CTL Response to Mycobacterium tuberculosis: Identification of an Immunogenic Epitope in the 19-kDa Lipoprotein¹

The importance of CTLs in protective immunity against Mycobacterium tuberculosis (M.tb)³ is supported by their presence in mycobacterium-infected animals and by their ability to prevent bacillary dissemination in these animals. Studies of experimental tuberculosis (TB) in mice indicate that CD8⁺ T cells contribute to the successful resolution of M.tb infection by releasing IFN and killing cells harboring tubercle bacilli (1, 2, 3, 4, 5, 6, 7, 8, 9). IFN enables monocytes to more efficiently restrict the replication of intracellular bacilli (10, 11). Direct contact between mycobacteria-reactive CD8⁺ CTLs and the infected cells can inhibit M.tb growth (3). In addition, organisms released from host cells that are unable to mobilize an effective antimycobacterial effector mechanism (e.g. endothelial and epithelial cells) are taken up and destroyed by infiltrating blood monocytes (2).

In keeping with findings from the murine models of TB, precursors of CD8⁺ CTLs that are capable of lysing autologous mycobacteria-infected monocytes are present in the peripheral blood and bronchoalveolar lavages of healthy tuberculin skin test-positive (TSP) individuals (12, 13). CD8⁺ class I-restricted T cells that are specific for epitopes in the M.tb early secretory antigenic target 6 have also been detected both in the peripheral blood of patients with active TB and in healthy contacts (14). Moreover, CD8⁺ CD1-restricted T cell lines derived from patients with active TB have been shown to lyse M.tb-infected monocytes and thereby restrict the growth of intracellular organisms (15). Surprisingly, however, no reports have yet appeared describing class I MHC-restricted M.tb-specific CTLs in humans. The current study represents an effort to identify and characterize such CTLs in M.tb-infected individuals. The CTL response to a single M.tb protein, the 19-kDa secreted lipoprotein (M.tb19) (16, 17), was studied, because M.tb19 and other secreted proteins have been implicated in immune protection against mycobacterial infection (5, 18). Moreover, M.tb19 induces CD4⁺ T cell responses in both mice and humans (19, 20), suggesting that it is a major target Ag of cellular immunity.

To screen for M.tb19 epitopes with the potential to serve as targets for class I MHC-restricted CTLs, we identified sequences within the protein containing the HLA-A*0201-binding motif (21). Synthetic peptides corresponding to these sequences were then tested for their ability to bind the HLA-A*0201 allele in an immunofluorescence-based peptide-binding assay (22); those peptides with the highest binding activity were studied as target Ag in CTL assays. The results indicate that CD8⁺ CTLs with specificity for residues 88 to 97 of M.tb19 (P_88–97) are present in the circulation of M.tb-sensitized individuals, including patients with active TB. Furthermore, a synthetic peptide corresponding to this region can elicit CTLs in vitro from the T cells of uninfected persons.

A total of 11 HLA-A*0201⁺ individuals participated in this study. The protocol and consent forms relating to our use of human subjects were approved by the California Pacific Medical Center Research Institute Administrative Panel on Human Subjects in Medical Research. Informed consent for blood donation was obtained from all donors. None of the subjects had evidence of infection with HIV. Two subjects were healthy TSP individuals, four had pulmonary TB, two were healthy tuberculin skin test-negative (TSN) individuals, and three were normal blood donors (Table I). Venous blood from healthy TSP and TSN individuals and patients with TB was collected into heparinized Vacutainer tubes (Becton Dickinson, Rutherford, NJ). White blood cell concentrates from HLA-A2⁺ normal donors were obtained from the Stanford Medical School Blood Center (Palo Alto, CA). HLA typing was performed by flow cytometry using anti-HLA-A2 mAb (BB7.2) as the first Ab and FITC-labeled goat anti-mouse F(ab′)₂ fragments as a second Ab. The HLA-A2 subtype of blood donors was determined by staining their B lymphoblastoid cell line (B-LCL) with an HLA-A2.1/A2.2-specific mAb (CR11–351) (23); the subtype was confirmed by the CTL response of their CD8⁺ T cells to a previously identified HLA-A*0201-restricted influenza A virus matrix peptide 58 to 66 (designated IMP) (24). The proliferative response of T cells to whole heat-killed M.tb was assayed as described below.

