The importance of being proline: the interaction of proline-rich motifs in signaling proteins with their cognate domains BRIAN K. KAY,*,1 MICHAEL P. WILLIAMSON,† AND MARIUS SUDOL‡ *Department of Pharmacology, University of Wisconsin-Madison, Madison, Wisconsin 53706-1532, USA; †Department of Molecular Biology and Biotechnology, University of Sheffield, Western Bank, Sheffield S10 2TN, United Kingdom; ‡Department of Biochemistry and Molecular Biology, Mount Sinai School of Medicine, New York, New York 10029-6574, USA ABSTRACT A common focus among molecular SH3 DOMAINS and cellular biologists is the identification of pro- teins that interact with each other. Yeast two- SH3 domains are 50–70 amino acids long and often hybrid, cDNA expression library screening, and present in eukaryotic signal transduction and cy- coimmunoprecipitation experiments are powerful toskeletal proteins such as Abl, actin-binding pro- methods for identifying novel proteins that bind to tein, Bem1, cdc25, cortactin, calcium channel ␤1B2 one’s favorite protein for the purpose of learning subunit, Grb2, myosin, Nck, PI3K regulatory subunit, more regarding its cellular function. These same PLC␥, ras GTPase activator, spectrin, and tight junc- techniques, coupled with truncation and mutagen- tion protein ZO-1. Computer-aided analysis of pro- esis experiments, have been used to define the tein sequences has even suggested that SH3 domains region of interaction between pairs of proteins. may also exist in bacteria (7a). For a long time, the One conclusion from this work is that many inter- function of the SH3 domain in eukaryotes was enig- actions occur over short regions, often less than 10 matic; however, since the domain was present in so amino acids in length within one protein. For many cytoskeletal proteins, it was assumed to play a example, mapping studies and 3-dimensional anal- role in mediating protein–protein interactions (8) or yses of antigen–antibody interactions have re- directing cell compartmentalization (9). The SH3 vealed that epitopes are typically 4–7 residues long domain clearly had an important function for some (1). Other examples include protein-interaction proteins, as mutations in the SH3 domains of cellular modules, such as Src homology (SH) 2 and 3 Src (10) and Abl (11, 12) are activating. domains, phosphotyrosine binding domains The ligand specificity of SH3 domains was first (PTB), postsynaptic density/disc-large/ZO1 revealed when a ␭-cDNA expression library was (PDZ) domains, WW domains, Eps15 homology screened with a glutathione S-transferase (GST)-Abl (EH) domains, and 14–3-3 proteins that typically SH3 fusion protein (13). Two different cDNA clones recognize linear regions of 3–9 amino acids. Each were isolated and the regions responsible for their of these domains has been the subject of recent binding to the Abl SH3 domain were shown to be reviews published elsewhere (2–7). Among the proline-rich segments (14). Examination of other primary structures of many ligands for protein– SH3 ligands identified in the same manner (15–17) protein interactions, the amino acid proline is has suggested that SH3 domains recognize proline- critical. In particular, SH3, WW, and several new rich sequences containing the core PxxP, where x protein-interaction domains prefer ligand se- denotes any amino acid (18, 19). quences that are proline-rich. In addition, even With the advent of combinatorial peptide libraries, though ligands for EH domains and 14–3-3 do- ligand specificity has been determined rapidly for a mains are not proline-rich, they do include a single large number of different SH3 domains. The SH3 proline residue. This review highlights the analysis domains of Src and the phosphatidyl inositol 3-ki- of those protein–protein interactions that involve nase (PI3K) regulatory subunit select proline-rich proline residues, the biochemistry of proline, and sequences from peptides displayed on beads (20) current drug discovery efforts based on proline peptidomimetics.—Kay, B. K., Williamson, M. P., 1 Sudol, M. The importance of being proline: the Correspondence: Department of Pharmacology, Univer- sity of Wisconsin, 1300 University Ave., Madison, WI 53706- interaction of proline-rich motifs in signaling pro- 1532, USA. E-mail: [email protected] teins with their cognate domains. FASEB J. 14, 2 The FASEB Journal cannot guarantee the availability of 231–241 (2000) references to the Web. 0892-6638/00/0014-0231/$02.25 © FASEB 231 TABLE 1. Protein-rich sequences that bind to specific protein complexes have shown that peptide ligands can bind a interaction modules in two orientations with respect to the SH3 domain (28, 29). The peptides, which bind in either an N to Proteins with SH3 domains Peptide ligand motif Reference C or C to N terminal orientation relative to the SH3 Src RPLPPLP (21–23, 28) domain, have been classified as either class I or class PPVPPR Yes RxLPxLP (26) Lyn RxxRPLPPLPxP (22) Abl PPP⌿PPPP␺P (22) PI3K RxxRPLPPLPP (22) PLC␥ PPVPPRP (26) Cortactin ϩPP␺P␺KP (26) p53BP2 RPx␺P␺R (26) Grb2 A PLPxLP (26) Crk A PxLPx(K/R) (134) Amphiphysin I PxRPx (R/H) (R/H) (27) Nck, SH3 B PxPPxRxxSL (25) CAP, SH3 C PxPPxRxSSL (135) Proteins with WW domains Class I WW’s YAP65 PPxY (53) Nedd-4 PPxY (66) Dystrophin PPxY (69) Class II WW’s FBPs PPPPPPL/R (56) FE65 PPPPPPL/R (65) Class III WW’s FBPs (PxxGMxPP)2* (58) Class IV WW’s Essl/Pinl (pS/pT)P (59, 62, 110) Nedd-4 (pS/pT)P (62) Other proteins Mena (EVH1) (D/E)FLPPPP (84) Homer (EVH1) PPxxFr (88) CD2-BP (BHB) (PPPGHR)2* (91) Profilin G/LPPPPPP(PP) (92) a Peptide sequences were deduced from screens of combinato- rial peptide libraries, analysis of various interacting proteins, or amino acid replacements of individual sequences. In some cases, only the core of the deduced motifs is shown. Additional ligand sequences Figure 1. A) Three-dimensional model of how class I and II can be found on the Internet for SH3 (http://www.pharmacia.mod- ligands bind to SH3 domains, based on the structure of the ules/SH3/ligands) and WW domains (http://www.bork.embl-heidel- berg.de/Modules.ww-gif.html). Residues labeled ⌿, ϩ, and x corre- Src SH3 domain (28). The protein’s topological surface is spond to aliphatic (A, I, L, V, and P), positively charged (R, K, or H), shown in gray, with acidic residues colored in blue. The and any residues, respectively. Lowercase letters refer to positions ligands are RALPPLPRY, a class I ligand (below) and AFAP- that are often (but not always) a certain residue. pS and pT refer to PLPRR, a class II ligand (above). All residues in the ligands phosphoserine and phosphothreonine, respectively. Residues in pa- are colored gray, except for the recognition prolines (yellow), rentheses have been shown to be equivalent. Proteins with multiple flanking prolines (cyan), leucine (brown), and arginine SH3 domains, such as Grb2, Nck, and CAP (Cbl-Associated Protein) (red). The side-chain of the first arginine in the class II ligand have the particular SH3 domains denoted with letters, starting at the has been omitted for clarity. The peptides interact with the amino terminus with A. 2* indicates a tandem repeat of the bracketed two LP dipeptide pockets (left and center) and the acidic sequence. specificity pocket (right), which form grooves in the surface separated by ridges. The NH2- and carboxyl-termini of the and phage (21–23). Phage libraries have also been peptides are denoted with N and C, respectively. B) Three- dimensional model of how peptide ligands bind to the YAP65 used to identify the specificity of SH3 domains from WW domain. The protein’s topological surface is shown in Abl, amphiphysin I, cortactin, Crk, Fyn, Grb2, Lyn, white, with three of the key residues shown in green (from left Nck, p53BP2, PI3K, PLC␥, Tsk, and Yes (22, 24–27). to right, Trp39, Tyr28, and Leu30). The peptide ligand is The optimal ligand preference for each domain shown in gray, except for the core prolines (yellow), which sit varies around the PxxP core (Table 1), demonstrat- in a shallow pocket formed by Trp39 and Tyr28, the flanking prolines (cyan), and the tyrosine (red), which sits over the ing that a great deal of protein–protein interaction hydrophobic pocket containing Leu30. C) A space-filling specificity can be encoded by proline-rich sequences. model of a polyproline helix, colored as green, carbon; red, Investigations into the structure of peptide-SH3 oxygen; blue, nitrogen; white, hydrogen. 232 Vol. 14 February 2000 The FASEB Journal KAY ET AL. II ligands, respectively. The orientation of the pep- Recently, a novel ligand sequence has been ob- tide is dictated by the location of a positively charged served for the SH3 domain of the epidermal growth residue, relative to the PxxP core, which forms a salt factor receptor substrate, Eps8, which has been bridge with an acidic residue in the SH3 domain discovered to form intertwined dimers (37). Because (Fig. 1A). Thus, peptides with the motif ϩxxPxxP the site of dimerization includes the interface of the and PxxPxϩ (where ϩ refers to a positively charged SH3 domain that typically interacts with PxxP motifs, amino acid) correspond to class I and class II motifs, the specificity of the intertwined dimers is altered. respectively. Some SH3 domains, like that present in Analysis of the ligand preferences of this dimer has Src, have the capacity to bind peptides of either class revealed it to recognize not PxxP, but PxxDY instead (i.e., RPLPPLP and PPVPPR; where the scaffolding (37a). It will be interesting to learn how common prolines are underlined), although the biological this unusual arrangement is among other SH3 do- implications of this are unclear. (One frequently mains. raised possibility is that ligand binding in two orien- Analysis of several SH3 domain interactions has tations could access different partners in multicom- suggested that some form stable complexes and others ponent protein complexes.) In both orientations, may be regulated in the cell.