Induction of Cytotoxic T-Cell Responses against Culture Filtrate ...

3 downloads 0 Views 310KB Size Report
Brett, S., J. M. Orrell, J. Swanson Beck, and J. Ivanyi. 1992. Inffuence of ... Cooper, A. M., D. K. Dalton, T. A. Stewart, J. P. Griffin, D. G. Russell, and. I. C. M. Orme.
INFECTION AND IMMUNITY, Feb. 1997, p. 676–684 0019-9567/97/$04.0010 Copyright q 1997, American Society for Microbiology

Vol. 65, No. 2

Induction of Cytotoxic T-Cell Responses against Culture Filtrate Antigens in Mycobacterium bovis Bacillus Calmette-Gue´rin-Infected Mice OLIVIER DENIS, EVELYNE LOZES,

AND

KRIS HUYGEN*

Laboratory of Mycobacterial Immunology, Pasteur Institute of Brussels, B-1180 Brussels, Belgium Received 15 August 1996/Returned for modification 13 September 1996/Accepted 15 November 1996

CD81 T cells are essential for protection against mycobacteria, as is clearly demonstrated by the fatal outcome of experimental infection of b-2 microglobulin knockout mice. However, the mechanisms and antigens (Ags) leading to CD81 T-cell activation and regulation have been poorly characterized. Here we show that, upon immunization of major histocompatibility complex (MHC)-congenic mice with Mycobacterium bovis bacillus Calmette-Gue´rin (BCG), a cytotoxic response against BCG culture filtrate (CF) Ags (CFAgs) is induced in H-2b and H-2bxd haplotypes but not in H-2d haplotype. This response is mediated by CD81 T cells and absolutely requires the activation of CD41 T cells and their secretion of interleukin 2. The lack of cytotoxic response in H-2d mice cannot be explained by impaired cytokine production or by a defect in Ag presentation by H-2d macrophages. Using the MHC class I mutant B6.C-H-2bm13 mouse strain, we demonstrate that cytotoxic T lymphocytes (CTLs) recognize CFAgs exclusively in association with Db molecules. These Ags are cross-reactive in mycobacteria, since BCG-induced CTLs also recognize macrophages pulsed with CF from Mycobacterium tuberculosis H37Rv and H37Ra and from two virulent strains of M. bovis. Moreover, immunization with Mycobacterium kansasii induces CTLs able to lyse macrophages pulsed with BCG CF. Finally, we have found that these Ags can be characterized as hydrophilic proteins, since they do not bind to phenylSepharose CL-4B. Our results indicate that MHC-linked genes exert a profound influence on the generation of CD81 CTLs following BCG vaccination. BCG infection in mice. Cytolytic CD81 T cells have already been described upon mycobacterial infection (16), but the use of mycobacterium-infected macrophages as target cells did not allow the identification of the recognized antigens (Ags). Some reports have suggested that g/d T cells also contribute to protective immunity against mycobacteria, since g/d T cells in the peripheral blood of healthy individuals but also in mice are selectively activated and expanded by mycobacterial ligands (31–33). Some of these ligands have been identified as protease-resistant phosphorylated metabolites of a thymidine-containing nucleotide (13, 50). In addition, CD42 CD82 a/b Tcell lines generated from normal donor peripheral blood lymphocytes were shown to recognize mycolic acid from M. tuberculosis in a CD1-restricted fashion (51), but the in vivo function of these cells remains unknown. Identification of the specific Ags recognized by T cells is essential for evaluating their role in the protective immune response. Since protective immunity against M. tuberculosis is generated most efficiently by immunizing with live bacilli (44), interest has focused in recent years on proteins actively secreted and released into the medium of mycobacterial cultures. Culture filtrate (CF) Ags (CFAgs) are major targets of CD41 T cells (45), and effective vaccination of guinea pigs (46) and mice (1) against M. tuberculosis could be achieved by immunization with CF and adjuvants. Some of these CFAgs have been identified and cloned and were shown to induce CD41 T-cell proliferation and interleukin 2 (IL-2) and IFN-g production (2, 27; reviewed in reference 61). The characterization of mycobacterium-specific CD81 cytotoxic T lymphocytes (CTLs) and the identification of their target Ags, however, have been hampered by difficulties due to their recognition in the context of MHC class I molecules, since exogenous soluble Ags are generally not introduced in this presentation pathway. Therefore, Ags recognized by CD81 T cells remain largely unknown. In

Human tuberculosis caused by Mycobacterium tuberculosis is responsible for approximately 8 million new cases and 3 million deaths annually and is by far the most prevalent infectious disease worldwide (5). The only available vaccine is the attenuated Mycobacterium bovis BCG. Although BCG is one of the most frequently used vaccines, its efficacy remains controversial, since human trials have demonstrated protection ranging from 0 to 80% (52). The increased incidence of tuberculosis and the lack of an effective vaccine have given a new impetus to tuberculosis research, since a comprehensive understanding of the mechanisms leading to protection is essential for the development of improved or new vaccines. Protective immunity against mycobacteria is considered to be essentially cell mediated, dependent on T lymphocytes locally secreting molecules such as gamma interferon (IFN-g), which stimulate the antimicrobial activity of infected macrophages, allowing intracellular bacterial killing (59). The crucial role of IFN-g in acquired protection has been clearly demonstrated by antibody neutralization experiments and by the fatal outcome of BCG and M. tuberculosis infection in IFN-g or IFN-g-receptor knockout mice (4, 14, 20, 34). Although T cells are known to be necessary for a protective immune response, the relative contribution of CD41 and CD81 T-cell populations in protection against M. tuberculosis infection is not clearly defined. Early studies with antibody depletion of T-cell subsets in vivo (26, 41, 43, 49) and the recent availability of major histocompatibility complex (MHC) class I and MHC class II knockout mice (21, 38) have shown that CD41 or CD81 T cells or both are required to control M. tuberculosis or * Corresponding author. Mailing address: Laboratory of Mycobacterial Immunology, Pasteur Institute of Brussels, Engelandstraat 642, B-1180 Brussels, Belgium. Phone: 32 2 373 33 70. Fax: 32 2 373 31 74. E-mail: [email protected]. 676