Patients with TB were managed at the City and County of San Francisco Division of TB Control, Department of Public Health. All of the TB patients had active pulmonary TB and were receiving standard chemotherapy. The diagnosis of TB was based on standard clinical parameters and bacteriologic identification of M. tuberculosis from sputum cultures.

Our laboratory generated the following mAbs using hybridomas that were obtained from the American Type Culture Collection (ATCC, Manassas, VA): Leu-4 (CD3), Leu-3a (CD4), Leu-2a (CD8), Leu-11a (CD16), w6/32 (class I MHC), L243 (HLA-DR), and BB7.2 (HLA-A2). For blocking studies, mAbs were used as purified reagents at concentrations of 10 μg/ml. The HLA-A2.1/A2.2-specific mAb (CR11–351) was a gift of Dr. S. Ferrone from Columbia University (New York, NY). FITC-conjugated mAbs directed against CD3, CD4, CD8, TCRαβ, HLA-DR, class I MHC (w6/32), and CD14 (Leu-M3) were purchased from the Becton Dickinson Monoclonal Center (Mountain View, CA). FITC-conjugated goat anti-mouse IgG F(ab′)₂ fragments and affinity-purified goat anti-mouse IgG (γ-chain-specific) Ab were purchased from Zymed Laboratories (San Francisco, CA).

A panel of 54 M.tb19-derived peptides (8–10 mer) containing HLA-A*0201-binding motifs were identified using a computer scoring program (25). A total of 28 peptides with a computed score between 48 and 69 were selected for testing in a T cell-binding assay. Two previously identified HIV-1-derived peptides, an HLA-A2-restricted HIV-1 gag peptide (amino acids (aa) 71–85, designated HIV gag A2) and an HLA-B8-restricted HIV-1 gag peptide (aa 253–267, designated HIV gag B8) (26) were used as controls, as was an HLA-A*0201-restricted IMP (aa 58–66) (24). Peptides were synthesized by F-moc chemistry at either University Hospital (Leiden, The Netherlands) or the Beckman Center at the Stanford University Medical Center. Before use, lyophilized, HPLC-purified peptides (>90% pure) were reconstituted at 40 mg/ml in DMSO and diluted to 1 mg/ml with Iscove’s modified Dulbecco’s medium (IMDM) (Life Technologies, Grand Island, NY).

T2, an HLA-A*0201⁺ Ag-processing defective cell line (22) which was a gift of Dr. M. Cheever (University of Washington, Seattle, WA), was propagated in RPMI 1640 containing nonessential aa, sodium pyruvate, and 10% FBS. Before use, the cells were incubated for 6 h at 37°C in serum-free IMDM; next, cells were washed once, suspended in serum-free IMDM containing 20 μM of 2-ME and 15 μg/ml of human β₂-microglobulin (β₂m) (Calbiochem, La Jolla, CA), and pulsed with 0 to 200 μM of M.tb19 peptide. Control cells were pulsed with either HIV gag A2 or HIV gag B8 peptide. After a 24-h incubation at 37°C, T2 cells were washed once with cold PBS containing 0.5% BSA and 0.02% NaN₃. They were then stained directly with FITC-conjugated w6/32 mAb or indirectly with anti-HLA-A2.1 mAb as first Ab and with goat anti-mouse FITC-labeled F(ab′)₂ fragments as a second-step Ab.

The percentage of FITC-positive cells as well as their staining intensity (mean fluorescence intensity (MFI)) were determined on an Epics Profile II (Coulter, Hialeah, FL). The Δ MFI for a particular mAb was calculated by subtracting the MFI of either the isotype-matched control mAb or the second-step Ab from each MFI value. The fluorescence ratio (FR) was calculated using the following formula: FR = (Δ MFI of peptide-treated T2 cells)/(Δ MFI of nontreated T2 cells).

