Cutting Edge: A Crucial Role for B7-CD28 in Transmitting T Help from APC to CTL¹

The conditional nature of activation requirements for the in vivo priming of CD8⁺ CTL presents an intriguing paradox for immunologists, with responses against some Ags requiring “help” from CD4⁺ Th cells, while others, notably those against certain viruses, can apparently be generated in their absence (1, 2, 3). However, once primed in the absence of T_h, these viral CTL responses wane over time and become ineffective in controlling viral persistence and replication (reviewed in Ref. 4). A deeper understanding of how help is transmitted would allow insight into the instructional program that guides the development and maintenance of CTL and may facilitate its manipulation in the clinical setting (5, 6). Two models have been proposed to explain the role of CD4⁺ T cells in providing help to CTL. The first, supported by early studies on the in vitro generation of CTL, specifies a role for CD4⁺ T cells in the paracrine secretion of cytokines such as IL-2 that are believed to be required by CTL during primary activation (7). This is envisioned to occur while both T_h and CTL are in sufficient proximity to each other at the surface of the same APC to allow the effective use of short-range factors like lymphokines. The alternative model suggests that T help proceeds via the CD4-dependent activation of APC to a state in which they are able to directly prime CTL in the absence of CD4⁺ T cells (8). This model allows the obligate interaction of three cell types to be accomplished through serial two-cell interactions, thus avoiding the temporal and statistical probability drawbacks implicit in the first model. This scenario also provides a possible explanation for Th-independent CTL responses induced by viruses such as lymphocytic choriomeningitis virus (LCMV)³ and vaccinia, which may achieve the same functional activation of APC through direct infection or replication within lymphoid tissues.

Indirect or “cross” priming clearly requires both bone marrow-derived APC and CD4⁺ T_h cells to generate cytotoxic effectors against cell-based Ags and, as such, provides a useful experimental platform for studying the cellular and molecular requirements of T help for CTL (3). In support of the second model, cross-priming studies have shown that the CD4 requirement in CTL priming can be overcome using an agonistic mAb to CD40 on APC and that blockade of CD40 ligand (CD154) on CD4⁺ T cells prevents them from providing help to CTL (9, 10). These results indicate that CD40-CD40L interactions play a central role in APC activation by CD4⁺ T cells and suggest that this event represents the initial step in transmitting T help to CTL. However, once activated by CD4⁺ T cells, it is unclear how the APC becomes competent to autonomously prime naive CTL. Activation induces a number of functional and phenotypic changes within APC, including their migration to lymphoid organs, secretion of cytokines/chemokines, and expression of a range of costimulatory molecules (reviewed in Ref. 11). Regulating the transmission of early costimulatory signals would seem an effective way for APC to control CTL priming, and there exist numerous potentially relevant ligands whose expression on APC is increased upon activation, including B7-1/B7-2 (CD80/CD86), ICAM-I (CD54), 4-1BBL, OX40L, and CD70, among others (12). The B7-1/B7-2:CD28/CTLA-4 costimulatory pathway has received considerable attention as a regulatory control point in T cell responses, as the low basal levels of both B7-1 and B7-2 on immature APC are up-regulated following activation and can interact with either positive (CD28) or negative (CTLA-4) ligands on T cells (13, 14). Previous studies using blocking Abs have demonstrated that either B7-1 or B7-2 are sufficient for cross-priming of Th-dependent CTL but did not address which CTL receptors were involved or whether APC activation altered the B7 requirement (15). We have now investigated the role of the costimulatory molecules B7-1 and B7-2 and their T cell ligands, CD28 and CTLA-4, in the transmission of help to CTL by CD40-activated APC. Our results show that both B7-1/B7-2 and CD28 are absolutely required for CTL cross-priming by CD40-activated APC and indicate that CD28 can function as a kind of molecular rheostat through which up-regulation of B7 molecules on APC is perceived as “help” by CTL, leading to their primary activation and functional development.

C57BL/6 (H-2^b) and CD28^−/− (H-2^b) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). B7-1^−/−, B7-2^−/−, and B7-1/2^−/− (H-2^b) strains have been previously described (16). All mice were maintained by in-house breeding at the La Jolla Institute for Allergy and Immunology (San Diego, CA), and were maintained under specific pathogen-free conditions in accordance with guidelines by the Association for Assessment and Accreditation of Laboratory Animal Care International.

Mouse embryonic cell (MEC) lines expressing the human adenovirus type 5 early region 1 (Ad5E1) were produced by transfection of both C57BL/6 and TAP^−/− MEC lines and have been previously described (17). The EL-4 thymoma was purchased from the American Type Culture Collection (Manassas, VA). MEC were cultured in DMEM, and EL-4 cells and splenocytes were cultured in IMDM. All media were supplemented with 10% FCS, 50 μM 2-ME, 2 mM l-glutamine, 20 U/ml penicillin, and 20 μg/ml streptomycin.

