B7 Costimulation Is Critical for Antibody Class Switching and CD8⁺ Cytotoxic T-Lymphocyte Generation in the Host Response to Vesicular Stomatitis Virus

Abstract

Antibody and cytotoxic T-lymphocyte (CTL) responses have critical roles in eliminating many viral infections. In addition to stimulation of the T-cell receptor, T cells require costimulatory signals to respond optimally. We evaluated the role of B7 costimulatory molecules (B7-1 and B7-2) in the immune response to viral infection using vesicular stomatitis virus (VSV) and mice lacking either B7-1 or B7-2 or both molecules. Mice lacking both B7-1 and B7-2 had essentially no anti-VSV immunoglobulin G1 (IgG1) response, decreased IgG2a responses, and normal IgM responses, while mice lacking either B7-1 or B7-2 had unaltered anti-VSV antibody responses compared to wild-type mice. Depletion of CD4⁺ cells further reduced the IgG2a response in mice lacking both B7 molecules, suggesting that CD4⁻ cells may supply help for IgG2a in the absence of B7 costimulation. The absence of both B7 molecules profoundly reduced generation of both primary and secondary VSV-specific class I major histocompatibility complex (MHC)-restricted CTL, whereas VSV-specific CTL responses in mice lacking either B7-1 or B7-2 were similar to those of wild-type animals. Class I MHC-restricted CTL in wild-type mice were not dependent on CD4⁺ cells, suggesting that the failure of CTL in the absence of B7s is due to a lack of B7 costimulation directly to the CD8⁺ CTL. These data demonstrate that B7-1 and B7-2 have critical, overlapping functions in the antibody and CTL responses to this viral infection.

Costimulation of T cells is important in the generation of immune responses. B7 costimulation enhances T-cell responses, and perhaps unique among the costimulators, the B7 molecules can prevent induction of anergy (5). The B7 molecules, B7-1 (CD80) and B7-2 (CD86), are expressed by antigen-presenting cells (APC); activation of APC via CD40 or soluble factors such as lipopolysaccharide increases expression of the B7 molecules (9, 17). The potential for manipulation of the immune response through manipulation of B7 costimulation has made these molecules the subject of intense study. We have made mice lacking B7-1, B7-2, or both of these molecules (B7-1^−/−, B7-2^−/−, or B7-1/2^−/− mice) to investigate the role of this pathway in vivo (2, 13).

T cells express two receptors for the B7 molecules, one of which is stimulatory (CD28) and the other of which is inhibitory (CTLA-4; also called CD152). CD28 is constitutively expressed on most T cells (15). B7 binding to CD28 stimulates T-cell responses by enhancing T-cell proliferation and interleukin-2 (IL-2) production; this accounts for the costimulatory activity of the B7 molecules (24). In contrast, CTLA-4 is upregulated following activation of T cells. Signaling through CTLA-4 inhibits T-cell responses, decreasing proliferation and blocking cell cycle progression at G₁/S (19, 33). The inhibitory effect of CTLA-4 is underscored by the phenotype of CTLA-4-deficient mice. These mice have pronounced expansion of lymphocytes and lymphocytic infiltration with tissue destruction in several organs, including heart, pancreas, and skeletal muscle (31, 34).

Previous studies have demonstrated the importance of the B7 pathway in the immune response to simple haptenated proteins (2), but infectious agents present a more complex array of antigenic stimuli to the immune system. Here, we have used vesicular stomatitis virus (VSV), a rhabdovirus related to rabies virus, to determine the role of B7 molecules in the immune response to viral infection. When injected outside the central nervous system in immunocompetent mice, VSV elicits a strong immune response. VSV stimulates a strong neutralizing antibody response, which is required for elimination of the infection (8). VSV also drives a strong T-cell response, eliciting viral reactive T helper cells and both CD4⁺ and CD8⁺ cytotoxic T lymphocytes (CTL) (restricted to class II and class I major histocompatibility complex [MHC] molecules, respectively), and thus provides a convenient model for studying many aspects of the immune response to viral infection (6, 7, 22, 32).

