The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sørensen, P.
Right arrow Articles by Björk, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sørensen, P.
Right arrow Articles by Björk, P.
Right arrowPubmed/NCBI databases
*Substance via MeSH
The Journal of Immunology, 2004, 172: 6803-6809.
Copyright © 2004 by The American Association of Immunologists

Identification of Protein-Protein Interfaces Implicated in CD80-CD28 Costimulatory Signaling

Poul Sørensen1,*, Martin Kussmann2,{dagger}, Anna Rosén*, Keiryn L. Bennett3,{dagger}, Dorthe da Graça Thrige*, Kristina Uvebrant*, Björn Walse*, Peter Roepstorff{dagger} and Per Björk4,*

* Active Biotech Research, Lund, Sweden; and {dagger} Institute for Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
The B7 ligands CD80 and CD86 on APCs deliver either costimulatory or inhibitory signals to the T cell when interacting with their counter-receptors CD28 and CD152 (CTLA-4) on the T cell surface. Although crucial for lymphocyte regulation, the structural basis of these interactions is still not completely understood. Using multivalent presentation and conditions mimicking clustering, believed to be essential for signaling through these receptors, and by applying a combined differential mass spectrometry and structural mapping approach to these conditions, we were able to identify a putative contact area involving hydrophilic regions on both CD28 and CD80 as well as a putative CD28 oligomerization interface induced by B7 ligation. Analysis of the CD80-CD28 interaction site reveals a well-defined interface structurally distinct from that of CD80 and CD152 and thus provides valuable information for therapeutic intervention targeted at this pathway, suggesting a general approach for other receptors.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
It is well known that T cells need a second, costimulatory signal in addition to the Ag peptide presented on MHC class II to get fully activated. The major and dominant costimulatory signal is delivered by the B7 ligands CD80 and CD86. They are expressed on the surface of APCs and have been shown to specifically interact with CD28 on the T cell. Hence, the B7 ligands represent very interesting therapeutic targets for preventing autoimmune disease, graft rejection, and promoting tumor immunity (1). Although CD86 is thought to act early in an immune response and in the primary lymphoid tissue, CD80 acts later and in the periphery at the site of inflammation (2, 3). Upon activation, T cells also up-regulate a homologue to CD28, CD152, or CTLA-4, which interacts with both B7 ligands but mediates a negative signal to the T cell leading to termination of the immune response (4). The proteins involved in this costimulatory pathway are all type I membrane glycoproteins belonging to the Ig superfamily. CD80 and CD86 consist of variable and constant Ig-like domains and exhibit ~20% sequence identity with each other, whereas CD28 and CD152 are disulphide-linked homodimeric proteins sharing ~30% sequence identity in their variable Ig-like chains (1). To date, structural information is available for human and murine CD152 (5, 6), human CD80 (7), and the complex of CD152 with CD80 and CD86 (8, 9). Despite the crucial role of CD28 in delivering the costimulatory signal to the T cell, the structure of CD28 has not yet been solved. Structural data of the CD152-CD80 and CD152-CD86 complexes as well as data from alanine mutants have revealed that the MYPPPY loop in the CDR3-like region of CD152 forms part of the contact area with the B7 ligands (8, 9, 10). Mutagenesis studies have shown that this strongly conserved hexapeptide sequence in CD28 and CD152 is essential also for binding of CD28 to the B7 ligands (5, 11, 12) but cannot explain the 10- to 100-fold lower affinity for the latter interaction (13, 14, 15).

Therefore, the aim of this study was to elucidate further the features that govern specificity for these receptors. We have used a method that does not suffer from the limitations that are inherent in e.g., mutagenesis-based approaches (16) in which one of the major caveats is the introduction of structural mutations that disrupt native interactions. An alternative approach that circumvents this is the use of combined chemical cross-linking and differential mass spectrometry methods that we have previously developed for unraveling protein-protein interfaces (17) and that have been described recently by others (18, 19). Importantly, this approach also allows the study of transient interactions that are otherwise hard to pick up by traditional methods. In the present study, special attention was drawn to conditions favoring multivalent CD80-CD28 complexes, as formation of the immunological synapse (also known as supermolecular activation cluster) is believed to be essential for costimulatory signaling (20, 21, 22, 23, 24, 25, 26, 27). By applying cross-linking under conditions mimicking clustering and multivalency (in this study "pseudo-clustering" conditions) and subsequent differential mass spectrometry and structural analysis, a contact area between CD28 and CD80 distinct from that of CD152 and CD80 was identified. Furthermore using this approach, a putative CD28 oligomerization interface induced by B7 ligation was indicated. These results provide valuable information for therapeutic intervention targeted at the CD80-CD28 pathway and suggest a general approach for the study of other cell surface ligand-receptor interactions.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Experimental materials

