Crystal structures of the SH2 domain of grb2: highlight on the binding of a new high-affinity...

doi:10.1006/jmbi.2001.5299 available online at http://www.idealibrary.com on J. Mol. Biol. (2002) 315, 1167±1177

Crystal Structures of the SH2 Domain of Grb2:Highlight on the Binding of a New High-affinityInhibitor

Pierre Nioche1, Wang-Qing Liu2, Isabelle Broutin1*Franck Charbonnier1, Marie-TheÂreÁse Latreille1, Michel Vidal2

Bernard Roques2, Christiane Garbay2 and Arnaud Ducruix1

1Laboratoire de Cristallographieet RMN biologiques; UMR8015 CNRS, FaculteÂ dePharmacie, UniversiteÂ ReneÂDescartes, 4, Avenue del'Observatorie, 75270 Pariscedex 06, France2Laboratoire de PharmacologieMoleÂculaire et Structurale,INSERM U266, UMR CNRS8600 FaculteÂ de Pharmacie,UniversiteÂ ReneÂ Descartes4, Avenue de l'Observatoire75270 Paris cedex 06, France

Present address: W.-Q. Liu, M. VPharmacie, UniversiteÂ ReneÂ Descart

Abbreviations used: Boc, tert-butyhexa¯uorophosphate; Fmoc, 9-¯uor1-hydroxybenzotriazole; MAZ,m-aminobenzyloxycarbonyl; MDPSphosphotyrosine-binding; SH2, Src

E-mail address of the correspond

0022-2836/02/051167±11 $35.00/0

The activation of growth factor receptors induces phosphorylation oftyrosine residues in its C-terminal part, creating binding sites for SH2domain-containing proteins. Grb2 is a protein that recruits Sos, theexchange factor for Ras. Recruitment of Sos allows for Ras activation andsubsequent signal transmission. This promotes translocation of MAPkinases into the nucleus and activation of early transcription factors.Grb2, a 25 kDa protein, is composed of one SH2 domain surrounded bytwo SH3 domains. The SH2 domain of Grb2 binds to class II phosphotyr-osyl peptides with the consensus sequence pYXNX. Thus, Grb2 is a goodexample of a bifunctional adaptor protein that brings proteins into closeproximity, allowing signal transduction through proteins located indifferent compartments.

To explore the interactions between Grb2 and phosphorylated ligands,we have solved the crystal structure of complexes between the Grb2-SH2domain and peptides corresponding to Shc-derived sequences. Two struc-tures are described: the Grb2-SH2 domain in complex with PSpYVNVQNat 1.5 AÊ ; and the Grb2-SH2 domain in complex with mAZ*-pY-(aMe)pY-N-NH2 pseudo-peptide, at 2 AÊ . Both are compared to an unligandedSH2 structure determined at 2.7 AÊ which, interestingly enough, forms adimer through two swapping subdomains from two symmetry-relatedmolecules. The nanomolar af®nity of the mAZ-pY-(aMe)pY-N-NH2pseudo-peptide for Grb2-SH2 is related to new interactions with non-conserved residues.

The design of Grb2-SH2 domain inhibitors that prevent interactionwith tyrosine kinase proteins or other adaptors like Shc or IRS1 shouldprovide a means to interrupt the Ras signaling pathway. Newly syn-thesized pseudo-peptides exhibit nanomolar af®nities for the Grb2-SH2domain. It will then be possible to design new inhibitors with similaraf®nity and simpler chemical structures.

# 2002 Elsevier Science Limited

Keywords: Grb2; inhibitor; SH2; X-ray structure; drug design
*Corresponding author
idal and C. Garbay, Laboratoire de Pharmacochimie, UMR 8638 FaculteÂ dees, 4, Avenue de l'Observatoire, 75270 Paris cedex 06, France.loxycarbonyl; BOP, (1H-benzotriazol-1-yloxy)tris(dimethylamino)phosphonium

enylmethyloxycarbonyl; HOBt,

E, (methylolphenylsilyl)ethyl; NOE, nuclear Overhauser enhancement; PTB,homology 2; TFA, tri¯uoroacetic acid.ing author: [email protected]

# 2002 Elsevier Science Limited

mailto:[email protected]

1168 Crystal Structures of Grb2-SH2 Domain Complexes

Introduction

Phosphorylation and dephosphorylation pro-cesses and their regulation are essential in themechanisms of signal transduction pathways.Deregulation of these pathways has dramaticeffects, especially in tumorigenesis. Cellular anar-chic proliferation, observed in some breast andovarian cancers as well as in leukaemia, has beenrelated to the dysfunction of the Ras signalingpathway.1 One of the ®rst steps of intracellular sig-nal transduction is the recognition of a tyrosine-phosphorylated target by the SH2 domain of Grb2.Overexpression of Grb2 has recently been relatedto liver tumorigenesis in mice2 and human breastcancer cells.3 Thus, interruption of signal at thelevel of Grb2-SH2 could serve as a useful strategyfor the treatment of hyperproliferative diseases.4

Many laboratories intend to design new inhibitorswith phosphorylated sequence residues.5,6 Twoknown domains interact speci®cally with phos-phorylated tyrosine: phospho-tyrosine bindingdomains (PTB) and the Src-homology 2 domain(SH2). Both domains have been studied exten-sively. Over a dozen structures of PTB domainshave been elucidated. Even more structures of SH2domains exist in their unliganded form or in com-plex with peptides or pseudo-peptides. SH2domains were found in a great variety of signalingproteins.7 All SH2 domains share the same top-ology with a central antiparallel b-sheet, ¯ankedby two helices.8 The binding site topology of SH2domains excludes the binding of phosphoserineand phosphothreonine residues. Their recognitionmotifs involve phosphorylated tyrosine (pY) resi-dues and subsequent amino acids on the C-term-inal side with speci®c recognition depending onthe nature of amino acid �2 or �3 relative to thepY residue. Phosphopeptides with a speci®city at�3 tend to interact in an extended conformationwith the surface of SH2 domains.8 In Grb2, thepresence of TrpEF1(W121) prevents the linear rec-ognition of phosphopeptides at the surfaceof the SH2 domain. This forces the adoption of aU-shape, as shown by X-ray9 and NMRstructures.10 Therefore, from the consensussequence pYXNX, the speci®city is not given bythe residue at position pY � 3 but by pY � 2.