We have used VSV in mice lacking one or both B7 molecules to investigate the role of B7 costimulation in antibody and class I MHC-restricted CTL responses to viral infection. The absence of both B7-1 and B7-2 profoundly reduced the antibody response, decreasing or abrogating class switching of the antibodies. The modest immunoglobulin G (IgG) response to VSV in the B7-1/2^−/− mice was further reduced in the absence of CD4⁺ cells. In contrast, the absence of either B7-1 or B7-2 did not alter the antibody response to the virus. The class I MHC-restricted CTL response against VSV was also dependent on B7 costimulation, as primary and secondary responses were profoundly reduced in the absence of both B7 molecules. However, the presence of either B7 molecule was sufficient to generate a strong class I-restricted CTL response to VSV infection. These results demonstrate that the B7 pathway plays an important role in stimulating humoral and CTL responses to this viral infection.

MATERIALS AND METHODS

Mice.

B7-1^−/− (13), B7-2^−/− and B7-1/2^−/− (2) mice have been described previously. Animals used in this study were inbred 129S4/SvJae or backcrossed from 129S4/SvJae onto the BALB/c background and then interbred to generate B7-deficient mice. B7-1^−/− BALB/c mice were backcross generation 10, and B7-2^−/− BALB/c mice were backcross generation 6. B7-1/2^−/− BALB/c mice were backcross generation 3 but were homozygous for H-2^d. Controls for the B7-1/2^−/− mice were H-2^d BALB F₃. Wild-type matches for the BALB/c B7-1^−/− or B7-2^−/− mice were commercial BALB/c mice obtained from Taconic. 129S4/SvJae wild-type mice were bred within our animal facility. Brigham and Women's Hospital and Harvard Medical School are Association for Assessment and Accreditation of Laboratory Animal Care-accredited institutions, and the mice were cared for in accordance with institutional guidelines.

Virus preparation.

An Indiana serotype (San Juan isolate) of VSV was the kind gift of Carol S. Reiss. Virus stocks were prepared on confluent BHK-21 monolayers inoculated at a low multiplicity of infection (MOI). The supernatant was harvested 16 to 20 h later, rapidly aliquoted on ice, and then frozen at −80°C. Viral titer was determined by plaque assay on Vero cells as described previously (10). Virus for use as antigen in the enzyme-linked immunosorbent assay (ELISA) was prepared by differential centrifugation as described elsewhere (10). The protein concentration in this virus antigen preparation was determined by the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, Calif.).

Determination of antibody responses to VSV and depletion of CD4⁺ cells.

For determination of antibody responses to VSV, groups of four to seven BALB/c mice were bled to obtain preimmune sera and then infected by intraperitoneal (i.p.) injection with 2 × 10⁶ PFU of VSV. Animals were bled on day 4, day 7, and usually weekly thereafter. Serum was prepared from the blood and frozen at −20°C until all samples were collected. BALB/c mice were used because only this strain was available in the quantities needed for these antibody experiments. The results obtained with wild-type BALB/c mice are very similar to those published for H-2^b mice (26).

Anti-VSV antibody titers were measured by ELISA. ELISA plates were coated with VSV (2.5 μg/ml) for 2 h at 37°C and then blocked overnight with 2% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) at 4°C. The mouse serum was diluted 1/300 in 2% BSA with 1% fetal calf serum in PBS. The serum was titrated in duplicate, and the plates were incubated at room temperature for 2 h. Next, goat anti-mouse Ig, or anti-IgM, anti-IgG1, or anti-IgG2a conjugated to alkaline phosphatase (from Southern Biotechnology Inc.), was added at 1/1,000 in 2% BSA–PBS, and plates were incubated for 2 h at room temperature. Then Sigma 104 phosphatase substrate, prepared as in the manufacturer's instructions, was added to each well. The maximum optical density was allowed to develop to between 0.5 and 1.5, and the reaction was stopped with 3 N NaOH. Optical density was read at 410 nm.

For some assays, the endpoint titer was defined for each sample as the dilution which gave 10% of the maximum signal within the assay, after subtraction of background from both sample and maximum values. For other assays, the raw data, with background level subtracted, are shown.