Recombinant human CD28 and human CD80 were obtained as fusion proteins with murine IgG1(Fc) (CD28Fc) and murine IgG2a(Fab) (CD80Fab) and were purified to homogeneity as described (28). CD28Fc was expressed as a single polypeptide chain in adherent Chinese hamster ovary (CHO)5 cells with the extracellular part of CD28 (134 aa long) connected in its C-terminal position to the Fc domain of mouse IgG1 (hinge plus the CH2 and CH3 domains). Cysteine to serine mutations were introduced in the hinge region of the Fc moiety to allow the natural disulphide-mediated dimerization of the CD28 part to occur. The CD80Fab protein was obtained by assembly of the extracellular domain of CD80 (length 208 aa), fused by C-terminal via a 18-aa linker with the {kappa}L chain of the C215 mAb (29), and the corresponding C215 mAb Fd domain (VH plus CH1) after coexpression in HEK 293 cells. Identity of both fusion proteins was confirmed by Western blotting (using Abs recognizing both the CD28/CD80 and the Ig moieties) and by N-terminal sequencing. Anti-mouse polyvinyl toluene and streptavidin-coated scintillation proximity assay (SPA) beads, L-4,5-[3H]leucine, specific activity 120–190 Ci/mmol, and N-succinimidyl(2,3-[3H]propionate), 97.0 Ci/mmol, were purchased from Amersham Biosciences (Piscataway, NJ). Hygromycin C was from Boehringer Mannheim (Indianapolis, IN) and 3,3'-dithiobis(sulfosuccinimidyl-propionate) (DTSSP) was obtained from Pierce (Rockford, IL). CHO cells, CHO cells transfected with human CD28, or human HLA-DR4 and CHO cells cotransfected with human CD80 and DR4 were obtained as described in (28). Expression was verified by FACS analysis (data not shown).

Metabolic labeling of cell surface proteins

The parental and transfected cells were cultured as described (28) except that hygromycin B (0.5 mg/ml) was used as selection agent for CD28-CHO cells. Incorporation of [3H]leucine was made according as previously described (30), but with the cells resuspended at a somewhat lower density; 1 x 105 cells/ml (CHO-DR4; CHO-CD80-DR4) and 2 x 105 cells/ml (CHO; CHO-CD28) assay buffer, after metabolic labeling.

SPA analysis

The cell-based SPA was performed as described (30) with either CD28Fc or CD80Fab captured on anti-mouse polyvinyl toluene SPA beads at final concentrations ranging from 0.4 to 50 nM or 0.6 to 70 nM, respectively, and at a cell density of 104 cells per well. For cell-free SPA, 50 µl of streptavidin-SPA beads (final concentration 0.7 mg/ml) was sequentially incubated at room temperature for 90 min with 50 µl of biotinylated anti-mouse {kappa} Abs (9 µg/ml; Southern Biotechnology Associates, Birmingham, AL) and varying concentrations of CD80Fab (1.5–100 nM). CD28Fc, radiolabeled with [3H]succinimidyl propionate to a specific activity of ~2 mCi/mg protein and serially diluted in PBS with 1% w/v BSA to final concentrations of 25, 50, and 100 nM, was added to the CD80Fab-coated SPA beads and incubated for 2 h at room temperature. The amount of beads was chosen so that all CD80Fab was captured even at the highest concentration used. For determination of nonspecific binding, the SPA beads were incubated with equimolar concentrations of the corresponding Fab instead of CD80Fab. Other conditions were as stated for the cell-based SPA with the affinity estimated as the half-maximal binding Bmax/2 (30).

CD80-CD28 interaction and cross-linking assay

Suspensions of CD80Fab- and CD28Fc-coated SPA beads were prepared by adding 50 µl of each fusion protein dissolved in PBS without BSA to 25 µl of anti-mouse SPA beads (40 mg/ml). The suspensions were then mixed and incubated for 3 h at room temperature at final concentrations of 100 nM CD80Fab and 50 nM CD28Fc. During incubation, sedimented beads were resuspended using a pipette. After incubation, the suspension was centrifuged at 2000 rpm for 30 s and the supernatant removed. The beads were resuspended in 150 µl of PBS, 5 µl of a freshly prepared 10 mM solution of DTSSP in PBS was added, and the suspension incubated at ambient temperature for 30 min. The reaction was quenched by the addition of 7.5 µl of 1 M Tris-HCl, pH 7.5, and incubated for a further 15 min. The beads were centrifuged at 2000 rpm for 30 s, the supernatant was removed, and the beads were washed with 200 µl of PBS. The cross-linked CD28-CD80 complex was released from the SPA beads by washing twice with 100 µl of 2% (v/v) acetic acid for 10 min. Subsequent tryptic digestion, matrix assisted laser desorption/ionisation-time of flight (MALDI-TOF) mass spectrometric, and data analyses were conducted as described (17).