T cells were cultured in IMDM that had been supplemented with 10% pooled heat-inactivated human serum, 2 mM of l-glutamine, 100 μg/ml of streptomycin, 100 U/ml of penicillin, and 2.5 μg/ml of fungizone (hereafter designated complete medium (CM)). Monocytes were cultured in antibiotic-free RPMI 1640 containing 10% pooled human serum and 2 mM of l-glutamine (antibiotic-free CM).

PBMCs from either patients with TB or healthy individuals were suspended in CM containing 10 μg/ml β₂m and stimulated with 6 μM of each M.tb peptide or IMP using a previously described procedure with some modifications (27). A total of 4 × 10⁶ cells/well were incubated at 37°C in 24-well plates containing 2 LfU/ml of tetanus toxoid (Department of Public Health, State of Michigan, East Lansing, MI) in 1 ml of CM. PBMCs from individuals that did not respond by proliferation to tetanus toxoid received 0.2 μg/ml of PHA (Wellcome Diagnostics, Research Triangle Park, NC) at the initiation of culture. After 3 days, 1 ml of CM supplemented with 10 U/ml of rIL-2 (Life Technologies) was added in each well. On day 7, the cultures were restimulated with the 6 μM of peptide in CM containing 10 μg/ml β₂m and 10 U/ml rIL-2 in the presence of 1 × 10⁶ irradiated (3000 rad), pooled, HLA-A2⁺, allogeneic PBMCs as feeder cells. On day 14, viable cells were recovered on Ficoll-Hypaque gradients, and CD8⁺ T cells were isolated by positive-selection panning using anti-CD8 mAb. After 1 day, the resulting cells (>95% CD8⁺, TCRαβ⁺ by flow cytometric analysis) were tested for CTL activity against autologous B-LCLs that had been pulsed with the stimulating peptide (see below).

Populations that had been enriched for either dendritic cells (DCs) or monocytes were separated from PBMCs on the basis of their differential densities (28, 29). Briefly, the PBMCs that were obtained by Ficoll-Hypaque gradient centrifugation were separated into low-density (monocytes) and high-density Percoll fractions. Monocytes that had been collected from the Percoll gradients (≥90% CD14⁺) were frozen in aliquots and thawed 1 day before use. To separate DCs, cells from the high-density Percoll fraction suspended in CM were incubated overnight in Teflon vessels at 37°C. Thereafter, the cells were layered onto 15% (w/v) metrizamide and centrifuged at 650 × g for 10 min at room temperature (30). For further enrichment, cells from the metrizamide interface (DC-enriched) were refloated on a 14% metrizamide solution. The DC-enriched population, which stained brightly with anti-HLA-DR mAb, was used as APCs in the in vitro vaccination system, as described below. CD8⁺ T cells were obtained from the high-density metrizamide fraction by positive-selection panning (31) using anti-CD8 mAb. The resulting cells were >95% CD8⁺ and TCRαβ⁺ cells by flow cytometric analysis.

Purified human CD8⁺ T cells (10⁵ cells suspended in 100 μl of CM) from healthy HLA-A*0201⁺, M.tb-unresponsive (stimulation index (SI) < 2; Table I) subjects were added to 10⁴ autologous DCs; the DCs had been pulsed with 6 μM of M.tb peptide by incubation for 2 h in 100 μl of CM supplemented with 20 μg/ml β₂m and 1 U/ml of rIL-1 (R&D Systems, Minneapolis, MN). The total volume per well was 200 μl, and plates were incubated at 37°C in a humidified atmosphere containing 10% CO₂. On day 3, a mixture of rIL-2 and rIL-4 (R&D Systems) at 5 U/ml was added to each culture. On day 7, and weekly thereafter, T cells were restimulated with 6 μM of original peptide in the presence of irradiated (3000 rad) autologous monocytes in CM supplemented with 5 U/ml of rIL-2 and rIL-4. After 4 to 6 wk of expansion, the CD8⁺ T cells that had been recovered by positive panning with anti-CD8 mAb were tested for their CTL activity against autologous B-LCLs that had been pulsed with the stimulating peptides. Cultures that displayed peptide-specific cytolytic activity were expanded by weekly restimulation and retested against autologous monocytes that had been infected with M.tb.

