prevention of pulmonary tuberculosis remains controversial, it is well documented that nearly all healthy individuals mount a sustained immune response following M. tuberculosis infection that is able to control bacterial replication.

An impediment to developing an optimum prophylactic vaccine is an incomplete understanding of which T cell subsets mediate protection and which mycobacterial Ags are the targets of antimicrobial immunity. To rigorously develop a vaccine for M. tuberculosis, one needs to demonstrate that immunization induces appropriate clonal activation of specific T cells, which, upon bacterial challenge, can rapidly and specifically recognize infected cells. Without such evidence, failure of any particular microbial Ag to induce protective immunity could simply reflect the technical difficulty in eliciting an appropriate T cell response, or that the Ag is not efficiently presented by infected cells.

The identification of peptide epitopes that are specifically recognized by class I MHC-restricted CD8⁺ T cells elicited during M. tuberculosis infection has made possible a more detailed assessment of the function of CD8⁺ T cells in host resistance to tuberculosis. For example, identifying the CFP10_32–39 epitope enabled us to show that CFP10-specific CD8⁺ T cells, which accumulate in the lungs of mice following infection, have cytolytic activity in vivo (1). Elucidating the minimal epitopes presented by class I MHC allows accurate enumeration and characterization of Ag-specific CD8⁺ T cells primed during infection. Such data can be used to design peptide-loaded class I MHC tetramers that can track Ag-specific CD8⁺ T cells elicited by infection by flow cytometry (1). This approach can provide a direct measurement of vaccine efficacy independent of host protection, which ultimately will allow a direct comparison of vaccine strategies (2).

Majlessi et al. (3) described an epitope found in two ESAT6-related proteins, TB10.3 (Rv0288) and TB10.4 (Rv3019c), which is recognized by H-2 K^d-restricted CD8⁺ T cells following infection with BCG or M. tuberculosis. We have used this epitope, hereafter referred to as TB10.4_20–28, and TB10.4_20–28-loaded H-2 K^d tetramers to characterize the distribution and function of TB10.4_20–28-specific CD8⁺ T cells following respiratory M. tuberculosis infection. Importantly, we observed that Ag-specific CD8⁺ T cells with features of central memory T cells (T_CM) develop even during chronic infection. By treating mice with antibiotics to resolve the infection, we were able to determine how the bacterial burden affects memory T cells specific for mycobacterial Ags. Finally, rechallenge of antibiotic-treated mice with M. tuberculosis was done to determine whether TB10.4_20–28-specific CD8⁺ T cells undergo clonal expansion, which is one of the defining features of memory T cells.

Materials and Methods

Mice

Age-matched female BALB/c mice were purchased from The Jackson Laboratory. Mice were housed in a biosafety level 3 facility under specific pathogen-free conditions at the Animal Biohazard Containment Suite (Dana-Farber Cancer Institute) and were used in a protocol approved by the institution.

Abs and peptide MHC class I tetramers

The following Alexa488 (A488)-, FITC-, PE-, PE-Cy5-, allophycocyanin-, and PerCP-conjugated anti-mouse mAbs were purchased from BD Pharmingen: anti-CD8α (FITC, PE-Cy5, PerCP), anti-CD3 (FITC), anti-CD69 (FITC), anti-CD44 (FITC), anti-CD45RB (FITC), anti-CD107A (FITC), CD107B (FITC), anti-CD62L (allophycocyanin), anti-IFN-γ (A488), and anti-TNF (A488). Anti-human granzyme B (allophycocyanin) was obtained from Caltag Laboratories. Class I MHC tetramers were produced from TB10.3/10.4_20–28-loaded H-2K^d biotinylated monomers, complexed with PE-conjugated streptavidin (National Institute of Allergy and Infectious Diseases tetramer facility, Emory University Vaccine Center, Atlanta, GA).

Peptide

The peptide, GYAGTLQSL (TB10.3/10.4_20–28), was identified by Majlessi et al. (3) as an Ag that is recognized by H-2 K^d-restricted CD8⁺ T cells. The synthetic peptide was commercially synthesized, and its identity was confirmed by mass spectrophotometry (BioSource International). Peptides used for immunological assays were unpurified. The purity of peptides used for tetramer production was >95%.

Aerosol infection with M. tuberculosis

Virulent M. tuberculosis (Erdman strain) was prepared, as previously described (4). Mice were infected via the aerosol route using a nose-only exposure unit (Intox) (5). The average lung inoculum was determined by plating serial dilutions of lung homogenates 1 day after aerosol infection and was 150 ± 50 bacteria per mouse.

Preparation of cells

Single-cell suspensions were prepared from spleens, pulmonary lymph node (PLN), axillary lymph nodes (LN), bone marrow, bronchoalveolar lavage (BAL), and lungs of infected mice, as previously described (1, 4). BAL was obtained by flushing of mouse lungs with 0.5 mM EDTA in PBS. Bone marrow was isolated from femur and tibia via flushing with complete medium. Briefly, tissue was dispersed by physical agitation, and RBC were lysed using lysis buffer (0.15 M NaCl, 1 mM KHCO₃, and 0.1 mM sodium-EDTA (pH 7.3)). After washing, the cells were resuspended in complete medium (RPMI 1640, 10% FCS, 2% HEPES, 1% l-glutamine, 1% penicillin-streptomycin, and 0.1% 2-ME). Lung mononuclear cells (MNC) were obtained by digesting tissue with collagenase type IV (Sigma-Aldrich) for 1–2 h at 37°C, followed by filtration through a 60-mesh metal strainer and 70 µM nylon strainer (Fisher Scientific). RBC lysis was performed, as described above. For cultured cells, rIL-2 (Chiron) was added to assay medium at 100 U/ml to promote T cell growth.