Among many potential targets, the Grb2-SH2domain recognizes tyrosine kinase receptors andother adaptors, such as Shc,11 with a very high af®-nity. Two sequences in the Shc adaptor have beenidenti®ed as recognition sequences for Grb2:(315PSYVNVQN322);12 (236DHQYYND242).13

SH2 domains have received particular attentionfrom the pharmaceutical companies and severalpeptides or peptidomimetics with potential thera-peutic activity have been prepared.6,14 Some ofthem showed inhibition activity on the growth ofHER2-dependent tumor cells.15 Most of theseinhibitors have been designed by molecular model-ing using the crystal structure of the complexGrb2-SH2/KPFpYVNV and sequential modi®-

cations of the recognition sequence Cap-pY-X � 1-X � 2-X � 3, with Cap being a protecting group ofthe amino-terminal residue. One of the most potentpeptides was synthesized by Garcia-Etchveria16

with a mAZ-pYXN-NH2 tetrapeptide, where X is ahydrophobic a,a-disubstituted cyclic a-amino aciddesigned to interact with the TrpEF1(W121) resi-due. A semi-rigid compound has been designed byCaravatti et al.,17 with an amino pyrimidine at pos-ition pY � 2 and a rigid aromatic ring at pY � 1.Another approach was to take bene®t from theproximity of the amino and carboxy termini to syn-thesize cyclic molecules. This reduces the numberof degrees of freedom for a given molecule to thepoint that a non-phosphorylated compound hasbeen shown to bind with micromolar af®nity.18,19

The structure of Grb2 is composed of one SH2domain facing two neighboring SH3 domains.Here, we describe the 3D structures of the unli-ganded Grb2-SH2 domain and of two complexeswith the following: (i) peptide 1, PSpYVNVQNderived from hp52Shc315-322 sequence; and (ii)pseudo-peptide 2, of chemical sequence mAZ-pY-(aMe)pY-N-NH2 derived from the hp52Shc239-241sequence. The latter was synthesized using X-raystructure-based modeling,20,21 with the insertion ofaMe-phosphotyrosine at position �1 aiming atstrengthening electrostatic and van der Waalsinteractions.22 The structure of the complex clearlydemonstrates the importance of both phosphategroups and van der Waals contacts with the CDloop (according to the nomenclature used by Ecket al.23), especially with residue GlnbD3(Q106).

Results

Unliganded Grb2-SH2 domain

The data presented here reveal that the unli-ganded SH2 domain contains swapping subdo-mains, with the point of rotation at TrpEF1(W121).Consequently, the EF loop folds as an extendedchain. The classical domain is reconstituted by asymmetry-related molecule (Figure 1). Therefore,the peptide recognition site is formed by twodifferent molecules. This Grb2-SH2 domain swap-ping has been observed in a Grb2-SH2 peptidecomplex,24 and cannot therefore be attributed tothe absence of ligand. This could be related to thecrystallization protocol. The unliganded crystalsare the only ones obtained without heating theprotein solution prior to crystallization as observedthat the SH2 domain was precipitated by heatingwhen not complexed. The interface between thetwo swapping domains is almost entirely hydro-phobic, as seen in Figure 1. This can explain theability of the C-terminal helix to open as it is notconstrained by hydrogen bonds, which represent ahigh energy barrier to overcome. Domain swap-ping has been reported for the SH3 domain ofEps8.25 Like the Grb2-SH2 domain presented here,Eps8-SH3 was expressed as a fusion protein withGst. It is well known that Gst has a tendency to

Figure 1. Unliganded SH2 structure. Structure of theunliganded SH2 domain (red) presented with the sym-metry related molecule (orange) that completes the clas-sical SH2 domain structure (green). The residuespresented are at the interface of the two swapping sub-domains.

Crystal Structures of Grb2-SH2 Domain Complexes 1169

induce dimerization of the protein partner. Furtherinvestigation should clarify whether this has a bio-logical relevance or is an artifact due to the isolatedSH2 domain. In order to be able to compare ourunliganded SH2 domain structure with other com-plexes, the structure formed by the swappingdomains coming from the two symmetry-relatedmolecules is considered as the reference.

Nine structures of the Grb2-SH2 domain havebeen solved. Among them, only two are unli-ganded structures and both have been resolved byNMR; 1GHU,26 and 1FHS.27 These two structurespresent an average rms deviation on Ca of 1.3 AÊ

with our reconstituted unliganded structure and1.1 AÊ between themselves. When comparing ourunliganded structure with the 24 structures from1GHU and the 18 structures from 1FHS, it can beseen that the main differences are located at the BCand BG loops, which form the pY to pY � 2 bind-ing pockets. All residues that are poorly de®ned inour crystallographic structure are very disorderedin the NMR structures. TrpEF1(W121), which isresponsible for the pY � 2 speci®city of Grb2, hasthe same orientation in 1GHU. In the 1FHS struc-tures, this residue is poorly de®ned, making thecomparison very dif®cult. The predominantposition seems to be a rotamer of TrpEF1(W121)frequently encountered in the Grb2-SH2 complexstructures. In the absence of a bound peptide,ArgBG4(R132) is not stabilized and has high B fac-tors with little or no density for the guanidiniummoiety. The entire BG loop is poorly de®ned in theelectron density, the only contacts with the rest ofthe domain being two hydrogen bonds betweenGlnBG6(Q144) and GlnbD3(Q106) and two van derWaals contacts (Figure 2(b)1). Those hydrogenbonds are not present in 1FHS structures, in whichthe BG loop is further away from the SH2 domain,thereby opening the secondary binding site, creat-ing a cavity. GlnBG6(Q144) is too far away to form

hydrogen bonds with GlnbD3(Q106). The side-chain of PhebC8(101) is poorly de®ned but seemsto enter the cavity. Regarding the 1GHU struc-tures, the different positions of PhebC8(F101)clearly cluster in two classes. One is superimposedon our PhebC8(F101) position. In this case, thehydrogen bonds between GlnBG6(Q144) andGlnbD3(Q106) are present. The other position ofPhebC8(F101) is equivalent to those of the 1FHSstructures. It enters the cavity and there are nohydrogen bonds any longer betweenGlnBG6(Q144) and GlnbD3(Q106).