To generate B-LCLs, PBMCs from each participant were transformed by EBV-containing supernatants from the marmoset line B-958 that was provided by Dr. S. K. H. Foung (Stanford University). Cells were incubated in 24-well plates (5 × 10⁶ cells/well) in IMDM that had been supplemented with 30% heat-inactivated FBS, 2 mM of l-glutamine, 100 μg/ml of streptomycin, and 100 U/ml of penicillin. After 14 days, the transformed cells were expanded in IMDM containing 10% FBS, and an aliquot was stained with CR11–351 mAb. The percentage of HLA-A2.1/A2.2⁺ cells ranged between 82 and 98 by flow cytometry. Before their use as targets, B-LCLs were incubated for 18 h with 12 μM of each synthetic peptide.

The M.tb strain H37Ra (ATCC) was used to infect monocytes. Inocula were prepared by culturing bacilli on 7H11 agar plates. Before their addition to monocytes, colonies were resuspended in cold PBS (pH 7.2) and washed twice by centrifugation. To prevent clumping, the bacterial suspension was vortexed vigorously after each centrifugation and left standing at room temperature for 5 min before the upper fraction was withdrawn. After the last wash, clumps of mycobacteria in the suspension were dispersed by multiple passages through a syringe with a 25-gauge needle, and the concentration of bacilli was adjusted spectrophotometrically at 600 nm with reference to McFarland Equivalent Turbidity Standards (Remel, Lenex, KS); the viability of the organisms was determined by colony count (32).

M.tb bacilli were incubated in polypropylene tubes with monocytes that had been suspended in antibiotic-free CM at a bacilli:monocyte ratio of ∼5:1. Following 4 h of incubation at 37°C in a humidified atmosphere containing 5% CO₂, monocytes were washed thoroughly in PBS warmed to 37°C without Ca²⁺ and Mg²⁺ by low-speed centrifugation and incubated for 1 day in suspension in antibiotic-free CM in Teflon vessels. Noninfected monocytes that were maintained in the suspension culture for the same length of time served as a control. Before being used as targets, both M.tb-infected and noninfected monocytes were harvested and then washed in warmed PBS (37°C) by low-speed centrifugation; cell viability was determined by the exclusion of trypan blue. An aliquot of infected monocytes was analyzed to verify infection by acid-fast staining and colony count. At 1 day after infection, 31 to 53% of monocytes contained multiple bacteria. The number of viable bacteria/cell ranged between 18 and 36 CFU. The cell viability of both monocyte populations was >90%.

Target cells (B-LCLs or monocytes) were labeled with 150 μCi of ⁵¹Cr (ICN, Costa Mesa, CA) for 1 h at 37°C. A total of 5000 cells were added to round-bottom microtiter wells containing variable numbers of effector cells. In Ab-blocking experiments, CD8⁺ T cells and target cells were treated for 1 h with 10 μg/ml of Leu-2a or w6/32 mAb, respectively, before the addition of the opposing cells. All assays were performed in triplicate at 37°C in a 10% CO₂-humidified atmosphere. ⁵¹Cr release was measured at 5 h after the addition of B-LCLs or at 18 h after the addition of monocytes. The percentage of cytotoxicity was determined using the following formula: % specific cytotoxicity = 100 × ([experimental release − spontaneous release]/[maximum release − spontaneous release]). The maximum release was determined by the lysis of targets with 1% Triton X-100. Spontaneous release was <20% of the maximum release in all assays. A positive CTL response was defined as ≥15% difference in the lysis of target cells that had been pulsed with peptide or infected with M.tb and the corresponding untreated target cells at an E:T of 40:1.

All proliferation assays were performed in round-bottom microtiter wells in a final volume of 200 μl of CM. For these experiments, 1 × 10⁵ PBMCs were incubated with either 10⁸ heat-killed M.tb or 2 LfU/ml of tetanus toxoid in quadruplicate. Control cells were incubated in medium alone. Stimulation with Ag was conducted for 6 days at 37°C in a humidified 10% CO₂ atmosphere. Cellular proliferation was measured on the basis of the incorporation of [³H]thymidine that was added 6 h before harvesting. The SI was calculated using the following formula: SI = (mean cpm of cells cultured in the presence of stimulus)/(mean cpm of cells cultured in the absence of stimulus).