Flow cytometry and tetramer staining

Purified total lymphocytes from the lymphoid and peripheral tissues were resuspended at a concentration of 1–5 million cells per sample in FACS buffer (2% FCS, 2 mM sodium azide in PBS). Cells were initially blocked with anti-FcγRII/III Ab (24G2) for 10 min at 4°C, washed, and then stained with fluorochrome-conjugated isotype-matched control IgG or Abs specific for mouse cellular markers. Cells were stained with Abs and tetramer mixtures at an optimum concentration in FACS buffer for 15–20 min at 4°C, and then washed and fixed in 1% paraformaldehyde overnight. Cells were analyzed using a FACSCanto (BD Biosciences), and FlowJo (Tree Star) software was used to analyze the data.

Intracellular cytokine staining

Five million total cells isolated from infected spleen or lung were cultured for 12 h ± TB10.3/10.4_20–28 peptide (10 µM), with IL-2 (100U/ml). Brefeldin A (10 µg/ml) was added after the first hour of culture. Similar data were obtained when the cells were stimulated in vitro with the peptide for 4–6 h (data not shown). After incubation, cells were first stained for cell surface markers, as described above. Cells were then washed thoroughly with FACS buffer, resuspended in permeabilization buffer (Perm/Wash kit; BD Biosciences), and incubated at room temperature for 20 min, as per the manufacturer’s instructions. Cells were washed and resuspended in permeabilization buffer containing labeled Ab (anti-IFN-γ or anti-TNF at 1–2 µg/ml) for half an hour on ice. The cells were then washed with the perm/wash buffer, fixed in 1% paraformaldehyde overnight, and analyzed by flow cytometry, as described above.

Intracellular granzyme B staining was performed ex vivo using unstimulated cells. Cells were stained for surface markers, as described above, and then stained for intracellular granzyme B using the intracellular cytokine-staining protocol. Cells were then fixed in 1% paraformaldehyde overnight and analyzed the next day. The ability of the anti-human granzyme B mAb to stain mouse cells was verified by culturing normal mouse splenocytes in 20 U/ml human rIL-2 (6). After 5 days, 25% of the CD8⁺ T cells expressed high levels of granzyme B (data not shown).

CFUs

Bacterial burden was determined for each mouse individually. Splenic bisections or left lung lobes of infected mice were homogenized, and serial dilutions were spread onto Mitchison 7H11 selective agar plates (REMEL). Colonies were enumerated after 3 wk, and total organ CFUs were calculated.

Antibiotic treatment

Mice were begun on antibiotic treatment 2–3 wk postaerosol infection. Rifabutin and isoniazid (INH) were added to drinking water, each at 0.1 g/L. Treatment water was changed twice weekly. Memory-immune mice were maintained on antibiotics throughout each experiment.

ELISPOT assay for IFN-γ

The ELISPOT method was used to detect IFN-γ secretion by individual CD8⁺ T cells from infected mice following stimulation with peptides in vitro using the BD Biosciences ELISPOT kit and protocol (BD Biosciences), as previously described (1). Briefly, ELISPOT plates were coated with capture IFN-γ Ab overnight at 4°C and blocked with complete medium for 2 h at room temperature. CD8⁺ T cells purified from individual infected spleens or lungs by positive selection using CD8α immunomagnetic beads and MACS LS⁺ columns, as per manufacturer’s protocol (Miltenyi Biotec), were stimulated in triplicate with TB10.3/10.4_20–28 peptide and irradiated naive splenocytes for 40–44 h at 37°C (7). In all experiments, the purity of CD8⁺ cells was >90%, as determined by flow cytometry. Postincubation, cells were discarded and plates were washed with deionized water and PBS/Tween 20. Secondary biotinylated Ab was added for 2 h and incubated at room temperature, followed by washing (PBS/Tween 20). Streptavidin-alkaline phosphatase was then added to the plates for 1 h, followed by washing (PBS/Tween 20, PBS) and then development of a color reaction was performed using the substrate 3-amino-9-ethylcarbazole substrate reagent kit (BD Biosciences). The reaction was stopped after the spots developed by running the plate under water. The spots were enumerated using a series A Immunospot plate reader, Image Acquisition v4.0, and Immunospot v3.2 analysis software (Cellular Technology). Ag-specific spots were determined by subtracting the average medium control values for each individual mouse.

Statistics

The t tests and one-way ANOVA with Bonferroni’s posttest were performed using GraphPad Prism version 4.02 for Windows (GraphPad; 〈www.graphpad.com〉).

