Molecular Analysis of the Epidermal Growth Factor-like Short Consensus Repeat Domain-mediated Protein-Protein Interactions

Abstract

Epidermal growth factor-like (EGF) and short consensus repeat (SCR) domains are commonly found in cell surface and soluble proteins that mediate specific protein-protein recognition events. Unlike the immunoglobulin (Ig) superfamily, very little is known about the general properties of intermolecular interactions encoded by these common modules, and in particular, how specificity of binding is achieved. We have dissected the binding of CD97 (a member of the EGF-TM7 family) to the complement regulator CD55, two cell surface modular proteins that contain EGF and SCR domains, respectively. We demonstrate that the interaction is mediated solely by these domains and is characterized by a low affinity (86 μm) and rapid off-rate (at least 0.6 s−1). The interaction is Ca2+ -dependent but is unaffected by glycosylation of the EGF domains. Using biotinylated multimerized peptides in cell binding assays and surface plasmon resonance, we show that a CD97-related EGF-TM7 molecule (termed EMR2), differing by only three amino acids within the EGF domains, binds CD55 with aKD at least an order of magnitude weaker than that of CD97. These results suggest that low affinity cell-cell interactions may be a general feature of highly expressed cell surface proteins and that specificity of SCR-EGF binding can be finely tuned by a small number of amino acid changes on the EGF module surface. Epidermal growth factor-like (EGF) and short consensus repeat (SCR) domains are commonly found in cell surface and soluble proteins that mediate specific protein-protein recognition events. Unlike the immunoglobulin (Ig) superfamily, very little is known about the general properties of intermolecular interactions encoded by these common modules, and in particular, how specificity of binding is achieved. We have dissected the binding of CD97 (a member of the EGF-TM7 family) to the complement regulator CD55, two cell surface modular proteins that contain EGF and SCR domains, respectively. We demonstrate that the interaction is mediated solely by these domains and is characterized by a low affinity (86 μm) and rapid off-rate (at least 0.6 s−1). The interaction is Ca2+ -dependent but is unaffected by glycosylation of the EGF domains. Using biotinylated multimerized peptides in cell binding assays and surface plasmon resonance, we show that a CD97-related EGF-TM7 molecule (termed EMR2), differing by only three amino acids within the EGF domains, binds CD55 with aKD at least an order of magnitude weaker than that of CD97. These results suggest that low affinity cell-cell interactions may be a general feature of highly expressed cell surface proteins and that specificity of SCR-EGF binding can be finely tuned by a small number of amino acid changes on the EGF module surface. epidermal growth factor short consensus repeat immunoglobulin epidermal growth factor module-containing seven transmembrane receptor regulators of complement activation calcium-binding EGF motif epidermal growth factor module-containing mucin-like receptor 2 surface plasmon resonance fluorescence-activated cell sorting response unit monoclonal antibody polymerase chain reaction high performance liquid chromatography Most cell surface proteins are highly modular in organization and are constructed from different combinations of a limited set of structural domains. In recent years high resolution structures of many of the most common domains have become available, and analysis of the kinetic characteristics of the interactions they mediate is increasing our understanding of processes that underlie the most basic cellular interactions. In particular, cell adhesion interactions mediated by proteins of the Ig superfamily are characterized by multiple weak binding events, each with a low affinity, but with the multivalent nature of the cell surface proteins resulting in a cell-cell interaction of high avidity (1Davis S.J. Ikemizu S. Wild M.K. van der Merwe P.A. Immunol. Rev. 1998; 163: 217-236Crossref PubMed Scopus (119) Google Scholar). Two of the most common structural modules used in cell surface proteins are the epidermal growth factor-like (EGF)1 and short consensus repeat (SCR) domains (Fig. 1). Recent analysis of the human genome (as of March 1, 2001) shows that the EGF family is the fifth most common protein family with 3% of all potential proteins containing EGF domains whereas SCR domains are found in 0.3% of all proteins. Both of these domain types are commonly used to mediate protein-protein interactions. The EGF module, which often occurs as multiple tandem repeats, is widely distributed among extracellular proteins involved in adhesion, receptor-ligand interactions, extracellular matrix structure, determination of cell fate, and blood coagulation (see Fig. 1). A subset of EGF domains contains a consensus sequence associated with calcium binding (cb): (D/N)X(D/N)(E/Q)Xm(D/N)*Xn(Y/F), where m and n are variable and the asterisk indicates possible ;-hydroxylation (2Handford P.A. Mayhew M. Baron M. Winship P.R. Campbell I.D. Brownlee G.G. Nature. 