The SH2 domain from the tyrosine kinase Fyn in complex with a phosphotyrosyl peptide reveals insights into domain stability and binding specificity

Research Article

1313

The SH2 domain from the tyrosine kinase Fyn in complex with a phosphotyrosyl peptide reveals insights into domain stability and binding specificity Terrence D Mulhern1†, Graeme L Shaw1, Craig J Morton1, Anthony J Day2 and Iain D Campbell1* Background: SH2 domains are found in a variety of signal transduction proteins; they bind phosphotyrosine-containing sequences, allowing them to both recognize target molecules and regulate intramolecular kinase activity. Fyn is a member of the Src family of tyrosine kinases that are involved in signal transduction by association with a number of membrane receptors. The kinase activity of these signalling proteins is modulated by switching the binding mode of their SH2 and SH3 domains from intramolecular to intermolecular. The molecular basis of the signalling roles observed for different Src family members is still not well understood; although structures have been determined for the SH2 domains of other Src family molecules, this is the first structure of the Fyn SH2 domain. Results: The structure of the Fyn SH2 domain in complex with a phosphotyrosyl peptide (EPQpYEEIPIYL) was determined by high resolution NMR spectroscopy. The overall structure of the complex is analogous to that of other SH2–peptide complexes. Noteworthy aspects of the structure are: the BG loop, which contacts the bound peptide, contains a type-I′ turn; a capping-box-like interaction is present at the N-terminal end of helix αA; cis–trans isomerization of the ValβG1–ProβG2 peptide bond causes conformational heterogeneity of residues near the N and C termini of the domain.

Addresses: 1Oxford Centre for Molecular Sciences, South Parks Road, Oxford, OX1 3QY, UK and 2Department of Biochemistry, University of Oxford, South Parks Road, Oxford, OX1 3QU, UK. †Present

address: Department of Biochemistry, University of Adelaide, Adelaide SA 5005, Australia. *Corresponding author. E-mail: [email protected] Key words: binding specificity, Fyn, NMR structure, peptide complex, SH2 Received: 30 June 1997 Revisions requested: 11 August 1997 Revisions received: 29 August 1997 Accepted: 29 August 1997 Structure 15 October 1997, 5:1313–1323 http://biomednet.com/elecref/0969212600501313 © Current Biology Ltd ISSN 0969-2126

Conclusions: Comparison of the Fyn SH2 domain structure with other structures of SH2 domains highlights several interesting features. Conservation of helix capping interactions among various SH2 domains is suggestive of a role in protein stabilisation. The presence of a type-I´ turn in the BG loop, which is dependent on the presence of a glycine residue at position BG3, is indicative of a binding pocket, characteristic of the Src family, SykC and Abl, rather than a binding groove found in PLC-γ1C, p85αN and Shc, for example.

Introduction Human pp59fyn (Fyn) is a member of the Src family of cellular non-receptor tyrosine kinases that, collectively, are involved in the cytoplasmic signal transduction cascade for a variety of membrane receptors (reviewed in [1]). These molecules are of a modular nature, consisting of a unique N-terminal sequence, three protein modules (the SH3, SH2 and kinase domains) connected by linker regions and a C-terminal tail (shown schematically in Figure 1a). Recently determined structures of the inactive state of essentially intact Src [2] and Hck [3] molecules have given great insight into the processes involved in the regulation of Src family kinase activity. Although some of the global aspects of kinase regulation are now clear [4], the features that give rise to the different biological roles of Src family members have yet to be elucidated.

Fyn interacts with the T-cell receptor ζ chain–CD3 complex [5], integrins via the focal adhesion kinase pp125FAK [6] and glycosyl phosphatidylinositol (GPI) anchored receptors lacking cytoplasmic domains [7]. The specificity of the Fyn SH2 domain for phosphotyrosine (pY)-containing sequences is similar to that of other members of the Src family; it binds, with highest affinity, peptide ligands in which both of the residues immediately C-terminal to the pY residue (the +1 and +2 positions) are glutamate and the +3 residue is isoleucine [8]. A number of crystal and solution structures of Src family SH2 domains have been determined (Src [2,9–12], Lck [13–16], Blk [17] and Hck [3]). This is the first report of a structure for the SH2 domain of Fyn, however. Structures have also been determined of a variety of SH2 domains

1314

Structure 1997, Vol 5 No 10

Figure 1

(a)

Myristate

SH3

SH2

Kinase

PPII

Fyn Palmitate

pTyr

(b)

150

160

170

180

190

200

210

220

230

240

SIQAEEWYFGKLGRKDAERQLLSFGNPRGTFLIRESETTKGAYSLSIRDWDDMKGDHVKHYKIRKLDNGGYYITTRAQFETLQQLVQHYSERAAGLSSRLVVPSHK

βA

αA

βB

βC

βD βD′ β E

βF

αB

[βG]

(c) 60

Intermolecular Long Range Medium Range Sequential Intraresidue

Summary of the Fyn domain structure, SH2 sequence and experimental restraints. (a) Schematic representation of the Fyn tyrosine kinase. The locations of the SH3, SH2 and kinase domains are shown along with the myristate and palmitate moieties, the autoregulatory PPII helix and phosphotyrosine. (b) The sequence of the Fyn SH2 construct used in this study with the position of the secondary structure elements underlined. (c) Graph of the residue-by-residue breakdown of experimental restraints used in the structure calculation. The phosphotyrosine containing peptide precedes the SH2 domain with its sequence shown below, where the pY is underlined.