VOL. 65, 1997

CTL RESPONSES IN MYCOBACTERIUM-INFECTED MICE

677

this report, we describe the generation and regulation of a cytotoxic CD81 T-cell response against CF-pulsed macrophages in mice immunized with BCG. Our results demonstrate the feasibility of using exogenously CF-pulsed macrophages for direct identification of the molecular targets of mycobacterium-specific CD81 T cells. MATERIALS AND METHODS Mice. BALB/c (H-2d Bcgs), DBA/2 (H-2d Bcgr), C3H (H-2k Bcgr), C57BL/6 (H-2b Bcgs), C57BL/10 (H-2b Bcgs), 129 Sv (H-2b Bcgr), BALB.B10 (BALB/c background, H-2b Bcgs), B6.C-H-2bm12 (H-2bm12 Bcgs), B6.C-H-2bm13 (H-2bm13 Bcgs), B10.A (H-2a Bcgs), and (C57BL/6 3 BALB/c)F1 (H-2bxd Bcgs) mice were obtained from the animal house of the Pasteur Institute from breeding pairs initially obtained from the Netherlands Cancer Institute or the University of Louvain, Brussels, Belgium (129 Sv). C.D2 (BALB/c background, H-2d Bcgr) mice were kindly given by E. Skamene (8). B6.C-H-2bm12 mutant and C57BL/6 mice differ by three productive nucleotides within a stretch of 14 nucleotides in the exon encoding the first extracellular domain of the MHC class II Ab gene, leading to three amino acid changes: Ile to Phe at position 67, Arg to Gln at position 70, and Thr to Lys at position 71 (24). B6.C-H-2bm13 mice carry mutations in the a2 domain of the MHC class I Db gene: Leu to Gln at position 114, Phe to Tyr at position 116, and Glu to Asp at position 119 (7). Female mice were 8 to 10 weeks old at the beginning of all experiments. BCG infection and CFAgs. Mice were inoculated intravenously in a lateral tail vein with 0.5 mg (4 3 106 CFU) of freshly prepared M. bovis BCG GL2 (Pasteur Institute of Brussels) as described previously (29) or 1 3 106 CFU of Mycobacterium kansasii. For secondary responses, mice were given an intravenous boost injection of 0.5 mg of BCG or 106 CFU of M. kansasii 2 months after the first injection. CFs were obtained from 14-day-old zinc-deficient cultures of M. bovis BCG GL2, from zinc-containing cultures of M. tuberculosis H37Rv and H37Ra, and from two virulent M. bovis strains (AN5 and VF61) grown as a surface pellicle culture on synthetic Sauton medium. Proteins were concentrated by precipitation with ammonium sulfate (80% saturation), extensively dialyzed against phosphate-buffered saline (PBS), and sterilized through a 0.2-mm-poresize filter. In vitro restimulation of CTLs in bulk cultures. Spleens were removed aseptically 4 weeks after the BCG injection, at the earliest, and spleen cells (2.5 3 106 cells/ml) were cultured for 1 week in RPMI 1640 medium supplemented with 10 mg of gentamicin per ml, 2 mM L-glutamine, 5 3 1025 M 2-mercaptoethanol, 10% fetal calf serum (FCS), and 15 mg of BCG CF per ml in 24-well plates. In some experiments, 10 U of IL-2 (CLB, Amsterdam, Holland) per ml was added to the culture medium. Cultures were maintained in 5% CO2 at 378C. Neutral red assay. Target cells were peritoneal exudate cells (PEC) elicited 1 week before harvesting by intraperitoneal thioglycolate (Difco Laboratories, Detroit, Mich.) injection. Macrophages were harvested by washing the peritoneal cavity twice with 8 ml of ice-cold PBS. PEC were washed and cultured for 4 h in round-bottom microplates (105 macrophages/well) in RPMI 1640 medium supplemented with 20 mg of CF per ml. Effector cells were harvested from bulk cultures, washed, and added in triplicate to target cells, at the effector cell/target cell ratios (E/T ratios) indicated in the figures, in RPMI containing 10% FCS. After 16 h of coculture, target lysis was analyzed by a slight modification of the technique described by Parish and Mullbacher (48). Briefly, microplates were gently washed to remove supernatants containing effector cells; then 0.036% neutral red in Hanks balanced salt solution was added. Fifteen minutes later, microplates were washed three times with PBS and the neutral red was released by the addition of 100 mM acetic acid and ethanol. Optical densities (ODs) were measured at 540 nm in an enzyme-linked immunosorbent assay microspectrophotometer. Results were expressed as percent lysis of target cells at each E/T ratio, based on the formula percent lysis 5 [(C-B)2(E-B)]/(C2B) 3 100, where C is the mean OD of macrophages without effector cells, B is the mean OD of wells without cells, and E is the mean OD of macrophages plus effector cells. T-cell subset depletion. Lymphocytes were purified on Lympholyte-M (Cedarlane, Hornby, Ontario, Canada), and 5 3 106 cells/ml were incubated with monoclonal antibody (MAb) RL172 (anti-CD4) or 83.12.5 (anti-CD8) culture supernatants for 45 min at 378C. Spleen cells were then pelleted and resuspended in low-toxicity rabbit serum (Cedarlane) as a complement source for 45 min at 378C. Finally, spleen cells were washed twice before use. Immunofluorescence staining. CD4-CD8 double staining was performed with fluorescein isothiocyanate-labelled rat anti-mouse L3T4 MAb (clone RM4-5; Pharmingen, San Diego, Calif.) and phycoerythrin-labelled rat anti-mouse CD8b MAb (clone 53-5.8; Pharmingen). One hundred microliters of cells (107 cells/ml) was incubated in PBS containing 5% FCS, 0.1% NaN3, and 1 mg of each antibody for 30 min at 48C. Cells were washed twice and fixed in 200 ml of PBS with 1% paraformaldehyde. Fluorescence was analyzed on a FACScan cytofluorometer (Becton Dickinson, Mountain View, Calif.) with Lysis II software. Live lymphocytes were gated on the basis of their characteristic forward and side scatter profile. CF fractionation with hydrophobic chromatography. Proteins in CF were fractionated on the basis of their hydrophobicity as previously described (15). Briefly, a phenyl-Sepharose CL-4B column (Pharmacia, Uppsala, Sweden) was

FIG. 1. BCG immunization induces a CF-specific cytotoxic response. Spleen cells from BALB.B10 mice were obtained 6 weeks after primary or secondary immunization with BCG and amplified with BCG CF. Effector cells were harvested 1 week later, and cytotoxic activity was determined on syngeneic BCG CF-pulsed or unpulsed PEC as described in Materials and Methods. Cultures were performed in triplicate. Standard errors of the mean (SEM) were below 8%. This experiment was performed twice with similar results.

equilibrated with 20 mM phosphate buffer containing 0.45 M NaCl. The salinity of the CF was adjusted to that of the column before injection. Unbound material was termed L0. The column was then washed with 1 column volume of the same buffer to remove unfixed material (L1) and successively with 20 mM (L2) and 4 mM (L3) phosphate buffer. Protein analysis was done by sodium dodecyl sulfatepolyacrylamide gel electrophoresis on 15% (wt/vol) acrylamide gels with Mini Protean II apparatus (Bio-Rad Laboratories, Richmond, Calif.). Gels were stained with Sypro-Orange (Bio-Rad) and visualized on a 302-nm-wavelength box with Molecular Analysist software (Bio-Rad).

