INTRODUCTION

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

Two types of costimulatory interactions seem critical in the development of productive immunity (Figure 1). These are the CD28-B7 and CD40-CD154 (CD40L) interactions. Whereas CD40 and CD154 appear to provide signals primarily in the direction of the CD40 bearing cell, the CD28-B7 system appears to work primarily in the opposite direction, with CD28 delivering a positive signal of costimulation to the T cell. There are important links between these two pathways and it has been shown that ligation of CD40 on the antigen-presenting cell (APC) results in upregulated expression of B7 proteins. The CD28 homolog CTLA4 is also critical in immune responses but appears to function primarily as a negative regulator of immunity. This review focuses on the CD28/CTLA4- B7 pathways.

View larger version (61K): [in this window] [in a new window]   Figure 1.   Costimulation of resting T cells. Schematic representation of the costimulatory pathways involved between an APC and a T cell after antigen recognition through the MHC-peptide-TCR interaction. CD40L- CD40 interaction activates the APC, which in turn upregulates CD86 and activates T cells through CD28 signaling. After activation and proliferation, CD80 and IgV-CD80 are upregulated on APCs and CTLA4 is upregulated on T cells. Signaling through CTLA4 then provides a negative regulatory signal.

	B7-CD28/CTLA-4 PATHWAY

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

The B7 genes cd80 and cd86, encode proteins that are an essential component in the regulation of T cell-mediated immune responses. Ligation of CD28 on T cells by CD80 (B7.1) or CD86 (B7.2) on APCs enhances the production and stability of several cytokine transcripts (1, 2), thereby providing a signal of costimulation to the T cell. Deliberate interruption of the CD28-B7 interaction has been shown to inhibit immune responses both in cell culture and whole animal systems and in some instances results in T cell tolerance (e.g., References 3-8, reviewed in Reference 9). A second ligand for the products of cd80 and cd86 genes, CTLA4, provides a negative signal to the T cell, thus limiting immune responses ([10-16]; reviewed in Reference 17). Whereas the need for costimulation in the responses of naive CD4⁺ T cells is well established, its requirement for reactivation of effector and memory T cells is less clear. Perhaps the discovery of new B7 homolog proteins described below might clarify this issue. However, in the absence of costimulation T cell clones may become unresponsive or anergic (4, 18, 19) and at least for helper T cell type 1 (Th1) responses and some Th2 responses a requirement for costimulation has also been demonstrated (reviewed in Reference 20). To begin to understand how these four proteins control the magnitude of immunity, it is necessary to consider the kinetics and level of their expression as well as the affinities of interaction between them. CD28 and CD86 are expressed constitutively (albeit at varying levels on different cells) whereas CTLA4 and CD80 appear as cell surface proteins only on activation of T cells and APCs, respectively ([6, 21-24], reviewed in Reference 25). Both CTLA4 and CD80 are expressed at approximately 10-fold lower levels than are CD28 or CD86, but because CD80 binds with higher affinity (due to its slow dissociation rate) to CTLA4 than to CD28 and with higher affinity to both than does CD86 (26) it is possible that CD80-CTLA4 binding prevails in the interaction of activated cells. It has therefore been suggested that normally in the generation of an immune response the interaction of CD86 with CD28 dominates in costimulation whereas that of CD80 and CTLA4 governs negative signaling (30, 31).

	cd80 AND cd86 GENE PRODUCTS

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

cd80 and cd86 are both members of the immunoglobulin gene superfamily and encode a glycoprotein comprising a signal peptide, extracellular immunoglobulin variable (IgV)-like and Ig constant (IgC)-like domains, a transmembrane domain, and a cytoplasmic domain (Figure 2) (32). Like numerous genes of the immune system cd80 and cd86 can encode various gene products through alternative splicing. We, like others (33, 34), have found through the use of the reverse transcription-polymerase chain reaction (RT-PCR), performed on RNA extracted from lipopolysaccharide (LPS)-stimulated mouse splenocytes, that a naturally occurring truncated form of CD80 is expressed that results from the splicing of exon 3, which encodes the IgC domain. This splice variant, herein called IgV-CD80 (in other studies also called B7-1a, MB7-2, or B7IgV), potentially encodes a protein possessing only the IgV-like domain outside the cell membrane, with normal transmembrane and intracellular domains. Studies using cell lines transfected with IgV-CD80 cDNA have demonstrated that this isoform can be expressed as a cell surface protein but have generated contradictory data regarding its binding to CD28 and CTLA4 (33, 34). It has hitherto been difficult to analyze the normal expression and function of this isoform because there is no antibody that binds exclusively to IgV-CD80. A further variant of CD80 has been identified (35) in which either exon 5 and/or exon 6 are used to generate two different intracellular domains, but no distinct functions have yet been established for the two variants. Alternative splicing is a common mechanism used in prokaryotes and eukaryotes in order to modulate the function of proteins by producing variants with differing activities (36). Indeed, the immune system shows numerous examples of this, for instance in the transcription of Fas (37) and the interleukin 1 receptor (IL-1R) antagonist (38). It is tempting to speculate therefore that there is an important role for the IgV-CD80 isoform in regulation of T cell responses or in modulating the activity(s) of the other B7 proteins.