CD28 homology modeling

The sequence of human CD28 (CD28_HUMAN) was retrieved from the SWISS-PROT database (31). The crystal structure of CD152, complexed with human CD86 in Protein Data Bank code 1I85 (9), was used as template for the homology model of human CD28. The sequences of human CD152 and human CD28 were automatically aligned and subsequently manually adjusted. The MODELER/InsightII version 2000 (32) was used for construction and evaluation of the human CD28 homology model and no misfolded areas were revealed.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Assessing the effect of receptor clustering on CD80-CD28 binding

Earlier studies using cell-free systems have shown that the interaction of soluble CD80 with its receptor CD28 is of low affinity (14, 15) and significantly weaker than in cell-based settings. This finding indicated that receptor clustering and avidity could be important for strengthening the interaction of these costimulatory cell surface receptors. To address the importance of avidity in more detail for the CD28-CD80 interaction, the homogeneous SPA technology was adopted. SPA is based on distance-related energy transfer to discriminate between bound and unbound ligand (33) and it is possible to configure the assay in both a cell-based (i.e., using metabolically labeled cells) and cell-free format thereby allowing a direct comparison of the two situations. In the cell-based format (Fig. 1) a dose-dependent binding with similar binding characteristics was obtained with a Bmax/2 of ~5 nM irrespective of whether CD28 or CD80 was presented on the cell surface. This is significantly stronger than observed in e.g., BIAcore studies in which the affinity was in the low micromolar range (14, 15). Furthermore, the pivotal role of receptor/ligand clustering and dynamics was strongly supported as fixation of cells before SPA analysis remarkably reduced binding to both CD28-CHO and CD80-CHO cells without affecting expression or integrity of CD28 or CD80 on the cell surface (data not shown).



View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 1. Interaction of CD28 and CD80 in a cell-based SPA. Binding of CD80Fab (A) and CD28Fc (B) coated SPA beads to [3H]labeled CHO-CD28 ({diamondsuit}) and CHO (•) cells and CHO-CD80-DR4 ({diamondsuit}) and CHO-DR4 (•) transfectants, respectively. Specific binding ({triangleup}) was obtained after subtraction of CD80Fab and CD28Fc binding to CHO and CHO-DR4 cells from that of the corresponding CD28 and CD80 transfectants. Bmax/2 of CD80Fab (A) and CD28Fc (B) was 3.2 and 6.2 nM, respectively.

 
Structural delineation of interaction surfaces

From the binding experiments previously described it appears that multivalency and clustering are important for the CD80-CD28 interaction and we therefore wanted to establish a cell-free experimental system that would mimic the situation on the cell surface and simultaneously allow the molecular analysis of the CD80-CD28 interaction. For this purpose a CD80-Fab fusion protein was captured on SPA beads through its Fab moiety and subsequently its binding to increasing concentrations of soluble radiolabeled CD28 was monitored. Although this setting cannot simulate the dynamics of ligand clustering in a membrane, it should however mimic clustering conditions via a spherical and high-density presentation of CD80. Tritiated CD28 was found to bind to CD80-coated beads in a specific and dose-dependent manner (Fig. 2) and, interestingly, only with a moderately diminished affinity than when CD80 was presented on CHO cells (Bmax/2 for CD80-coated beads and CD80 on CHO cells {approx} 13 and 6 nM, respectively). Thus, as on cell transfectants, multivalent presentation of CD80 on beads significantly increases the strength of its interaction with CD28 and suggests that the CD80-CD28 interaction is avidity driven. For the subsequent protein-protein interaction studies we designed the strategy to cross-link the CD80-CD28 complex under these pseudo-clustering conditions and subsequently dissect the interaction surface in structural terms applying our previously developed mass spectrometry-based protein footprinting method for direct assessment of protein-protein interaction surfaces in solution (17). Using this approach, possible alterations in mass of trypsin-digested peptide fragments derived from a preformed and subsequently chemically cross-linked CD80-CD28 complex are detected. Hence, this analysis allows delineation of the boundaries of the CD80-CD28 interaction and subsequently the deduction of the specific interactions. Picomole quantities of CD80 and CD28 were immobilized on beads as previously described and cross-linked by adding the DTSSP linker. This linker has two active esters directed against lysine-{epsilon}-amino groups and has a net spacer length after incorporation into a protein complex of ~12 Å or a total length of 25 Å including the side chains. All reactions were monitored by nonreducing SDS-PAGE (data not shown). The 200-kDa CD80-CD28 heterodimeric band (~100 ng of protein) obtained after DTSSP cross-linking of the receptors, estimated to represent the combined masses of the glycosylated CD28 and CD80 fusion proteins, ~110 and ~90 kDa, respectively (17), was excised and digested in situ with trypsin.



View larger version (16K):
[in this window]
[in a new window]
  		FIGURE 2. Specific binding of CD80 to CD28 in a cell-free SPA. CD80Fab captured on anti-mouse {kappa}-chain SPA beads was allowed to bind to soluble [3H]CD28Fc at three concentrations: 25 ({blacktriangleup}), 50 ({square}), and 100 ({diamondsuit}) nM. Specific binding was obtained by subtracting [3H]CD28Fc bound to beads coated with the corresponding Fab instead of CD80Fab. The Bmax/2 value is ~13 nM for all three concentrations.