RESULTS b

BCG-immunized H-2 but not H-2d mice can mount cytotoxic responses specific for CFAgs. In preliminary experiments, C57BL/6 and BALB.B10 mice were immunized intravenously with BCG and the cytotoxic response was analyzed after a primary and a secondary immunization. Cytotoxic activity after bulk culture was monitored on syngeneic PEC that were unpulsed or pulsed with BCG CF. Cultured BALB.B10 (but also C57BL/6) spleen cells developed strong cytolytic activity against CF-pulsed macrophages, particularly after a secondary BCG immunization, as shown by analysis with the neutral red uptake assay (Fig. 1). Therefore, only mice immunized twice with BCG were used throughout the rest of this study. This response was also monitored by a standard 4-h 51Cr release assay with EL-4 cells as targets. EL-4 cells pulsed with CF could also be lysed efficiently by effector cells, although the percentages of cytotoxicity observed were lower than with the neutral red uptake assay (data not shown). Specific cytolytic activity could be measured beginning 4 to 5 weeks after immunization and up to at least 25 weeks. A weak cytolysis of unpulsed target cells was frequently observed early after immunization and was probably due to NK cell or g/d T-cell activation. No response was found directly ex vivo before bulk culture or after restimulation of cells from nonimmune mice (data not shown). We analyzed various strains of mice for their ability to mount cytotoxic responses against CFAgs upon immunization with BCG. Cultured spleen cells from BALB.B10, C57BL/6, C57BL/10, 129 Sv, (C57BL/6 3 BALB/ c)F1, and B6.C-H-2bm12 mice all demonstrated comparable cytolytic activity against CF-pulsed syngeneic macrophages,

678

DENIS ET AL.

INFECT. IMMUN.

FIG. 2. Analysis of the ability of mice of different strains to mount cytotoxic responses upon BCG immunization. Spleen cells from C57BL/10, (C57BL/6 3 BALB/c)F1, B6.C-H-2bm12, B6.C-H-2bm13, and BALB/c mice, obtained 5 weeks after secondary immunization, and spleen cells from 129 Sv mice, obtained 7 weeks after secondary immunization, were cultured for 1 week with BCG CF. Cytotoxicity in bulk cultures was assessed on BCG CF-pulsed or unpulsed syngeneic PEC. Cultures were performed in triplicate. SEM were below 12%. As for BALB/c mice, no cytotoxicity was detected in C.D2 and DBA/2 mice even 8 weeks after immunization. Each experiment was repeated at least twice.

whereas bulk cultures of B6.C-H-2bm13 (C57BL/6 mice carrying mutations in the MHC class I D molecule) spleen cells were devoid of cytotoxic activity (Fig. 2). When supernatants from the cytolysis assay were harvested and tested for their cytolytic potential on fresh syngeneic peritoneal macrophages, no lysis was detected, indicating that cytolysis was genuinely cell mediated (data not shown). Furthermore, we were unable to detect a specific cytotoxic response in BALB/c, C.D2, and DBA/2 mice (Fig. 2 and data not shown). The lack of cytotoxic response in these H-2d strains was not due to a peculiar resistance of target cells to cell lysis, since H-2bxd, H-2d, H-2b, and mutant H-2 haplotype macrophages were equally lysed by CTLs generated during H-2b 3 H-2d or H-2d 3 H-2b mixed lymphocyte cultures (data not shown). Cytotoxic response is mediated by CD81 T cells and requires CD41 T-cell activation and IL-2 secretion. The phenotype of the effector cells mediating cytotoxicity was determined by complement-mediated lysis of T-cell subsets. Spleen cells were harvested from bulk cultures and were left untreated, incubated with complement only, or depleted of CD41 or CD81 T cells and assessed for CTL activity on syngeneic CFpulsed PEC. After depletion, the remaining population contained less than 3% CD41 or CD81 T cells, as shown by analysis by cytofluorometry. Figure 3 shows that depletion of CD81 T cells totally abrogated the cytotoxic activity of effector cells, while CD41 T-cell depletion had no effect. This experiment clearly demonstrated that BCG immunization induces a CTL response that is mediated by CD81 T cells and is specific for CFAgs. We next investigated the role of IL-2 secretion during bulk culture for efficient CTL generation. Spleen cells were cultured with IL-2, CF, or both, and cytotoxic activity in these bulk

cultures was analyzed on unpulsed or CF-pulsed macrophages. Effector cells cultured in the presence of IL-2 showed increased nonspecific cytotoxic activity against unpulsed macrophages at higher E/T ratios, probably as a result of NK and/or g/d T-cell activation (Fig. 4A). Culturing with IL-2 in the presence of CF increased the specific cytolysis of CF-pulsed mac-

FIG. 3. Cytotoxic response to BCG CFAgs is mediated by CD81 T cells. Effector cells from C57BL/6 were generated 12 weeks after secondary immunization. Effector cells were harvested from bulk cultures, washed, divided into five equal parts, and then counted to determine the E/T ratio. Effector cells were left untreated, and their cytolytic activity was tested on BCG CF-pulsed or unpulsed syngeneic PEC. Effector cells were also incubated with rabbit complement (Ctreated) or depleted of CD41 or CD81 T cells upon incubation with MAb RL172 or MAb 83.12.5, respectively, and rabbit complement. Effective cell depletion was confirmed by cytofluorometry. Cultures were performed in triplicate. SEM were below 10%. These results are representative of three experiments.