Using the MHC-binding motif for HLA-A*0201 (21), 28 peptides derived from M.tb19 were selected and synthesized for screening in a cell-based binding assay. Two previously described CTL epitopes, HIV gag A2 and HIV gag B8, were used as positive and negative controls, respectively. HLA-A*0201⁺ T2 cells, which have a defect in the assembly and transport of class I molecules (33, 34, 35), were used in these assays, because exogenously added HLA-A*0201-binding peptides can increase the number of properly folded HLA-A2 molecules on the cell surface. The increase in the surface expression of HLA-A2 molecules was measured by flow cytometry using mouse mAb to class I MHC (w6/32) and HLA-A2 (BB7.2) molecules.

A total of 5 of 28 candidate peptides stabilized HLA-A*0201 expression on T2 cells (FR > 2.0 at 100 μM peptide) (Table II and Fig. 1). Two peptides, corresponding to residues 14 to 22 and 88 to 97, bound to the cells with affinities that were comparable with that of the control HIV gag A2 peptide. As expected, HIV gag B8 had no effect on HLA-A*0201 expression by T2 cells (FR = 1.1) (Fig. 1).

To determine whether the selected HLA-A*0201-binding peptides were recognized by CD8⁺ T cells that were primed in vivo, PBMCs from an HLA-A*0201⁺ TSP subject were stimulated separately with each peptide; after 2 to 4 wk of expansion, CD8⁺ T cells that had been isolated by positive selection were analyzed for their cytotoxic activity against peptide-pulsed autologous B-LCLs. After 2 wk of stimulation, a CTL response (33% specific lysis at a 40:1 E:T ratio) was detected against B-LCLs that had been pulsed with P_88–97 (Fig. 2,A), and the intensity of lytic activity increased after each round of restimulation with the peptide, indicating a time-dependent enrichment of Ag-specific effectors (Fig. 2, B and C). The cultured effectors lysed P_88–97 pulsed B-LCLs in a dose-dependent manner at all three timepoints. Only weak lysis was observed for nonpulsed B-LCLs or for target cells that had been pulsed with either HIV gag A2 (Fig. 2) or two other 19-kDa-derived peptides, P_14–22 or P_101–108 (data not shown). The cytotoxic activity of these Ag-specific CTLs was markedly inhibited by Abs to CD8 or class I MHC (Fig. 2,D) but was not inhibited by Abs to CD4 (Fig. 2,D) or HLA-DR (data not shown). These results were confirmed in two other experiments that tested the responsiveness to P_88–97 of PBMCs that were obtained 11 mo later from the same donor (TSP-1) (Fig. 2,E) or PBMCs from a second TSP individual (donor TSP-2; 38% specific lysis at an E:T ratio of 40:1). In parallel experiments, CD8⁺ T cells isolated from the PBMCs of TSP-1 and TSP-2 subjects were incubated with an IMP. The cells from both subjects exhibited ≥42% lysis (Fig. 2 F and data not shown) of autologous targets that had been pulsed with IMP at an E:T ratio of 40:1. By contrast, P_14–22 and P_101–108 did not elicit CTL activity when tested repeatedly on PBMCs that had been derived from the same TSP participants (data not shown).

To determine whether P_88–97-specific CTLs are present in M.tb-infected persons, PBMCs from four HLA-A*0201⁺ patients with active TB were expanded in vitro in the presence of P_88–97 peptide. After 4 wk of culture, CD8⁺ T cells from three patients lysed autologous B-LCLs that were pulsed with P_88–97 (Fig. 3). The specificity of the response was established by the absence of cytotoxic activity against B-LCLs that were pulsed with HIV gag A2 (Fig. 3), and the MHC restriction was indicated by ≥64% inhibition of CTL-mediated lysis by Ab to either class I MHC or CD8 molecules (data not shown). T cells from one patient (TB-4) did not yield a strong CTL response against peptide-pulsed autologous B-LCLs or a vigorous proliferative response against M.tb (SI = 4) (Table I).