PSpYVNVQN hp52Shc315-322 derivedsequence in complex with theGrb2-SH2 domain

A structure of Grb2-SH2 in complex with a Shc-derived peptide (DDPSpYVNVQNLDK) has beensolved by NMR.10 Only residues located in the bturn, i.e. pYVNV, gave rise to enough NOEs to beobserved in the 2D spectrum. In our crystallo-graphic structure, the electron density clearlyshows residues PSpYVNV, but only the mainchain of the QN residues at the carboxy terminusis observed. The proline residue at the amino-term-inal extremity of the Shc-derived peptide does notshow any interaction with the SH2 domain or anysymmetry-related molecule. The carbonyl group ofserine at position pY ÿ 1 presents a bi®d hydrogenbond with the conserved ArgaA2(R67)(Figure 2(a)2). Each of the oxygen atoms of thephosphate group, except the phenolic oxygen atomO4, is bound two or three times to an acceptor: O1is bound to SerBC2(S90) and a water molecule;O2 has a link to conserved ArgbB5(R86) andArgaA2(R67); O3 is bound to the NH2 ofArgbB5(R86), to SerbB7(S88) and SerbC3(S96).Valine in position pY � 1 (Figure 2(b)2) displaysvan der Waals interactions with PhebD5(F108),TrpEF1(W121) and GlnbD3(Q106). Its peptidicamide group is bound to the carbonyl group ofHisbD4(H107). Nd2 of asparagine in positionpY � 2 (Figure 2(c)2) interacts with the carbonylgroups of LysbD6(K109) and LeubE4(L120), andOd1 interacts with the NH peptidic group ofLysbD6(K109). Ca and Cb of Asn � 2 are in vander Waals contact with TrpEF1(W121), which isresponsible for the type I b turn conformation ofthe inhibitor peptides. Valine in position pY � 3 isclose to LysbD6(K109), whereas the C-terminal glu-tamine and asparagine residues point toward thesolvent. TrpEF1(W121) of the SH2 domain, whichhas the same orientation in most of the unligandedstructures, presents a different conformation in theShc-peptides-SH2 domain complexes. The aromaticmoiety is rotated 120 � clockwise and now capsAsn � 2, reinforcing the complex (Figures 2(b)and (c), and 3). ArgBG4(R142) rotates accordingly,its guanidinium head being stacked onTrpEF1(W121). This involves the movement of theentire BG loop, which comes closer to the CD loop.Thus, it forms hydrogen bonds through water mol-

Figure 2. Details of the Grb2-SH2peptides interactions. (a) pY(0) site; (b) pY � 1 site; (c) pY � 2 site. 1.Grb2-SH2 unliganded structure; 2.Grb2-SH2/PSpYVNVQN complex;3. Grb2-SH2/mAZ-pY(aMe)pYN-NH2 complex. The broken linescorrespond to hydrogen bonds; thecontinuous lines correspond to vander Waals contacts. The completenomenclature is indicated onlyonce for each site for reasons ofclarity. In (c)2, residues pY � 3 topY � 6 (VQN) are omitted.


ecules acting as a zipper between BG-CD loops,reinforced by the GlnBG6(Q144)-GlnbD3(Q106)contact (Figure 2(b)2).

mAZ-pY-(aaaMe)pY-N-NH2 pseudo-peptide 2 incomplex with the Grb2-SH2 domain

The mAZ-pY-(aMe)pY-N-NH2 pseudo-peptide 2(derived from the 239-241 hp52Shc sequence) incomplex with the Grb2-SH2 domain was mod-

Figure 3. Superposition of the three Grb2-SH2 structures. SSH2 domain (brown) with the Grb2-SH2/mAZ-pY(aMePSpYVNVQN complex (yellow/green). The residues indicatnition.

eled21 prior to the X-ray structure determination.The localization of the pseudo-peptide is in fullagreement with the modeled structure, the rmsdbeing 1. AÊ for the atoms of the peptide chain. Itwas postulated that the negative charge of thephosphate group of aMe-phosphorylated tyrosinein position �1 would be paired with the positiveNH2

� group of ArgBG4(R142) (Figure 4). In accord-ance with this hypothesis, the X-ray structureshows that two (O2 and O3) of the three oxygen

tereo view of the superposition of the unliganded Grb2-)pYN-NH2 complex (blue/red) and the Grb2-SH2/ed are those responsible for the speci®city of the recog-

Figure 4. GRASP representation of the Grb2-SH2domain in complex with the mAZ-pY-(aMe)pY-N-NH2peptide.


atoms of the phosphate group are bound toArgBG4(R142) (Figure 2(b)3). O1 is linked toSerBG3(S141), AsnBG5(N143) and a water mol-ecule; O2 is bound to NH2 of ArgBG4(R142) andto one water molecule; O3 is linked toArgBG4(R142) via its backbone amide group andits guanidinium head NH. The aromatic ring isengulfed in a hydrophobic pocket consisting ofPhebC8(F101), PhebD5(F108), TrpEF1(W121) and,to a lesser extent, Cg of GlnbD3(Q106). The amideproton of pY � 1 is hydrogen bonded to the carbo-nyl group of HisbD4(H107).

The oxygen atoms of the phosphate group ofpTyr(0) as well as the atoms of asparagine in pos-ition �2 are involved in the same interactions asobserved in the peptide 1-Grb2-SH2 complex withan additional water molecule. The NH2 at the C-terminal end of the peptide is internally hydrogenbonded to the carbonyl group of pY(0) residue.The mAZ chemical group is internally boundthrough its amino group to two oxygen atoms (O1and O4) of the pY(0) phosphotyrosine residue. Thearomatic moiety is parallel with the guanidiniumgroup of ArgaA2(R67) but not completely stacked.The carbonyl group of the carbamate group has abi®d link to ArgaA2(R67) (Figure 2(a)3).

The two SH2 domains of the crystal asymmetricunit present the same set of interactions with themAZ-pY-(aMe)pY-N-NH2 peptide; the rmsd valuebetween molecule A and B is 0.35 AÊ . In moleculeA, the pseudo-peptide is surrounded by the sol-vent, whereas in molecule B it is sandwiched witha B0 molecule. Although there are intermolecularinteractions, this does not affect the intramolecularcontacts.

The main difference with modeling21 is the dis-placement of the SH2 domain BG loop by 2 AÊ dueto the interaction with the mAZ-pY-(aMe)pY-N-NH2 peptide. This movement, in turn affects theconformation of the CD loop. Arising from thisdisplacement, Ca FhebC8(F101) is displaced by�2 AÊ , whereas the moiety is moved away by�5 AÊ .

Discussion

Specificity of tryptophan EF1(W121) fromGrb2-SH2 domain

It was expected from the structure of Grb220 thatTrpEF1(W121) would prevent binding of linearpeptides due to steric hindrance with residues inpositions �3 and �5 of a peptide in an extendedconformation. This is in agreement with resultsfrom ¯uorescent studies.28 The ®rst structure of acomplex between the SH2 domain and theKPFpYVNV peptide derived from BCR-Abl9

demonstrated clearly that the phosphopeptideadopts a U shape, although only the central resi-dues were visible in the electron density. Also, themutation W-EF1-T transforms the speci®city ofGrb2-SH2 recognition from a pY � 2 position to apY � 3 amino acid, allowing extended phospho-peptide recognition, as observed in the Src-SH2domain.29 In the present work (Figure 3), therotation of TrpEF1(W121) by 120 �, as observed inthe complexes with the two peptides, allows forthe capping of Asn � 2, thus reinforcing the stabil-ization of the complex through van der Waalscontacts.