VOL. 65, 1997

CTL RESPONSES IN MYCOBACTERIUM-INFECTED MICE

679

FIG. 4. CTL generation in bulk culture requires IL-2 secretion by CD41 T cells. Spleen cells from C57BL/6 mice were obtained 13 weeks after secondary immunization. Cells were cultured with IL-2, BCG CF, or both, and cytotoxic activity in bulk culture was analyzed on unpulsed (A) or BCG CF-pulsed (B) syngeneic PEC. Spleen cells were also treated with complement only (C-treated) and cultured with BCG CF or depleted of CD41 T cells and cultured with BCG CF alone or supplemented with IL-2 (C). Cytotoxicity in bulk culture was assessed on BCG CF-pulsed syngeneic PEC. Cultures were performed in triplicate. SEM were below 10%.

rophages, indicating a requirement for IL-2 secretion for optimal CTL generation (Fig. 4B). However, this increase could also simply reflect an additive effect of CF-specific CTL response and nonspecific NK activity increased by IL-2. To further demonstrate the importance of IL-2, CD41 T cells (the main T-cell subset synthesizing IL-2) were depleted before bulk culture and the remaining cells (consisting of less than 4% CD41 T cells, as analyzed by cytofluorometry) were cultured with CF alone or supplemented with IL-2. CD41 T-cell depletion before bulk culture totally abrogated the cytotoxic response. However, addition of IL-2 during culture restored this response even above that of the control culture (Fig. 4C). Depletion of CD81 T cells before bulk culture abolished the lysis of CF-pulsed macrophages, even when IL-2 was added in the culture (data not shown). Therefore, although some nonspecific lymphokine-activated killer cell-mediated CTL activity may be generated by IL-2 treatment, in our experimental system, this will account for a minimal percentage compared to the major contribution of CD81 T cells. These results show that efficient CD81 CTL generation requires the activation of IL-2-producing CD41 T cells, at least during the amplification process in bulk culture. The inability of H-2d mice to mount a CTL response upon immunization with BCG cannot be explained by a defect in Ag presentation. Since (C57BL/6 3 BALB/c)F1 mice mounted a CTL response, we investigated the MHC restriction of this response by using CF-pulsed PEC from different H-2 haplotypes. Figure 5 shows that syngeneic (C57BL/6 3 BALB/c)F1 macrophages but also C57BL/6 and B6.C-H-2bm12 PEC were efficiently lysed by CTLs generated in H-2bxd mice. However, B6.C-H-2bm13 macrophages were not recognized by CTLs

from H-2bxd mice, indicating that CTL peptides were exclusively presented in association with Db molecules. Moreover, these class I mutant macrophages were not recognized by C57BL/6 CTLs either (data not shown). Finally, BALB/c PEC

FIG. 5. MHC restriction of the CTL response generated in (C57BL/6 3 BALB/c)F1 mice. Spleen cells were obtained 7 weeks after secondary immunization and cultured with BCG CF. Cytotoxicity in bulk culture was analyzed on BCG CF-pulsed PEC from (C57BL/6 3 BALB/c)F1 [(C573 Ba)F1], C57BL/6, BALB/c, C3H, B6.C-H-2bm12 (bm12), and B6.C-H-2bm13 (bm13) mice. Cultures were performed in triplicate. SEM were below 8%. This experiment was conducted twice with similar results.

680

DENIS ET AL.

INFECT. IMMUN.

In order to know whether this CTL response could also be induced upon infection with other mycobacteria, not belonging to the M. bovis-M. tuberculosis complex, C57BL/6 mice were immunized with M. kansasii. These mycobacteria were chosen because their protective efficacy against an M. tuberculosis challenge more or less equals that of BCG (47). CD81 T cells from M. kansasii-immunized mice cultured in vitro with BCG CF efficiently lysed BCG CF-pulsed macrophages (Fig. 7B), showing that these mycobacteria can also activate in vivo CTLs specific for cross-reactive Ags in CF. Partial purification of Ags recognized by CTLs. Proteins in BCG CF were fractionated on the basis of their hydrophobicity on phenyl-Sepharose CL-4B. The first two fractions collected (L0 and L1) were very similar in composition and contained weakly hydrophobic proteins, while fractions L2 and L3 showed different patterns of proteins with increasing hydrophobicity (Fig. 8A). These four fractions were used to sensitize macrophages for CTL recognition. As shown in Fig. 8B, only PEC pulsed with fractions L0 and L1, but not PEC pulsed with fractions L2 or L3, were killed. This lysis was even higher than

FIG. 6. Recombinant B10.A mice mount a CTL response restricted to the H-2d allele. B10.A mice (Kk Ak Ek Dd Ld) were immunized twice with BCG. Ten weeks later, spleen cells were cultured with BCG CF and the cytotoxic activity in bulk cultures was assessed on unpulsed or BCG CF-pulsed syngeneic PEC (A). Cytotoxic activity was also analyzed on BCG CF-pulsed syngeneic, BALB/c, or C3H macrophages (B). Cultures were performed in triplicate. SEM were below 7%. These results are representative of three experiments.

were not lysed by CTLs from H-2bxd mice, showing that (C57BL/6 3 BALB/c)F1 mice mounted an exclusively H-2brestricted CTL response. One explanation of these results could be that H-2d macrophages are unable to present Ags recognized by CTLs. However, recombinant B10.A mice (Kk Ak Ek Dd Ld) mounted a CTL response after two injections of BCG (Fig. 6A). The H-2 restriction of this response was investigated by using CF-pulsed PEC from BALB/c (H-2d) and C3H (H-2k) mice as target cells. C3H macrophages were not lysed, while BALB/c PEC were killed by B10.A CTLs, although to a lesser extent than syngeneic macrophages (Fig. 6B). Therefore, in B10.A mice, CTLs seem to recognize Ags in association with molecules common to B10.A and BALB/c macrophages, most likely Dd or Ld molecules. CTLs generated in mycobacterium-infected mice recognize cross-reactive mycobacterial CFAgs. To determine whether Ags recognized by CTLs generated upon BCG immunization are specific for this vaccine or not, we analyzed the cytolysis of PEC pulsed with CF from two virulent M. bovis strains (AN5 and VF61) and from M. tuberculosis H37Rv and H37Ra, using BCG-induced CTLs. Figure 7A shows that not only BCG CF but also M. tuberculosis and M. bovis CF were able to sensitize PEC for CTL recognition, clearly demonstrating that the Ags recognized are shared by all these mycobacteria. Several purified native or recombinant mycobacterial antigens (39, 61), including Ag85, hsp65, hsp10, and PstS-2, were analyzed, but none was able to sensitize PEC for cell lysis.

FIG. 7. Ags recognized by BCG-induced CTL are ubiquitous in mycobacteria. (A) Effector cells from BCG-immunized BALB.B10 mice were generated 7 weeks after secondary immunization. Cytotoxicity was assessed on syngeneic macrophages that were unpulsed or pulsed with CF from BCG (20 mg/ml), M. tuberculosis H37Rv (40 mg/ml), M. tuberculosis H37Ra (40 mg/ml), M. bovis AN5 (50 mg/ml), or M. bovis VF61 (20 mg/ml). (B) Spleen cells from M. kansasiiimmunized C57BL/6 mice were restimulated in culture with BCG CFAgs for 1 week. Effector cells were then treated with complement only (C-treated) or depleted of CD41 or CD81 cells, and cytotoxicity was then analyzed on unpulsed or BCG CF-pulsed syngeneic macrophages. Cultures were performed in triplicate. SEM were below 12%. These experiments were conducted twice with similar results.