View larger version (33K): [in this window] [in a new window]   Figure 2.   The extended B7 family. The B7 family comprises several genes, all members of the immunoglobulin super family. cd80 and cd86 genes encode proteins that bind to both CD28 and CTLA4, delivering a positive and negative signal, respectively. B7h/B7RP-1 binds to ICOS, providing a positive signal to memory type T cells. B7-H1, whose ligand is not described yet, provides a positive signal to T cells.

To assess the function of IgV-CD80, we have used a gene-targeting strategy that prevents expression of full-length CD80 but leaves the short isoform, IgV-CD80, intact and normally expressed (Figure 3). Our experiments in a transplantation model using our targeted animals have revealed an unexpected negative regulatory role for IgV-CD80. Targeted C3H mice show a prolonged survival of an allogeneic C57BL/10 heart transplant when compared with the graft survival in wild-type recipients. The presence of IgV-CD80 was responsible for the prolongation of survival observed because treatment with an antibody blocking IgV-CD80 could abrogate its beneficial effect. The mechanism of action of IgV-CD80 appears to involve some interaction with CTLA4 because we could reverse the beneficial effect of IgV-CD80 by blocking access to CTLA4 with antibody (L. Bugeon and M. J. Dallman, unpublished data).

View larger version (15K): [in this window] [in a new window]   Figure 3.   Knockout mice: targeting CD80. A disruptive neo DNA cassette was inserted in exon 3 by gene targeting. After alternative splicing of exon 3, a naturally occurring event, IgV-CD80 is still normally expressed while full-length CD80 can no longer be expressed.

Isoforms in Different Species

Interestingly, a variety of CD80 and CD86 isoforms are produced and these differ between species. For example, secreted forms of CD80 and CD86 have been identified in dog (39). Interestingly, distinct molecular mechanisms are used to generate these two secreted forms. While mRNA for secreted dog CD80 is generated by a differential use of the transcription termination signal and the exclusion of both transmembrane and cytoplasmic domains, secreted dog CD86 is generated by deleting a transmembrane exon through alternative splicing. In human, soluble CD80 can be detected in synovial fluids (40), although the mechanism by which the soluble protein is formed is unknown. In mouse no soluble forms of either CD80 or CD86 have yet been identified. Conversely, no isoform equivalent to IgV-CD80 has been found in human or dog. Thus, it appears that the production of isoforms for both cd80 and cd86 can differ in their structure between species. Nevertheless, these different shorter gene products such as IgV-CD80 and secreted CD80, for example, could indeed achieve the same physiological role, that is, downregulate the immune response, by providing a weaker signal or even by blocking a costimulatory signal.

Extended B7 Family

Two novel genes have been identified that belong to the B7 family (Figure 2). The first, in mouse, called B7h for B7 homolog, or B7RP-1 for B7-related protein 1, was described by Swallow and co-workers (41) and Yoshinaga and coworkers (42), respectively. The second, in human, called B7-H1 for B7 homolog 1, was described by H. Dong and coworkers (43). Because human B7-H1 and mouse B7h/B7RP-1 share only 22% amino acid identity, it is unlikely that these proteins are orthologs. Furthermore, human B7-H1 does not appear to bind to ICOS (inducible T cell costimulator), unlike B7h/ B7RP-1. A search for the orthologs of mouse B7-H1 and human B7h may clarify this question.