mAZ (3-aminobenzoylcarbonyl) acting as anN-terminal protecting group

It has been suggested30 that the presence of amAZ protecting group at the amino extremity of apeptide would reinforce interactions with theGrb2-SH2 domain, both through the amino andthe aromatic moiety. In the crystal structure ofmAZ-pY-(aMe)pY-N-NH2 in complex with theGrb2-SH2 domain, it appears that the N-terminalprotecting group mAZ, and especially the meta-amino group of the aromatic ring, does not play anessential role in the interactions between the twopartners. To con®rm this view, other N-terminalprotecting groups such as Z (benzyloxycarbonyl)or Ac (acetyl), which both present the advantage ofan easier chemical synthesis, were tested (Table 1).The similar binding af®nity of peptide Z-pY-(aMe)pY-N-NH2 (Kd 2.5 nM), lacking the meta-amino group of mAZ, and of peptide Ac-pY-(aMe)pY-N-NH2, without an N-terminal aromaticgroup (Kd �3.0 nM), indicates that the intramole-cular H-bond between the amino group of mAZand the phosphate group does not contribute tothe peptide af®nity signi®cantly. Thus, the N-term-inal residue of our series of molecules plays aminor role in the Grb2-SH2 domain recognition, in

Table 1. Comparison of Kd and IC50 values of phospho-tyrosine-containing peptides af®nity for Grb2

Peptide Kd (nM)IC50 (nM)/

ELISA

Ac-pY-V-N-NH2 [33] 4300 � 900Ac-pY-Ac6c-N-NH2 [33] 210 � 20PSpYVNVQN (Shc-peptide 1) 18 � 2 71.5mAZ-pY-Ac6c-N-NH2 [21] 30 � 5 120 � 14mAZ-pY-Y-N-NH2 [21] 155 � 30 497 � 50mAZ-pY-pY-N-NH2 [21] 60 � 10 235 � 42mAZ-pY-(aMe)pY-N-NH2 [21] 2 � 1.5 13 � 2Z-pY-(aMe)pY-N-NH2 2.5 � 2.5 44 � 19Ac-pY-(aMe)pY-N-NH2 3.0 � 2.0 47 � 11

mAZ, m-aminobenzyloxycarbonyl.Ac6c, 1-aminocyclohexanecarboxylic acid.ELISA33 was performed using the C-terminal region of EGFR

instead of PSpYVNVQN derived from Shc for all the other pep-tides of the Table.


contrast to compounds described by researchersfrom the Novartis Pharmaceutical Company.30

Mapping the pY�1 binding site

In order to increase the binding af®nity of ourcompounds, we modi®ed the pY � 1 residue toachieve a high level of speci®city. The presence ofa tyrosine repeat in the sequence of Shc warranteda test of various compounds with a tyrosine resi-due in position �1 of the consensus sequencepYXN. The af®nity of the peptide mAZ-pYYN-NH2 is tenfold weaker as compared to the Shc-peptide 1. Gay et al. have shown that a heptamerpeptide with two adjacent phosphorylated tyrosineresidues displayed a 25-fold greater af®nity thanthe monophosphorylated analog.31 We synthesizedthe mAZ-pY-pY-N-NH2 peptide,21 which exhibitsa 2.5-fold increase in af®nity as compared to itsmonophosphorylated pYYN analog (Table 1), butis still less ef®cient than the Shc-peptide 1. Underthe hypothesis that the orientation of the pY � 1side-chain should be constrained to adopt theproper conformation, it was conjectured that a-methylation would restrict the backbone of pY � 1,reinforcing the type I b-turn conformation of thepeptide.21 Indeed, the af®nity of the methylatedcompound presents a 30-fold increase as comparedto that of the non-methylated analog. Althoughthis is the best-known af®nity observed so far forthe Grb2-SH2 domain (2 nM), it amounts to only aninefold increase as compared to the Shc-peptide1. This can be explained by comparing the pY � 1binding site of the three structures from the presentwork (Figure 2(b)). In the unliganded structure, theBG loop is quite loose, having limited anchoragewith the rest of the domain (two hydrogen bondsbetween GlnBG6(Q144) and GlnbD3(Q106), onevan der Waals contact between GlnBG6(Q144) andGlnbD3(Q106), and one van der Waals contactbetween IleBG8(I146) and PhebC8(F101)). Thisexplains the poor electron density and high tem-perature factors. In the peptide 1-SH2 complex, the

rotation of TrpEF1(W121) and ArgBG4(R142)brings the BG loop closer to the SH2 domain, thusallowing hydrogen and van der Waals contactswith the N � 1 residue and the CD loop via fourwater molecules. These water molecules seem toact as a zipper, involving GlnBG6(Q144)-GlnbD3(Q106) interactions (Figure 2(b)2) tightly®xing the BG loop in position and explaining thehigh af®nity of the peptide. In the peptide 2-SH2complex, aMepY was expected to ®ll the cavity,thereby stabilizing the loop. Even if this is partlytrue, as can be seen in Figure 2(b)3 with the highnumber of additional hydrogen bonds, the pre-sence of the tyrosine residue modi®es the pY � 1binding pocket drastically. The additional phos-phate group generates a large movement of the BGloop, displacing it 1.7 AÊ away from its position inthe peptide 1-SH2 complex. In a correlated move-ment, the CD loop comes 1 AÊ closer to the SH2domain, enlarging the cavity between the twoloops. GlnBG6(Q144) and GlnbD3(Q106) moveapart, allowing PhebC8(F101) to insert and formthe hydrophobic pocket with PhebD5(F108) andTrpEF1(W121). This new position of PhebC8(F101),which is observed only in NMR structures,10,26,27 isnot due to steric constraints, the SH2 domain beingsurrounded by the solvent in this region. TheAsnBG5(N143) and IlebD3(I146) positions aremodi®ed and form the new contacts with thepY � 1 phosphate group and with PhebC8(F101).Residue AspbD1(D104), which is in contact withwater molecules in the peptide 1-SH2 complex isstacked onto the CD loop, forming hydrogenbonds between both Od1 and Od2, and the carbonylgroup of AsnCD7(N103). GlnbD3(Q106), whichloses its contact with GlnBG6(Q144), adopts a newconformation; the side-chain turns to letPhebC8(F101) enter the cavity. This new position ismaintained only by van der Waals interactionswith PhebC8(F101) and with pY � 1 from peptide2 (Figure 2(b)3).