VOL. 65, 1997

CTL RESPONSES IN MYCOBACTERIUM-INFECTED MICE

681

FIG. 8. CFAgs recognized by CTLs are hydrophilic proteins. CF from zinc-deficient stressed BCG cultures was fractionated on a phenyl-Sepharose CL-4B column and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (A). L0 fraction contains proteins which do not bind to phenyl-Sepharose. L1, L2, and L3 were fractions recovered with decreasing salt concentrations in 20 mM phosphate buffer containing 0.45 M NaCl, 20 mM phosphate buffer, and 4 mM phosphate buffer, respectively. These four fractions were used to sensitize macrophages for recognition by CTLs from BALB.B10 mice immunized twice with BCG (B). Cultures were performed in triplicate. SEM were below 12%. This experiment was conducted twice with similar results.

that of whole CF-pulsed macrophages, demonstrating that target Ags were mostly hydrophilic proteins. DISCUSSION It is now well established that CD81 CTLs play a major role in the control of M. tuberculosis and M. bovis BCG infection (21, 38, 41, 43, 49). CTLs are currently assumed to provide surveillance for mycobacteria surviving in macrophages, destroy these chronically infected cells by apoptosis, and thereby purge the host of a reservoir of pathogens. CD81 T cells also secrete IFN-g, which can stimulate the antimicrobial activity of infected macrophages. Identification of target Ags recognized by CD81 T cells is therefore important for understanding host immune regulation of infection and for development of more effective vaccination strategies. It was previously reported that mycobacterium-specific cytolytic CD81 T-cell lines could be obtained after several rounds of restimulation with mycobacterially infected macrophages in vitro (16). However, because of the use of whole bacteria, characterization of recognized Ags was impossible. In this study, we have used soluble CFAgs for CTL amplification in bulk culture and for sensitization of target cells paired with colorimetric detection of cell lysis to demonstrate that BCG-immunized H-2b mice mount a robust CTL response specific for soluble Ags in CF. We found that CF-specific CTLs generated upon BCG immunization were of genuine CD81 T-cell phenotype and that Ags were recognized in association with MHC class I molecules, more precisely Db molecules, since MHC class I mutant B6.C-H-2bm13 mice were unable to mount a CTL response and CF-pulsed B6.C-H-2bm13 macrophages were not recognized by H-2b or H-2bxd CTLs. This exclusive presentation in association with Db molecules is surprising, since in vitro amplification and target cell sensitization was done with a complex mixture of CF proteins (12). It could have been expected that presentation of so many proteins would have involved both Db and Kb molecules. Therefore, our results strongly suggest that relatively few (or maybe only one) proteins serve as efficient tar-

gets for CTL recognition in these BCG-infected mice. However, we should point out that our in vitro system using macrophages pulsed with exogenous Ags could also favor unique Ags displaying special features facilitating their internalization, processing, or binding to MHC class I molecules (see below). Similar results have been observed with mice infected with Toxoplasma gondii, where CTLs restimulated in vitro with irradiated tachyzoites efficiently lysed soluble Agpulsed macrophages (17). In that study, it was also hypothesized that CTLs recognized only a single Ag. It is well known that CTLs can secrete low amounts of IL-2 and use it as an autocrine growth factor (3), but there remains substantial controversy as to whether MHC class I-restricted CTLs can act autonomously or whether they require MHC class II-restricted CD41 helpers to differentiate and/or expand. Indeed, CTLs have been generated both in vitro (30, 42) and in vivo (11, 23) in the absence of CD41 T cells. Our results demonstrated that IL-2-producing CD41 T cells were necessary for amplification of BCG-specific CTLs in vitro; however, their requirement for CTL activation in vivo remains to be elucidated. Another finding of this study was the complete absence of CF-specific CTLs in three H-2d mouse strains analyzed, i.e., BALB/c (Bcgs), C.D2 (Bcgr), and DBA/2 (Bcgr). We have previously reported that CFAgs selectively induce a strong Th1 response in BCG-infected C57BL/6 and BALB.B10 mice, with elevated levels of IL-2 and IFN-g, whereas in BALB/c mice this Th1 response is lower and is partly counterbalanced by Th2 cells secreting IL-4 (27). However, it seems unlikely that the lower Th1 reactivity generally observed in the latter strain or a differential cytokine synthesis in vivo (during priming) or in vitro (during bulk cultures) can account for the lack of CTLs, at least for BALB/c mice, since in (C57BL/6 3 BALB/c)F1 mice, H-2b-restricted CTLs were observed while H-2d-restricted CTLs were absent. Moreover, we recently succeeded in inducing a strong protective immune response, including cytolytic CD81 T-cell activation, in BALB/c mice vaccinated with plasmid DNA encoding the 32-kDa fibronectin-binding Ag85A

682

DENIS ET AL.