Animals were depleted of CD4⁺ cells by injection with 100 μg of GK1.5 anti-CD4 antibody once a day for 3 days (prepared as ammonium sulfate-precipitated ascites fluid, dialyzed extensively against saline and then PBS), followed by challenge with virus on the next day; 100 μg of antibody was injected every 4 to 5 days throughout the experiment. At the end of each experiment, splenocytes were stained for CD4⁺ cells with RM4-4-fluorescein isothiocyanate (Pharmingen), an anti-CD4 antibody that is not blocked by GK1.5 (manufacturer's description, which we have confirmed) to confirm CD4⁺ cell depletion by flow cytometry. The depletion of CD4⁺ cells was always >95%. Control animals were treated with rat IgG.

Determination of class I MHC-restricted CTL responses to VSV.

For determination of CD8⁺ CTL responses to VSV, 129S4/SvJae mice were infected by intravenous (i.v.) injection with 2 × 10⁶ PFU of VSV or left untreated as controls. 129S4/SvJae mice were used for CTL measurements because primary CTL responses are reproducibly vigorous in this strain, unlike BALB/c mice (reference 11 and our unpublished observations).

Primary CD8⁺ CTL responses were measured 6 days after infection. Mice were killed, and the spleens were removed, dissociated, and depleted of erythrocytes. The effectors were plated and titrated in 100 μl of medium in V-bottom 96-well plates. Targets were EL4, an H-2^b cell line, infected with VSV at an MOI of 50 during labeling with 200 μCi of ⁵¹Cr at 37°C for 90 min. Control targets were uninfected EL4 cells. EL4 express class I but not class II MHC molecules, and so only class I-restricted (CD8⁺) CTL are detected in this assay. The targets were washed, counted, and adjusted to 5 × 10⁴/ml, and then 100 μl of cells was added to the effectors. Spontaneous and maximum release were measured by incubating targets with medium and 2% Triton X-100, respectively. Plates were incubated at 37°C with 5% CO₂ for 5 h, and then 100 μl of supernatant was removed and counted in a gamma counter. Percent specific lysis was calculated as [(mean sample cpm − mean spontaneous cpm)/(mean total cpm − mean spontaneous cpm)] × 100.

Secondary class I MHC-restricted CTL were generated from 129S4/SvJae mice primed by i.v. infection with 2 × 10⁶ PFU of VSV 21 days before sacrifice; 3 × 10⁶ APC and responding cells were plated in 2 ml of medium in 24-well plates. The APC were erythrocyte-depleted splenocytes from naive wild-type or B7-1/2^−/− 129S4/SvJae mice which were pulsed with UV (50 μJ)-inactivated VSV at an MOI of 10 for 2 h at 37°C, washed, and then irradiated with 2,000 rads. In some experiments, IL-2 (10 U/ml) was added to the culture, as indicated in the appropriate figure legends. Five days later, the medium was gently aspirated off the cells and 2 ml of fresh medium was added. The effectors were then resuspended, plated onto a 96 V-bottom plate, and titrated for the cytotoxicity assay. Secondary CD8⁺ CTL were measured in 4-h assays, with target preparation and calculations done as described above.

RESULTS

B7 costimulators are important for isotype switching of the anti-VSV antibody response.

To determine the role of B7 molecules in the antibody response to VSV, groups of mice were immunized with VSV and bled subsequently to obtain sera to test for VSV-specific antibodies in an isotype-specific ELISA. An endpoint titer for each isotype in each animal was determined.

The absence of either B7-1 or B7-2 did not significantly alter the antibody response to VSV (Fig. 1). The kinetics and magnitude of the response in the B7-1^−/− and B7-2^−/− animals were very similar to those for the wild-type control animals. In contrast, mice lacking both B7-1 and B7-2 had reduced antibody responses compared to wild-type mice (Fig. 2). The total anti-VSV Ig responses were reduced approximately 5- to 10-fold in the B7-1/2^−/− mice at 7 to 42 days after infection. The VSV-specific IgM response was not significantly altered by the absence of both B7 molecules. However, the isotype-switched antibodies (IgG) were reduced in the B7-1/2^−/− mice. This reduction was profound for VSV-specific IgG1, which was essentially undetectable in the B7-1/2^−/− mice, while wild-type mice had an anti-VSV IgG1 response detectable from days 14 to 42. The decrease in VSV-specific IgG2a was statistically significant but smaller: about 10-fold less in B7-1/2^−/− than wild-type mice at days 7 and 14. Together, these data demonstrate that there is a reduction in class switching of the antibody response to VSV, and this deficit varies between the isotypes of IgG.