Results

TB10.4_20–28 is an immunodominant Ag following respiratory M. tuberculosis infection

Following respiratory M. tuberculosis infection of BALB/c mice, as many as 30–40% of the CD8⁺ T cells in the lung are specific for TB10.4_20–28 based on staining with TB10.4_20–28-loaded K^d tetramers (Fig. 1A). In addition to their enrichment in lung tissue, TB10.4_20–28-specific CD8⁺ T cells are recovered in the BAL fluid (Fig. 1A). The high frequency of TB10.4_20–28-specific CD8⁺ T cells in the BAL indicates that they are recruited as part of the inflammatory response into the lung airspace. In addition to the lung, discrete populations of TB10.4_20–28-specific CD8⁺ T cells are found in the spleen, draining PLN, and nondraining peripheral LN (Fig. 1A). These Ag-specific CD8⁺ T cells are also found in the bone marrow, which has been described as a niche occupied by memory CD8⁺ T cells (8). The rapid expansion of TB10.4_20–28-specific CD8⁺ T cells in the lung, PLN, and spleen indicates that the TB10.3/TB10.4 Ag elicits an immune response early following respiratory infection (Fig. 1B). The absolute number of TB10.4_20–28-specific CD8⁺ T cells increases in the lungs, and the peak response is 4–5 wk following infection, after which there is a modest decline in the frequency of TB10.4_20–28-specific CD8⁺ T cells in the lung (Fig. 1, B and C). Finally, many TB10.4_20–28-specific CD8⁺ T cells are found in the blood (Fig. 1D). The frequency in peripheral blood is of particular interest because it is the most feasible compartment to sample in people infected with M. tuberculosis. Although there was not a consistent linear relationship between the frequency in blood and lung in individual mice, the average frequency of TB10.4_20–28-specific CD8⁺ T cells in the lung tissue was ~7-fold greater than in peripheral blood (Fig. 1D).

FIGURE 1.

Distribution of TB10.4_20–28-specific CD8⁺ T cells during respiratory M. tuberculosis infection. A, The frequency of TB10.3/10.4_20–28-specific CD8⁺ T cells in different tissues from M. tuberculosis-infected mice. Lymphocytes were gated by size and by CD8⁺ staining. The percentage of CD8⁺ cells that stain with the TB10.3/10.4_20–28-loaded H-2 K^d tetramer (TB10.4 K^d tetramer) is indicated in each dot plot. This value is referred to as the percentage of CD8⁺tet⁺ throughout the following figures. Representative FACS plots are shown for a cohort of mice that were analyzed 11 wk after aerosol infection. Ax, axillary. B, The percentage of CD8⁺ cells in the lung, spleen, and PLN that stain with the TB10.4 K^d tetramer at various time points after respiratory M. tuberculosis infection. Each data point represents the mean ± SE of between three and eight determinations for each time point (except for weeks 11 and 32, for which there was only a single determination), representing a total of 82 determinations. The data were generated from the analysis of individual mice and from pooled tissue. C, The absolute number of TB10.3/10.4_20–28-specific CD8⁺ T cells in the lungs of infected mice from a representative experiment. Each data point represents the mean of three mice ± SE. D, The relationship between Ag-specific CD8⁺ T cells detected in blood and the lung. The frequency of TB10.4_20–28-specific CD8⁺ T cells in the blood and the lung of M. tuberculosis-infected mice was determined by flow cytometry using tetramer analysis. Individual mice were analyzed 4–11 wk after infection. The lines indicated paired samples from the same individual.

Thus, just as CFP10_32–39 is an immunodominant epitope recognized by CD8⁺ T cells in mice with the H-2^k haplotype (1), TB10.4_20–28 is an immunodominant epitope in H-2^d BALB/c mice. TB10.4_20–28-specific CD8⁺ T cells are found both in primary and secondary lymphoid organs, and accumulate in large numbers in the lung and in the bronchoalveolar compartment following aerosol infection with M. tuberculosis.

TB10.4-specific CD8⁺ T cells produce IFN-γ and TNF

IFN-γ production by CD8⁺ T cells can provide protection to susceptible hosts against M. tuberculosis infection (9). Therefore, we wished to determine which cytokines were produced by TB10.4_20–28-specific CD8⁺ T cells. IFN-γ was produced by splenocytes from M. tuberculosis-infected BALB/c mice after in vitro culture with the TB10.4_20–28 peptide (data not shown). To further characterize cytokine production by TB10.4_20–28-specific CD8⁺ T cells, intracellular cytokine staining was performed. IFN-γ- and TNF-producing TB10.4_20–28-specific CD8⁺ T cells were detected when purified pulmonary CD8⁺ T cells were stimulated with irradiated splenocytes and the TB10.4_20–28 peptide (Fig. 2A). A similar percentage of TB10.4_20–28-specific CD8⁺ T cells produces IFN-γ and TNF, emphasizing the potential importance of CD8⁺ T cells producing these cytokines during M. tuberculosis infection (10–12). The percentage of TB10.4_20–28-specific CD8⁺ T cells producing IFN-γ and TNF was ~2- to 3-fold greater in the spleen compared with the lung (Fig. 2B). Therefore, TB10.4_20–28-specific CD8⁺ T cells elicited following infection produce both IFN-γ and TNF, which are critical for cellular immunity to M. tuberculosis.

FIGURE 2.