B7h/B7RP-1. The amino acid identity both between B7h/ B7RP-1 and B7.1 (CD80) and between B7h/B7RP-1 and B7.2 (CD86) is 20%. Because the amino acid identity between B7.1 and B7.2 is similar and in the same extracellular portion, it makes B7h/B7RP-1 a new member of the B7 family of the immunoglobulin superfamily. However, while B7.1 and B7.2 share the same ligands, that is, CD28 and CTLA-4, B7h/ B7RP-1 binds to a newly described CD28 homolog ICOS (44) and does not bind to CD28 or CTLA4.

These studies describe interesting characteristics for this new B7 molecule that differ from those of both CD80 and CD86. Indeed, whereas CD80 and CD86 appear to provide an essential costimulatory signal to T cells in the initiation of the immune response, this does not seem to be the case for B7-H1 and B7h/B7RP-1.

B7h is constitutively expressed on B cells (41) but does not appear to be expressed on dendritic cells (DCs) (42). It can function as a costimulatory ligand for T cells. Perhaps more interestingly, Swallow and co-workers reported that B7h could be induced by tumor necrosis factor  (TNF-) at the surface of fibroblasts or on nonlymphoid tissues after LPS activation (41) This is different from B7.1, which requires the presence of both interferon  (IFN-) and TNF- to be upregulated (45). When Yoshinaga and colleagues (42) analyzed the effect of a B7RP-1-Ig fusion protein in vitro, they found a pronounced stimulatory effect in the rechallenge phase rather than in the sensitization phase of oxazolone-induced contact hypersensitivity. In the same model, blocking the CD28/CD80/CD86 pathway shows an effect only in the sensitization phase (46). ICOS, the ligand for B7h/B7RP-1, is not expressed on naive T cells, but is induced on T cells on activation (44). ICOS expression is observed at the surface of cells carrying markers for resting memory cells. While interactions between CD28 and CD80/CD86 trigger the production of both IFN- and interleukin 2 (IL-2) by T cells, B7h-Ig or anti-ICOS monoclonal antibodies (MAbs) induce the production only of IFN- by T cells.

Taken together, these observations suggest that B7h/B7RP-1 could be involved in sustaining local inflammatory responses, by providing costimulatory signals to T cells previously activated through the CD28/CD80, CD86 pathway in the context of professional antigen-presenting cells.

B7-H1. B7-H1 enhances the proliferative response of T cells to anti-T cell receptor (TCR) stimuli or allogeneic antigens (43). This costimulation pathway induces the production mainly of IL-10 but also of lesser amounts of IL-2. Interestingly B7-H1 transcripts are found at high levels in organs such as placenta, lung, and testis.

	STRUCTURE AND EVOLUTION OF THE B7 FAMILY

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

The B7 genes seem to belong to a distinct subgroup of the immunoglobulin superfamily that includes three MHC-encoded molecules, namely, butyrophilin (BT), myelin oligodendrocyte glycoprotein (MOG), and the chicken B-G antigen (47). These molecules all have an IgV domain and share 40 consensus residues. Whereas B-G and MOG possess an IgV domain only, BT has IgV and IgC domains like full-length CD80/ CD86. The IgC domain of these molecules belongs to the IgC1 group. IgC1 has been found in MHC class I and II molecules, 2-microglobulin, immunoglobulins, and the T cell receptor, suggesting that this group of genes was once linked.

	WHY ARE THERE SO MANY COSTIMULATORY MOLECULES?

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

The fact that there appear to be so many costimulatory molecules, in other words that the system seems so redundant, could suggest that the existence of costimulation is so critical to our survival that multiple receptor ligands are required. Alternatively, the different ligands may mediate subtly different effects through binding to the different receptors. There is evidence that CD80 and CD86 preferentially induce Th1 and Th2 cytokine-producing cells, respectively, but this is not a universal finding. Indeed, in our own work, in which costimulation delivered to T cells is defined by the use of CD80- and CD86-Ig, we could find no evidence of preferential cytokine production when using either ligand. As might be expected, given the affinities of each for CD28, however, CD80 induced much higher levels of proliferation and cytokine (both Th1 and Th2) than did CD86.

Given the expression patterns of CD28, CTLA4, CD80, and CD86 (Figure 4), it has been suggested that CD28 reacts preferentially with CD86 to provide costimulation and CD80 reacts with CTLA4 to resolve immune responses. However, there is little evidence that this is the case and we have data both from studies using the fusion proteins and from our knockout mice to suggest that CD80 is critical in maintaining T cell responses.