Number of NOEs

40

20

EPQYEEIPIYL

150

160

170

180

190

200

210

220

230

240

Structure

from molecules that are not members of the Src family (e.g. p85αN [18]; PLC-γ1C [19]; Abl [20]; Grb2 [21]; ZAP70 [22]; SykC [23]; SypN [24]; and Shc [25]). Many of these SH2 domains are notable in that they bind the residues C-terminal to the pY residue in a groove, rather than in a tight pocket as occurs for Src family members. The relatively large number of different SH2 structures available allows detailed comparisons to be made of subtle structural similarities and differences. Indeed, although some of the important structural determinants of peptide binding have been elucidated [13], our comparison of the Fyn SH2 with other SH2 domains has highlighted the critical role of residue GlyBG3 (the third residue of the BG loop) in the formation of a tight hydrophobic pocket rather than a binding groove.

Results NMR spectroscopy data

The solution structure of the SH2 domain from human p59fyn (residues 143–248) in complex with a pY-containing peptide EPQpYEEIPIYL, corresponding to residues 321–331 of hamster polyomavirus middle T antigen [26], was determined by high-resolution NMR spectroscopy.

Structures were generated using dynamical simulated annealing, employing 1608 experimentally derived restraints, composed of 1505 NOE restraints, 39 dihedral angle restraints and 32 hydrogen bonds (with two restraints per hydrogen bond). NMR experiments were recorded on unlabelled, 15N-labelled and 13C/15N-labelled protein samples containing unlabelled peptide. Of the 1505 NOE restraints used in the calculations, 368 were intraresidue, 414 were sequential, 213 were medium range (i to i + j, j ≤ 4), 371 were long range (i to i+j, j ≥ 5) and 34 were intermolecular (SH2–peptide). In addition, there were 105 distance restraints that remained ambiguous, with unresolved contributions from two or more groups or protons (five of these restraints involved intermolecular interactions). The sequence of the expressed clone, the location of secondary structure elements and a residue-by-residue breakdown of experimental restraints are shown in Figure 1. Final structures (22) were selected on the basis of having the lowest overall energy, and they had no NOE violations > 0.3 Å and no dihedral angle violations > 5°. The Cα trace for the minimised average coordinates are shown in Figure 2. Statistics for the entire family and the energy minimised average structure are shown in Table 1.

Research Article Structure of a Fyn SH2–peptide complex Mulhern et al.

1315

Figure 2 Stereoview of the Cα trace of the energy minimised average coordinates. The Cα atom of every 10th residue is labelled and the nonhydrogen atoms of the peptide residues which interact with the protein (pY, +1E, +2E and +3I) are shown.

C 240

C

170

240

190 230

+1E +3I N

+2E

190

200 160 230 pY

220

+3I

N+1E

200 160

+2E

pY

220

180 210

170

180 210 Structure

Isotope-filtered experiments

The use of X-filtered experiments for differentiating between protons attached to 13C or 15N and those attached to unlabelled nuclei has been demonstrated in a number of experiments [27–29]. The X-filtered NOESY experiment [27] provides a convenient way of assigning NOEs in complexes between labelled and unlabelled species. X-filtered NOESY and TOCSY experiments were used here to obtain the 1H chemical shift assignments of the unlabelled peptide in the complex and to identify NOEs from the unlabelled peptide to the labelled protein; they were also useful in identifying several slowly exchanging threonine hydroxyl protons (Thr174 and Thr223) on the protein. The X-filtered experiments were modified to include a double-pulse field gradient spin echo (DPFGSE) at the end of the pulse sequence to achieve water suppression [30]. In addition, the 90x90±x pulses used in the X-filter were replaced with the composite pulse 90±x180–x270x [31]. By alternating the phase of the 90° pulse, the composite pulse acts as either a 0° pulse or a 180° pulse on z-magnetisation. The larger band width of the composite pulse makes the X-filter more effective when used to suppress protons attached to 13C spins; similar improvements have been observed by incorporating a hyperbolic secant pulse into the X-filter [28]. Description of the solution structure

The SH2 domain of Fyn is similar to that of other Src family members. It consists of a large triple-stranded antiparallel β sheet (βB, βC and βD) flanked by two α helices (αA and αB), with an additional short β strand (βA) arranged parallel to the first strand of the large β sheet and

a small triple-stranded antiparallel β sheet (βD′, βE and βF) involving the C-terminal end of the third strand of the large β sheet (βD′). In other Src family SH2 structures, an additional β strand (βG), corresponding to residues Val244–Ser246 in Fyn, has been found; however, these residues were not well-defined in the Fyn structure. A superposition of the 22 final structures and a MOLSCRIPT [32] representation of the energy minimised average coordinates showing the secondary structure elements is presented in Figure 3. Three cysteine residues in the native sequence in, and close to, β-strand βG (CysBG5, CysBG6, of loop BG, and CysβG3) were mutated to serine residues for this study to eliminate the requirement for the sample to be maintained under reducing conditions. Comparisons of 1H–15N-HSQC NMR data for the wild-type sequence [33] with that of this Cys → Ser mutant suggested that these mutations have a minimal impact on the overall protein conformation. The lack of a well-defined conformation for strand βG in the structures of this Fyn mutant, however, could be due to local unfolding as a result of the reduction in hydrophobic contacts that would otherwise be expected for the cysteine residues in the wild-type sequence. The Fyn construct used in this study has six residues preceding βA and two residues following the residues corresponding to βG. In common with many other SH2 domains, the N and C termini of the polypeptide chain are close together in space, but despite the presence of several NOEs between residues near the N and C termini, the first six residues (Ser143–Glu148) and the last nine residues (Ser240– His248) are not particularly well-defined with respect to the protein as a whole.