(28), demonstrating that this mouse strain is indeed capable of mounting strong CTL responses against other CFAgs. Inefficient Ag processing or presentation of generated peptides in association with Kd, Dd, or Ld molecules also seems unlikely, since BALB/c macrophages could efficiently present CFAgs to B10.A CTLs, demonstrating that at least Dd or Ld molecules can associate with the relevant peptides. Since the number of CFAgs recognized appears to be very low, another possibility is that CTLs are not activated in BALB/c mice as a consequence of cross-reactivity between these Ags and host Ags, leading to a hole in the repertoire of these mice. Up to now, the relevance of CF-specific CTL induction with respect to immune protection against mycobacterial infection in various mouse strains is unknown. To our knowledge, comparative studies, using mouse strains with various H-2 haplotypes and analyzing the generation of protective immunity against an M. tuberculosis challenge following BCG vaccination, have not yet been performed. However, it is well known that MHC-linked genes influence the outcome of infection. Particularly, resistance to M. tuberculosis in the lungs at the later stages of infection seems to be associated with the lack of expressed I-E molecules and/or with the expression of Db molecules (9). Moreover, this lack of CTLs in H-2d mice correlates well with the lower immune memory, Th1 cytokines, and NO radical production in BALB/c mice upon BCG immunization compared to C57BL/6 or BALB.B10 mice (26, 60). Ags recognized by CTLs were present not only in CF from BCG but also in CF from M. tuberculosis and from virulent M. bovis. Moreover, CTLs were activated in vivo not only by BCG vaccination but also after infection with M. kansasii, and the CTL response was increased after a second immunization. Separation of these Ags on a phenyl-Sepharose column indicated that they are hydrophilic proteins. Moreover, contaminating small peptides were not the target of CTLs, since the macrophage-sensitizing activity was recovered in the excluded volume when CF was subjected to gel filtration chromatography on Sephadex G-25 (data not shown). One intriguing question is the mechanism of CTL activation by exogeneous CF proteins. It is currently admitted that MHC class I-restricted epitopes are exclusively generated endogenously in the host cell cytoplasm, whereas MHC class II-restricted presentation involves endocytosis of exogenous soluble Ags and degradation in a low-pH endosomal compartment (22). According to this view, neither infection with BCG, which is thought to reside exclusively in the phagosome, nor macrophage sensitization with soluble Ags would be capable of activating class I-restricted CD81 T cells. However, a number of recent studies have demonstrated that this segregation of cytoplasmic and soluble Ags into class I and class II presentation pathways, respectively, is not as absolute as initially thought (reviewed in reference 53). For instance, injection of soluble ovalbumin in adjuvants or of antigen-presenting cells pulsed in vitro with hepatitis B surface Ag particles has been shown to induce CD81 CTL responses (6, 35). Other studies further dissecting the class I presentation pathway of soluble Ags have demonstrated that macrophages are able to present phagocytosed exogenous particulate Ags to class I-restricted hybridomas and stimulate their IL-2 secretion (37), but the mechanisms and compartments where Ag degradation occurs remain controversial (25, 36, 55). Soluble heat shock protein-chaperoned polypeptides were also shown to be efficiently presented on macrophage MHC class I molecules (58). However, in these studies, very few primary macrophages exposed to exogenous Ags were killed by cloned CTLs (54). Taken together, these results have led to the concept that there exists a specialized macrophage subpopulation that shuttles soluble Ags into the

INFECT. IMMUN.

class I-restricted presentation pathway. However, our findings that the nonphagocytic lymphoma EL-4 could also be efficiently lysed by CTLs in a standard 51Cr release assay indicate that the presentation of CFAgs is probably not related to this alternative class I presentation. This presentation was not due to EL-4 membrane leakage, since these cells were unable to present soluble Ag85 to CTLs induced in vivo by injection of naked DNA encoding Ag85 (reference 28 and data not shown). Other exceptions to the conventional view of Ag presentation to CD81 T cells relate to the unique ability of some bacterial toxins to be delivered in the cytosolic compartment. Some bacterial toxins, such as listeriolysin, can disrupt membranes, allowing escape of the bacteria into the cytoplasm and generation of CD81 T-cell-specific epitopes (10). Some other toxins can insert themselves into planar lipid bilayers and penetrate into the cytoplasm of eucaryotic cells (40, 57). Moreover, polypeptides genetically fused to some bacterial toxins have been shown to efficiently sensitize target cells for MHC class I-restricted CTL recognition (18, 56) or induce CTL activation in vivo (19). As discussed earlier, our results indicate that very few (or maybe only one) CF proteins are the targets of CTLs, favoring the hypothesis that special biochemical features displayed by recognized CFAgs, rather than a special presentation pathway, could explain the very efficient CTL detection by using soluble Ag-pulsed macrophages. Maybe a kind of cell-invasive protein is secreted by mycobacteria in the endosomal compartment, penetrates into the macrophage cytoplasm, and subsequently is processed as a classical class Irestricted Ag. Our further work will focus on the purification and characterization of this Ag. ACKNOWLEDGMENTS Part of this work was supported by grant 3.0020.89 from the NFWO (Belgian National Research Foundation). We thank Muriel Moser and Veronique Cabiaux (ULB, Brussels, Belgium) for helpful comments and discussions and Jacques Urbain (ULB, Brussels, Belgium) for providing FACScan utilities and for his interest in this work. The anti-CD4 (RL172) and anti-CD8 (83.12.5) hybridomas were gifts from Muriel Moser. C.D2 mice were provided by E. Skamene (Montreal General Hospital Research Institute, Montreal, Quebec, Canada). The M. tuberculosis and M. bovis CFs were prepared by Jean Nyabenda (Pasteur Institute, Brussels, Belgium). The excellent technical assistance of Camille Palfliet and Albert Vanonckelen is also gratefully acknowledged. REFERENCES 1. Andersen, P. 1994. Effective vaccination of mice against Mycobacterium tuberculosis infection with a soluble mixture of secreted mycobacterial proteins. Infect. Immun. 62:2536–2544. 2. Andersen, P., A. B. Andersen, A. L. Sorensen, and S. Nagai. 1995. Recall of long-lived immunity to Mycobacterium tuberculosis infection in mice. J. Immunol. 154:3359–3372. 3. Andrus, L., A. Granelli-Piperno, and E. Reich. 1984. Cytotoxic T cells both produce and respond to interleukin 2. J. Exp. Med. 59:647–652. 4. Banerjee, D. K., A. K. Sharp, and D. B. Lowrie. 1986. The effect of gammainterferon during Mycobacterium bovis (BCG) infection in athymic and euthymic mice. Microb. Pathog. 1:221–229. 5. Bloom, B. R. 1992. Tuberculosis: commentary on a reemergent killer. Science 257:1055–1064. 6. Bo¨hm, W., R. Schirmbeck, A. Elbe, K. Melber, D. Diminky, G. Kraal, N. Van Rooijen, Y. Barenholz, and J. Reimann. 1995. Exogenous hepatitis B surface antigen particles processed by dendritic cells or macrophages prime murine MHC class I-restricted cytotoxic T lymphocytes in vivo. J. Immunol. 155: 3313–3321. 7. Boog, C. J. P., J. Boes, and C. J. M. Melief. 1988. Role of dendritic cells in the regulation of class I restricted cytotoxic T lymphocyte responses. J. Immunol. 140:3331–3337. 8. Bourassa, D., A. Forget, M. Pelletier, and E. Skamene. 1985. Cellular immune response to Mycobacterium bovis (BCG) in genetically-susceptible and resistant congenic mouse strains. Clin. Exp. Immunol. 62:31–38.