Cytokine production by TB10.4_20–28-specific CD8⁺ T cells. A, Intracellular cytokine production after stimulation with the TB10.4_20–28 peptide. Three months after M. tuberculosis infection, lung MNC were cultured in vitro in the absence (top row) or presence (bottom row) of 10 µM TB10.4_20–28 for 12 h. The cells were gated by size, and CD8⁺tet⁺ cells were analyzed for their production of IFN-γ or TNF. The percentage of CD8⁺tet⁺ cells producing IFN-γ or TNF is indicated in each contour plot. This experiment is representative of three independent experiments. B, Cytokine production by TB10.4_20–28-specific CD8⁺ cells from infected lung (left) and spleen (right) after in vitro culture in the absence (none) or presence (peptide) of the TB10.4_20–28 peptide. Intracellular cytokine staining was performed, as described. Nonspecific staining determined using an isotype control was subtracted from the specific staining obtained with or without the peptide. The data are an average of three independent experiments, and the bars represent the mean ± SE. The percentage of CD8⁺tet⁺ T cells secreting cytokine in the absence or presence of peptide was significantly different, as determined by a two-way ANOVA with Bonferroni’s posttest (*, p < 0.05; **, p < 0.01). The percentage of CD8⁺tet⁺ T cells producing IFN-γ vs TNF was not significantly different. □, IFN-γ; ■, TNF.

TB10.4-specific CD8⁺ T cells express cytolytic effector molecules

The TB10.4_20–28 epitope was originally shown to be recognized by CD8⁺ T cells obtained from BCG-infected mice using a CTL assay (3). Subsequently, we showed that that TB10.4_20–28-specific CD8⁺ T cells have cytolytic activity in vivo (1). Therefore, we were particularly interested in determining which cytolytic effector molecules are expressed by these Ag-specific CD8⁺ T cells. CD8⁺ T cells were analyzed ex vivo for their intracellular expression of granzyme B. Granzyme B was expressed by a small subset of TB10.4_20–28-specific CD8⁺ T cells in the lungs of infected mice (Fig. 3A). In a study by Irwin et al. (2), a similar proportion of 32C-specific CD8⁺ T cells from M. tuberculosis-infected C57BL/6 (H-2^b) mice expresses granzyme B. We also stained these cells with the anti-perforin mAb (KM585); perforin expression could not be convincingly demonstrated by flow cytometry, although we suspect that this may be due to a technical issue with the Ab.

FIGURE 3.

Expression of granzyme B and CD107A/B by CD8⁺ T cells. A, Expression of granzyme B and CD107A/B by pulmonary CD8⁺ T cells. Lung MNC were obtained from mice 4 wk after infection, and CD8⁺tet⁻ (left and middle panels) and CD8⁺tet⁺ (right panel) T cells were analyzed for their ex vivo expression of intracellular granzyme B and cell surface CD107A/B. The staining with the isotype control (left) is compared with the expression of granzyme B and CD107A/B by CD8⁺tet⁻ cells (middle panel) and CD8⁺tet⁺ cells (right panel). B, Expression of CD107A/B by pulmonary and splenic CD8⁺ T cells. Lung or spleen MNC were obtained from infected mice, and the expression of CD107A/B by CD8⁺tet⁺ cells was determined by flow cytometry. The results from three independent experiments are shown. Each symbol represents individual or pooled tissue samples obtained from mice 4 to 17 wk after infection. □, Isotype control; ■, CD107A/B. Line, mean.

Cytotoxic granule exocytosis leads to the decoration of CD107A and CD107B (also known as lysosome-associated membrane protein- 1 and -2, referred to hereafter as CD107A/B) on the surface of CD8⁺ T cells (13). Otherwise, CD107A/B expression is limited to intracellular lysosomal compartments. We observed a reciprocal expression of surface CD107A/B and intracellular granzyme B, which may be due to granzyme depletion in cells having undergone recent granule exocytosis. Surface expression of CD107A/B and intracellular expression of granzyme B were enriched in TB10.4-specific CD8⁺ T cells, over the nontetramer⁺ population of CD8⁺ T cells, in both the lung and spleen (p = 0.0003 and p = 0.0209, respectively) (Fig. 3). The percentage of TB10.4_20–28-specific CD8⁺ T cells expressing CD107A/B was slightly higher in the lung than in the spleen, although this was not statistically significant. Thus, a subset of CD8⁺ T cells expresses effector molecules associated with the cytotoxic function of these T cells. We believe that these data strongly support the hypothesis that Ag-specific CD8⁺ T cells mediate cytotoxic activity in vivo.

A subset of TB10.4_20–28-specific CD8⁺ T cells in the lymphoid tissue of M. tuberculosis-infected mice expresses phenotypic markers of central memory cells

The majority of both the TB10.4_20–28-specific CD8⁺ T cells (CD8⁺tet⁺ T cells) and the CD8⁺tet⁻ in the lung express activation markers CD69 and CD44, but lack high expression levels of CD62L and CD45RB (Fig. 4; data not shown) (12, 14, 15). The phenotype of these pulmonary TB10.4_20–28-specific CD8⁺ T cells is consistent with either effector or effector memory T cells (16). In contrast, the majority of CD8⁺ T cells found in the LN and spleens of uninfected and chronically infected mice express CD62L and CD45RB, but lack CD69 and CD44 (Fig. 4; data not shown). This pattern of markers is most consistent with naive T cells. As discussed above, TB10.4_20–28-specific CD8⁺ T cells are widely distributed throughout the host, which may be due to the systemic nature of M. tuberculosis infection. Thus, it is not surprising that Ag-specific T cells with an effector phenotype (e.g., CD8⁺tet⁺CD62L^low or CD45RB^low) are found in the lymphoid organs. However, in addition to these CD8⁺ T cells with an effector phenotype, there exists a distinct population of TB10.4_20–28-specific CD8⁺ T cells in the secondary lymphoid organs that have features of T_CM.