1991; 351: 164-167Crossref PubMed Scopus (246) Google Scholar, 3Mayhew M. Handford P. Baron M. Tse A.G. Campbell I.D. Brownlee G.G. Protein Eng. 1992; 5: 489-494Crossref PubMed Scopus (48) Google Scholar, 4Rees D.J. Jones I.M. Handford P.A. Walter S.J. Esnouf M.P. Smith K.J. Brownlee G.G. EMBO J. 1988; 7: 2053-2061Crossref PubMed Scopus (186) Google Scholar). Calcium is thought to perform a key role in the orientation of cbEGF pairs by restricting conformational flexibility of interdomain linkages (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar, 6Werner J.M. Knott V. Handford P.A. Campbell I.D. Downing A.K. J. Mol. Biol. 2000; 296: 1065-1078Crossref PubMed Scopus (61) Google Scholar) resulting in tandem EGF repeats that are highly resistant to proteolysis (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar, 7Reinhardt D.P. Ono R.N. Sakai L.Y. J. Biol. Chem. 1997; 272: 1231-1236Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar, 8Reinhardt D.P. Ono R.N. Notbohm H. Muller P.K. Bachinger H.P. Sakai L.Y. J. Biol. Chem. 2000; 275: 12339-12345Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar). SCR domains are frequently found among proteins of the complement system with many of the complement regulatory proteins (e.g. CD55 and CD46, see below) and complement receptors (e.g. CD21 and CD35) consisting entirely of repeated SCR domains. Structural studies (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar,9Casasnovas J.M. Larvie M. Stehle T. EMBO J. 1999; 18: 2911-2922Crossref PubMed Scopus (134) Google Scholar, 10Rao Z. Handford P. Mayhew M. Knott V. Brownlee G.G. Stuart D. Cell. 1995; 82: 131-141Abstract Full Text PDF PubMed Scopus (312) Google Scholar) reveal that both of these small modules (<60 amino acids) fold to form all β-strand domains strengthened by disulfide bonds (two in the case of SCRs and three for EGFs). An understanding of why these domains are so well suited for protein interactions has been hindered by the fact that the disulfide-rich multidomain constructs required for structural analyses are difficult to produce in large yields in the native form. We therefore have very few structures of complexes involving these common modules and no structures of EGF-SCR complexes or kinetic characterization of the interaction. CD55 (or Decay Accelerating Factor; DAF) is a member of the regulators of complement activation (RCA) family. It protects host cells from complement system attack by binding to C3b and C4b preventing formation of the membrane attack complex. It has a widespread tissue distribution and is expressed at high levels on many different cell types. The protein consists of an N-terminal extracellular portion of four SCR domains linked via a heavily O-glycosylated spacer to a C-terminal glycosylphosphatidylinositol anchor (Fig.1). SCR domains 2–4 are involved in regulation of complement and also in binding to a variety of viral and bacterial pathogens. The most N-terminal SCR (domain 1) also provides the site of interaction for some viruses but until recently the native role of domain 1 was unknown. However, identification of CD97 as a cellular ligand for the N-terminal domains of CD55 J. van van J. 1996; PubMed Scopus Google Scholar, J. van J. Immunol. 1998; PubMed Scopus Google Scholar) has a associated with portion of the CD97 is a member of the EGF-TM7 characterized by the in which tandem EGF repeats are to a receptor via a mucin-like S. Immunol. 1996; Full Text PDF PubMed Scopus Google Scholar, S. J. Biol. 1998; PubMed Scopus Google Scholar). CD97 is expressed on and and is on and It is known to in a variety of containing different of the EGF domains M. J. Immunol. 1996; Google Scholar, J. D. J.M. M. van J. Immunol. 1995; Google Scholar) each of which binds CD55 with different The CD55 binding with the affinity three EGF domains 1, and two of which 2 and are to calcium (Fig. 1). the role of the interaction is the structure, the of and the of both CD97 and CD55 in J. van 1999; PubMed Scopus Google Scholar) suggest possible for the interaction in adhesion and within the and a EGF-TM7 highly EGF domains with CD97 but to show an interaction with CD55 in assays was M. J. S. 2000; PubMed Scopus Google Scholar). In we the and properties of the CD55 and CD97 interaction and the for CD55 We show that the interaction is mediated solely by the EGF domains of CD97 and is characterized by a low affinity and rapid Ca2+ is for the formation of the CD55 binding on but glycosylation of EGF domains from CD97 is The three amino acid within the EGF domains of that from CD97 the affinity for CD55 by at least an order of The of these for general properties of cell surface interactions and specificity of EGF domain interactions are and from and from CD97 monoclonal and by of of The for CD97 was a of of The three extracellular EGF domains of the CD55 binding of CD97 and of expressed a system used for the of EGF domains 1, and from CD97 and and used for the are The also contains a site and a acid spacer sequence to the sequence of CD97 and affinity peptides and to a well in (2Handford P.A. Mayhew M. Baron M. Winship P.R. Campbell I.D. Brownlee G.G. Nature. 