1316

Structure 1997, Vol 5 No 10

Table 1 Structural statistics for the 22 Fyn SH2–peptide complex structures*. Deviations from restraints



Rmsd (Å) from (1505) distance restraints

0.036 ± 0.001

Rmsd (°) from dihedral angle restraints (39)

0.38 ± 0.13

Deviations from covalent geometry bonds (Å) ×

103

angles (°) impropers (°) X-PLOR energies

(SA)r

(SA)r

3.28 ± 0.08

3.21

0.565 ± 0.008

0.547

0.420 ± 0.008

0.409



(SA)r

356.4 ± 13.3

333.3

(kcal/mol)†

Etot Ebond

20.91 ± 0.97

20.04

Eangle

169.62 ± 5.18

159.31

Eimpr

20.12 ± 0.84

18.67

Erepel

46.12 ± 4.45

42.79

EL–J

–1062.0 ± 85.6

–1170.29

ENOE

99.24 ± 5.41

92.17

Ecdih

0.39 ± 0.22

0.28

N, Ca, C¢

Heavy atoms

well defined‡

0.57 ± 0.08

0.98 ± 0.09

all protein

0.96 ± 0.13

1.41 ± 0.13

all protein + peptide

1.23 ± 0.16

1.72 ± 0.15

site§

0.36 ± 0.09

0.73 ± 0.16

Cartesian coordinate rmsds (Å) for versus (SA)

binding

* is the ensemble of 22 final simulated annealing structures, (SA) is the mean coordinate set calculated after superposition of the backbone N, Cα and C′ atoms of the well-defined SH2 residues and (SA)r is the representative structure generated by restrained energy minimisation of the mean coordinate set. †The van der Waals repulsion term Erepel was calculated with a force constant of 4 kcal mol–1 Å–4 and the atomic radii set to 0.75 times those in the allhdg parameter set. The distance restraint energy term ENOE and dihedral angle restraint term Ecdih were calculated with force constants of 50 kcal mol–1 Å–2 and 200 kcal mol–1 rad–2, respectively. ‡The welldefined residues (having both AOPφ and AOPψ ≥ 0.9) were residues TyrβA2–GluαA6, GlnαA8–LeuαA10, ProAB5–GlyAB7, IleβB4–ArgβC7, AspCD2–AsnDE2, TyrβE2–SerBG5, +1E and +2E. Three other well-defined residues ValBG9, +4P and +6Y were excluded from the superposition as they are locally well-defined, but their positions in the protein and peptide C-terminal tails makes them poorly defined with respect to the rest of the structure. §The bindingsite residues (SH2 and peptide residues which displayed intermolecular NOEs) were ThrBC2, SerβC3, HisβD4–LysβD6, IleβE4–ThrEF1, GlyBG3–SerBG5, pY, +1E and +3I.

Quality of the NMR structure

In the family of 22 final structures, 77.8% of the backbone φ and ψ angles are well-defined with angular order parameters (AOPs; [34]) of ≥ 0.9 (94 φ angles and 88 ψ angles). Superposition of the backbone N, Cα and C′ atoms of the

well-defined residues (having both φ and ψ well-defined) on their mean coordinates gave a root mean squared deviation (rmsd) of 0.57 ± 0.08 Å for their backbone N, Cα, C´ atoms and 0.98 ± 0.09 Å for heavy atoms. A superposition of the backbone atoms of the peptide and protein residues for which intermolecular NOEs were detected (pY, +1E, +3I, ThrBC2, SerβC3, HisβD4–LysβD6, IleβE4, ThrEF1 and GlyBG3–SerBG5, where +1E and +3I refers to the residues in the +1 and +3 positions immediately C-terminal to pY) on their mean coordinates gave an rmsd of 0.36 ± 0.09 Å for the backbone atoms and 0.73 ± 0.16 Å for their heavy atoms. Plots of the residue-by-residue AOPs and rmsds are shown in Figure 4. Using the Ramachandran regions defined by Morris et al. [35], the angular averages for non-glycine/proline residues (100 of the 117 residues in the complex) place 67 residues in the most favoured regions (A, B and L), 30 in the additional allowed regions (a, b, l and p) and three in the generously allowed regions (~a, ~b, ~l and ~p). No angular average values occur in the disallowed regions of the Ramachandran plot. Peptide-binding interaction