VOL. 65, 1997

CTL RESPONSES IN MYCOBACTERIUM-INFECTED MICE

9. Brett, S., J. M. Orrell, J. Swanson Beck, and J. Ivanyi. 1992. Influence of H-2 genes on growth of Mycobacterium tuberculosis in the lungs of chronically infected mice. Immunology 76:129–132. 10. Brunt, L. M., D. A. Portoy, and E. R. Unanue. 1990. Presentation of Listeria monocytogenes to CD81 T cells requires secretion of hemolysin and intracellular bacterial growth. J. Immunol. 145:3540–3546. 11. Buller, R. M. L., K. L. Holmes, A. Hu ¨gin, T. N. Frederickson, and H. C. Morse. 1987. Induction of cytotoxic T-cell responses in vivo in the absence of CD4 helper cells. Nature 328:77–79. 12. Closs, O., M. Harboe, N. H. Axelsen, K. Bunch-Christensen, and M. Magnusson. 1980. The antigens of M. bovis BCG studied by crossed immunoelectrophoresis: a reference system. Scand. J. Immunol. 12:249–263. 13. Constant, P., F. Davodeau, M. Peyrat, Y. Poquet, G. Puzo, M. Bonneville, and J. Fournie´. 1994. Stimulation of human g/d T cells by nonpeptidic mycobacterial ligands. Science 264:267–270. 14. Cooper, A. M., D. K. Dalton, T. A. Stewart, J. P. Griffin, D. G. Russell, and I. C. M. Orme. 1993. Disseminated tuberculosis in interferon g gene-disrupted mice. J. Exp. Med. 178:2243–2247. 15. De Bruyn, J., K. Huygen, R. Bosmans, M. Fauville, R. Lippens, J. P. Van Vooren, P. Falmagne, M. Weckx, H. G. Wiker, M. Harboe, and M. Turneer. 1987. Purification, characterization and identification of a 32 kDa protein antigen of Mycobacterium bovis BCG. Microb. Pathog. 2:351–366. 16. De Libero, G., I. Flesch, and S. H. E. Kaufmann. 1988. Mycobacteriareactive Lyt-21 T cell lines. Eur. J. Immunol. 18:59–66. 17. Denkers, E. Y., R. T. Gazzinelli, S. Hieny, P. Caspar, and A. Sher. 1993. Bone marrow macrophages process exogenous Toxoplasma gondii polypeptides for recognition by parasite-specific cytolytic T lymphocytes. J. Immunol. 150:517–526. 18. Donnelly, J. J., J. B. Ulmer, L. A. Hawe, A. Friedman, X. Dhi, K. R. Leander, J. W. Shiver, A. I. Oliff, D. Martinez, D. Montgomery, and M. A. Liu. 1993. Targeted delivery of peptides to class I major histocompatibility molecules by a modified Pseudomonas exotoxin. Proc. Natl. Acad. Sci. USA 90:3530–3534. 19. Fayolle, C., P. Sebo, D. Ladant, A. Ullmann, and C. Leclerc. 1996. In vivo induction of CTL responses by recombinant adenylate cyclase of Bordetella pertussis carrying viral CD81 T cell epitopes. J. Immunol. 156:4697–4706. 20. Flynn, J. L., J. Chan, K. J. Triebold, D. K. Dalton, T. A. Stewart, and B. R. Bloom. 1993. An essential role for interferon-g in resistance to Mycobacterium tuberculosis infection. J. Exp. Med. 178:2249–2254. 21. Flynn, J. L., M. M. Goldstein, K. J. Triebold, B. Koller, and B. R. Bloom. 1992. Major histocompatibility complex class I-restricted T cells are required for resistance to Mycobacterium tuberculosis infection. Proc. Natl. Acad. Sci. USA 89:12013–12017. 22. Germain, R. N., and D. H. Margulies. 1993. The biochemistry and cell biology of antigen-processing and presentation. Annu. Rev. Immunol. 11: 403–450. 23. Grusby, M. J., and L. H. Glimcher. 1995. Immune responses in MHC class II-deficient mice. Annu. Rev. Immunol. 13:417–435. 24. Hansen, T. H., and H. Y. Tse. 1987. Insights into immune-response gene function using an Ia mutant mouse strain. Crit. Rev. Immunol. 7:169–192. 25. Harding, C. V., and R. Song. 1994. Phagocytic processing of exogenous particulate antigens by macrophages for presentation by class I MHC molecules. J. Immunol. 153:4925–4933. 26. Hubbard, R. D., C. M. Flory, and F. M. Collins. 1991. Memory T cellmediated resistance to Mycobacterium tuberculosis infection in innately susceptible and resistant mice. Infect. Immun. 59:2012–2016. 27. Huygen, K., D. Abramowicz, P. Vandenbussche, F. Jacobs, J. De Bruyn, A. Kentos, A. Drowart, J.-P. Van Vooren, and M. Goldman. 1992. Spleen cell cytokine secretion in Mycobacterium bovis BCG-infected mice. Infect. Immun. 60:2880–2886. 28. Huygen, K., J. Content, O. Denis, D. L. Montgomery, A. M. Yawman, R. Randall Deck, C. M. Dewitt, I. M. Orme, S. Baldwin, C. D’Souza, A. Drowart, E. Lozes, P. Vandenbussche, J. Van Vooren, M. A. Liu, and J. B. Ulmer. 1996. Immunogenicity and protective efficacy of a tuberculosis DNA vaccine. Nat. Med. 2:893–898. 29. Huygen, K., K. Palfliet, F. Jurion, J. Hilgers, R. Ten Berg, J. P. Van Vooren, and J. De Bruyn. 1988. H-2-linked control of in vitro gamma interferon production in response to a 32-kilodalton antigen (P32) of Mycobacterium bovis bacillus Calmette-Gue´rin. Infect. Immun. 56:3196–3200. 30. Inaba, K., J. W. Young, and R. M. Steinman. 1987. Direct activation of CD81 cytotoxic lymphocytes by dendritic cells. J. Exp. Med. 166:182–194. 31. Inoue, T., Y. Yoshikai, G. Matsuzaki, and K. Nomoto. 1991. Early appearing g/d-bearing T cells during infection with Calmette-Gue´rin bacillus. J. Immunol. 146:2754–2762. 32. Janis, E. M., S. H. E. Kaufmann, R. H. Schwartz, and D. M. Pardoll. 1989. Activation of gd T cells in the primary immune response to Mycobacterium tuberculosis. Science 244:713–716. 33. Kabelitz, D., A. Bender, S. Schondelmaier, B. Schoel, and S. H. E. Kaufmann. 1990. A large fraction of human peripheral blood g/d1 T cells is activated by Mycobacterium tuberculosis but not by its 65-kD heat shock protein. J. Exp. Med. 171:667–679. 34. Kamijo, R. J., J. Le, D. Shapiro, E. A. Havell, S. Huang, M. Aguet, M.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48. 49.

50.

51. 52.

53. 54.

55.

56.

57. 58.