FIG. 1.

The absence of either B7-1 or B7-2 does not significantly alter the antibody response to VSV. Groups of four to seven BALB/c mice were bled for the day 0 assay and then injected with 2 × 10⁶ PFU of VSV i.p. Mice were then bled at the indicated days after VSV infection. Antibody response to VSV was measured by ELISA, and the mouse total Ig or indicated isotype was determined. The mean titers for wild-type, B7-1^−/−, and B7-2^−/− mice are shown, with the standard deviation of each group indicated. There were no statistically significant differences between any two groups of mice on any day (all P values > 0.05 by two-tailed Student t test). This experiment was performed three times with similar results.

FIG. 2.

The absence of both B7-1 and B7-2 reduces class switching of the antibody response to VSV. Groups of four to seven BALB/c mice were bled and challenged with VSV as described for Fig. 1. The mean titers for wild-type and B7-1/2^−/− mice are shown, with the standard deviation of each group indicated. Days at which the responses between wild-type and B7-1/2^−/− mice were significantly different are indicated by asterisks (P values < 0.05 by two-tailed Student t test). This experiment was performed four times with similar results.

VSV-specific IgG2a production in B7-1/2^−/− mice is partially dependent on CD4⁺ cells.

Haptenated proteins elicit antihapten IgM, but not IgG, in B7-1/2^−/− mice (2). However, it has been found that γ/δ T cells (which are CD4⁻) can promote class switching to IgG2a against VSV (26). To determine the role of CD4⁺ cells in the VSV-specific IgG2a response in the absence of B7-1 and B7-2, we depleted wild-type and B7-1/2^−/− mice of CD4⁺ cells and tested the antibody response to VSV. Depletion of CD4⁺ cells was maintained by periodic injection of anti-CD4 for 21 days, after which the animals were bled and sacrificed. Control animals were injected with rat Ig on the same schedule as the anti-CD4.

Depletion of CD4⁺ cells reduced the anti-VSV antibody response in both wild-type and B7-1/2^−/− mice (Fig. 3). The VSV-specific total Ig and IgG2a were reduced profoundly (approximately 100-fold) in the wild-type mice and more modestly (approximately 10-fold) in the B7-1/2^−/− mice. In contrast, depletion of CD4⁺ cells from the wild-type animals abrogated any detectable VSV-specific IgG1 response. Whether depleted of CD4⁺ cells or not, the B7-1/2^−/− mice do not make detectable VSV-specific IgG1. IgM was not measured since IgM wanes by day 21 (Fig. 1). For all isotypes examined, the antibody responses in the anti-CD4-treated wild-type and anti-CD4-treated B7-1/2^−/− mice were virtually identical. These results demonstrate CD4⁺ cell-independent generation of anti-VSV IgG2a, but not IgG1, and are consistent with previous studies (3, 26).

FIG. 3.

Depletion of CD4⁺ cells reduces the antibody response to VSV in both wild-type and B7-1/2^−/− mice. Groups of three to four BALB/c mice were treated with anti-CD4 or control rat Ig and then injected with 2 × 10⁶ PFU of VSV i.p. Antibody treatment was repeated every 4 to 5 days for 21 days, and then the animals were bled. Antibody response to VSV was measured by ELISA. Panels A and B show antibody responses in four wild-type mice (■, control IgG; □, anti-CD4); and three B7-1/2^−/− mice (●, control IgG; ○, anti-CD4), respectively, for the indicated antibody isotypes. This experiment was performed twice with similar results.

Anti-VSV class I MHC-restricted CTL responses are profoundly reduced in the absence of B7 molecules.