The peptide-binding interaction is of the classical twopronged-plug type [13], with the pY and +3I sidechains inserted into two discrete binding pockets on either side of the main β-sheet structure. The intervening peptide between pY and +3I is in a largely extended conformation and lies orthogonal to the β sheet. The pY sidechain sits in a pocket, the sides of which are formed by residues SerβB7, ThrBC2, SerβC3, HisβD4, TyrβD5 and LysβD6. Intermolecular NOEs were observed between the pY residue and ThrBC2, SerβC3, HisβD4, TyrβD5 and LysβD6. The guanidinium group of ArgβB5 is located at the bottom of the pocket and forms salt bridges with the pY phosphate group in 64% of the final structures; this interaction is well-formed in these structures, despite the fact that no NOEs were detected between these two residues. In the X-ray structures of Src SH2–peptide complexes, ArgαA2 has also been implicated in a guanidino– aromatic interaction with the pY ring, and in a salt-bridge formation with the pY phosphate oxygens [9–10]. NMR spectroscopic studies on the Src SH2–pYEEI complex, however, did not reveal evidence for this interaction in solution [12]. In the Fyn SH2 structures, the sidechain of ArgαA2 is poorly defined and makes no consistent ArgαA2–pY interaction in the calculated structures, however, ArgαA2 and the residues around it do show backbone NH chemical shift changes upon peptide binding. The +3I sidechain sits in a pocket on the other side of the main β sheet. At this pocket (composed of residues TyrβD5, IleβE4, ThrEF1, GlyBG3 and LeuBG4) the peptide has been described as being held between the ‘jaws’ formed by the EF and BG loops [13]. Intermolecular

Research Article Structure of a Fyn SH2–peptide complex Mulhern et al.

1317

Figure 3 NMR solution structures of the Fyn SH2–peptide complex. (a) A superposition of the 22 final structures showing the backbone N, Cα and C′ atoms of the SH2 domain and the backbone N, Cα and C′ atoms of peptide pY (green) to +3I (red) residues as well as the pY and +3I sidechains. (b) A schematic representation of the energy minimised average coordinates, showing the location of the secondary structure elements (see text). Colour scheme: β strands, yellow; α helices, magenta; and loops, blue.This figure was prepared using the programs MOLSCRIPT [32] and Raster3D [62,63].

NOEs were observed between the +3I residue and TyrβD5, IleβE4, ThrEF1 GlyBG3, LeuBG4 and SerBG5. Although intermolecular NOEs were observed between the +1E residue and HisβD4 and TyrβD5, no intermolecular NOEs were detected involving +2E. Conformational heterogeneity

For nine residues, an alternate set of minor peaks were also assigned. These residues were located near the N and C termini, in βA (TrpβA1, TyrβA2 and PheβA3), the AB loop (GlyAB1) and in αB (LeuαB2 and GlnαB3). The minor sets of resonances were, on average, two thirds as intense as the major form. The source of the multiple conformations was identified as a cis–trans isomerization of the ValβG1– ProβG2 peptide bond near the C terminus of the molecule. In the major (trans) form, the N and C-terminal regions show relatively few structurally significant NOEs and, as a result, are disordered in the final structures. In the minor (cis) form, however, NOEs could be detected between SerβG3 and TyrβA2. It appears that the conformational heterogeneity is transmitted from ProβG2 via steric contacts between the C-terminal and N-terminal tails to βA and the AB loop. There is no spectral evidence for multiple conformations, as such, for LeuαB2 and GlnαB3, but it appears that they show two sets of peaks due to the different ring current shielding, they receive from TrpβA1 in its two alternative positions. Interestingly, NMR data for the SH3–SH2 module pair under identical solution conditions (TDM, unpublished results) indicate that this conformational heterogeneity is absent in the larger construct and that the single conformation in the SH3–SH2 corresponds to the more highly populated trans-isomer in the isolated SH2.

aB helix capping interaction As mentioned previously, the hydroxyl group of ThrαB1 is protected from solvent exchange. The hydroxyl hydrogen

is not hydrogen bonded itself, but is buried due to involvement of its hydroxyl oxygen in a helix capping interaction, reminiscent of the capping-box motif [36,37]. The capping box generally involves a sequence motif of the form S/T–X–X–E/Q, in which the serine or threonine is located at the N-cap position of the helix and the glutamate or glutamine is located at the N+3 position. A reciprocal hydrogen-bonding arrangement occurs between the N-cap and N+3 residues, in which the sidechain of the N-cap forms a hydrogen bond with the backbone NH of N+3 and the sidechain of the N+3 residue forms a hydrogen bond with the backbone NH of the N-cap. In the capping box, the N-cap residue adopts a characteristic backbone conformation with φ = –94 ± 15° and ψ = 167 ± 5° [36]. As a result, capping boxes also display characteristic deviations of the Cα and Cβ chemical shifts from random coil values [38]. In Fyn SH2, although the 13C chemical shifts match the expected pattern, the capping box hydrogen-bonding arrangement is somewhat distorted, with a hydrogen bond formed between the sidechain hydroxyl oxygen of the N-cap (ThrαB1) and the backbone NH of the N+2 residue (GlnαB3) rather than the N+3 position (GlnαB4). The backbone conformation of ThrαB1 reflects this distortion having φ = –143 ± 6° and ψ = 179 ± 6°. No reciprocating hydrogen bonds were detected between the sidechains of either GlnαB3 or GlnαB4 with any backbone NH groups.

Discussion Helix capping in SH2 domains

A hitherto unexplored aspect of SH2 structure is the presence of helix capping motifs on αA and αB. Inspection of SH2 sequences (Figure 5) shows that many SH2 domains contain putative capping box or capping box-like sequence motifs on both αA and αB.