 
The chemical modification status of lysines in the isolated receptors and in the CD80-CD28 complex provides valuable information on putative interaction surfaces. In Fig. 3 the data obtained from the CD80-CD28 peptide mapping studies are schematically outlined. The most remarkable result is that only one peptide signal at a mass to charge ratio (m/z) 2383.9 (expected m/z 2383.2) was obtained from the digest of the CD80-CD28 200 kDa band that corresponds in mass to an intermolecular cross-link formed between peptide 110–120 from CD28 and peptide 87–94 from CD80 (Fig. 4A). Following DTT reduction the ion at m/z 2383.9 was no longer observed (Fig. 4B). Inspection of the spectrum at lower mass range, in which the thiol-cleaved peptides should have been observed (data not shown), however, failed to reveal the presence of peptide signals corresponding to the theoretical masses (m/z 1241.7 and 1144.6). This may be due to low ionization efficiency of the peptides or signal suppression effects. Nonetheless, the disappearance of a putatively assigned cross-link after thiol treatment is highly indicative that the peptides were linked by the DTSSP cross-linker. The analysis of the amino acid sequences of the peptides revealed an apparent dual ambiguity because two possible attachment sites were apparent in each peptide, but only one cross-link was demonstrated. Hence, either K118 or K120 from CD28 may specifically cross-link with K89 or K93 from CD80. This ambiguity may be partially eliminated as the C-terminal residue (K120) in peptide 110–120 from CD28 would most likely have been blocked from cleavage by trypsin if it had been cross-linked. Residues that are surface-labeled but not cross-linked (K6, K34, K36, K37, K54, K86, and K201 in CD80, and K2 and K39 in CD28) are not expected to be involved in an intermolecular surface.



View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 3. Linear schematic representation of the cross-linked CD28-CD80 complex. The only intermolecular cross-link between CD28 and CD80 is shown in its nonreduced form. The modification status of lysine residues are displayed with their respective amino acid sequence number and according to the following color codes: cross-linked lysines (red); lysines modified by the cross-linking reagent in free CD28 or CD80, but not in the complex (magenta); and surface-labeled, not cross-linked lysines (green). The Ig-like V-domains (dark gray), Ig-like C2 domain (light gray), hinge region (dotted area), and stalk regions (hatched area) are also shown together with the position of the MYPPPY loop (black area) and potential Asn glycosylation sites (orange area).

 


View larger version (21K):
[in this window]
[in a new window]
  		FIGURE 4. MALDI-TOF mass spectrometry spectrum before (A) and after (B) reductive cleavage of the disulphide bridge in the DTSSP cross-linker. The only intermolecular cross-link formed between peptide 110–120 in CD28 and peptide 87–94 in CD80 (m/z 2383.9) and a cross-link between the Ig fusion domains in CD28Fc (peptide 233–237) and CD80Fab (peptide 212–250) with a m/z of 4973.1 (open arrow) (A). Note that neither peptide is observed after DTT reduction (B). Peptides derived from CD28Fc (A) and CD80Fab (B) are labeled. Peptide plus a reduced or hydrolyzed DTSSP moiety is indicated with ({blacktriangledown}) or (•).

 
In addition, a few interactions between the Ig fusion partners were observed (see spectrum in Fig. 4). Compared with our previous study based on in-solution interactions (17), no cross-links were seen between CD28 and the Ig moiety of CD80Fab and the total number of receptor-Ig domain interactions was reduced from 10 to 4 by capturing and preorientating the receptors on the beads via the fusion partners (data not shown). As SPA analyses indicated strong and highly specific binding between CD80 and CD28, the latter interactions are therefore ascribed to being weak nonspecific interactions originating from molecules detaching from the beads.

The exclusive finding of only one cross-link between CD80 and CD28 (Fig. 3) suggests that this interaction is mediated through a well-defined interface. Therefore, lysines that are unmodified in the complex, although being positioned close in three-dimensional space to the cross-linked residues on CD80 and CD28 but surface-accessible and modified on the individual soluble receptors, were identified as the most likely residues involved in an interaction surface. As is shown in Fig. 5, the only lysine in CD80 that may interact with CD28 is the residue in peptide 87–94 that is not cross-linked because the only remaining lysine candidates, K9 and K105, are too distant from the putative interaction interface. Applying the same approach to a homology model of CD28 leads to the identification of K6, K63, K71, K95, and K109 as the nearest structurally neighboring nonmodified lysines that have not been cross-linked (Fig. 5). These residues map to the same central region but are situated on opposite sides of CD28 with K6, K95, and K109 on one side and K63 and K71 on the other. The accessibility of K6, however, would be limited due to glycosylation of the nearby N111 (the corresponding residue in CD152, N110, is known to be glycosylated) (5), which makes it unlikely that this residue is part of the interaction surface. As to K120 in the only cross-linked peptide from CD28, this lysine was assumed not to be the cross-linking residue because it would block cleavage by trypsin at this position. Even if this assumption should turn out to be incorrect, K120 is still not a likely contact residue due to its distant localization in the stalk region. Finally, residues expected to be surface accessible but nevertheless do not become modified in either the isolated receptor or in the CD80-CD28 complex (K169 and K185 in CD80 and K23 in CD28) are all situated in close vicinity of N-linked glycosylation sites, which may protect them from surface labeling.



View larger version (47K):
[in this window]
[in a new window]
  		FIGURE 5. Structural mapping of cross-linking data on CD80 and CD28. A ribbon representation of human CD80 (Protein Data Bank code 1DR9) and the homology model of human CD28 with the {beta}-strands in CD28 labeled by yellow letters. The modification status of lysine residues is displayed by color-coding as described in Fig. 3. Also shown are acidic residues near the hot spot (cyan) and Asn glycosylation sites (orange).