To determine the importance of the B7 molecules in primary CD8⁺ CTL generation, 129S4/SvJae mice were injected with VSV or left untreated as controls. Six days later, the mice were sacrificed and spleen cells were tested as primary effectors in a conventional cytotoxicity assay against VSV-infected EL4 cells or control EL4 cells.

While the absence of either one of the B7 molecules did not alter CTL generation, the absence of both molecules greatly reduced primary CTL generation as detected in this assay (Fig. 4). Killing by primed wild-type, B7-1^−/−, or B7-2^−/− splenocytes was easily detected against the VSV-pulsed target cells but not against the negative control cells. In contrast, splenocytes from VSV-infected B7-1/2^−/− mice had only very slight cytotoxicity against the VSV-infected targets. Lysis of the VSV-infected targets by the B7-1/2^−/− splenocytes was slightly greater than for the negative control target, suggesting that there was indeed a small CTL response.

FIG. 4.

The absence of both B7-1 and B7-2 reduces primary CTL generation, while the absence of either one of the B7 molecules does not. Groups of three 129S4/SvJae mice were injected with 2 × 10⁶ PFU of VSV i.v. or left untreated as controls (data for a single mouse of each type are shown for clarity; results for all mice within a group were similar). Six days later, the mice were killed and the spleen cells were tested for killing of VSV-infected EL4 targets (top) or uninfected control EL4 targets (bottom) in a 5-h cytotoxicity assay. The VSV-primed mice are indicated by filled symbols, and the uninfected mice are indicated by open symbols. The mice tested were wild type (squares), B7-1^−/− (triangles), B7-2^−/− (circles), and B7-1/2^−/− (diamonds). This experiment was performed three times with similar results.

We next wished to determine if secondary class I MHC-restricted CTL responses from B7-1/2^−/− mice were also diminished. Secondary CTL responses are often more sensitive than primary responses, and in some systems primary and secondary CTL responses are not coupled (25). Secondary responses in wild-type or B7-1/2^−/− mice primed with VSV were tested. Splenocytes were restimulated with irradiated syngeneic splenocytes (wild type or B7-1/2^−/−) that had been pulsed with inactivated VSV. This was done in the presence or absence of added recombinant IL-2. IL-2 was added to enhance any weak CTL responses. Five days later, cultures were tested for VSV-specific cytotoxicity. Cytolysis was the only CTL function measured, and so an increase in noncytolytic VSV-specific CD8⁺ cells would not be detected.

Secondary class I MHC-restricted CTL responses were greatly reduced in B7-1/2^−/− mice (Fig. 5). There was no detectable killing of VSV-infected targets by B7-1/2^−/− cells restimulated without added IL-2. When IL-2 was added to the culture, the effectors from the B7-deficient mice had modest CTL activity (maximum cytolysis of 38%, compared to 72% for the matched primed wild type). Killing of the VSV-infected cells was reproducibly threefold higher than killing of the control cells; no such difference was found if naive B7-1/2^−/− cells were used as the responders (data not shown). With added IL-2, the presence of B7 on the APC enhanced CTL generation in the culture threefold above that generated by B7-1/2^−/− APC. Wild-type responders had a brisk cytotoxic response, whether stimulated with wild-type or B7-1/2^−/− APC. In the wild-type cultures, there was also a small primary CTL response, whether the stimulator cells expressed B7 or not. It is important to note that B7 costimulation could be provided by APC in the responder population in these cultures, and so a role for B7 in the restimulation cannot be eliminated.

FIG. 5.

The absence of both B7-1 and B7-2 reduces, but does not abrogate, secondary CTL generation. Wild-type (top) or B7-1/2^−/− (bottom) 129S4/SvJae mice (two in each group) were injected with 2 × 10⁶ PFU of VSV i.v. Mice were killed 21 days later, and the pooled spleen cells were stimulated with UV-inactivated VSV-pulsed irradiated splenocytes in the absence or presence of recombinant IL-2 (10 U/ml). Five days later, cultures were tested for killing of VSV-infected targets or uninfected control targets in a 4-h cytotoxicity assay, as indicated on each panel. Symbols: ■, VSV-infected animal with no IL-2 added to culture; ●, VSV-infected animal with IL-2 added to culture; □, uninfected animal with no IL-2 added to culture; ○, uninfected animal with IL-2 added to culture. This experiment was performed three times with similar results.

The reduction of CTL in B7-1/2^−/− mice is not due to a reduced T helper response.

The reduced anti-VSV CTL in the B7-1/2^−/− mice could be due to the absence of B7 costimulation directly to the CD8⁺ cells or to a lack of costimulation to CD4⁺ cells, leading in turn to reduced help for the CTL. To distinguish between these possibilities, we tested whether CTL generation in this strain of mice is dependent on CD4⁺ cells. If the CD8⁺ CTL generation is independent of T helper cells, this would suggest that the reduced CTL response in the absence of B7 molecules is due to a lack of costimulation to the developing CD8⁺ CTL.

Both primary and secondary anti-VSV class I MHC-restricted CTL in 129S4/SvJae mice are independent of CD4⁺ cells. Mice were depleted of CD4⁺ cells or treated with rat Ig as described above. There was a similar primary CD8⁺ CTL response in the rat Ig-treated and anti-CD4-treated mice (Fig. 6). Secondary CD8⁺ CTL responses from anti-CD4-treated animals were dependent on added IL-2. With added IL-2, the generation of secondary CTL was not significantly altered by the absence of CD4⁺ cells (data not shown). These data suggest that development of class I MHC-restricted CTL against VSV is not dependent on T help in this strain of mice. This is consistent with the observation that C57BL/6 mice, which have the same MHC haplotype as 129S4/SvJae mice, also generate helper independent anti-VSV CD8⁺ CTL (20).

FIG. 6.

Generation of primary anti-VSV CTL does not depend on CD4⁺ helper cells. 129S4/SvJae mice (two in each group) were treated with anti-CD4 or control rat Ig and then injected with 2 × 10⁶ PFU of VSV i.v. Antibody treatment was repeated 3 days after infection. Six days after infection, the mice were killed and the pooled spleen cells were tested for killing of VSV-infected targets or uninfected control targets in a 5-h cytotoxicity assay. Symbols: ■, control IgG treated with VSV-pulsed target cells; □, control IgG treated with unpulsed target cells; ●, anti-CD4 treated with VSV-pulsed target cells; ○, anti-CD4 treated with unpulsed target cells. This experiment was performed three times with similar results.

DISCUSSION

Our studies demonstrate an important role for B7 molecules in the immune response to VSV. The absence of both B7-1 and B7-2 reduced class switching of the antibody response without any effect on IgM. Mice lacking either one of the B7 molecules did not have an altered antibody response to VSV. The class I MHC-restricted CTL response was greatly reduced in the absence of both molecules but unchanged in mice lacking either one of the B7 molecules.

Previous studies from this laboratory have shown that B7-1^−/− mice have normal antibody responses to haptenated proteins (2). B7-2^−/− mice have a lack of class switching to antigens given without adjuvant, but they will class switch if antigen is given with alum or with Freund's incomplete or complete adjuvant. Finally, mice lacking both molecules do not class switch to antigen given with any of these adjuvants (2). The antibody responses to haptenated proteins and VSV differ markedly. While no adjuvant was given with the VSV, B7-2^−/− mice (as well as B7-1^−/− mice) had an antibody response similar to that of the wild-type mice. It is likely that the infection with VSV leads to upregulation of B7-1 expression in the B7-2^−/− mice due to the response to the vigorous infection. The important role of B7-1 in the absence of B7-2 is underscored by the decreased class switching in the B7-1/2^−/− mice.

Class switching to IgG2a was partially dependent on CD4⁺ cells. Previous studies have suggested that the closely packed G protein spikes on the surface of VSV form a crystalline-like array that can act as a type 2 T-independent antigen for the IgM response (1). It is possible that the crystalline-like array also facilitates IgG2a production in the absence of CD4⁺ cells (3). Production of IgG2a against VSV, without production of IgG1, has also been seen in other models, including mice deficient in CD40 ligand (CD154) and mice lacking T cells with the α/β T-cell receptor (TCR) (3, 26, 35). In mice lacking α/β T cells, production of VSV-reactive IgG2a depends on T cells which express the γ/δ TCR (26). The γ/δ T cells produce gamma interferon, which is important in driving B cells to produce IgG2a (26, 29). These γ/δ T cells may provide the T help to elicit anti-VSV IgG2a in the CD4-depleted animals as well (Fig. 3 and reference 35). Together, the previous data suggest that the T-cell help that these γ/δ T cells provide to B cells does not depend on CD40-CD40 ligand interaction. Our data further suggest that the γ/δ T cells may not need B7 costimulation to respond to VSV infection.

The role of B7 molecules in CTL generation has been controversial. There have been some studies suggesting differences in the ability of B7-1 and B7-2 to elicit CTL, while other studies have found the two molecules to be similar (12, 14, 21, 28). Our own data demonstrate that either B7-1 or B7-2 is sufficient to generate a normal primary CD8⁺ CTL response against VSV in vivo (Fig. 4). A recent study found that B7-1 is more potent than B7-2 at stimulating proliferation and cytokine production by CD8⁺ T cells (12). Although, B7-2 is expressed at higher levels than B7-1 on resting splenocytes (17, 23), B7-1 expression is induced by 2 days after activation of APC (30). Therefore, either B7-1 or B7-2 could be induced rapidly enough to generate CTL after 6 days, as measured here (Fig. 4). As CD4⁺ cells are not required for the primary anti-VSV CTL response, it is very likely that either B7 molecule can directly costimulate class I MHC-restricted CTL, as has been demonstrated for B7-1 (16).

The decrease in VSV-reactive CD8⁺ CTL in the B7-1/2^−/− mice is consistent with a similar decrease seen in CD28^−/− animals. A study using CD28^−/− mice demonstrated that primary anti-lymphocytic choriomeningitis virus (LCMV) CTL responses were unchanged by the absence of CD28, while primary anti-VSV CTL were profoundly decreased (20, 27). The brief duration of VSV infection compared to LCMV was likely the cause of the difference. After restimulation with added IL-2, secondary anti-VSV CTL from CD28^−/− mice could be detected. Similarly, we have found that primary anti-VSV class I MHC-restricted CTL responses in the B7-1/2^−/− mice are greatly reduced, but a small secondary response can be detected if IL-2 is added to the culture. We have also demonstrated that CD8⁺ CTL generated in the absence of B7 are responsive to B7 costimulation in secondary culture, which could not be evaluated in CD28^−/− mice. The presence of the small CTL response to VSV from B7-1/2^−/− mice suggests that the CTL are not rendered anergic by VSV infection in the absence of B7 costimulation. The reduction of class I MHC-restricted CTL generation in B7-1/2^−/− mice was not due simply to a lack of costimulation to T helper cells, as depletion of CD4⁺ cells did not alter the primary CD8⁺ CTL response in wild-type mice. It is possible that γ/δ T cells provide B7-independent help to CTL, as they do for B cells producing IgG2a. Taken in total, these data indicate that B7 costimulation is important in generation of antiviral CTL against some viruses, and the B7 molecules may directly costimulate the developing CD8⁺ CTL.

It is interesting that although B7-1/2^−/− mice are similar to CD40L^−/− mice in the antibody response to VSV, the CTL response is markedly different between these strains of mice. The primary anti-VSV and anti-LCMV CTL responses in CD40L^−/− mice are normal, although anti-LCMV memory CTL are reduced (3, 4). We believe that the importance of B7 molecules in the antibody response depends on B7 costimulation of CD4⁺ T-cell help. In turn, the helpers stimulate B cells via cytokines, CD40-CD40L interaction, or both. In contrast, costimulation of the CD8⁺ CTL by B7 molecules, along with TCR stimulation, can be sufficient to elicit CD8⁺ CTL without T-cell help. These data suggest that despite the absence of CD40-CD40L interactions in the CD40L^−/− mice, B7 is expressed at levels adequate to costimulate CTL. It is well established that CD40-CD40L interaction is one pathway leading to induction of B7 expression (9, 18). Our data, along with those of others, suggest that there are additional pathways for inducing expression of B7 and that these mechanisms may be activated during viral infection.