1318

Structure 1997, Vol 5 No 10

Figure 4 Backbone rmsd and angular order parameters. The portions of the graphs corresponding to the peptide are lightly shaded, whereas those for the SH2 domain are more heavily shaded. (a) Backbone rmsd values. (b–d) Angular order parameters for φ, ψ and χ1, respectively. The least well-defined regions are the peptide and the protein N and C termini along with the SH2 AB loop (residues 165–171), CD loop (residues 192–198) and DE loop (residues 209–211).

8.0 (a) Rmsd (Å)

6.0 4.0 2.0 0.0 1.0 (b)

AOPφ

0.8 0.6 0.4 0.2 0.0 1.0 (c)

AOP ψ

0.8 0.6 0.4 0.2 0.0 1.0 (d) AOPχ

1

0.8 0.6 0.4 0.2 0.0 pY

+3I

150

160 170

180 190 200

210 220

230

240 Structure

Helix αA in Fyn lacks the N-cap serine or threonine required for a capping-box formation, however, many other SH2 domains have the correct sequence motif, and partly formed capping boxes can be found in Lck [13], peptide-free Src [10] and peptide-bound Src [12] in which the αA1 sidechain hydrogen bonds to the backbone NH of αA4, but the reciprocal interaction is lacking. Helix αB of Fyn has a distorted capping box in which the ThrαB1 sidechain hydrogen bonds to the backbone NH of GlnαB3, rather than GlnαB4. This type

of distorted partial-capping box can also be found in SykC [23]. The solution structure of the Src SH2–pYEEI complex [12] contains a quite different distorted capping box, in which the correct SerαB1–GluαB4 hydrogen bond is present, but the reciprocal interaction is distorted, such that the sidechain of GluαB4 hydrogen bonds to the backbone NH of PheFB1, the residue preceding αB1. The peptide-free Src SH2 structure [10] contains the GluαB4–PheFB1 hydrogen bond, but lacks the αB1–αB4 hydrogen bond.

Research Article Structure of a Fyn SH2–peptide complex Mulhern et al.

1319

Figure 5 βA WYF WYF WFF WYF WYF WYF WFF WFF WFF

AA GKI GKM KGI GKL GKL GKL KNL KDI RTI

αA TRRESERLLL GRKDAERLLL SRKDAERQLL GRKDAERQLL GRKDAERQLL GRKDAERQLL SRKDAERQLL TRKDAERQLL SRKDAERQLL

AB NPENPRG NPGNQRG APGNMLG SFGNPQG SFGNPRG SFGNPRG APGNTHG APGNSAG APMNKAG

βB TFLVRES IFLVRES SFMIRDS AFLIRES TFLIRES TFLIRES SFLIRES AFLIRES SFLIRES

BC ET TKG ET TKG ET TKG ET TKG ET TKG ET TKG ES TAG ET LKG ES NKG

zap70C

WFH WYH FFY WYH

GKI GPV GSI SSL

SREESEQIVL SRNAAEYLL SRAEAEEHLK TREEAERKLY

IGSKTNG SSG ING LAGMADG SGAQTDG

KFLIRAR SFLVRES LFLLRQC KFLLRPR

DN NG SYALCLLH ES SPG QRSISLRY LR SLG GYVLSLVH KE QG TYALSLIY

E E D G

PLCg1C GRB2 p85aN sypN shc

WYH WFF WYW WFH WFH

ASL GKI GDI PNI GKL

TRAQAEHMLM PRAKAEEMLS SREEVNEKLR TGVEAENLLL SRREAEALL

RVP KQR DT TRG Q

RDG HDG ADG VDG LNG

AFLVRKR AFLIRES TFLVRDA SFLARPS DFLVRES

NE PN ES APG STKMHG KS NPG TT TPG

SYAISFRA DFSLSVKF DYTLTLRK DFTLSVRR QYVLTGLQ

E G G N S

βA ABA aBA ABb ABB

AA -bB -bB AAB bbB

αA BAAAAAAAAb -AAAAAAAAA BAAAAAAAAA BAAAAAAAAA

AB BaaB-BAA-B-bB--BbBB-AbBb-

βB BBBBBBB bBBBbBB BBBBBBB BBBbBBB

BC AA BBAA bAAA BBAA Ba-

βC BBBBBBBB bbBbBBBB BBBBBBBB BbBBBBBB

(a) IA Src family Type I

src yes hck fgr fyn yrk lck lyn blk sykC abl

IB zap70N

Type II

(b) IA Type I IB

Type II

src fyn lck blk

X N X N

βC AYCLSVSD AYSLSIRD SYSLSVRD AYSLSIRD AYSLSIRD AYSLSIRD SFSLSVRD SFSLSVRD AFSLSVKD

sykC N abB -ab aaAAAAAAAA A-DDAl- BBBBBBb BB b- BBBBBBBa abl N aBa --B BAAAAAaAb Ab- bA- BBBBBBB ba B-- BBBBBBBb