FIGURE 4.

Appearance of TB10.4_20–28-specific CD8⁺ T cells with a central memory phenotype in the lymphoid organs of M. tuberculosis-infected mice. A, The expression of CD62L by CD8⁺tet⁻ (top row) and CD8⁺tet⁺ (bottom row) T cells in the lung, PLN, and spleen of M. tuberculosis-infected BALB/c mice. The numbers indicate the percentage of gated CD8⁺ cells that express high levels of CD62L. B, The expression of CD62L and CD45RB by CD8⁺tet⁻ (●) and CD8⁺tet⁺ (○) T cells from the lung, PLN, and spleen of M. tuberculosis-infected BALB/c mice. The percentage of CD8⁺tet⁻ and CD8⁺tet⁺ T cells that express high levels of CD62L (CD62L^high) or CD45RB (CD45RB^high) is represented in the graphs. Data represent 10–17 determinations from seven experiments, done 4–13 wk after infection (some of the symbols are superimposed). Bar, mean. A t test was done to determine whether the expression of these markers was statistically different between CD8⁺tet⁻ and CD8⁺tet⁺ T cells.

Although the majority of Ag-specific CD8⁺ T cells in the PLN and spleen had an activated phenotype (e.g., CD62L^low and CD45RB^low; Fig. 4A), a variable fraction (between 27 and 43%) of the CD8⁺tet⁺ T cells in the PLN and the spleen expresses high levels of CD62L and CD45RB (Fig. 4). This constellation of findings, namely the expression of high levels of CD62L and CD45RB by Ag-experienced (i.e., tetramer⁺) T cells in lymphoid organs, is typical of central memory CD8⁺ T cells (16, 17). Thus, by using TB10.4_20–28-loaded K^d tetramers, a population of Ag-specific CD8⁺ T cells that have characteristics of T_CM cells has been identified.

Resolution of infection leads to a decline in the frequency of TB10.4_20–28-specific CD8⁺ T cells and a shift in the proportion of T_CM cells

Murine tuberculosis is a chronic infection. Although the murine T cell response initially controls the infection by inhibiting bacterial replication, mice never clear the infection and eventually die of tuberculosis. This chronicity complicates the study of the memory-immune response as persistent bacteria maintain chronic activation of the immune system. Data from other infection models indicate that chronic infection can inhibit the development of T_CM (18, 19). To study T_CM development in M. tuberculosis chronic infection, we infected BALB/c mice by the aerosol route, and 3 wk later divided the mice into two groups. One group received antibiotics to cure the infection and establish memory-immune mice (20). Within 3–4 wk of antibiotic administration, there was already significant resolution of the infection, and this was accompanied by a decline in the frequency and absolute number of TB10.4_20–28-specific CD8⁺ T cells in the lung, PLN, and spleen (Fig. 5, A and B; data not shown). These data indicate that persistent M. tuberculosis infection has a role in maintaining a high frequency of Ag-specific T cells at the site of disease, and contraction of the pool of Ag-specific CD8⁺ T cells occurs following resolution of infection.

FIGURE 5.

Reduction in the frequency of TB10.4_20–28-specific CD8⁺ T cells and shift toward T_CM phenotype during resolution of M. tuberculosis infection. A, The frequency of pulmonary TB10.4_20–28-specific CD8⁺ T cells in infected mice and antibiotic-treated memory-immune mice. The percentage of CD8⁺ T cells that stained with the TB10.4 K^d tetramer was determined. Data represent nine determinations from three experiments. No Rx, untreated infected mice; Abx, antibiotic-treated mice (e.g., memory-immune mice); bar, mean. The difference between the groups was tested for statistical significance using a t test (p < 0.0001). B, The absolute number of TB10.4_20–28-specific CD8⁺ T cells in the lungs of infected mice and antibiotic-treated memory-immune mice. Data represent six determinations from two experiments. No Rx, untreated infected mice; Abx, antibiotic-treated mice; bar, mean. The difference between the groups was tested for statistical significance using the Mann-Whitney U test (p = 0.0022). C, Representative flow cytometric analysis of CD8⁺ T cells from the spleen of an infected mouse (No Rx; top row) and an antibiotic-treated mouse (Abx; bottom row). Mice were infected by the aerosol route and, after 3 wk, half of the mice were started on INH and rifabutin. The data represented in this figure were obtained 12 wk later. CD8⁺tet⁻- and CD8⁺tet⁺-staining cell populations were gated (left panels), and the CD62L and CD45RB expression by these CD8⁺tet⁻ (middle panels), and CD8⁺tet⁺ (right panels) T cells was determined. The percentages in each quadrant are an average of four individual subjects that were analyzed. D, The expression of CD62L and CD45RB by CD8⁺tet⁺ T cells from the lung, PLN, and spleen of M. tuberculosis-infected mice. Mice were infected for 3 wk with M. tuberculosis and then were treated with antibiotics (Abx) for 4 (top row) or 12 wk (bottom row), or as a control were left untreated (No Rx). Differences in the phenotype of CD8⁺tet⁺ T cells in Abx-treated and untreated mice were tested for their statistical significance with a t test.