The only Grb2-SH2 complex structure showingthis rearrangement of PhebC8(F101) is thatdetermined by Ogura et al (Grb2-SH2/DDPSpYVNVQNLDK).10 Therefore, we comparedboth structures. The rms over main-chain atoms is1.4 AÊ . The main chains of the CD and BG loops arevery close to our peptide 2-SH2 main chains, butthe side-chains are not oriented in a similar way.This is not surprising, as in the NMR structure theresidue at the pY � 1 position is valine, which isshort enough to keep the BG loop side-chainsinside the cavity. Concerning the CD loop,PhebC8(F101) is superimposed on ours, breakingthe GlnBG6(Q144)-GlnbD3(Q106) contact,GlnbD3(Q106) being unmodi®ed. This is madepossible by the movement of PhebD5(F108) in theNMR structure. The aromatic ring of this residue isperpendicular to that in the mAZ-pY(aMe)pYN-NH2 complex. In the peptide 2-SH2 structure,PhebD5(F108) is stacked on the aromatic ring ofthe pY � 1 phosphotyrosine residue, which prohi-bits this movement.

Figure 5. Superimposition on all Ca of all the Grb2-SH2 complexes from the PDB: red, Grb2-SH2/mAZ-pY(aMe)pYN-NH2 complex; green, Grb2-SH2/PSpYVNVQN complex; orange, NMR Grb2-SH2/DDPSpYVNVQNLDK complex (1QG110); blue, Grb2-SH2/mAZ-pY-Ac6c-N-NH2 complex (1CJ132); the othercolors correspond to the following RCSB PDB codes:1BMB and 1BM2,19 1ZFP42 and 1TZE.9 The van derWaals contact between the aMe and Q106 is presented.

Figure 6. Mapping of the pY � 1 binding site: red,mAZ-pY(aMe)pYN-NH2; blue, mAZ-pY-Ac6c-N-NH2;white, calculated solvent-accessible surface of the grb2-SH2-peptide 2 complex after omission of the peptide.


In light of this comparison, it is hard to under-stand the importance of the methyl group additionto the pY � 1 phosphotyrosine residue. In order toexplain why the addition of the methyl groupcauses a 30-fold increase of af®nity, all the Grb2-SH2 complex structures from the RCSB PDB weresuperimposed (Figure 5). It appears that the pos-ition of GlnbD3(Q106) (in red in Figure 5) from ourpeptide 2-SH2 complex structure is different fromall the others. It is no longer hydrogen bondedwith GlnBG6(Q144) but gives rise to van derWaals contacts with PhebC8(F101) and the methylgroup of (aMe)pY (Figure 2(b)3). This couldexplain the af®nity increase for the pY(aMe)pYNpeptide as compared to pYpYN. We can assumethat in the absence of the methyl group,GlnbD3(Q106) is not constrained. In all the PDBstructures, there is a van der Waals contactbetween Cg2 of Val � 1 and Cb of GlnbD3(Q106),whereas in our peptide 2-SH2 structure, it isreplaced by two van der Waals contacts. Onebetween aMe and NH2 of GlnbD3(Q106) and onebetween Cd1 from the pY � 1 aromatic ring and Cg

of GlnbD3(Q106). This hypothesis is strengthenedby the fact that aMe points in the same direction asthe C29-C34 bond of the 1-aminocyclohexane-carboxylic acid (Ac6c) from the 1CJ1 structure.32

Garcia-Echeverria et al. showed that the af®nity ofAc-pY-Acnc-N-NH2 peptides are related to thenumber of van der Waals contacts between the

Acnc and residues PhebD5(F108) andGlnbD3(Q106).33 They showed that the Ac-pY-Ac6c-N-NH2 peptide has a 20-fold higher af®nityfor Grb2-SH2 as compared to Ac-pYVN-NH2, withAc6c and Val � 1 making eight and seven van derWaals contacts, respectively.

Consequently, in order to optimize the af®nity ofa phosphopeptide for the Grb2-SH2 domain, itappears important to increase (i) van der Waalscontacts with the CD loop, and (ii) hydrogenbonds with the BG loop, particularly with the posi-tively charged NH2

� of ArgBG4(R142) and thehydroxyl group of SerBG3(S141). As can be seen inFigure 6, the mAZ-pY(aMe)pYN-NH2 peptidealready ®ts the binding pocket well but there isstill a protrusion pointing toward GlnbD3(Q106).The orientation of aMe as well as this un®lled cav-ity suggest that replacing aMe by aCH2OH mightallow the presence of a hydrogen bond betweenthe hydroxyl group and Ne of GlnbD3(Q106), andtherefore may increase the af®nity. This hypothesiswould imply a movement of the GlnbD3(Q106)side-chain. In the present structure, Oe ofGlnbD3(Q106) is involved only in van der Waalscontacts, therefore it is free to accommodate newinteractions.

Ongoing studies aim at replacing the aminoacids by metabolically stable mimetics. In order toobtain phosphatase-resistant peptides, the phos-phate group of (aMe)pY � 1 was replaced by a car-boxylate or a carboxymethyl group.21 Such


substitutions resulted in a 15-fold decrease of pep-tide binding af®nity. The lower af®nity may belinked to the mono-ionizing carboxylate group. Inour crystal structure, the oxygen atoms of thephosphate group of pY (�1) form multiple hydro-gen bonds with the residues of the Grb2-SH2domain. The loss of a negative charge in the acidfunction lowers the capacity of the inhibitors tointeract with the residues of Grb2-SH2.

This successful example of peptidomimeticdesign is a ®rst step towards the development ofmore potent, low molecular mass compounds ableto inhibit the cellular Ras activation pathway. Itconstitutes a progress in our efforts to identifyinhibitors of the Grb2-SH2 domain suitable forpharmacological investigations.

Materials and methods

Cloning and expression of Grb2-SH2 protein

The cDNA corresponding to residues 60-151 of Grb2(with an additional four N-terminal residues, Gly-Ser-Met-Ala) was ampli®ed by PCR with primers 50-ATGAAACCCATGGCGTGGTTTTTTGGCAAAATCCC-CAGA-30 and 50-CTGTGGTACCTGTTATATGTCCCG-CAGGAATATCTGCTG-30. The NcoI and HindIII cleavedfragment was cloned into the expression vector pGEX-TT (derived from pGEX-2T; Pharmacia, France). The par-ticular feature of this vector lies in the sequence encodedbetween GST and the fusion partner that contains twothrombin cleavage sites separated by a nine amino acidresidue linker.34 After DNA sequence con®rmation, theplasmid was transformed into Escherichia coli BL21-DE3cells and induced by 250 mM IPTG, with 100 mg/ml ofampicillin, for ®ve hours at 28 �C. Cells were centrifugedat 7000 g for 20 min and frozen at ÿ80 �C.

Purification

The cell pellet (20 g) from a six litre culture in Fern-bach was sonicated in 120 ml of lysis buffer A (16 mMNa2HPO4, 500 mM NaCl, 5 mM EDTA (pH 7.0)).Addition of 0.1 % (v/v) Triton X-100 for 30 minutes atroom temperature, and then a 5� dilution in buffer B(16 mM Na2HPO4, 150 mM NaCl, 5 mM EDTA (pH 7.0))were applied before centrifugation at 7000 g. Super-natant was then applied to a 10 ml column of gluta-thione (GST) attached to cross-linked agarose (Sigma,G4510). The column was washed with a buffer contain-ing 50 mM Tris-HCl (pH 8.0) and the protein elutedfrom the column with 10 mM reduced glutathione(Sigma, G4251) in the same buffer. Fractions containingthe pure Gst-SH2 protein were pooled (160 ml at2.3 mg/ml). It was then digested for three hours at 28 �Cin the presence of ten units of thrombin (DiagnosticaStago) per milligram of GST-SH2. The solution wasapplied to an ion-exchange column Source 15Q (Amer-sham Biotech) and 120 mg of the pure SH2 was obtainedin the ¯ow-through. Finally, the solution was dialyzedagainst a 50 mM succinate buffer (pH 6.0). The proteinwas characterized using N-terminal sequence determi-nation and mass spectrometry.

Peptide synthesis

Syntheses of phosphopeptides were carried out by thesolid-phase procedure on an Applied Biosystems (ABI)431A peptide synthesizer with ABI small-scale 9-¯uoro-enylmethyloxycarbonyl (Fmoc) strategy. The peptidePSpYVNVQN issued from the Shc sequence at Tyr317was synthesized by the method described by Ottingeret al.35 The C-terminal ®ve suitably protected residueswere introduced successively to the HMP resin by classi-cal dicyclohexylcarbodiimide/1-hydroxybenzotriazole(DCC/HOBt) coupling and the Fmoc protecting groupswere removed by 20 % (v/v) piperidine. The Fmoc-pTyr-OH, with a free phosphate group, was then coupled bytris(dimethylamino)phosphonium hexa¯uorophosphate(BOP)/HOBt/N,N-dissopropylethylamine (DIEA). Thefollowing two residues were introduced by the sameBOP coupling method. The synthesis of the pseudo-pep-tides was performed as described.21 The Fmoc-Asn(Trt)-OH was coupled to the Fmoc pre-deprotected RinkMBHA amide resin. The Fmoc-(aMe)pTyr(MDPSE2)-OHand the Fmoc-pTyr(MDPSE2)-OH, with the phosphateprotected by MDPSE groups36 were then introduced by¯uoro-N,N,N0,N0-tetramethylformamidinium hexa¯uoro-phosphate (TFFH)/DIEA coupling37 due to the sterichindrance induced by the methyl group of the(aMe)pTyr residue. For the peptide mAZ-pTyr-(aMe)pTyr-Asn-NH2, the N-terminal group was coupledin the form of the active ester Boc-mAZ-ONp.30 For theother two peptides, the Fmoc deprotected peptidyl resinswere capped with the benzyl chloroformate Z-Cl oracetic anhydride Ac2O. The ®nal peptides were cleavedfrom the resins and the protecting groups were removedby a solution of tri¯uoroacetic acid (TFA)/triisopropylsi-lane (TIPS)/water (9.5:2.5:2.5 by vol.). It was further pur-i®ed by semi-preparative HPLC on a Nucleosil C18column (Vydac, 5 mm, 4.6 250 mm) with linear gradientsof mobile phase consisting of solutions A (0.1 % (v/v)TFA in water) and B (0.09 % TFA in 70 % aqueous aceto-nitrile solution). The structure of peptides was con®rmedby electrospray mass spectrometry and NMR spec-troscopy.

Competition assay

Precoated streptavidin plates (Boehringer) were incu-bated with 100 ml/well of biotin-Aha-PSpYVNVQN pep-tide (100 nM concentration in PBS buffer) overnight at4 �C. Nonspeci®c binding was blocked with PBS, 3 %(w/v) bovine serum albumin (BSA) during four hours at4 �C. Competitors were incubated, at the appropriateconcentrations, in PBS, 0.01 % (v/v) Tween 20, 3 % (w/v)bovine serum albumin containing 40 ng of GST-Grb2protein (100 ml/well) during one night at 4 �C. Revel-ation is performed, after anti-GST (Transduction labora-tories; 1/500 in PBS, 0.01 % Tween 20, 3 % bovine serumalbumin) and peroxidase-coupled anti-mouse (Amer-sham; incubation 1/1000 in PBS, 0.01 % Tween 20, 3 %bovine serum albumin) incubations, using TMB solution(Interchim). After coloration was stopped with 10 %(v/v) H2SO4 absorbance, at 550 nm was measured. Doseresponse relationships were constructed by non-linearregression of the competition curves with Origin 40 soft-ware.


Crystallization of unliganded and complexed formsof the Grb2-SH2 domain

Thin plate crystals of unliganded SH2 were obtainedin PEG4000 solution. In order to optimize the crystalliza-tion, glycerol was added and crystals of size0.35 mm � 0.35 mm � 0.12 mm were obtained in 6 %(w/v) PEG4000, 22 % (v/v) glycerol, 50 mM sodiumcacodylate buffer (pH 6.0). All attempts to cryo-freezecrystals of unliganded SH2 failed. Therefore, data had tobe quickly recorded with crystals mounted in a capillary.Attempts at soaking peptides 1 and 2 in crystals of unli-ganded SH2 domain failed. Therefore, the complexeswere co-crystallized using a molar ratio of SH2 domainto phosphopeptide of 1:1.5. Crystals were observed butwere of poor quality unless the solution of the SH2-pep-tide complex was heated at 50 �C for ten minutes beforecrystallization.9

For the peptide 1 complex, the best crystals wereobtained in 2 M (NH4)2HPO4, 4 % (w/v) PEG400,50 mM sodium cacodylate (pH 6.0) with 46 mg/ml ofSH2 domain (solution A). Crystals as large as0.6 mm � 0.2 mm � 0.2 mm were obtained repeatedlyand used for data collection.