 
Residues that are unmodified in the complex but apparently distant from the CD80-CD28 intermolecular interface may indicate additional interaction surfaces that exist in the complex (see Fig. 3 for a linear representation of these residues). This is the case for K9 and K105 in CD80, which map to the described CD80 dimerization surface (K9) and the interdomain region (K105) (7) and is consistent with previous findings (20) that CD80 interacts with CD28 in its homodimeric state (Fig. 5). Interestingly, K63 and K71 in CD28 are both unmodified upon ligation with CD80 and may indicate a putative oligomerization interface consisting of residues on the B, E, and D {beta}-strands (Fig. 5), which is induced when CD28 is bound to CD80.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
In the absence of a three-dimensional structure of CD28, the present study provides the first experimental non-mutagenesis- based study of the CD28-CD80 interaction. By combining cross-linking of the CD80-CD28 complex under pseudo-clustering conditions and structural mapping using differential mass spectrometry, only one single intermolecular cross-link was observed indicating that the CD80-CD28 interaction is governed by a well-defined interaction surface. It is worth noticing, however, that failure to demonstrate cross-links of surface-accessible residues may reflect impaired reaction rates due to local environmental factors and therefore absence of modification should be interpreted with care and preferably be confirmed using different cross-linking agents (18). Because the cross-linking agents used (17) produced about the same reaction pattern as DTSSP when tested under pseudo-clustering conditions (data not shown), these residues may nevertheless give reliable information on putative interaction interfaces in CD80 and CD28. The nearest potentially reactive residues that were not modified, K89 or K93 on CD80 and K95 and K109 on CD28, were identified as likely key residues in this interaction interface. Due to the ambiguity whether K89 or K93 is involved in the cross-linking we can only speculate which of these residues actually interacts with CD28. Both lysines are located on the very long FG loop, which is conserved among all B7 molecules (7). Interestingly, this region is situated close to the hydrophobic contact area observed in the CD80-CD152 complex (8). The corresponding key residues on CD28, K95 and K109, are situated on both sides of the conformationally constrained and conserved MYPPPY motif in the complementarity determining region 3 loop. In contrast to the highly conserved N-terminal extended region to the MYPPPY loop, the C-terminal region is one of the regions in CD28 that are most distinctive compared with CD152 and has previously been found to be essential for binding of both CD80 and CD86 (11). Notably, this region is not only one residue longer but also very hydrophilic compared with the corresponding region in CD152 in which hydrophobic residues dominate (5). In fact, database surveys have shown that the majority of protein-protein interactions can be described in terms of such "hot-spots" (Refs. 16 , 34 and references therein). These are typically characterized by either a highly uneven distribution of amino acids with a preference for amino acids that can make multiple types of favorable interactions such as tryptophan, arginine and tyrosine (34) or hydrophobic amino acid residues that are flanked with hydrophilic residues providing structural epitopes and charge complementarity (16). Accordingly, the lysine residues we have deduced for the CD80-CD28 interaction would be situated in the flanking region of the core of the interaction site. This theory is confirmed by the presence of additional acidic and exposed residues observed in the vicinity, namely E81, E88, D90 and E99 in CD80 and E32, E46, E97 and D106 in CD28 (Fig. 5). One or more of these acidic residues may be potential interaction partners for the identified lysines on the corresponding interacting protein. In the crystal structure of CD80-CD152 it was revealed that this interaction is characterized by an unusual high degree of shape complementarity with the hydrophobic residues of the MYPPPY region at the core of the interaction (8). Although it is known from earlier mutagenesis studies that this motif is essential for B7 binding, alanine mutants of this motif actually retain binding to CD80 (but not CD86) at high concentrations (10). Thus, this indicates that the interaction site is composite and specificity is mediated by additional determinants. Based on the data presented in this study we propose that the C-terminal extended region of the MYPPPY loop (in which K109 is localized) guides CD80 specificity toward either CD28 or CD152 with the former being dominated by hydrophilic interactions and high charge complementarity and the latter by hydrophobic interactions and high surface complementarity. This would also explain the 10- to 100-fold higher affinity of the CD80-CD152 interaction (13, 14, 15).

