Vaccination with Mouse Mammary Adenocarcinoma Cells Coexpressing B7-1 (CD80) and B7-2 (CD86) Discloses the Dominant Effect of B7-1 in the Induction of Antitumor Immunity¹

The interaction between B7-1 or B7-2 molecules with their T cell counter-receptors CD28/CTLA-4 plays a critical role in controlling T cell responses (1, 2). Although B7-1 and B7-2 bind with comparable low avidity to CD28 and high avidity to CTLA-4, they have distinct binding sites on either T cell counter-receptor (3, 4, 5). Furthermore, B7-1 and B7-2 display different kinetics of binding to either counter-receptor, B7-2 having a faster dissociation kinetics (3). The expression of B7-1 and B7-2 is also differently regulated on APCs: B7-2 is expressed constitutively on monocytes, dendritic cells, and resting B cells, and when both B7 molecules are up-regulated upon APC activation, B7-2 is up-regulated more rapidly (6, 7, 8, 9, 10, 11). Altogether, these findings would support different costimulatory roles for each B7 molecule. A number of experimental evidences obtained both in the human and mouse systems have indeed indicated that the two B7 molecules exert nonredundant costimulatory effects. B7-2 but not B7-1 stimulated the production of IL-4 in vitro (12, 13, 14). B7-2 also plays a critical role in directing CD4⁺ Th2 differentiation in an animal model of helminth infection (15). Furthermore, blocking in vivo B7-1 with specific mAbs significantly accelerated the development of diabetes in nonobese diabetic mouse (16), whereas it prevented or cured established experimental allergic encephalomyelitis (17, 18). On the contrary, the treatment with anti-B7-2 mAb blocked the development of diabetes (16) while it exacerbated experimental allergic encephalomyelitis (17, 18). Thus, these data suggest that B7-1 and B7-2 ought to differentially activate the Th1/Th2 development pathways (19, 20). However, other experimental systems failed to reveal any significant difference between B7-1 and B7-2 in the induction of proliferation, cytokine production, and cytolytic activity (21, 22). Thus, additional factors, such as Ag dose, anatomical distribution, APCs, and genetic background may influence the outcome of B7-mediated costimulation.

Since B7-dependent costimulatory signals play a central role in T cell activation, it has been proposed that the lack of immunogenicity of many tumor types could be due to the lack of B7 expression (23, 24). Indeed, it was proved that transfection of B7-1 genes into different experimental mouse tumors greatly improved their immunogenicity (23, 24). However, when the efficacy of B7-1- and B7-2-dependent costimulation were compared on a more extended panel of mouse tumors, heterogeneous results were obtained. B7-1 and B7-2 equally induced protective and curative antitumor immunity when expressed into both immunogenic mastocytoma P815 (25, 26) and lymphoma RMA cells (27), but B7-1 was found to be superior to B7-2 when transfected into the 32Dc13 myeloid cell line (28) or into a subclone of P815 (29, 30). We have also found that B7-1 and B7-2 exert nonredundant costimulatory effects when expressed on TS/A mouse mammary adenocarcinoma cells. When used as nonreplicating cell vaccines, TS/A cells expressing B7-2 are more immunogenic than those expressing B7-1 (27). This differential effect between B7-1- and B7-2-dependent costimulation observed in the TS/A tumor model gave us the possibility to investigate whether the lower costimulatory efficacy of B7-1 was due to an active inhibitory effect by B7-1, or more simply to a weaker costimulatory activity of B7-1. For that, we coexpressed B7-1 and B7-2 on TS/A cells and compared nonreplicating double and single B7-1 or B7-2 transfectants for their capacity to 1) induce both protective and therapeutic immunity and 2) differentially induce IL-4 and IFN-γ in vivo. Our data show that B7-1 modulates B7-2-dependent antitumor response in an active, dominant, and noncompetitive way.