View larger version (13K): [in this window] [in a new window]   Figure 4.   Expression kinetics of costimulatory receptors and ligands. CD86 on APCs and CD28 on T cells are constitutively expressed at low levels and upregulated on activation whereas CD80 and CTLA4 are detectable at the cell surface only after activation.

	THERAPEUTIC STRATEGIES BASED AROUND BLOCKING COSTIMULATION

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

CTLA4-Ig, which blocks all interactions between CD28, CTLA4, and the two original members of the B7 family, CD80 and CD86, can provide extremely potent immunosuppression (Figure 5). Indeed, we, like others (7, 48), have shown that administration of a short course or even a single injection of CTLA4-Ig can induce transplantation tolerance. In our experiments delivery of the fusion protein either by adenovirus or by injection of a single dose of protein can prevent rejection of fully allogeneic rat liver transplants and result in antigen-specific tolerance. However, because CD28 and CTLA4 deliver opposing signals and because blocking CTLA4 alone can inhibit tolerance induction, a more effective strategy to prevent immune responses would seem to necessitate blocking CD28 alone. This type of strategy may be less effective than total blockade induced with CTLA4-Ig CD28 knockout mice can rapidly reject grafts (49). The reasons for this are not completely clear, but it may be that signals of costimulation delivered by, for instance, ICOS, can compensate for the lack of CD28 signaling. Why this is not the case when CTLA4-Ig is used (because CTLA4 has been reported not to bind to B7h/ B7RP-1) is unclear, but may indicate that absence of CD28 during development allows the emergence of a developmental compensatory mechanism that is not manifest in short-term experiments in adult animals. Alternatively, CTLA4-Ig may bind to other B7 homologs that have yet to be described. However, we do not believe that this is can be the case because treatment with a combination of CD80 and CD86 antibodies can be as efficient as a CTLA4-Ig treatment at prolonging graft survival (49).

View larger version (42K): [in this window] [in a new window]   Figure 5.   Blocking costimulation through CD28. CTLA4-Ig fusion protein is a potent blocking agent of the costimulatory pathway. It has been well described for its beneficial effect in various models of transplantation and autoimmune disease where anergy and/or specific tolerance can be induced.

The timing of delivery of the costimulatory block may be critical. Indeed, it is the view of some workers that time for the immune system to receive an initial activating signal in the presence of costimulation must be allowed in order to achieve tolerance. In our hands delivery of a single shot of CTLA4-Ig 2 d after transplantation allowed the majority of animals to develop tolerance in the rat liver transplant model, a result in keeping with those of Lin and coworkers (48). Other workers, however, have found that CTLA4-Ig delivered from the time of transplantation can induce tolerance (7).

	CONCLUSION

TOP ABSTRACT INTRODUCTION B7-CD28/CTLA-4 PATHWAY cd80 AND cd86 GENE... STRUCTURE AND EVOLUTION OF... WHY ARE THERE SO... THERAPEUTIC STRATEGIES BASED... CONCLUSION REFERENCES

The CD28/CTLA4-B7 pathways are critical in the generation and resolution of productive immunity. The extended family of costimulatory proteins is complex and clearly our understanding of the integration of signals delivered by and received from each of the players is incomplete. A further complication, which we have not addressed here, stems from the finding that B7 proteins may be expressed by T lymphocytes. The function of these proteins on the T cell is unclear, although they apparently do not provide costimulation with the efficiency of that delivered by the APC, because in experiments using purified T cells stimulated by CD3 antibody, exogenous costimulation is required for proliferation to be observed. A view of how costimulatory signals may result in activation or tolerance in the immune system is given (Figure 6) and attempts to tie together our understanding of this pathway with other signals that may also be critical.

View larger version (22K): [in this window] [in a new window]   Figure 6.   CD28 and CTLA4 in activation and tolerance. After an interaction with antigen and APCs, T cells may become activated to induce productive immunity or adopt an alternative fate, that of proliferative unresponsiveness and tolerance. In this latter case abortive activation (probably dependent on CD28 signaling) appears to be required but is associated with cell death and the emergence of regulatory cells. Because Notch signaling can prevent apoptosis in T cells we suggest that delivery of this signal at the time of APC-T cell interaction allows T cells to survive and adopt a regulatory phenotype, which is determined by CTLA4 (with or without cytokines) signaling.