In addition to the identification of a well-defined epitope in the CD80-CD28 interaction, this combined approach identifies residues that upon complex formation become resistant to modification by chemical cross-linkers although being located distant from the presumed primary interaction surface. It has been shown that the V-domain of CD80 is involved in CD80 dimerization both in solution and in crystals (7). The significance of the V-domain for dimerization of CD80 was confirmed by our data as K9, which is located close to the structurally determined dimerization interface (7), is unmodified even when CD80 is not complexed with CD28. Another residue that is not modified also in the absence of CD28 ligation is K105. This lysine is located in the adjacent hinge region between the V- and C-domains of CD80 (Fig. 5) and has been suggested to play a role in stabilization of the two domains (35, 36). The side chain of K105 is partly buried in the dimerization interdomain and may explain why it does not react with the DTSSP cross-linker. For CD28 K63 and K71 may define a putative oligomerization surface and it is intriguing that these residues, which become resistant to chemical modification upon complex formation with CD80, have been shown to map to the region recently identified as a "superagonistic site" for Ab-mediated activation of CD28 (37, 38). As these lysines are accessible on the isolated CD28 receptor it follows that this oligomerization is B7 induced. A B7-dependent aggregation of CD28 on the T cell has previously been shown (39), in which the evenly dispersed CD28 on activated T cells is redistributed following CD80/CD86 engagement. Hence, depending on the degree of B7-induced clustering of CD28 on the T cell surface, qualitatively different signals through the CD28 receptor may be delivered to the T cell. In an attempt to present a working model for the proposed B7-induced oligomerization of CD28, a schematic representation of multimeric CD80-CD28 complexes is depicted in Fig. 6, in which the two arms of homodimeric CD28 are bent aside and the oligomeric interface is correctly positioned for oligomerization when CD28 binds to CD80 dimers clustered on the APC surface. As the amino acids mediating dimerization in CD152 are not conserved in CD28, further studies are however needed to verify whether CD28 like CD152 actually forms oligomers in complex with the B7 ligands.



View larger version (27K):
[in this window]
[in a new window]
  		FIGURE 6. Schematic view of the supermolecular activation cluster resulting from CD80-CD28 interaction. Binding of preformed clusters of CD80 dimers on the APC bind to CD28 homodimers dispersed on the T cell surface is depicted (A). This induces a dynamic change that positions an oligomerization interface in CD28 leading to formation of multimeric CD80-CD28 complexes (B). Symbols are color-coded as follow: K89/K93 in CD80 and K118 in CD28 (red); K95 and K109 in CD28 (magenta); K63 and K71 in CD28 (green); and residues in the MYPPPY loop (black). The putative CD80(K93)-CD28(K118) cross-link and the disulphide bond in CD28 are shown as a red and gray line, respectively.

 
In conclusion, this study has underscored the importance of two important facets when studying interactions by costimulatory receptors and ligands in a functional context: clustering/multivalency and specificity obtained through well-defined hot-spot interaction surfaces. Our data, together with the crystal structures for B7-CD152 complexes, would support a model of how specificity is achieved in the interaction of CD28 with the B7 ligands being dominated by dynamic, weaker, hydrophilic interactions, whereas the CD152 interactions are tighter and driven by hydrophobic interactions. The identified contact area in this study may guide efforts for therapeutic intervention targeted at this pathway. In support of this, we (with International patent application WO 03/004495 and the granted US Patent serial number 6,642,249), by applying the experimental settings used in this study, and others (40) have described small molecules of non-peptide nature that inhibit the binding of human CD80 to human CD28 in a potent and highly specific manner. This motivates further studies to find out whether these compounds block the CD80-CD28 interaction by mapping to the interface identified. The general applicability of our approach for direct identification of dynamic interactions of even larger protein complexes, although formed in solution, has recently been supported (18). Hence, protein foot-printing approaches like the one used in this study should be considered for protein complexes that turn out refractory to structural determination by traditional means.


   Acknowledgments
 
We thank Simon Davis for fruitful discussions throughout this work. We are indebted to Helena Arozenius for skilful help with the SPA analyses.


   Footnotes
 
1 Current address: Micromet AG, Staffelseestrasse 2, 81477 Munich, Germany. Back

2 Current address: Nestlé Research Center, Vers-chez-les-Blanc, PO Box 44, 1000 Lausanne 26, Switzerland. Back

3 Current address: MDS Proteomics, Denmark, Stærmosegaardsvej 6, DK-5230 Odense M, Denmark. Back

4 Address correspondence and reprint requests to Dr. Per Björk, Active Biotech Research, Box 724, SE-220 07 Lund, Sweden. E-mail address: per.bjork{at}activebiotech.com Back

5 Abbreviations used in this paper: CHO, Chinese hamster ovary; DTSSP, 3,3'-dithiobis(sulfosuccinimidyl-propionate); SPA, scintillation proximity assay. Back