1991; 351: 164-167Crossref PubMed Scopus (246) Google Scholar). The that the multidomain constructs the native fold was but Both peptides the in for all cbEGF constructs on (e.g. the for is in Fig. calcium binding is a of the native fold of cbEGF domains, of these domains in CD97 and was by the proteolysis and the Ca2+ of interaction (see CD97 and and at the The of protein was the at from the amino acid on the Full Text PDF PubMed Scopus Google Scholar). The of the four SCR domains of CD55 as a soluble in the has been in T. D.J. J. 1997; PubMed Google Scholar). The protein was to and to a The of protein was the at Full Text PDF PubMed Scopus Google plasmon resonance on a CD55 was to the matrix on the surface of via the as T. D.J. D. Stuart van der Merwe P.A. J. Biol. Chem. 1998; Full Text Full Text PDF PubMed Scopus Google Scholar) with the the activation CD55 was at in for in the of response of CD55 to the surface. CD97 was to the surface via to the by by of the the surface at a of at and all for changes by of a from a all in a the constructed in The and CD97 containing and three domains, CD97 and CD97 have been M. J. S. 2000; PubMed Scopus Google Scholar). The constructs for the by the of the domain of or CD97 to that of the of CD97 or which was in with the of the and CD97 and by the domains with the of The domains of and CD97 by and the of and CD97 the with and with respectively. in the to the of and CD97 the domains of and CD97 as respectively. was to the by the The resulting and with the of CD97 in by the of the of soluble biotinylated the the consensus by the van der Merwe P.A. V. P.A. J. 1998; PubMed Scopus Google was by and and as P.A. 1996; PubMed Scopus Google Scholar). the was of the of in or domains of or CD97 the of the The constructs therefore domains by the a and a with 2 and and cells in a cells in cells in and cells in cells to the cells with of calcium the was with of and for a assays as M. J. S. 2000; PubMed Scopus Google Scholar). cells with the at 1 and for to human blood cells human blood cells in and to cells for at blood cells by The of blood cell adhesion was by the of of blood cells the The of cell surface protein on cells was by analysis CD97 cells with and CD97 as well as the and and which the domain of both and and which the of used for cells with the and CD97 The blood cell binding of cells was by the of by the of cell surface protein by from cells was to a with and with 1 of at was by with containing and The biotinylated proteins and at by analysis basic as binding assays to biotinylated proteins as van der Merwe P.A. V. P.A. J. 1998; PubMed Scopus Google Scholar). In of and to 2 of biotinylated protein in a of The and protein was at for 1 and at for 1 proteins by with and the in of The was to cells in a The was at at for for a at and in of for and to the cells the of the complex. studies have that the EGF domains of CD97 are for interaction J. van J. Immunol. 1998; PubMed Scopus Google Scholar) and that is to with CD55 M. J. S. 2000; PubMed Scopus Google Scholar). CD97 and highly EGF domains but are within the the possible of the to the CD55 binding of and (Fig. a cell (see containing the domains of CD97 and the of are in CD55 binding whereas constructs consisting of the EGF domains of and the of CD97 are no CD55 binding in of the CD97 the of CD97 to CD55 but the binding by We the binding to CD55 on of the CD97 as by in the of the CD97 the EGF domains are to the for from our surface plasmon resonance studies (see below) that show that constructs of CD97 with or the have for soluble used in that different of CD97 have different CD55 binding with the CD97 CD97 the affinity whereas the CD97 shows only of the CD55 binding no binding to CD55 in M. J. S. 2000; PubMed Scopus Google Scholar). plasmon resonance was used to the interaction domains from CD55 and CD97 in Protein constructs consisting of four SCR domains from CD55 and the three EGF domains from CD97 with or the (Fig. 1) expressed and (see for proteins in both to a surface or in The CD97 used in the soluble of the EGF domains and was was from an system (see In the CD97 to the containing the and was was a cell shows an of CD55 for a with of CD97 whereas shows an CD97 for a with CD55 A response is in the of each the the the surface and therefore changes in the of the The response soluble protein is ligand is than the response in the in each of both of that the