To determine whether the in vitro expansion of Ag-specific CD8⁺ CTLs reflects in vivo priming by exposure to mycobacteria rather than in vitro priming by the peptide in this experimental system, we measured the CTL response of CD8⁺ T cells that had been isolated from the PBMCs of two HLA-A*0201⁺ TSN individuals (blood donors N-4 and N-5) following 3 wk of in vitro incubation with M.tb P_88–97. In parallel, we assessed the cytolytic activity of PBMCs that had been stimulated with IMP. As depicted in Figure 4, A and C, P_88–97-specific CD8⁺ CTLs were not recovered when PBMCs from these TSN subjects were cultured in the presence of M.tb P_88–97. In contrast, incubating the PBMCs with IMP resulted in an expansion of peptide-specific CTLs (Fig. 4, B and D). These results, taken together with those detailed above, indicate that the peptide-specific CTL precursors that expanded in this experimental system had been primed in vivo.

Finally, to determine whether P_88–97 can induce a class I-restricted CTL response in naive T cells, CD8⁺ T cells isolated from the PBMCs of two HLA-A*0201⁺, M.tb-unresponsive subjects (N-1 and N-2) were stimulated with P_88–97 using an in vitro sensitization system that uses DCs as APCs (28, 36). After 4 wk of restimulation, the expanded CD8⁺ T cells were tested for CTL activity against autologous peptide-pulsed B-LCLs. As shown in Figure 5, P_88–97 induced a specific CTL response in both normal donors. These results confirm our earlier observation that human peripheral DCs, when appropriately pulsed with synthetic peptides, can sensitize naive CD8⁺ T cells and consequently enable the generation of Ag-specific CTL lines in vitro (36).

We examined the ability of peptide-specific CTLs to recognize endogenously synthesized epitopes by measuring their cytolytic activity against autologous monocytes that were acutely infected with tubercle bacilli. The results in Figure 6 show that CTLs that were derived from both in vivo- and in vitro-primed CD8⁺ T cells were able to recognize and lyse M.tb-infected monocytes in a class I MHC-restricted manner. As shown, only weak lysis was observed for uninfected monocytes. These results suggest that P_88–97 is generated by natural processing within the infected cells.

In summary, we used 8 to 10 aa synthetic peptides with a relatively high affinity for HLA-A*0201 molecules (Fig. 1) and two independent experimental systems to identify epitopes that elicit class I MHC-restricted CD8⁺ T cell responses against tubercle bacilli. One system detected peripheral blood CD8⁺ CTL populations that were primed in vivo (27), while the other used peptide-pulsed DCs as APCs to sensitize naive CD8⁺ T cells in vitro (28, 36). The results establish that class I MHC-restricted CD8⁺ CTL precursors with a specificity for P_88–97 are present in patients with active pulmonary TB as well as in TSP individuals without a history of TB. Moreover, peptide-specific CTLs that were expanded from in vivo-primed CD8⁺ T cells or generated from naive CD8⁺ T cells by in vitro sensitization with peptide-pulsed DCs recognized and lysed autologous targets that were acutely infected with tubercle bacilli in a class I MHC-restricted manner. Peripheral blood DCs that have been pulsed with exogenous peptides are potent stimulators of quiescent T cells (28, 37). In natural infection, DCs residing in airway epithelium and lung parenchyma (38, 39, 40) are among the first cells to encounter the bacillus and, presumably, to acquire and process mycobacterial Ags for presentation to naive T cells (41, 42). A recent report indicates that M.tb enters human DC progenitors in vitro, resulting in the activation and maturation of DCs (43).

These findings are potentially relevant for both vaccine development and adoptive immunotherapy. Epitopes that are generated by the intracellular processing of endogenously synthesized Ags are appropriate candidates for inclusion in the design of a peptide-based M.tb vaccine. The in vitro generation of M.tb-specific, biologically active effector cells potentially permits a large scale ex vivo expansion of CTLs for adoptive immunotherapy.

We thank the individuals who donated blood for this study as well as the personnel of the Division of TB Control, Department of Public Health (San Francisco, CA) whose cooperation made this work possible.