plcg1C GRB2 p85aN Nsyp shc

N N N X X

aBB ABa ABA ABB ABa

AbB -lB -lb -AB -bB

BAaAAAAAAA -AAAAAAAAA BAAAAAaAAA B-AAAAAAAA BAAAAAAAB

ABABA bB AAa

lBBBbbBBBb-

BBBBBBb BBBBBBB BbbBBBB BBBBB-B BBbBBBb

ab -b AA b-bAAaaAA b-aA b--

bbBBBBBB BBBBBBBB abBBBBBb BBBBBBBB bBBBB-Bb

βD NVKHYKI NVKHYKI TVKHYKI HVKHYKI HVKHYKI HVKHYKI VVKHYKI VIKHYKI VVKHYKI

βD’ RKL RKL RTL RKL RKL RKL RNL RSL RSL

DE DSG DNG DNG DMG DNG DSG DNG DNG DNG

βE GFYI GYYI GFYI GYYI GYYI GYYI GFYI GYYI GYYI

EF TSR TTR SPR TTR TTR TTR SPR SPR SPR

βF TQF AQF STF VQF AQF AQF ITF ITF ITF

FB αB N SLQQLVAYYSKH D TLQKLVKHYTEH S TLQELVDHYKKG N SVQELVQHYMEV E TLQQLVQHYSER D TIQQLVQHYIER P GLHELVRHYTNA P CISDMIKHYQKQ P TLQALVQHYSKK

ADGL ADGL NDGL NDGL AAGL AAGL SDGL ADGL GDGL

CHRLT CHKLT CQKLS CNLLI CCRLV CCRLA CTRLS CRRLE CQKLT

TVC TVC VPC APC VPC VPC RPC KAC LPC

G G V K

KVLHYRI RVYHYRI RFHHFPI TVYHYLI

DKD NTA ERQ SQD

KTG SDG LNG KAG

KLSI KLYV TYAI KYCI

PEG SSE AGG PEG

KKF SRF KAH TKF

D N C D

TLWQLVEHYSYK TLAELVHHHSTV CPAELCEFYSRD TLWQLVEYLKLK

ADGL ADGL PDGL ADGL

LRVLT ITTLH PCNLR IYCLK

VPC YPA KPC EAC

G N G G G

KIKHCRV DVQHFKV NNKLIKI AVTHIKI QPKHLLL

QQE GQ LRD GAG FHR DG QNT GD VDP EG

TVML KYFL KYGF YYDL VVRT

GN WV SDP YGG KD

SEF VKF LTF EKF HRF

D N N A E

SLVDLISYYEKH SLNELVDYHRST SVVELINHYRNE TLAELVQYYMEH SVSHLISYHMDN

PLYR KMKLR SVSRN QQIFLR SLAQYN PKLDVKLL HGQLKEKNGDVIELK HLPIIS AGSELCLQ

YPI DIE YPV YPL QPV

βD BBBBBBB bbbBBbB BBBBBBB bbBBBBB

βD’ DE BBB AABBB AABBB BDBBB AA-

βE -BBa -BBA -BBA -BBA

EF BAA BAA B-A B-A

βF FB BBB A bBb A BBB BBB -

αB BAAAAAAAAAAB BAAAAAAAAAAB -AAAAAAAAAAb BAAAAAAAAAaB

Al-B Al-b Al-B -

CD FDNAKGL WDEIRGD YDPQRGD WDQTRGD WDDMKGD WDEAKGD FDQNQGE FDPVHGD ITT QGE

CD BBAbD-B BBaaa-B BBAAA-B BBA A-B

BG

BG

a a

-

bBBBBBB bba lA- BBBB BBBBBBB BBa aa- BBBA

-A- bBB A BAAAAAAAAAAb Al-b baA bBB A BAAAAAAAAAab ba-B

a L l

a -

BBBBBBB BBBBBBB BBBBBBB BBBBBBB B-BbBBB

-A aa aBl-Aa

BBB -a BBA -aBBa aBBA -Ab Bb- b-

bBBA BBBa bB-a bBbb abBb

BBB BBB laa BBB BBb

A A b A A

bAAAAAAAAAAB BAAAAAAAAAAB BAAAaAAAAAAa BAAAAAAAAAAb bAAAAAAAAAAA

AbBBa AbbbB ABBBa lBBBa

BBB a-b B-b B-D

ABbBa B-A abBba b-B

-abA aBBDA BABAa BbBBBB bbAAAB -aBBBBba A-ABBBDa-BBBBbA lB-BBB L-BBBBBa

b-B BB b-B B-B B-b

Structure

Sequence and Ramachandran alignments for SH2 domains. (a) Sequence alignments of SH2 domains. The four N-terminal residues on αA and αB are highlighted in yellow. Where a putative capping box sequence motif (S/TXXE/Q) is present, the residues are shown in red. Residues BG1–BG4, in those SH2 domains that exhibit type-I binding, are highlighted in blue. (b) Where structures were available, the Ramachandran region occupied by each residue is shown. For convenience, the allowed regions, a, b, l and p, include the generously

allowed regions, ~a, ~b, ~l and ~p. The type of structure is indicated by X (X-ray crystal) or N (NMR solution). For NMR solution structures, the entire ensemble was analysed using an in-house program (ANGORDER) to yield all the dihedral angles, their respective angular averages and AOPs, the Ramachandran region occupied by each residue in the ensemble and the Ramachandran region occupied by the angular average for each residue. The less well-defined residues (those having an AOP for φ or ψ < 0.8) are shown in magenta.

It is interesting that some SH2 modules exist without a capping box, or with distorted or partially formed capping boxes. It has been suggested that helix capping interactions enhance protein stability [39] and delineate helix termination points during folding [36]. The presence of capping-box sequence motifs, but not completely formed capping-box interactions, suggests that the stabilising influences of the capping-box hydrogen bonds may act early on in the folding process, but are not required for protein stability when tertiary contacts increase as the protein approaches the native fold.

members of the Src family as well as some other SH2s, suggestive of a shared +3 hydrophobic pocket conformation. They noted that the BG-loop conformation was constrained by a hydrogen bond between the sidechain of TyrβD5 and the backbone NH of BG5. In a complex between Src and the same peptide as studied in our work presented here, Waksman et al. [10] noted that the Cα atom of GlyBG3 makes close contact with the +3I sidechain and that substitution of leucine for GlyBG3 and a six-residue insertion in this region of the p85αN SH2 domain [18] resulted in selection for methionine at the +3 position over isoleucine.

BG-loop conformation

An alignment of SH2 sequences is shown in Figure 5. Where the coordinates were available, the Ramachandran region occupied by each non-glycine and non-proline residue is also shown. The sequences and Ramachandran alignments have been grouped according their peptide ligand sequence preferences [8]. When analysing the peptide binding of Lck, Eck et al. [13] highlighted the strict conservation of tyrosine at βD5 and a characteristic Asp–Gly–Leu sequence for BG2–BG4 in

Both the tyrosine residue at βD5 and the βD5–BG5 hydrogen bond are present in Fyn, however, Fyn does not have the Asp–Gly–Leu sequence in the BG loop. Both Fyn and Yrk have alanine, not aspartate at position BG2; the Ramachandran alignment shows that in SH2s which display type IA specificities (and SykC, which displays type IB specificity) it is not necessarily the residue type at BG2, but the presence of a positive φ angle that is conserved. An analysis of these structures shows that residues BG1–BG4 form a type-I′ turn with the concomitant hydrogen bond

1320

Structure 1997, Vol 5 No 10

Figure 6

Biological implications SH2 domains are found in a variety of signal transduction proteins. The ability of SH2 domains to bind polypeptide sequences containing phosphorylated tyrosine residues allows them to be involved in both the recognition of target molecules and the regulation of intramolecular kinase activity. We have determined the NMR solution structure of the SH2 domain from the Src family tyrosine kinase Fyn in complex with a phosphotyrosine-containing peptide. The structure and binding properties of the Fyn SH2 domain are similar to those of other members of the Src family.

Comparison of BG-loop conformation, and relative peptide and TyrβD5 positions. Fyn (orange), Blk (red) [17], Lck (yellow) [13], Src (green) [10], and SykC (cyan) [23]. The backbone N, Cα and C′ atoms are shown for residues BG1–BG4, TyrβD5 and peptide residues pY to +3I. Also displayed are the non-hydrogen sidechain atoms of TyrβD5, pY and +3I, the N and HN atoms of BG5, the HN of BG4 and the carbonyl O of BG1. The BG1–BG4 and TyrβD5–BG5 hydrogen bonds are shown as dashed lines for Src. Residue BG5 is not shown for Blk as the orientation of this residue is poorly defined in the solution structures. Structures were superimposed over the backbone atoms of the displayed residues.

between the NH of LeuBG4 and the backbone carbonyl of BG1. A strict requirement for the formation of a type-I′ turn is a glycine residue at the third position in the turn [40]. Figure 6 shows a superposition of BG-loop residues, residue TyrβD5 and, where present, the bound peptide for Fyn, Lck [13], Src [10], Blk [17] and SykC [23]. All of the molecules have very similar BG-loop conformations. It therefore would appear that another important determinant of the formation of a hydrophobic +3 pocket, and thus type-I binding (and specifically type-1A binding), is a glycine residue at position BG3, such that a hydrogen bond can form between BG1 and BG4 as part of a type-I′ turn. This hydrogen bond, in conjunction with the βD5–BG5 hydrogen bond, constrains the conformation and positioning of the BG loop with respect to the other peptide binding site residues. An interesting exception to this is the Abl SH2 domain. It has a glycine residue at BG3 and displays type-1B specificity, selecting for proline at the +3 position. Although the solution structure of the Abl SH2 domain does not show a type-I′ turn in the BG loop, the polypeptide chain follows a very similar path to the other examples above and it does have a hydrophobic +3 pocket, rather than a groove. The availability of more structures of SH2 domains that display variants of type-1B binding should be able to resolve why these differences exist and further extend our ability to predict the binding preferences of new SH2 domains.

In recent years, structures have been determined of a wide range of SH2 domains and this large number of SH2 domain structures provides a useful pool of structural data from which sequence diversity can be analysed in terms of subtle structural differences. We have highlighted two aspects of SH2 domain structure, namedly, helix capping interactions and binding-loop conformation. Many SH2 domains appear to have capping-box sequence motifs, but not completely formed capping-box interactions; it is therefore apparent that the role of helix capping interactions in SH2 domain folding and stability warrants further investigation. Our work also suggests that a significant marker of SH2 specificity is the presence, or otherwise, of a glycine residue at position three in the BG loop (BG3). In SH2 domains which display type-I two-pronged-plug peptide binding, the conformation of the BG loop, and thus the positioning of a critical leucine residue that makes major contacts with the hydrophobic ‘prong’ of the binding sequence, is determined by the presence of a tyrosine residue at position five in β-sheet D and a glycine at BG3. Intimate knowledge of the determinants of SH2 domain stability and substrate specificity will facilitate the design of specific SH2 ligands or SH2 containing molecules for therapeutic purposes.