of the interaction are binding within of the of the and is within of the of the Fig. show a of by a of the soluble protein CD55 or the surface. The are by of the and so that the and of the are The of the levels of protein to the of protein in and for that of the CD55 was to with the soluble CD97 whereas of the CD97 was to with soluble In the case of from the the was from a on which an protein was than a surface. binding of at least of the soluble protein by at least order of and the of a of the to these (Fig. and for the was also from of the results of the order of to high or that both the CD55 and CD97 for the of the the of from different protein expressed in different and the of the to the The of all the a for of 1 The of the in orientation the was or CD97 on the that the of provides a of affinity and is to is the for both the soluble and of glycosylation and the are required for CD55 of from the binding to the of analysis shows that the off-rate is at least 0.6 for by to binding as in Fig. on shows the from of shows the from of in a studies of a cbEGF domain from have that calcium is for of a interdomain (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar). of to a in the properties of which may be by an in the of the EGF to proteolysis Knott V. Handford P.A. Mol. 2000; PubMed Scopus Google Scholar). of the sequence of the soluble of CD97 that two of the three EGF domains 2 and are of the Ca2+ binding the potential role of Ca2+ in the we Ca2+ of the interaction. The of CD97 was CD55 in in the or of (Fig. In the of was to the interaction so that no was the response from the and In was to for no binding was CD97 was CD55 in the of and (Fig. binding was by the of of of interaction. The the of CD97 in a of with the of the the are for the from the surface and surface A specific interaction CD55 and CD97 is where the and (as they in 1 and The of the from to the of the different properties of the with and interactions by the of are only by the of M. J. S. 2000; PubMed Scopus Google Scholar) from CD97 by only three amino acid changes within EGF domains 1, and two of which in EGF domain 1 and in EGF domain 2 sequence in the three in (Fig. In assays M. J. S. 2000; PubMed Scopus Google Scholar) and has no interaction with CD55 (Fig. plasmon resonance was used to the of the three amino acid on CD55 The three EGF domains of expressed and an system (see and in the soluble a surface. by high of a specific interaction CD55 and be and the binding was weak that determination of the was possible with the of protein is difficult to the of a of the interaction of CD55 with CD97 and may be these proteins are of the and therefore produce the on binding to the surface. Fig. shows the response CD97 and the for CD97 are from two of CD97 and show that the in response different CD97 is small by to the in response CD97 and we that the binding of CD55 for is the as to CD97 we can the where the of or CD97 is to the of protein to an of the affinity of for CD55 to that of CD97. of the of the that the affinity of for CD55 is at least an order of magnitude weaker than that of CD97 for is to a in the binding the from to the of the three amino acid to CD55 two assays of CD97 and analysis of cells that the of the and all with that of CD97 cells with the or constructs CD55 binding surface protein in cells was by and to be for all proteins (see with the and of the cells proteins A ligand binding of soluble extracellular domains of CD97 and proteins was also to the binding properties of proteins extracellular domains of CD97 and to to form the protein for in a system (Fig. S. J. Immunol. 1995; PubMed Scopus Google Scholar). human cell a cell surface CD55 J.M. D. D. Google was used as a of CD55 in the CD97 complexes to cells and a in In complexes the binding of complexes to cells was found to be and mediated by CD55 as the of with or a the binding (Fig. and In with the biotinylated and increasing binding for whereas CD97 and binding (Fig. and Both show that the of of the three amino acids the CD97 sequence in a in CD55 and of of the three CD97 amino acids the sequence CD55 The of with at of CD97 the in CD55 binding whereas the with at of in the in CD55 In we have the of the cell surface interaction and is mediated by protein-protein interactions of the SCR and EGF module at the of each plasmon resonance studies of the EGF-SCR interaction show that is characterized by a low affinity by a rapid CD55 and CD97 are known to be expressed at high levels on the surface of cells J.M. D. D. Google Scholar, D. J. Immunol. PubMed Scopus Google Scholar, J. 1997; PubMed Scopus (48) Google our an in interaction of high by has that cell-cell interactions mediated by proteins of the Ig superfamily are also characterized by low affinity interactions. results are therefore of general they suggest may to be a widespread of cell-cell interactions mediated by a variety of protein of cell-cell binding affinity may be by of the proteins on the cell surface. the of the interaction is at least an order of magnitude weaker than the characterized interaction CD55 and The of the interaction to be of a affinity than the interaction the fact that an at binding for receptor and therefore have a high affinity for receptor to a interaction. However, an have many than of CD97 on the cell and a high avidity interaction may therefore be by a protein-protein interaction of a both CD97 CD97 to and CD97 is in the soluble the for glycosylation of CD97 is involved in the specificity or affinity for has been about the role of glycosylation in the specificity and affinity of cell adhesion interactions S.J. S. Jones Stuart van der Merwe P.A. J. Biol. Chem. 1995; Full Text Full Text PDF PubMed Scopus Google Scholar, M. J. Mol. Biol. 1999; PubMed Scopus Google and recent has that for family of the of the glycosylation of the EGF domains by of the is used to the interactions of with D.J. J. P. Nature. 2000; PubMed Scopus Google Scholar). interaction has that the interaction of with is D.J. J. P. Nature. 2000; PubMed Scopus Google Scholar). of the glycosylation of the interaction shows that interactions are an that the and that glycosylation of extracellular domains is required to protein-protein interactions. the binding assays in no assays have a weak but specific interaction CD55 and It is possible that the weaker interaction has the of the which are than It be to high levels of CD55 proteins be to the interaction the both CD97 and are expressed by and provides a the CD55 binding of these cells be by the cell surface levels of a of EGF-TM7 proteins. The of the interaction indicates that the Ca2+ binding within the and domains of CD97 are for intermolecular interactions. Structural from a cbEGF have that Ca2+ binding is required for the of interdomain (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar, 6Werner J.M. Knott V. Handford P.A. Campbell I.D. Downing A.K. J. Mol. Biol. 2000; 296: 1065-1078Crossref PubMed Scopus (61) Google Scholar). a tandem repeats of cbEGF domains or domains with of are to form which specific protein for protein-protein interactions. of the of and pairs from CD97 and suggest they are of the or (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google they have the of the N-terminal cbEGF and the calcium binding of the C-terminal cbEGF (Fig. In also in the of are in CD97 and Calcium binding to cbEGF domains is therefore in interaction by an of the three EGF domains. The of the interaction in the of Ca2+ and that the protein surface on CD97 by CD55 the domain than on a and is by the of the amino acid changes within domains 1 and which CD97 and how of these amino acid changes a in CD55 that of of a chain which CD55 or of a within the have to be from changes that binding by a in A small number of in cbEGF domains associated with human are in variable and has been that in these at used for protein-protein interactions (5Downing A.K. Knott V. Werner J.M. Cardy C.M. Campbell I.D. Handford P.A. Cell. 1996; 85: 597-605Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar). EGF modules are found to be highly in structure, the of by than all EGF structures to are The known of an EGF can therefore be used to the of the on the two domains of CD97. of the three amino acid changes CD97 and the of domains and (Fig. a and indicates that they are to be in these variable and may therefore in the with It that the of for the to at is a at is in the of a cbEGF domain from factor Z. Handford P. Mayhew M. Knott V. Brownlee G.G. Stuart D. Cell. 1995; 82: 131-141Abstract Full Text PDF PubMed Scopus (312) Google Scholar). The of the interaction CD55 and CD97 to in domains 1 and 2 of CD97 to be in to that have that of CD97 than the form CD55 (Fig. and J. van J. Immunol. 1998; PubMed Scopus Google Scholar). the binding site for CD55 is within domains 1 and 2 that the (e.g. 1, also CD55 with the The of our with these results that the of domain 2 to domain than the domain so the CD55 recognition It is to that variable for Ca2+ have been within cbEGF domains from different proteins P. Handford P.A. Downing A.K. J. Mol. Biol. 1999; PubMed Scopus Google Scholar). The within tandem repeats of therefore have the potential to the interdomain may have a high affinity for Ca2+ with that the surface within domains 1 and 2 and so In our demonstrate that the interaction is mediated solely by EGF and SCR domains at the of each protein and is The characteristics of binding affinity, high mediated by of the Ig superfamily and may to be a general feature of protein-protein interactions mediated by highly expressed cell surface proteins. The binding characteristics of how a of can be by a small number of amino acid changes within the EGF module, a for the widespread of module in extracellular proteins involved in protein-protein interactions. We der Merwe for and for in