For the peptide 2 complex, crystallization was ham-pered by polycrystallinity due to phase separation. The

Table 2. Statistics on data collection and structure re®nemen

Unligande

A. Data collectionSynchrotron line BM30 (EResolution 2.75Temperature (K) 290Crystal to detector distance (mm) 280Wavelength ( AÊ ) 0.980No. images 145Oscillation range (deg.) 1Time/image (seconds) 15No. observations 62,58No. unique reflections 3698Completeness (%) 98.5 (97Rsym

a 7.1 (3

B. CrystalSpace group I422Crystal cell edges (AÊ ) a � b � 81.3No. molecules per asymmetric unit 1

C. RefinementResolution (AÊ ) 8.0-2.7Mean temperature factor (AÊ 2) 37Rfree

b 28.7Rfactor

c 23.1

D. Final modelNo. protein atoms 783No. ligand atoms -No. water molecules 12

E. r.m.s. deviationIdeal bond lengths (AÊ ) 0.014Ideal bond angles (deg.) 2.1

a Rsym � (�hkl�ijIi(hkl) ÿ hI(hkl)ij)/(�hkl�i Ii(hkl)) where Ii(hkl) is theis the mean intensity for this set of hkl.

b Rfree � (�testjjFobsj ÿ kjFcalcjj)/�testjFobsj, where Fobs and Fcalc are tk is a scale coef®cient.

c R-factor � (�workjjFobsj ÿ kjFcalcjj)/�workjFobsj, where Fobs and Fcal

ing set.

best conditions were 2.3 M (NH4)2HPO4, 4 % (w/v)PEG400, 50 mM sodium cacodylate and 19 mg/ml ofSH2 domain (solution B). Crystals of size0.3 mm � 0.1 mm � 0.1 mm were isolated from twins.

Data collection

X-ray diffraction data for the unliganded SH2 domainand the two complexes are given in Table 2. For peptide2, crystals were frozen into a cryoprotectant solution(25 % (v/v) glycerol in solution B) and then transferredin a stream of liquid nitrogen at 100 K. All data wereintegrated with DENZO38 and SCALEPACK.39 Data stat-istics are given in Table 2.

Structure determination and refinements

Using coordinates of the uncomplexed SH2 domain(Parke Davis, personal communication) and programAMoRe,40 the unliganded and Shc-complexed structureswere positioned in the asymmetric unit. For the peptide2-SH2 complex, we used the coordinates reported byEttmayer et al.19 and the no rotate option of AMoRe tolocalize the molecules within the asymmetric unit andthen the rigid body function for the translation.

t

d SH2 Peptide 1 complex Peptide 2 complex

SRF) X31 (DESY) ID14 (ESRF)1.55 2.00290 100150 165

1.057 0.933121 2481 0.75

300 21 155,449 434,272

17,213 14,016.4) 99.7 (99.3) 99.8 (100)

8) 5.2 (37) 5.7 (9.5)

P41212 P6122c � 80.7 a � b � 50.8 c � 89.4 a � b � 61.0 c � 177.0

1 2

5 8.0-1.55 8.0-2.019.6 17.422.8 25.119.0 18.5

783 156674 10690 166

0.035 0.0272.9 2.2

scaled intensity of the ith re¯ection with hkl values and hI(hkl)i

he calculated and observed structure factors for the test set and

c are the calculated and observed structure factors for the work-


The CNS program (version 0.5)41 was used to re®neall structures. The Rfree value was computed so that 1000re¯ections were selected. Statistics are given in Table 2.Some regions of the electron density were reinterpretedafter omit maps and Fourier difference. The twomolecules of the asymmetric unit of the peptide 2-SH2complex were re®ned independently.

Protein Data Bank accession numbers

The RCSB Protein Data Bank accession numbers of thethree structures are 1JYU, 1JYR and 1JYQ for theunliganded SH2, the SH2-peptide 1 complex andSH2-peptide 2 complex, respectively.

Acknowledgments

P.N. was supported by a fellowship from the Minis-teÁre de l'eÂducation, de la Recherche et de la Technologie(MERT) followed by one from La Ligue Nationale contrele Cancer. We also thank the latter for ®nancial support.

References

1. Schlessinger, J. (1994). SH2/SH3 signaling proteins.Curr. Opin. Genet. Dev. 4, 25-30.

2. Diwan, B. A., Ramakrishna, G., Anderson, L. M. &Ramljak, D. (2000). Overexpression of Grb2 inin¯ammatory lesions and preneoplastic foci andtumors induced by N-nitrosodimethylamine in Heli-cobacter hepaticus-infected and non-infected A/Jmice. Toxicol. Pathol. 28, 548-554.

3. Yip, S. S. & Daly, R. J. (2000). Up-regulation of theprotein tyrosine phosphatase SHP-1 in human breastcancer and correlation with GRB2 expression. Int. J.Cancer, 88, 363-368.

4. Vidal, M., Gigoux, V. & Garbay, C. (2001). SH2 andSH3 domains as targets for anti-proliferative agents.Crit. Rev. Oncol. Hematol. 40, 175-186.

5. Cody, W. L., Lin, Z., Panek, R. L., Rose, D. W. &Rubin, J. R. (2000). Progress in the development ofinhibitors of SH2 domains. Curr. Pharm. Des. 6, 59-98.

6. Garbay, C., Liu, W.-Q., Vidal, M. & Roques, B. P.(2000). Inhibitors of Ras signal transduction as anti-tumor agents. Biochem. Pharmacol. 60, 1165-1169.

7. Koch, C. A., Anderson, D., Moran, M. F., Ellis, C. &Pawson, T. (1991). SH2 and SH3 domains: elementsthat control interactions of cytoplasmic signalingproteins. Science, 252, 668-674.

8. Kuriyan, J. (1997). Modular peptide recognitiondomains in eukaryotic signaling. Annu. Rev. Biophy.66, 259-287.

9. Rahuel, J. & Grutter, M. G. (1996). Structural basisfor speci®city of Grb2-SH2 revealed by a novelligand binding mode. Nature Struct. Biol. 3, 586-589.

10. Ogura, K. & Inagaki, F. (1999). Solution structure ofthe SH2 domain of Grb2 complexed with the Shc-derived phosphotyrosine-containing peptide. J. Mol.Biol. 289, 439-445.

11. Rozakis-Adcock, M., McGlade, J., Mbamalu, G.,Pelicci, G., Daly, R., Li, W. et al. (1992). Associationof the Shc and Grb2/Sem5 SH2-containing proteinsis implicated in activation of the Ras pathway bytyrosine kinases. Nature, 360, 689-692.

12. Salcini, A. E., McGlade, J., Pelicci, G., Nicoletti, I.,Pawson, T. & Pelicci, P. G. (1994). Formation of Shc-Grb2 complexes is necessary to induce neoplastictransformation by overexpression of Shc proteins.Oncogene, 9, 2827-2836.

13. Harmer, S. L. & DeFranco, A. L. (1997). Shc containstwo Grb2 binding sites needed for ef®cient for-mation of complexes with SOS in B lymphocytes.Mol. Cell. Biol. 17, 4087-4095.

14. Fretz, H., Furet, P., Garcia Echeverria, C., Rahuel, J.& Schoepfer, J. (2000). Structure-based design ofcompounds inhibiting Grb2-SH2 mediated protein-protein interactions in signal transduction pathways.Curr. Pharmaceut. Des. 6, 1777-1796.

15. Gay, B., Suarez, S., Caravatti, G., Furet, P., Meyer, T.& Schoepfer, J. (1999). Selective Grb2 SH2 inhibitorsas anti-ras therapy. Int. J. Cancer, 83, 235-241.