TS/A is a spontaneous, moderately differentiated mammary adenocarcinoma of BALB/c origin. The tumor cell line was maintained in vitro at 37°C in a humidified 5% CO₂ atmosphere in air in complete medium (RPMI 1640 with 5% FCS, 100 U/ml penicillin, 100 U/ml streptomycin, and 2.5 × 10⁻⁵ M 2-ME). Cells were not cultured for longer than 3–4 wk and were routinely screened for Mycoplasma contamination.

Vectors expressing mouse B7-1or mouse B7-2 cDNAs were transfected by electroporation into TS/A expressing mouse B7-2 or mouse B7-1, respectively, as described (27). Double-transfected cells were selected in complete medium containing 2 mg/ml of G418, 50 μg/ml xantine, and 20 μg/ml of mycophenolic acid. Surviving cells expressing variable levels of the two B7 molecules were stained with anti-mouse B7-1 1G10 or anti-mouse B7-2 GL1 mAbs (PharMingen, San Diego, CA), selected by cell sorting using a FACStar^Plus flow cytometer (Becton Dickinson, Mountain View, CA), and cloned by limiting dilution in 96-well plates. Wells displaying single clones growing were reanalyzed by flow cytometry to verify the expression of the desired amount of each B7 costimulatory molecule. Several TS/A-2/1 clones displaying high levels of B7-2 and low, medium, or high surface expression of B7-1 were found. Three clones expressing high levels of B7-1 and one representative clone expressing negative, low, or medium levels of B7-1 were selected for this study. Several TS/A-1/2 clones, coexpressing levels of both B7 molecules as high as those expressed by the single-transfected TS/A-1.26 and TS/A-2.22 clones, were also found. Four of them were selected for this study.

Female BALB/c mice (4- to 8-wk old), were purchased from Charles River Breeding Laboratories (Calco, Italy).

All of the in vivo studies were approved by the Ethical Committee of the Istituto Scientifico San Raffaele and performed according to its guidelines. Nonreplicating cells were obtained by incubating 1 × 10⁷ cells/ml with 60 μg/ml of mitomycin C (Mit. C;⁴ Sigma, St. Louis, MO) in RPMI 1640 for 30 min at 37°C. For vaccination experiments, mice were challenged in the left flank with a single inoculum of 1 × 10⁶ Mit. C-treated cells. Control vaccinations were done with nontransfected parental TS/A cells or TS/A cells transfected with empty vectors encoding for G418, guanine-xanthine-phosphoribosyl transferase (gpt), or both. After 2 wk, mice were challenged s.c. in the opposite flank with 10⁵ living nontransfected parental TS/A cells. Mit. C-treated cells maintained in vitro for 4 days did not show changes on the level of B7-1 and/or B7-2 expression (data not shown).

To block the effects of either B7-1 or B7-2 on double-transfected TS/A cells in vivo, mice were vaccinated s.c. with 10⁶ nonreplicating TS/A-2/1.4 cells, which were mixed in vitro for 10 min at 4°C with 25 μg of purified anti-B7-1 1G10 mAb (PharMingen), or anti-B7-2 GL1 mAb (obtained from Dr. A. Rosato, University of Padova, Padova, Italy), or nonimmune rat IgG (Pharmacia, Uppsala, Sweden). Twelve hours later, mice received 100 μg i.p. of the appropriate Ab, and 15 days later they were challenged s.c. in the opposite flank with 10⁵ living nontransfected TS/A cells .

To determine the effects of vaccination by mixed TS/A-1 and TS/A-2 cells, a total of 10⁶ nonreplicating TS/A-1.26 and TS/A-2–22 mixed at a 1:1, 2:1, or 1:2 ratio were injected s.c. into each mouse, followed 15 days later by a contralateral s.c. challenge with 10⁵ living parental TS/A cells.