Groups of mice were immunized s.c. in the right flank with 1 × 10⁷ irradiated (3000 rad) Ad5E1-TAP^−/− MEC in 200 μl PBS. Mice were sacrificed 14 days following immunization, and splenocytes were restimulated for 6 days in vitro in 24-well plates with irradiated syngeneic Ad5E1-B6 MEC (5 × 10⁶ splenocytes:5 × 10⁵ MEC). Effectors were then purified from culture debris over Lympholyte-M (Cedarlane Laboratories, Hornby, Ontario, Canada) and evaluated for cytotoxicity by the JAM assay (18). Percentage of specific killing of [³H]thymidine-labeled EL-4 cells loaded with either a control peptide (OVA_257–264; SIINFEKL) or with the E1B_192–200peptide (VNIRNCCYI) and serially diluted at various E:T ratios was calculated as ((S − E)/E) × 100, where S represents spontaneous and E represents experimental [³H] retention values. All data shown are representative of at least two experiments.

Hybridomas were cultured in Life Technologies Protein-Free Hybridoma Medium-II (Invitrogen, San Diego, CA), and mAbs were isolated by dialysis of supernatants. To deplete CD4⁺ cells, 150 μg of GK1.5 Ab was given i.p. on the first 3 days before immunization and every third day thereafter. For CD40 activation, 300 μg of FGK45 Ab (19) or control Ab (rat IgG2a) was given i.v. on day 0. For CD28 activation, 100 μg of PV1 Ab (Southern Biotechnology Associates, Birmingham, AL) or control Ab (hamster IgG) was given i.v. on day 0. For in vivo blockade of CD28, 150 μg of 37.51 Ab was given i.p. the day before immunization and on days 0, 1, 4, and 7. For in vivo blockade of CTLA-4, 100 μg of UC10 Ab (20) was given i.p. 1 day before immunization and on days 0, 3, 7, and 11. All mAbs were administered in 200 μl PBS.

Groups of mice were infected i.p. with 1 × 10⁵ PFU LCMV (Armstrong strain). Mice were sacrificed 7 days after infection, and splenocytes were tested for effector function by the JAM assay as described above against [³H]thymidine-labeled EL-4 cells loaded with either OVA_257–264 as a control peptide or with the LCMV GP_33–41 peptide (KAVYNFATM) and serially diluted at various E:T ratios.

We used a robust and well-characterized system of cross-priming to investigate the costimulatory requirements for activation of naive Th-dependent CTL by CD40-activated APC in vivo. Immunization of C57BL/6 mice with TAP^−/− Ad5E1-MEC results in cross-priming of E1B_192–200-specific effector CTL by host APC that is dependent on either CD4⁺ Th cells or administration of an agonistic CD40 mAb (Fig. 1,a) (10, 17). Immunization of “knockout” mice lacking either B7-1/B7-2 or CD28 reveals that the requirement for these molecules in CTL cross-priming cannot be overcome by ligation of CD40 (Fig. 1, b and c). Similar results were observed in knockout mice depleted of CD4⁺ cells and in mice lacking a functional MHC class II gene (I-A^b−/− mice; data not shown). Cross-priming in this system required either B7-1 or B7-2, as immunization of single knockout mice resulted in responses that were comparable to wild-type mice (data not shown). The requirement for B7-1/B7-2 and CD28 was specific for CTL cross-primed against cell-associated Ag, as primary CTL responses against an H-2D^b-restricted epitope of LCMV GP_33–41 were clearly detectable in LCMV-infected wild-type, B7-1/B7-2, or CD28 knockout mice, consistent with previous studies (Fig. 1, d–f; Ref. 21). These data indicate that B7-1/B7-2 and CD28 are genetically required for cross-priming of helper-dependent CTL, whether help is provided by CD4⁺ Th cells or APC activation via CD40.

The failure of anti-CD40 Abs to restore cross-priming of Th-dependent CTL in CD28-deficient mice suggests that CD28 transmits a key signal through which help is provided to CTL by activated APC. We investigated this directly by blocking CD28 with a mAb under conditions of CD40-mediated APC activation. Treatment of CD4-depleted mice with CD28-blocking, but not control, Abs prevented CTL cross-priming by CD40 activated APC (Fig. 2). This indicates that CD28 represents a critical conduit through which help is transmitted to CTL by CD40-activated APC.