In some infections, resolution is accompanied by a transition of Ag-specific CD8⁺ T cells into the T_CM compartment. We examined the PLN and spleen, as these tissues are predicted to contain a higher proportion of Ag-specific CD8⁺ T_CM cells. As observed in the lung, there was a lower frequency of Ag-specific CD8⁺ T cells in the spleen and PLN of memory-immune mice. However, there was an increase in the proportion of the remaining TB10.4_20–28-specific CD8⁺ T cells that expressed high levels of CD45RB and CD62L (Fig. 5, C and D; data not shown). An increase in the proportion of TB10.4-specific CD8⁺ T cells expressing high levels of CD45RB and CD62L was consistently observed in the lung, PLN, and spleen, following treatment of infected mice with antibiotics (Fig. 5D). As the absolute increase in TB10.4-specific CD8⁺CD62L^high T cells is relatively modest, their re-expression of CD45RB may reflect a change in their activation status instead of acquisition of a conventional T_CM phenotype. Nevertheless, high levels of CD45RB have been associated with T_CM in other systems (see Discussion), and this may suggest that resolution of M. tuberculosis infection is associated with a contraction of the Ag-specific CD8⁺ T cell pool and an increase in the proportion of T_CM-like cells.

The development of memory CD8⁺ T cells

The detection of TB10.4_20–28-specific CD8⁺CD62L^highCD45RB^high T cells suggested that CD8⁺ T_CM cells are generated during M. tuberculosis infection and persist after clearance of infection. To determine whether Ag-specific CD8⁺ T cells developed into bona fide memory T cells following infection, memory-immune mice (i.e., antibiotic-treated mice) were rechallenged with M. tuberculosis by the respiratory route, and their response to TB10.4_20–28 was compared with mice from the same cohort that had been maintained on antibiotics. Concurrently, previously uninfected age-matched BALB/c mice were infected so that the memory-immune response could be compared with the primary immune response. Splenic bacterial burden was assessed to confirm the continued suppression of CFU in antibiotic-treated mice, and (re)infection of primary and secondary challenged mice (Fig. 6A).

FIGURE 6.

Expansion of TB10.4_20–28-specific CD8⁺ T cells following challenge of memory-immune mice with M. tuberculosis. As described in Materials and Methods, mice were infected by the aerosol route and, after 3 wk, half of the mice were treated with INH and rifabutin. After 3 wk, half of the antibiotic-treated mice (memory) were rechallenged (secondary) with aerosolized M. tuberculosis together with age-matched previously uninfected mice (primary). Four mice were analyzed per group, and these results are representative of three experiments. These results were analyzed by a one-way ANOVA with Bonferroni’s post test: *, p < 0.05; **, p < 0.01; ***, p < 0.001. A, The splenic CFU were simultaneously determined in the three experimental groups. Each symbol represents an individual mouse, and the line represents the mean. B, The frequency of TB10.4_20–28-specific CD8⁺ T cells in the lungs of mice following a primary or secondary response, and compared with memory-immune mice. Ag-specific CD8⁺ T cells were enumerated by an IFN-γ ELISPOT assay using the TB10.4_20–28 synthetic peptide (left) and by staining lung cells with TB10.4_20–28 K^d tetramers. The bars represent the mean ± SE. C, The absolute number of total (right panel) and TB10.4_20–28-specific (left panel) CD8⁺ T cells isolated using anti-CD8 immunomagnetic beads from lung MNC obtained from mice following a primary or secondary response, and compared with memory-immune mice. Ag-specific CD8⁺ T cells were enumerated using TB10.4_20–28 K^d tetramers. Bars, Mean ± SE. D, PLN pooled from four mice were analyzed by flow cytometry. CD8⁺tet⁻- and CD8⁺tet⁺-staining cell populations were gated, and the CD62L and CD45RB expression by these CD8⁺tet⁻ (left panels) and CD8⁺tet⁺ (right panels) T cells was determined.

Memory-immune mice challenged with M. tuberculosis had an increase in the frequency of pulmonary TB10.4_20–28-specific CD8⁺ T cells compared with unchallenged memory-immune mice (7.9- and 11.9-fold increase by ELISPOT and flow cytometry, respectively) (Fig. 6B). Furthermore, the frequency of TB10.4_20–28-specific CD8⁺ T cells was higher during the secondary response compared with the primary response (5.6- to 9.9-fold greater 2 wk after infection; 2.4- to 2.9-fold greater 3 wk after infection) (Fig. 6B; data not shown). This indicates that the expansion of Ag-specific CD8⁺ T cells is more rapid during the secondary (i.e., memory) response than during the primary response, as was predicted by studies examining bulk CD8⁺ T cell populations (14). The tetramer and ELISPOT data reveal the percentage of CD8⁺ T cells that are specific for TB10.4_20–28. However, it is impossible to know whether the other CD8⁺ T cells recruited to the lung are specific for other mycobacterial Ags or whether some of them are recruited to the lung in an Ag-independent manner (21). Therefore, to gain a better sense of the true expansion of TB10.4_20–28-specific CD8⁺ T cells, we used our flow cytometry data to calculate the absolute number of CD8⁺tet⁺ T cells in the lungs of the infected mice. After 3 wk of infection, there were 7-fold more TB10.4_20–28-specific CD8⁺ T cells in the lungs of mice with a secondary response compared with primary infection (Fig. 6C). Compared with memory-immune mice, rechallenged mice had nearly 27-fold more TB10.4_20–28-specific CD8⁺ T cells in their lungs.