Received for publication November 14, 2003. Accepted for publication March 19, 2004.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Greenfield, E. A., K. A. Nguyen, V. K. Kuchroo. 1998. CD28/B7 costimulation: a review. Crit. Rev. Immunol. 18:389.[Medline]
  2. Zhang, Y., R. J. Joost van Neerven, S. W. Van Gool, L. Coorevits, M. de Boer, J. J. Ceuppens. 1997. B7-CD28 interaction is a late acting co-stimulatory signal for human T cell responses. Int. Immunol. 9:1095.[Abstract/Free Full Text]
  3. Salomon, B., J. A. Bluestone. 2001. Complexities of CD28/B7: CTLA-4 costimulatory pathways in autoimmunity and transplantation. Annu. Rev. Immunol. 19:225.[Medline]
  4. Leach, D. R., M. F. Krummel, J. P. Allison. 1996. Enhancement of antitumor immunity by CTLA-4 blockade. Science 271:1734.[Abstract]
  5. Metzler, W. J., J. Bajorath, W. Fenderson, S.-Y. Shaw, K. L. Constantine, J. Naemura, G. Leytze, R. J. Peach, T. B. Lavoie, L. Mueller, P. Linsley. 1997. Solution structure of human CTLA-4 and delineation of a CD80/CD86 binding site conserved in CD28. Nat. Struct. Biol. 4:527.[Medline]
  6. Ostrov, D. A., W. Shi, J. C. Schwartz, S. C. Almo, S. G. Nathenson. 2000. Structure of murine CTLA-4 and its role in modulating T cell responsiveness. Science 290:816.[Abstract/Free Full Text]
  7. Ikemizu, S., R. J. C. Gilbert, J. A. Fennely, A. V. Collins, K. Harlos, E. Y. Jones, D. I. Stuart, S. J. Davis. 2000. Structure and dimerization of a soluble form of B7-1. Immunity 12:51.[Medline]
  8. Stamper, C. C., Y. Zhang, J. F. Tobin, D. V. Erbe, S. Ikemizu, S. J. Davis, M. L. Stahl, J. Seehra, W. S. Somers, L. Mosyak. 2001. Crystal structure of the B7-1/CTLA-4 complex that inhibits human immune responses. Nature 410:608.[Medline]
  9. Schwartz, J. C., X. Zhang, A. A. Fedorov, S. G. Nathenson, S. C. Almo. 2001. Structural basis for co-stimulation by the human CTLA-4/B7-2 complex. Nature 410:604.[Medline]
  10. Morton, P. A., X.-T. Fu, J. A. Steward, K. S. Giacoletto, S. L. White, C. E. Leysath, R. J. Evans, J.-J. Shieh, R. W. Karr. 1996. Differential effects of CTLA-4 substitutions on the binding of human CD80 (B7-1) and CD86 (B7-2). J. Immunol. 156:1047.[Abstract]
  11. Peach, R. J., J. Bajorath, W. Brady, G. Leytze, J. Greene, J. Naemura, P. S. Linsley. 1994. Complementarity determining region 1 (CDR1)- and CDR3-analogous regions in CTLA-4 and CD28 determine the binding to B7-1. J. Exp. Med. 180:2049.[Abstract/Free Full Text]
  12. Kariv, I., A. Truneh, R. W. Sweet. 1996. Analysis of the site of interaction of CD28 with its counter-receptors CD80 and CD86 and correlation with function. J. Immunol. 157:29.[Abstract]
  13. Greene, J. L., G. M. Leytze, J. Emswiler, R. Peach, J. Bajorath, W. Cosand, P. S. Linsley. 1996. Covalent dimerization of CD28/CTLA-4 and oligomerization of CD80/CD86 regulate T cell costimulatory interactions. J. Biol. Chem. 271:26762.[Abstract/Free Full Text]
  14. van der Merwe, P. A., D. L. Bodian, S. Daenke, P. Linsley, S. J. Davis. 1997. CD80 (B7-1) binds both CD28 and CTLA-4 with a low affinity and very fast kinetics. J. Exp. Med. 185:393.[Abstract/Free Full Text]
  15. Collins, A. V., D. W. Brodie, R. J. C. Gilbert, A. Iaboni, R. Manso-Sancho, B. Walse, D. I. Stuart, P. A. van der Merwe, S. J. Davis. 2002. The interaction properties of costimulatory molecules revisited. Immunity 17:201.[Medline]
  16. DeLano, W. L., M. H. Ultsch, A. M. de Vos, J. A. Wells. 2000. Convergent solutions to binding at a protein-protein interface. Science 287:1279.[Abstract/Free Full Text]
  17. Bennett, K. L., M. Kussmann, P. Björk, M. Godzwon, M. Mikkelsen, P. Sørensen, P. Roepstorff. 2000. Chemical cross-linking with thiol-cleavable reagents combined with differential mass spectrometric peptide mapping: a novel approach to assess intermolecular protein contacts. Protein Sci. 9:1503.[Medline]
  18. Trester-Zedlitz, M., K. Kamada, S. K. Burley, D. Fenyö, B. T. Chait, T. W. Muir. 2003. A modular cross-linking approach for exploring protein interactions. J. Am. Chem. Soc. 125:2416.[Medline]
  19. Davidson, W. S., G. M. Hilliard. 2003. The spatial organization of apolipoprotein A-I on the edge of discoidal high density lipoprotein particles. J. Biol. Chem. 278:27199.[Abstract/Free Full Text]
  20. van der Merwe, P. A., S. J. Davis. 2003. Molecular interactions mediating T cell antigen recognition. Annu. Rev. Immunol. 21:659.[Medline]
  21. Symington, F. W., W. Brady, P. S. Linsley. 1993. Expression and function of B7 on human epidermal Langerhans cells. J. Immunol. 140:1286.