For optimal neutralization of IL-4 and IFN-γ in vivo during the priming phase of the antitumor immunity, 2 and 1 days before vaccination with nonreplicating TS/A cells, mice received 500 μg i.p. of the rat anti-IL-4 mAb 11B11 (31) or the rat anti-IFN-γ mAb AN18 (32) or nonimmune rat IgG (Sigma). The treatment continued 2 and 7 days after vaccination by injecting i.p. 500 μg of each neutralizing or control Abs. To titrate the relative amount of IL-4 and IFN-γ induced by each nonreplicating TS/A-B7 vaccine, suboptimal amounts of neutralizing mAbs were injected into mice: 400 μg 2 and 1 days before vaccination, followed by 150 μg 2 and 7 days after vaccination. Fifteen days after vaccination, mice received a challenge with 10⁵ living TS/A parental cells. By this time, rat Ig were undetectable in the sera of treated mice, indicating that this neutralization protocol did not affect the effector phase of the antitumor immunity.

To determine the persistence of the anti-B7-1 and B7-2 mAbs on their target molecules, 2 × 10⁶ nonreplicating TS/A-2/1.4 high cells were stained with 2 μg of either 1G10 or GL1 mAb for 20 min at +4°C. After washing and seeding in complete medium, aliquots of cells were drawn after 0, 1, 2, 4, 16, and 24 h of culture at 37°C, stained with a FITC-conjugated goat anti-rat antiserum (Southern Biotechnology Associates, Birmingham, AL), and analyzed by flow cytometry. A regression curved was generated by interpolating the progressively decreased mean fluorescent intensities, which allowed calculation of a 50% dissociation time of about 12 h and 9 h for the anti-B7-1 and anti-B7-2 mAb, respectively.

To determine the cytokine produced by T cells, primed in vivo by the different TS/A-B7 vaccines, three mice per group were vaccinated with 10⁶ nonreplicating parental TS/A, TS/A-1.26, TS/A-2.22, or TS/A-2/1 high-4 cells, respectively, injected s.c. in the left flank. Three days later, mice were sacrificed, and the T cells extracted from pooled draining left inguinal and cervical lymph nodes were activated in vitro at 10⁶/ml using a combination of 50 ng/ml PMA (Sigma) and 1 μg/ml inomycin (Sigma) in complete medium. After 2 h at 37°C, brefeldin A was added at 10 μg/ml for another 2 h at 37°C, followed by fixation in 2% for 15 min at +4°C, and permeabilization in PBS containing 1% BSA and 0.5% saponin (Sigma). Fixed/permeabilized T cells were stained with FITC-conjugated anti-IL-4 BVD4–1D11 mAb, PE-conjugated anti-IFN-γ XMG1.2 mAb, and Cy-Chrome-conjugated anti-CD4 RM4–5, or anti-CD8 53–6.7 mAb (PharMingen). Control staining for cytokine production was performed using FITC- or PE-conjugated isotype-matched mAbs (PharMingen). Cells were analyzed on a FACScan analyzer (Becton Dickinson, Mountain View, CA) by acquiring 3 × 10⁴ CD4⁺ or CD8⁺ T cells in each file.

Unless specified in the text, all experiments in vivo were performed twice with groups of 5–15 mice, and the data were pooled. Frequencies were compared by the Fisher exact test or the χ² test for trend, when appropriate.

To determine the immunogenicity of TS/A adenocarcinoma cells expressing both B7-1 and B7-2 costimulatory molecules, we supertransfected B7-1 cDNA into the highly immunogenic TS/A-2.22 clone. Clones expressing increasing surface levels of B7-1, with constant levels of both B7-2 and MHC molecules, were selected (Fig. 1) and directly compared as nonreplicating cell vaccines for their protective efficacy against a lethal challenge with the nontransfected parental TS/A cells. As shown in Table I, TS/A-2/1 double transfectants that do not express surface B7-1 (clone.1) or express low (clone.2) or intermediate (clone.6) levels of B7-1 are as immunogenic as the single TS/A-2.22 clone. On the contrary, the protection obtained with clone TS/A-2/1 high.4, which expresses high levels of B7-1, decreases to 40%, a value comparable to that obtained with the single TS/A-1.26 clone. Two other TS/A-2/1 high clones, tested in independent experiments, gave similar results (data not shown). Since these data were suggesting a possible dominance of B7-1 in the induction of antitumor response, we supertransfected B7-2 cDNA into the TS/A-1.26 clone and tested double-transfected TS/A-1/2 clones coexpressing comparable high levels of both B7 molecules as nonreplicating cell vaccines. Interestingly, both the bulk culture and four independent TS/A-1/2 clones protected 40% of the mice from a challenge with living parental TS/A cells (data not shown), therefore displaying the lower immunogenicity typical of the TS/A-1 cells. Expression of comparable levels of B7-1 and B7-2 on TS/A cells also decreased the efficacy of the original TS/A-2.22 clone to cure a 24-h-old tumor generated by the s.c. injection of 4 × 10⁴ living TS/A cells (1 × minimal tumorigenic dose (MTD)). Whereas nonreplicating TS/A-2 cells cured 40% of the mice, nonreplicating TS/A-2/1 and TS/A-1 cells similarly cured 10% of the mice (data not shown).

Taken together, the above data suggested that B7-1 plays an active role in modulating costimulation of antitumor immunity by B7-2. To confirm this hypothesis, we studied the effects of blocking either B7 molecule with specific mAbs on nonreplicating TS/A-2/1 vaccines at the time of injection into mice. Nonreplicating TS/A-2/1 cells were mixed in vitro with a saturating dose of either anti-B7-1 or anti-B7-2 mAb and injected s.c. Anti-B7-1 and anti-B7-2 mAbs displayed comparable stability of binding to transfected B7 molecules at 37°C (50% dissociation time at 37°C: 12 h and 9 h, respectively). Twelve hours later, mice received another i.p. injection of 100 μg of the appropriate Ab, and 2 wk later they were challenged contralaterally with a tumorigenic dose of living nontransfected TS/A cells. As shown in Fig. 2, blocking B7-1 on TS/A-2/1 cells neutralized its dominant effects and restored the higher immunogenicity typical of the single B7-2 transfectant, indicating that B7-1 is directly involved in the decreased immunogenicity of the double-transfected TS/A clone.

We also determined whether B7-1 could modulate B7-2-dependent costimulatory effects when expressed in trans. Therefore, mice were vaccinated with a 1:2, 1:1, or 2:1 mixtures of nonreplicating TS/A-1 and TS/A-2 cells. As shown in Table II, the protective effects of TS/A-2 vaccines were never substantially modified by the presence of various doses of TS/A-1 cells, indicating that B7-1 exerted a dominant effect on B7-2 only when coexpressed on the same cell.

Recent data have suggested that B7-1- and B7-2-dependent costimulations lead to different patterns of type 1 or type 2 cytokines produced during the immune response (19, 20). We therefore studied whether we could detect a difference in the balance of endogenous IFN-γ and IL-4 induced during the priming of the protective responses by nonreplicating TS/A-1, TS/A-2, or TS/A-2/1 vaccines. High amounts of either anti-IFN-γ– or anti-IL-4–neutralizing mAb were injected into mice only within the first 7 days after vaccination, corresponding to the priming phase of the antitumor immunity. Fifteen days after vaccination, when no anti-cytokine mAb was detectable in their sera, mice were challenge with 10⁵ living parental TS/A cells. Priming of a protective response by both nonreplicating TS/A-1 or TS/A-2 vaccines was completely abolished by the administration of either anti-IFN-γ or anti-IL-4 mAbs (Fig. 3), establishing 1) the optimal dose of each neutralizing mAbs, and 2) a critical role for both cytokines in the priming of a protective response by B7-expressing TS/A vaccines. Administration of a reduced dose of neutralizing mAbs allowed to titrate in vivo the relative amount of IFN-γ and IL-4 induced by nonreplicating TS/A-1 or TS/A-2/1 and TS/A-2 vaccines. As shown in Fig. 4, the reduced dose of anti-IL-4 but not of anti-IFN-γ mAb completely abrogates priming of protective responses by TS/A-1 or TS/A-2/1 cells. On the contrary, priming of a protective response by TS/A-2 vaccines was inhibited when the reduced dose of anti-IFN-γ, but not of anti-IL-4 mAb, was administered to mice. Thus, this data would indicate that nonreplicating TS/A-1 and TS/A-2/1 vaccines induced in vivo more IFN-γ than IL-4, whereas the opposite quantity of these cytokines is induced TS/A-2 vaccines, demonstrating further that B7-1 actively modulates the costimulatory effects of B7-2 in this tumor model.

To determine the cellular basis for the unbalanced production of cytokines induced by TS/A-1, TS/A2–1, and TS/A-2 vaccines, mice were vaccinated s.c. with a million of each type of nonreplicating B7 transfectant. Three days later, T cells purified from the draining lymph nodes were reactivated in vitro with a polyclonal stimulus, and the pattern of IFN-γ and IL-4 produced by either CD4⁺ or CD8⁺ T cells was determined by intracellular staining and flow cytometry analysis. As shown in Fig. 5, priming in vivo with TS/A-1.26 or TS/A-2/1 high.4 induced more CD4⁺ T cells to produce IFN-γ and fewer cells to produce IL-4 than did TS/A-2.22 vaccines, thus substantially confirming the data obtained by titrating in vivo the relative amount of cytokines induced by nonreplicating TS/A vaccines. At variance with CD4⁺ T cells, all three TS/A-B7 vaccines induced equivalent fractions of CD8⁺ T cells to produce IFN-γ, but never IL-4 (data not shown).

Collectively, these data would therefore suggest that the expression of B7-1 modifies quantitatively the balance of endogenous IFN-γ and IL-4 induced in vivo by TS/A-2 vaccines.

We have previously shown that nonreplicating TS/A mammary adenocarcinoma cells expressing B7-2 are more immunogenic than those expressing B7-1 (27). This tumor system therefore provided us with a useful tool to investigate the basis for the nonredundancy of B7-1 and B7-2. In this study, we have addressed whether B7-1 is simply less efficient than B7-2 in eliciting antitumor immunity, or whether it is actively regulating the immune response in a negative way relative to B7-2. By analyzing three parameters of the immune response induced by TS/A cells expressing one or both B7 costimulatory molecules, namely, protective immunity, therapeutic immunity, and endogenous IFN-γ and IL-4 balance, we show that B7-1 modulates in a dominant way the B7-2-dependent costimulation. The expression of increasing levels of B7-1 into TS/A-2 cells decreases proportionally their immunogenicity, making TS/A-2/1 cells inducing both protective and therapeutic immunity as low as those induced by TS/A-1 cells. On the contrary, the expression of increasing levels of B7-2 in TS/A-1 cells cannot improve their lower immunogenicity, even though the surface level of B7-2 is as high as that of immunogenic TS/A-2 vaccines. The dominant effects of B7-1 over B7-2 is strongly supported by the Ab-blocking experiment in which neutralization of B7-1 in nonreplicating double-transfected TS/A cells before in vivo injection restores the higher immunogenicity typical of TS/A-2. The lack of modulation of B7-2 effects by B7-1 expressed in trans, as observed in vaccination experiments performed by mixing single-transfected TS/A cells, further suggests that the dominance of B7-1 in TS/A-2/1 cells may be due to a direct competition between the two costimulatory molecules for their ligand(s).

As a last criteria to measure the relative role exerted by each B7 molecule during the induction of antitumor immunity, we have determined the pattern of endogenous IL-4 and IFN-γ induced by each B7-expressing tumor vaccine. Also by this criteria, B7-1 is found to exerts a dominant role over B7-2 and makes the pattern of cytokines induced by double-transfected TS/A-2/1 vaccines similar to that induced by single TS/A-1 vaccines. In particular, in the TS/A tumor system, B7-2 induces more IL-4 and less IFN-γ than B7-1. Although these data indicate quantitative rather than qualitative differences in the pattern of cytokine induced by either B7 molecule, they appear to be substantially consistent with a number of other demonstrations, showing that B7-1 and B7-2 costimulate the induction of different patterns of cytokines in several experimental systems (12, 15, 17).

Interestingly, B7-1- and B7-2-expressing vaccine appears to induce unbalanced IFN-γ and IL-4 production in CD4⁺ but not CD8⁺ T cells. Since TS/A is MHC class II negative and poorly responsive to IFN-γ up-regulation of MHC genes in vitro (M. Moro, unpublished observation), CD4⁺ T cells are likely to be induced by indirect presentation of tumor-derived Ags by endogenous APCs, although they may still remain sensitive to B7-dependent costimulation provided by tumor cells in trans (33, 34).

Taken together, these findings may suggest a model of noncompetitive inhibition between two ligands for their receptor (35) in which B7-1 could displace B7-2 from binding to their T cell ligand, but not the other way round. Several molecular data generated in vitro indicate that B7-1 and B7-2 indeed recognize CD28 or CTLA-4 in a nonidentical manner and would support this hypothesis (4, 5, 36, 37, 38). Since CD28 activates while CTLA-4 inhibits T cell activation (20, 39, 40, 41, 42), and the engagement of CD28 and CTLA-4 regulates positively and negatively, respectively, the induction of a Th2 response (43, 44), the differential ligation of CD28 or CTLA-4 by B7-1 or B7-2, would affect both clonal expansion and cytokine production by T cells, thus providing a molecular basis to explain our findings.

Other studies, performed on different antigenic systems, have also suggested that B7-1-dependent costimulatory signal exerts a somehow down-regulatory effect on the course of the immune response. Sethna et al. (45) showed that transgenic mice constitutively expressing B7-1 on B cells have a markedly reduced humoral response. Kearny et al. (46) showed that expansion of adoptively transferred CD4⁺ transgenic T cells upon Ag challenge was consistently enhanced by the concomitant injection of anti-B7-1-blocking mAb, whereas it was partially inhibited by anti-B7-2 mAb . Furthermore, Ab blocking in vivo of B7-1 but not B7-2 increases the germinal center formation in H. polygirus-inoculated mice (47).

Whether the superior immunogenicity of nonreplicating TS/A-2 cells may relate to the unbalance induction of IL-4 and IFN-γ remains to be established. We can only speculate that a different balance of these cytokines influences the development of T cell subsets that are more efficient antitumor effectors than those induced by TS/A-B7-1 vaccines. Because a burst of IL-4 is required for both priming and expansion of CTLs (48, 49), as well as for the development of a protective Th1 response against intracellular parasites (50, 51), nonreplicating TS/A-2 cells may be more immunogenic than TS/A-1 cells due to their superiority in eliciting an initial burst of this cytokine. On the other hand, it is possible that reduced induction of IFN-γ by TS/A-2 vaccines may also relate to their higher immunogenicity. In fact, exposure to IFN-γ may lead to depletion of recently activated T cells, leading to a smaller clonal expansion (52, 53).

Altogether, these findings confirm the nonredundancy of the two B7 molecules in regulating the immune response directed against some tumors and show a model in which B7-1 modulates B7-2-dependent costimulatory effects in a dominant noncompetitive way.

We thank Drs. M. Bellone, P. Protti, and M. G. Roncarolo for the critical reading of this manuscript and suggestions.