Finally, the majority of TB10.4_20–28-specific CD8⁺ T cells in the draining PLN of M. tuberculosis memory-immune mice were CD45RB^highCD62L^high (Fig. 6D). After reinfection with M. tuberculosis, there was significant down-modulation of CD45RB and CD62L on these Ag-specific CD8⁺ T cells (Fig. 6D). These data show that TB10.4_20–28-specific CD8⁺ T cells have functional properties of memory T cells because they are present at an increased frequency after the resolution of infection, and undergo a dramatic expansion following re-exposure to M. tuberculosis. Furthermore, our data suggest that central memory CD8⁺ T cells in the draining LN give rise to effector or effector-memory T cells following rechallenge with M. tuberculosis.

Discussion

CD8⁺ T cells that recognize mycobacterial Ags are elicited following M. tuberculosis infection of people and experimental infection of animals (22, 23). In mouse models, CD8⁺ T cells are critical for optimum immunity, including particular importance in models of latency and immunodeficiency (9, 24). Additionally, data exist that CD8⁺ T cells elicited by vaccination can mediate protection against M. tuberculosis challenge (25–27). Some debate continues about the absolute requirement for CD8⁺ T cells in immunity to tuberculosis, in large part because of a greater role for CD4⁺ T cells compared with CD8⁺ T cells in some systems (28). Resolution of this issue is complicated because the CD8⁺ T cell response is dependent on CD4⁺ T cells, which could potentially exaggerate the relative role of CD4⁺ T cells and minimize the role of CD8⁺ T cells (29). A detailed understanding of the role of CD8⁺ T cells in the immune response to M. tuberculosis infection has been hindered in part because of the lack of well-defined Ags that can be used to track, enumerate, and monitor the function of Ag-specific CD8⁺ T cells. This situation is beginning to change, and a number of Ags have been defined during the past few years that are recognized by human and murine CD8⁺ T cells. In this current study, we have used a 9-aa peptide epitope found in the ESAT6-related mycobacterial proteins TB10.3 and TB10.4 to characterize Ag-specific CD8⁺ T cells that are elicited following aerosol infection with M. tuberculosis (3).

Using TB10.4_20–28-loaded H-2 K^d tetramers, we found that TB10.4_20–28-specific CD8⁺ T cells are widely distributed and found at high frequencies in draining and nondraining LN, spleen, bone marrow, and blood following M. tuberculosis infection. Strikingly, TB10.4_20–28-specific CD8⁺ T cells are enriched in the lung and comprise up to 40% of the CD8⁺ T cells in the lung tissue and in the bronchoalveolar compartment. Although recruitment of bystander CD8⁺ T cells is observed following Mycobacterium avium infection (21), our data argue that the majority of the CD8⁺ T cells in the lungs of M. tuberculosis-infected mice are Ag specific and have the potential to participate in host defense against the pathogen. Whether the accumulation of Ag-specific CD8⁺ T cells results from their continuous recruitment into the lung or from cell division after CD8⁺ T cells enter the lung is still being determined. However, the net result is a dramatic enrichment of Ag-specific CD8⁺ T cells in the lungs of mice following respiratory infection. The TB10.4 epitope represents the second defined immunodominant epitope for which this is true.

We found that Ag-specific CD8⁺ T cells from infected mice produce various cytokines following stimulation with the TB10.4_20–28 peptide ex vivo. In addition to IFN-γ, which has been shown previously to play a role in the protective function of CD8⁺ T cells, we also detected TNF. These cytokines have important effector roles in activating macrophages, inducing apoptosis of infected cells, and contributing to granuloma formation; other immunoregulatory roles are possible as well (30–33). Interestingly, only a subset of the Ag-specific CD8⁺ T cells produces cytokines. Similarly, we have noted previously that tetramer analysis detects nearly 10-fold more Ag-specific T cells than does an IFN-γ ELISPOT, using the H-2 K^k-restricted CFP10 epitope (34). Thus, when using independent assay systems and mycobacterial Ags in different mouse strains, the frequency of Ag-specific T cells determined by tetramer staining is far greater than that predicted by cytokine production. Whether there is a biological explanation for this observation (e.g., exhaustion or anergy of Ag-specific CD8⁺ T cells) or whether it is simply a reflection of a technical issue (e.g., cell viability or efficiency of in vitro stimulation) remains to be determined.

In addition to their potential importance as cytokine-producing T cells, TB10.4_20–28-specific CD8⁺ T cells were originally identified based on their ability to lyse peptide-pulsed target cells in vitro (3). Using an in vivo cytotoxicity assay, we previously reported that TB10.4-specific CD8⁺ T cells elicited by M. tuberculosis infection eliminate TB10.4_20–28-pulsed splenic target cells in vivo (1). In this study, we show that a subset of TB10.4_20–28-specific CD8⁺ T cells expresses granzyme B, a component of cytotoxic granules that mediates target cell lysis. Our finding that only a subset of CD8⁺ T cells expresses granzyme B ex vivo may be due to heterogeneity in granzyme expression, or because it is depleted secondary to degranulation (13, 35). CD107A and CD107B are two intracellular proteins that are normally found in lysosomes, but are also a structural component of cytotoxic granules. Following exocytosis of cytotoxic granules, CD107A/B is transiently expressed on the cell surface of CTLs (13). The detection of cell surface CD107A/B expression by a subset of TB10.4_20–28-specific CD8⁺ T cells ex vivo is strong evidence that these T cells recognize and kill M. tuberculosis-infected cells in vivo. Supporting this contention is the reciprocal relationship between the intracellular expression of granzyme B and the cell surface expression of CD107A/B. Because CD8⁺ T cells that express cell surface CD107A/B have undergone degranulation, the contents of their cytotoxic granule, including granzyme B, are depleted (13, 36). Although granzyme B and CD107A/B are sometimes found to be highly expressed by Ag-specific CD8⁺ T cells, particularly after acute viral infection, the expression of these markers of CTL activity can be significantly less during chronic or persistent infection (2, 37, 38).

The evolution of T cell memory during M. tuberculosis infection is poorly understood, and even the question of whether or not memory T cells develop normally following infection remains controversial. Chronic viral infection inhibits the development of T_CM (18, 19). Although it is indisputable that memory T cells are generated following BCG vaccination, a response that results in protection in experimental animal models, sterilizing immunity to the degree observed with vaccines against other intracellular bacteria is not observed. The failure of the memory T cell response to completely protect the host against disease could result from evasion of immunity by M. tuberculosis, or alternately, could be an indication that the memory T cell response elicited by BCG is impaired.

To address some of these questions surrounding the differentiation of memory T cells during M. tuberculosis infection, we took advantage of TB10.4_20–28-loaded H-2 K^d tetramers to track Ag-specific CD8⁺ T cells. This allowed us to detect TB10.4_20–28-specific CD8⁺ T cells expressing cell surface markers associated with a T_CM phenotype in the secondary lymphoid organs of chronically infected mice. Resolution of infection, which has been found to be a necessary prerequisite for the development of central memory in other infectious disease models, led to the contraction of Ag-specific CD8⁺ T cell pool. The TB10.4_20–28-specific CD8⁺ T cells that persisted following this contraction were more likely to have a phenotype characteristic of T_CM cells. To test whether the TB10.4_20–28-specific CD8⁺ T cells that remain after antibiotic treatment behave functionally like memory T cells, we challenged memory-immune mice with M. tuberculosis. We observed an expansion of both the percentage and absolute number of TB10.4_20–28-specific CD8⁺ T cells that was far greater than the primary response of previously uninfected mice.

Ag-specific and polyclonal CD4⁺CD62L^high memory T cells express high levels of CD45RB (39), and although CD45RB expression by CD8⁺ memory T cells has not been typically measured after viral infection (18), a population of CD8⁺CD45RB^high CD62L^high T cells thought to represent a T_CM population is found in the liver after exposure to Plasmodium berghei (17, 40). Although we detect CD8⁺tet⁺CD62L^highCD45RB^low T cells in the lymphoid organs of chronically infected mice, treatment with antibiotics, a strategy used to experimentally induce memory immunity, led to the re-expression of CD45RB by TB10.4_20–28-specific CD8⁺ T cells. Thus, whereas most descriptions of T_CM emphasize their expression of CD62L and CCR7, we detect coexpression of CD62L and CD45RB on TB10.4_20–28-specific CD8⁺ T cells in memory-immune mice (40). Whether these CD8⁺tet⁺CD62L^highCD45RB^high T cells are bona fide T_CM will require additional studies to clarify the lineage relationships between these various Ag-specific T cell populations. Functional experiments are also needed to determine whether these different subsets are analogous to the different memory compartment elicited following acute viral infection (18).

The development of OVA-specific CD8⁺ T cells during mycobacterial infection has been studied using rBCG expressing OVA (41). Naive OVA-specific CD8⁺ T cells differentiate into T_CM cells, albeit with protracted kinetics, following infection. These CD8⁺ T cells differentiate into a stable population with a CD44^highCD62L^high phenotype and the capacity to produce IFN-γ and IL-2 after stimulation in vitro. Although these T cells have many of the features of typical memory CD8⁺ T cells, they respond slowly following a secondary challenge with BCG-OVA (41). This may be explained by the slow replication of BCG and/or the low level of OVA produced because challenge of the same memory-immune mice with rOVA-expressing Listeria monocytogenes results in a greater expansion of OVA-specific CD8⁺ T cells. This potential relationship between the Ag load during challenge and the secondary T cell response may have important implications for understanding why vaccines have only a limited efficacy against M. tuberculosis. How immunity is recalled during M. tuberculosis infection may be more critical than the strategy used to prime T cell immunity in determining the efficacy of vaccination. Although vaccination with BCG consistently enhances host control of M. tuberculosis in the lung and other organs, and can provide a long-term survival advantage, it is still uncertain whether prior infection with M. tuberculosis provides a similar degree of protection.

Although immunological memory induced by M. tuberculosis can provide short-term protection as indicated by a reduction in the bacterial burden following rechallenge, some data suggest that long-term protective memory to M. tuberculosis does not efficiently develop in experimentally infected animals (20, 42). The reinfection of patients previously treated for tuberculosis also challenges the notion that prior exposure to M. tuberculosis induces protective immunity (7, 43–45). Clearly, delineating how memory T cells develop during M. tuberculosis infection, defining which memory T cell subsets are critical for the control of infection, and understanding how the latent phase of the infection affects the maintenance of memory T cells are all important goals if we are to develop an effective vaccine against tuberculosis.