In the absence of APC activation, selective engagement of the negative B7-receptor CTLA-4 by limiting surface levels of B7-1/B7-2 has been proposed as a mechanism through which T cell tolerance can be achieved (22). In this model, activated APC expressing higher B7-1/B7-2 levels would allow coengagement of CD28 and lead to T cell activation. We therefore investigated whether blockade of CTLA-4 would leave the available B7-1/B7-2 molecules on APC free to interact with CD28 and thereby result in CTL priming in the absence of Th. Ab blockade of CTLA-4 in CD4-depleted mice was unable to overcome the requirement for CD40 ligation in cross-priming of E1B_192–200-specific CTL (Fig. 3). These data demonstrate that the levels of B7 available on nonactivated APC are insufficient to induce CTL cross-priming, even when CTLA-4 is blocked, and further suggest that if B7 levels play a direct role in the commitment to CTL cross-priming, it is likely to be mediated through the strength of signals received via CD28, rather than through differential interactions with CTLA-4 vs CD28.

A key prediction from our results is that if, in the absence of T help or CD40 ligation, the comparatively weaker CD28 signals generated by the low levels of B7-1/B7-2 molecules expressed on nonactivated APC could be augmented, CTL priming would result. In this way, the signals transmitted through CD28 by a fully activated APC could be mimicked in the absence of help. This possibility was examined directly using stimulatory Abs against CD28 or, as a positive control, Abs against CD40 to restore CTL cross-priming in CD4-depleted mice. Using a stimulatory mAb to potentiate signals transmitted through CD28 resulted in priming of Th-dependent effector CTL in the absence of CD4⁺ T cells (Fig. 4). Although clearly cytotoxic in our experiments, the lytic activity of secondary cultures primed in response to the agonistic CD28 Ab was not as great as that induced by CD40-ligation. This may be due to differences in the strength and/or duration of CD28 signals transmitted by the monoclonal anti-CD28 Ab vs those mediated by up-regulation of both B7-1/B7-2 molecules on activated APC. Conversely, additional costimulatory signals generated by CD40-activated APC may influence the strength of lytic activity detected in our in vitro assays. Despite these variations, our results clearly demonstrate that differences in the strength of signals transmitted through CD28 can be integrated into the cellular response of Th-dependent CTL and thereby lead to their priming in the absence of CD4⁺ T cells.

Cross-presentation by bone marrow-derived APC is an important mechanism through which naive T cells are exposed to both self and foreign peripheral Ags in vivo (23, 24). The ability of such APC to exist in functionally distinct states of activation has profound implications for the regulation of immune responses, as cross-presentation can result in either priming or tolerance (3, 25). Accordingly, there has been substantial interest in understanding the pathways that mediate these disparate outcomes. With respect to cross-priming of cytotoxic CD8⁺ T lymphocytes, APC activation via CD40-CD40L interactions is believed to underlie the requirement for CD4⁺ T help, but it has been unclear how activation leads to CTL priming. Our present data clearly identify B7-1/B7-2 and CD28 as critical downstream components of this pathway. This is supported by the findings that CD40-activated APC nonetheless require B7-1/B7-2 to prime CTL and that regulation of CTL priming can be controlled at the level of CD28 signaling either by blockade or stimulation. The same is not true of the negative B7 receptor CTLA-4, as Ab blockade of this molecule did not lead to CTL priming in the absence of T help or APC activation. It is unlikely that low levels of CD28, the assumed transducer of B7-mediated activating signals in this scenario, is the limiting factor as the stimulatory CD28 mAb leads to CTL priming under the same circumstances. This suggests that, in the absence of activation, the B7-1/B7-2 levels on APC are insufficient to transmit a priming costimulus through CD28.

Our results favor a model in which it is the strength of signal through CD28, rather than competition with CTLA-4 for available B7 molecules, that is functionally integrated by the CD8⁺ T cell into its commitment to prime. This interpretation assumes that CD28 functions not as an “on-off” switch but rather as a kind of “rheostat” which, depending on the strength and/or duration of its engagement by B7 molecules, can display a degree of plasticity in the intracellular signals it generates. This view is supported by the presence of distinct functional domains within CD28 as well as its capacity to assume differential roles in T cell activation (26, 27). Although our data define a critical role for B7-1/B7-2 and CD28 in transmitting help from APC to CTL, they do not rule out a role for other interactions involving B7-1/B7-2 or for the subsequent involvement of other cytokine or costimulatory pathways in CTL priming (28, 29).

The requirement for B7 and CD28 in CTL cross-priming by CD40-activated APC contrasts with that observed for viruses such as LCMV that can apparently generate primary CTL in the absence of CD4⁺ cells, CD28, B7-1/B7-2, or even bone marrow-derived APC (21, 30). It is unclear whether this discrepancy is due to the increased duration and intensity of antigenic stimulation that characterizes LCMV infection in rodents or whether the host response to the infection influences the activation threshold of CTL against this and other viruses. As we move forward in our understanding of how T help contributes to CTL priming, it will be important to learn not only the mechanism through which it is provided, but also to understand the circumstances under which it can be bypassed.

We thank K. Banks and the animal facility staff of the La Jolla Institute for Allergy and Immunology for their animal screening and maintenance services.