16. Garcia Echeverria, C. & Caravatti, G. (1998). Potentantagonists of the SH2 domain of Grb2: Optimiz-ation of the X � 1 position of 3-Amino-Z-Tyr(PO3H2)-X � 1-Asn-NH2. J. Med. Chem. 41, 1741-1744.

17. Caravatti, G., Rahuel, J., Gay, B. & Furet, P. (1999).Structure-based design of a non-peptidic antagonistof the SH2 domain of Grb2. Bioorg. Med. Chem.Letters, 9, 1973-1978.

18. Oligino, L. & King, C. R. (1997). Nonphosphorylatedpeptide ligands for the Grb2 Src homology 2domain. J. Biol. Chem. 272, 29046-29052.

19. Ettmayer, P. & Bair, K. W. (1999). Structural andconformational requirements for high-af®nity bind-ing to the SH2 domain of Grb2. J. Med. Chem. 42,971-980.

20. Maignan, S., Guilloteau, J. P., Fromage, N., Arnoux,B., Becquart, J. & Ducruix, A. (1995). Crystal struc-ture of the mammalian Grb2 adaptor. Science, 268,291-293.

21. Liu, W.-Q., Vidal, M., Gresh, N., Roques, B. P. &Garbay, C. J. (1999). Small peptides containing phos-photyrosine and adjacent alpha Me-phosphotyrosineor its mimetics as highly potent inhibitors of Grb2SH2 domain. J. Med. Chem. 42, 3737-3741.

22. Marshall, G. R., Gorin, F. A. & Moore, M. L. (1978).Peptide conformation and biological activity. InAnnual Reports in Medicinal Chemistry (Clarke, F. H.,ed.), vol. 13, pp. 227-238, Academic Press, Orlando.

23. Eck, M. J., Atwell, S. K., Shoelson, S. E. & Harrison,S. C. (1994). Structure of the regulatory domains ofthe Src-family tyrosine kinase Lck. Nature, 368, 0-769.

24. Schiering, N., Casale, E., Caccia, P., Giordano, P. &Battistini, C. (2000). Dimer formation throughdomain swapping in the crystal structure of theGrb2-SH2-Ac-pYVNV complex. Biochemistry, 39,13376-13382.

25. Kishan, K. V., Scita, G., Wong, W. T., Di Fiori, P. P.& Newcomer, M. E. (1997). The SH3 domain ofEps8 exists as a novel intertwined dimer. NatureStruct. Biol. 4, 739-743.

26. Thornton, K., Mueller, W., McConnell, P., Zhu, G.,Saltiel, A. & Thanabal, V. (1996). Nuclear magneticresonance solution structure of the growth factorreceptor-bound protein 2 Src homology 2 domain.Biochemistry, 35, 11852-11864.

27. Senior, M. M. & Wang, Y. S. (1998). The three-dimensional solution structure of the Src homologydomain-2 of the growth factor receptor-boundprotein-2. J. Biomol. NMR, 11, 153-164.


28. Cussac, D., Frech, M. & Chardin, P. (1994). Bindingof the Grb2 SH2 domain to phosphotyrosine motifsdoes not change the af®nity of its SH3 domains forSos proline-rich motifs. EMBO J. 13, 4011-4021.

29. Marengere, L. E. & Pawson, T. (1994). SH2 domainspeci®city and activity modi®ed by a single residue.Nature, 369, 502-505.

30. Furet, P. & Caravatti, G. (1997). Discovery of 3-ami-nobenzyloxycarbonyl as an N-terminal group con-ferring high af®nity to the minimal phosphopeptidesequence recognized by the Grb2-SH2 domain.J. Med. Chem. 40, 3551-3556.

31. Gay, B. & Caravatti, G. (1997). Dual speci®city of srchomology 2 domains for phosphotyrosine peptideligands. Biochemistry, 36, 5712-5718.

32. Furet, P., Garcia-Echeverria, C., Gay, B., Schoepfer,J., Zeller, M. & Rahuel, J. (1999). Structure-baseddesign, synthesis, and X-ray crystallography of ahigh-af®nity antagonist of the Grb2-SH2 domaincontaining an asparagine mimetic. J. Med. Chem. 42,2358-2363.

33. Garcia-Echeverria, C., Gay, B., Rahuel, J. & Furet, P.(1999). Mapping the X(�1) binding site of the Grb2-SH2 domain with alpha,alpha-disubstituted cyclicalpha-amino acids. Bioorg. Med. Chem. Letters, 9,2915-2920.

34. Guilloteau, J. P. & Becquart, J. (1996). Puri®cation,stabilization, and crystallization of a modular pro-tein: Grb2. Protein: Struct. Funct. Genet. 25, 112-119.

35. Ottinger, C. A., Shekel, L. L., Bernlohr, D. A. &Barany, G. (1993). Synthesis of phosphotyrosine-

containing peptides and their use as substrates forprotein tyrosine phosphatases. Biochemistry, 32,4354-4361.

36. Chao, H.-G., Bernatowicz, M. S., Reiss, P. D. &Matsueda, G. R. (1994). Synthesis and application ofbis-silylethyl-derived phosphate-protected Fmoc-phosphotyrosine derivatives for peptide synthesis.J. Org. Chem. 59, 6687-6691.

37. Carpino, L. A. & El-Faham, A. (1995). Tetramethyl-¯uoroformamidinium hexa¯uorophosphate: a rapid-acting reagent for solution and solid phase peptidesynthesis. J. Am. Chem. Soc. 117, 5401-5402.

38. Otwinowski, Z. (1993). Proceedings of the CCP4 StudyWeek End, Data Collection and Processing (Sawyer, L.,Isaacs, N. & Bailey, S., eds), pp. 56-62, DaresburyLaboratory, Warrington.

39. Otwinoski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307-326.

40. Navaza, J. (1992). AMoRe: a new package formolecular replacement. In Molecular Replacement,Proceedings of the CCP4 Study Weekend, pp. 87-90,SERC, Daresbury, Warrington, UK.

41. Brunger, A. & Warren, G. (1998). Crystallography &NMR system: a new software suite for macromol-ecular structure determination. Acta Crystallog. sect.D, 54, 905-921.

42. Rahuel, J. & Gay, B. (1998). Structural basis for thehigh af®nity of amino-aromatic SH2 phosphopeptideligands. J. Mol. Biol. 279, 1013-1022.

Edited by R. Huber

(Received 25 June 2001; received in revised form 19 November 2001; accepted 26 November 2001)

Crystal structures of the SH2 domain of grb2: highlight on the binding of a new high-affinity...

Documents

Transcript of Crystal structures of the SH2 domain of grb2: highlight on the binding of a new high-affinity...