  22. Linsley, P. S., W. Brady, L. S. Grosmaire, A. Aruffo, N. K. Damle, J. A. Ledbetter. 1991. Binding of the B cell activation antigen B7 to CD28 costimulates T cell proliferation and interleukin 2 mRNA accumulation. J. Exp. Med. 173:721.[Abstract/Free Full Text]
  23. Damle, N. K., L. V. Doyle, L. S. Grosmaire, J. A. Ledbetter. 1988. Differential regulatory signals delivered by antibody binding to the CD28 (Tp44) molecule during the activation of human T lymphocytes. J. Immunol. 140:1753.[Abstract/Free Full Text]
  24. Doty, R. T., E. A. Clark. 1998. Two regions in the CD80 cytoplasmic tail regulate CD80 redistribution and T cell costimulation. J. Immunol. 140:2700.
  25. Linsley, P. S., J. Bradshaw, J. Greene, R. Peach, K. L. Bennett, R. S. Mittler. 1996. Intracellular trafficking of CTLA-4 and focal localization towards sites of TCR engagement. Immunity 4:535.[Medline]
  26. Turley, S. J., K. Inaba, W. S. Garrett, M. Ebersold, J. Unternaehrer, R. M. Steinman, I. Mellman. 2000. Transport of peptide-MHC class II complexes in developing dendritic cells. Science 288:522.[Abstract/Free Full Text]
  27. Dustin, M. L.. 2002. The immunological synapse. Arthritis Res. 4:(Suppl. 3):S119.
  28. Olsson, C., E. Michaëlsson, E. Parra, U. Pettersson, P. Lando, M. Dohlsten. 1998. Biased dependency of CD80 versus CD86 in the induction of transcription factors regulating the human IL-2 promoter. Int. Immunol. 10:499.[Abstract/Free Full Text]
  29. Dohlsten, M., L. Abrahmsén, P. Björk, P. A. Lando, G. Hedlund, G. Forsberg, T. Brodin, N. R. Gascoigne, C. Förberg, P. Lind, T. Kalland. 1994. Monoclonal antibody-superantigen fusion proteins: tumor-specific agents for T-cell-based tumor therapy. Proc. Natl. Acad. Sci. USA 91:8945.[Abstract/Free Full Text]
  30. Cavallin, A., H. Arozenius, K. Kristensson, P. Antonsson, D. E. Otzen, P. Björk, G. Forsberg. 2000. The spectral and thermodynamic properties of staphylococcal enterotoxin A, E, and variants suggest that structural modifications are important to control their function. J. Biol. Chem. 275:1665.[Abstract/Free Full Text]
  31. Bairoch, A., R. Apweiler. 1997. The SWISS-PROT protein sequence data bank and its supplement TrEMBL. Nucleic Acids Res. 25:31.[Abstract/Free Full Text]
  32. Sali, A., T. L. Blundell. 1990. Definition of general topological equivalence in protein structures: a procedure involving comparison of properties and relationships through simulated annealing and dynamic programming. J. Mol. Biol. 212:403.[Medline]
  33. Hart, H. E., E. B. Greenwald. 1979. Scintillation proximity assay (SPA): a new method of immunoassay. Direct and inhibition mode detection with human albumin and rabbit antihuman albumin. Mol. Immunol. 16:265.[Medline]
  34. Bogan, A. A., K. S. Thorn. 1998. Anatomy of hot spots in protein interfaces. J. Mol. Biol. 280:1.[Medline]
  35. Ellis, J. H., M. N. Burden, D. V. Vinogradov, C. Linge, J. S. Crowe. 1996. Interactions of CD80 and CD86 with CD28 and CTLA4. J. Immunol. 156:2700.[Abstract]
  36. Guo, Y., Y. Wu, M. Zhao, X. P. Kong, Y. Liu. 1995. Mutational analysis and an alternatively spliced product of B7 defines its CD28/CTLA4-binding site on immunoglobulin C-like domain. J. Exp. Med. 181:1345.[Abstract/Free Full Text]
  37. Margulies, D. H.. 2003. CD28, costimulator or agonist receptor?. J. Exp. Med. 197:949.[Free Full Text]
  38. Lühder, F., Y. Huang, K. M. Dennehy, C. Guntermann, I. Müller, E. Winkler, T. Kerkau, S. Ikemizu, S. J. Davis, T. Hanke, T. Hünig. 2003. Topological requirements and signaling properties of T cell-activating, anti-CD28 antibody superagonists. J. Exp. Med. 197:955.[Abstract/Free Full Text]
  39. Linsley, P. S., J. Ledbetter, R. Peach, J. Bajorath. 1995. CD28/CTLA-4 receptor structure, binding stoichiometry and aggregation during T-cell activation. Res. Immunol. 146:130.[Medline]
  40. Erbe, D. V., S. Wang, Y. Xing, J. F. Tobin. 2002. Small molecule ligands define a binding site on the immune regulatory protein B7.1. J. Biol. Chem. 277:7363.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 J Biomol ScreenHome page
K. Uvebrant, D. Da Graca Thrige, A. Rosen, M. Akesson, H. Berg, B. Walse, and P. Bjork
Discovery of Selective Small-Molecule CD80 Inhibitors
J Biomol Screen, June 1, 2007; 12(4): 464 - 472.
[Abstract] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sørensen, P.
Right arrow Articles by Björk, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sørensen, P.
Right arrow Articles by Björk, P.
Right arrowPubmed/NCBI databases
*Substance via MeSH

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS