Cell, Vol. 72, 767–778, March 12, 1993, Copyright 1993 by Cell PressSH2 Domains Recognize SpecificPhosphopeptide Sequencesin SH2 domain sequences at likely sites of contact pro-vides a structural basis for the phosphopeptide selec-tivity of these families. Possible in vivo binding sitesZhou Songyang,1,2Steven E. Shoelson,3Manas Chaudhuri,3Gerald Gish,4Tony Pawson,4Wayne G. Haser,5Fred King,5Tom Roberts,5Sheldon Ratnofsky,6Robert J. Lechleider,7of the SH2 domains are discussed.Benjamin G. Neel,7Raymond B. Birge,8J. Eduardo Fajardo,8Margaret M. Chou,8IntroductionHidesaburo Hanafusa,8Brian Schaffhausen,9and Lewis C. Cantley1The discovery that Src homology 2 (SH2) domains provide1Cellular and Molecular Physiologyphosphorylation-dependent and sequence-specific con-Harvard Medical Schooltacts for assembly of receptor signaling complexes hasand Department of Medicineprovided a breakthrough in understanding signal trans-Beth Israel Hospitalduction (Cantley et al., 1991; Koch et al., 1991). This 苲 100Boston, Massachusetts 02115amino acid sequence was first pointed out as a nonkinase2Department of Physiologydomain conserved between the src and fps gene productsTufts University School of Medicinelikely to be involved in targeting (Sadowski et al., 1986).Boston, Massachusetts 02111Now more than 20 cytosolic proteins likely to be involved3Research Divisionin signaling have been shown to contain SH2 domains.Joslin Diabetes CenterRecombinant SH2 domains from several different pro-Department of Medicineteins, including Crk, phosphoinositide-specific phospholi-Brigham and Women’s Hospitalpase C type ␥ (PLC-␥), Ras GTPase-activating protein (Rasand Harvard Medical SchoolGAP), and Abl, have been shown to bind specifically toBoston, Massachusetts 02115tyrosine-phosphorylated cellular proteins (Anderson et al.,4Division of Molecular and Developmental Biology1990; Margolis et al., 1990; Mayer and Hanafusa, 1990;Samuel Lunenfeld Research InstituteMayer et al., 1991; Moran et al., 1990). Evidence that theMount Sinai Hospitalbinding of a particular SH2 domain to tyrosine-phosphor-Toronto, Ontarioylated proteins is dependent on the primary sequenceM5G 1X5, Canadaaround the phosphotyrosine (pTyr) came from a compari-6BASF Corporationson of the sequences of the regions of polyoma middle T195 Albany Streetand the platelet-derived growth factor (PDGF) receptorCambridge, Massachusetts 02139that bind phosphatidylinositol 3-kinase (Cantley et al.,8Rockefeller University1991). The sequence pTyr-Met/Val-X-Met was found atNew York, New York 10021sites known to be critical for phosphatidylinositol 3-kinase5Division of Cellular and Molecular Biologybinding to these proteins (Cohen et al., 1990; KazlauskasDana Farber Cancer Instituteand Cooper, 1989; Talmage et al., 1989; Whitman et al.,Harvard Medical School1985), and this sequence has been predictive for otherBoston, Massachusetts 02115receptors or receptor substrates that bind phosphatidyl-7Molecular Medicine Unitinositol 3-kinase (Lev et al., 1992; McGlade et al., 1992;Beth Israel HospitalSun et al., 1991; Reedijk et al., 1992). Synthetic phospho-Boston, Massachusetts 02215peptides based on this sequence have been found to block9Biochemistry Departmentphosphatidylinositol 3-kinase binding to the PDGF recep-Tufts University School of Medicinetor (Escobedo et al., 1991; Fantl et al., 1992) and to poly-Boston, Massachusetts 02111oma middle T (Auger et al., 1992; Yoakim et al., 1992;Carpenter et al., 1993). In addition, mutational studieshave shown that the SH2 domains of phosphatidylinositol3-kinase, Ras GAP, and PLC-␥ recognize distinct phos-Summaryphopeptide sequences in the PDGF receptor (Fantl et al.,1992; Kazlauskas et al., 1990, 1992).A phosphopeptide library was used to determine thesequence specificity of the peptide-binding sites of These studies have identified peptide sequences withhigh affinity for a few SH2 domains, yet systematicSH2 domains. One group of SH2 domains (Src, Fyn,Lck, Fgr, Abl, Crk, and Nck) preferred sequences with searches for optimal sequences for SH2 domains areneeded. For example, peptides containing the sequencethe general motif pTyr-hydrophilic-hydrophilic-lle/Prowhile another group (SH2 domains of p85, phospho- pTyr-Met/Val-X-Met have high affinity for phosphatidylino-sitol 3-kinase, and peptides with Gly, Ala, or Pro in placelipase C-␥, and SHPTP2) selected the general motifpTyr-hydrophobic-X-hydrophobic. Individual members of the final Met have low affinity (Fantl et al., 1992). How-ever, the possibility that a peptide with 1 of the other 16of these groups selected unique sequences, exceptthe Src subfamily (Src, Fyn, Lck, and Fgr), which all amino acids at this position has an equal or higher affinityhas not been tested. Here we demonstrate direct bindingselected the sequence pTyr-Glu-Glu-lle. The variabilityCell768Table 1. Degenerate Phosphorylated Peptide GDGY*XXXSPLLLAmino AcidCycle A R N D E Q G H I L K M F P S T Y V1(G)––––––458–––––––––––2(D)–––285––––––––––––––3(G)––––––394–––––––––––4 (Y*) –––––––––––––––– 3.3–5 (X) 19.5 5.8 15.7 18.8 14.1 19.5 29.9 5.8 12.0 17.5 13.7 9.5 13.2 16.6 11.1 13.1 22.0 13.26 (X) 15.6 10.9 15.0 18.7 14.7 16.9 19.3 6.9 7.7 12.1 13.1 9.3 9.3 10.9 6.8 9.2 16.8 9.77 (X) 11.5 13.4 12.1 14.8 13.1 14.2 15.5 6.6 5.7 8.9 9.6 7.0 6.6 8.9 8.0 8.7 16.5 7.08(S)––––––––––––––46.6 – – –Abbreviations for amino acid residues are: A, Ala; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R,Arg; S, Ser; T, Thr; V, Val; Y, Tyr. X indicates a degeneracy of all the above acids (e.g., all amino acids except Trp and Cys). Y* indicates pTyr (the phosphate on this tyrosine prevents tyrosine being sequenced). Data (in pmoles) show the amount of each amino acid at different sequencingcycles. A dash indicates a background signal of less than 3 pmol after the lag from the previous cycle was corrected.of short phosphopeptides to recombinant SH2 domains pTyr where Met was highly preferred (Figure 2C, cycle 7).There was a slight preference for large hydrophobic aminoand present a technique that selects the optimal phospho-peptides from a degenerate mixture. The likely sites of acids at the first residue after pTyr (Figure 2A), but theselectivity at this residue was not nearly as great as