683

Bosland, and J. Vilcek. 1993. Mice that lack the interferon-g receptor have profoundly altered responses to infection with bacillus Calmette-Gue´rin and subsequent challenge with lipopolysaccharide. J. Exp. Med. 178:1435–1440. Ke, Y., Y. Li, and J. A. Kapp. 1995. Ovalbumin injected with complete Freund’s adjuvant stimulates cytolytic responses. Eur. J. Immunol. 25:549– 553. Kovacsovics-Bankowski, M., and K. L. Rock. 1995. A phagosome-to-cytosol pathway for exogenous antigens presented on MHC class I molecules. Science 267:243–246. Kovacsovics-Bankowski, M., K. Clarck, B. Benacerraf, and K. L. Rock. 1993. Efficient major histocompatibility complex class I presentation of exogenous antigen upon phagocytosis by macrophages. Proc. Natl. Acad. Sci. USA 90:4942–4946. Ladel, C. H., S. Daugelat, and S. H. E. Kaufman. 1995. Immune response to Mycobacterium bovis bacille Calmette Gue´rin infection in major histocompatibility complex class I- and class II-deficient knock-out mice: contribution of CD4 and CD8 T cells to acquired resistance. Eur. J. Immunol. 25:377–384. Lefevre, P., M. Braibant, L. De Wit, M. Kalai, D. Roeper, J. Grotzinger, J. P. Delville, P. Peirs, J. Ooms, K. Huygen, and J. Content. Three different putative phosphate transport receptors are encoded by the Mycobacterium tuberculosis genome and are present at the surface of Mycobacterium bovis BCG. Submitted for publication. Marnell, M. H., S. Shia, M. Stookey, and R. K. Draper. 1984. Evidence for penetration of diphtheria toxin to the cytosol through a prelysosomal membrane. Infect. Immun. 44:145–150. Mu ¨ller, I., S. P. Cobbold, H. Waldmann, and S. H. E. Kaufmann. 1987. Impaired resistance to Mycobacterium tuberculosis infection after selective in vivo depletion of L3T41 and Lyt-21 cells. Infect. Immun. 55:2037–2041. Nonacs, R., C. Humborg, J. P. Tam, and R. M. Steinman. 1992. Mechanisms of mouse spleen dendritic cell function in the generation of influenza-specific cytolytic T lymphocytes. J. Exp. Med. 176:519–529. Orme, I. C. M. 1987. The kinetics of emergence and loss of mediator T lymphocytes acquired in response to infection with Mycobacterium tuberculosis. J. Immunol. 138:293–298. Orme, I. C. M. 1988. Characteristics and specificity of acquired immunologic memory to Mycobacterium tuberculosis infection. J. Immunol. 140:3589– 3593. Orme, I. C. M., E. S. Miller, A. D. Roberts, S. K. Furney, J. P. Griffin, K. M. Dobos, D. Chi, B. Rivoire, and P. J. Brennan. 1992. T lymphocytes mediating protection and cellular cytolysis during the course of Mycobacterium tuberculosis infection. Evidence for different kinetics and recognition of a wide spectrum of protein antigens. J. Immunol. 148:189–196. Pal, P. G., and M. A. Horwitz. 1992. Immunization with extracellular proteins of Mycobacterium tuberculosis induces cell-mediated immune responses and substantial protective immunity in a guinea pig model of pulmonary tuberculosis. Infect. Immun. 60:4781–4792. Palmer, C. E., and M. W. Long. 1966. Effects of infection with atypical mycobacteria on BCG vaccination and tuberculosis. Am. Rev. Respir. Dis. 94:553–568. Parish, C. R., and A. Mullbacher. 1983. Automated colorimetric assay for T cell cytotoxicity. J. Immunol. Methods 58:225–237. Pedrazzini, T., K. Hug, and J. A. Louis. 1987. Importance of L3T41 and Lyt-21 cells in the immunologic control of infection with Mycobacterium bovis strain bacillus Calmette-Gue´rin in mice: assessment by elimination of T cell subsets in vivo. J. Immunol. 139:2032–2037. Pfeffer, K., B. Schoel, H. Gulle, S. H. E. Kaufmann, and H. Wagner. 1990. Primary responses of human T cells to mycobacteria: a frequent set of g/d T cells are stimulated by protease-resistant ligands. Eur. J. Immunol. 20:1175– 1179. Porcelli, S., C. T. Morita, and M. B. Brenner. 1994. Recognition of a lipid antigen by CD1-restricted ab1 T cells. Nature 372:691–694. Roche, P. W., J. A. Triccas, and N. Winter. 1995. BCG vaccination against tuberculosis: past disappointments and future hopes. Trends Microbiol. 3:397–401. Rock, K. L. 1996. A new foreign policy: MHC class I molecules monitor the outside world. Immunol. Today 17:131–137. Rock, K. L., L. Rothstein, S. Gamble, and C. Fleischacker. 1993. Characterization of antigen-presenting cells that present exogenous antigens in association with class I MHC molecules. J. Immunol. 150:438–446. Schirmbeck, R., K. Melber, and J. Reimann. 1995. Hepatitis B virus small antigen particles are processed in a novel endosomal pathway for major histocompatibility complex class I-restricted epitope presentation. Eur. J. Immunol. 25:1063–1070. ˇ ebo, P., C. Fayolle, O. D’Andria, D. Ladant, C. Leclerc, and A. Ullmann. S 1995. Cell-invasive activity of epitope-tagged adenylate cyclase of Bordetella pertussis allows in vitro presentation of a foreign epitope to CD81 cytotoxic T cells. Infect. Immun. 63:3851–3857. Spangler, B. D. 1992. Structure and function of cholera toxin and the related Escherichia coli heat-labile enterotoxin. Microbiol. Rev. 56:622–647. Suto, R., and P. K. Srivastava. 1995. A mechanism for the specific immu-

684

DENIS ET AL.

nogenicity of heat shock protein-chaperoned peptides. Science 269:1585– 1588. 59. Sypek, J. P., S. Jacobson, A. Vorys, and D. J. Wyler. 1993. Comparison of gamma interferon, tumor necrosis factor, and direct cell contact in activation of antimycobacterial defense in murine macrophages. Infect. Immun. 61: 3901–3906.

Editor: S. H. E. Kaufmann

INFECT. IMMUN. 60. Yoshida, A., Y. Koide, M. Ushijima, and T. Yoshida. 1995. Dissection of strain difference in acquired protective immunity against Mycobacterium bovis Calmette-Gue´rin bacillus (BCG). J. Immunol. 155:2057–2066. 61. Young, D. B., S. H. E. Kaufmann, P. W. M. Hermans, and J. E. R. Thole. 1992. Microreview. Mycobacterial antigens: a compilation. Mol. Microbiol. 6:133–145.