Materials and methods Expression vector construction The pGEX-2T plasmid encoding residues 82–252 of human Fyn (nucleotides 823–1323 [41]) was a kind gift of CE Rudd (Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA). This vector provided the template for amplification of DNA encoding amino acids 82–248 using Tli polymerase (Promega). The 5′ primer (5′-AATTAATCATATGACAGGAGTGACACTCTTTGTGGC) introduced a NdeI site and an ATG start codon (underlined) in frame with codon 82 of Fyn. The 3′ primer (5′-AAGAATTCTTATTTGTGAGAGGGAACTACTAGGCGGGAGGAGAGACC) mutated Cys239, Cys240 and Cys246 to serine residues, and introduced a TAA stop codon followed by an EcoRI restriction site. The PCR product was digested with NdeI and EcoRI, subcloned into the corresponding sites in pRK172 [42], and confirmed as having no amplification errors. This vector (denoted fynIII6), which expresses the SH3–linker–SH2 polypeptide, was subsequently used to generate an SH2 construct (expressing residues 143–248 of Fyn) by deletion mutagenesis. The mutagenesis was performed using the Transformer site-directed mutagenesis kit (Clontech), according to the manufacturer’s instructions. In this procedure, the

Research Article Structure of a Fyn SH2–peptide complex Mulhern et al.

selection primer (5′-GGTTTCCCCCTAGGAATAATTTTGTT) changed a unique XbaI site in the pRK172 cloning cassette into an AvrII site, and the mutation primer (5′-GAAGGAGATATACATATGTCTATCCAGGCAGAAG) deleted the sequence encoding amino acids 82–142 leaving the ATG start site (underlined) in frame with the codon for Ser143. The mutant plasmid, designated Fyn4a-2, was transformed into the E. coli expression strain BL21(DE3)pLysS [43].

Protein expression and purification Unlabelled protein was produced from cell culture in Luria broth. The production of 15N- and 15N/13C-labelled protein was achieved by culturing the E. coli cells in vitamin supplemented M8 minimal media with 15NH Cl as the sole nitrogen source, or 15NH Cl and 13C -glucose as 4 4 6 the sole nitrogen and carbon sources, respectively. Gene expression was induced by the addition of 0.1 mM isopropyl-D-thiogalactopyranoside to the culture. Purification of both labelled and unlabelled protein was carried out as follows. Cells were harvested by centrifugation at 4°C, 4 h after the induction of expression, then stored at –20°C. For processing, the cells were resuspended in TTBS (50 mM Tris-HCl, pH 8.5, 150 mM NaCl with 0.1% (v/v) Triton X-100) at a ratio of 10 ml TTBS per gram of cells. To prevent protein degradation by E. coli protease enzymes, phenylmethyl-sulfonylflouride was added to a final concentration of 2 mM. The cell suspension was sonicated on ice (6 ± 30s) to effect lysis of the cells. Cell debris was removed by centrifugation (30 min, 20,000 g) at 4°C and the pellet discarded. The supernatant was passed through a column of phosphotyrosine agarose beads (Sigma). After extensive washing of the resin with TTBS followed by TBS (TTBS without Triton X-100), the bound Fyn SH2 was eluted from the column by a wash of TBS containing additional NaCl to 1.15 M. Analysis of the eluted material by SDS-PAGE and mass spectrometry (for unlabelled protein, 12281.9 Da compared with a theoretical value of 12278.7 Da) indicated that no contaminating proteins were present in the eluate, the affinity purification step produced clean Fyn SH2 that did not require further purification.

Peptide production The peptide EPQpYEEIPIYL, residues 321–331 from the hamster middle T antigen, which has been shown to bind to Fyn with high affinity was prepared by solid-state synthesis and purified by reverse-phase HPLC as described previously [33].

NMR sample preparation Initial NMR samples containing only SH2 displayed broad line widths [33] and were found to precipitate rapidly at 30°C and over time at 4°C (results not shown). By comparison, SH2 samples containing the peptide showed reduced line widths and were indefinitely stable in solution. Hence, all subsequent NMR samples contained peptide and were prepared according to the following protocol. The SH2 eluted from the pY–agarose was concentrated by ultrafiltration over YM3 membranes (Amicon) to a protein concentration of approximately 50 mM and then buffer exchanged into 10 mM phosphate, 0.02% NaN3, pH 6.0. Into this solution, a 1.5 × molar excess of peptide was added and the solution concentrated using centricon concentrators (YM3) to a protein concentration of ~0.5–1 mM. At this point, further phosphate buffer was added and the sample reconcentrated, iteratively, in order to remove excess unbound peptide. Finally, D2O was added to 10% and the pH adjusted to 6.0 by small additions of 0.1 M NaOD and DCl. The protein concentrations in the resulting NMR samples were in the range 0.8–1.5 mM (measured spectrophotometrically at 280 nm using ε = 11777 M–1cm–1 which was determined from amino acid analysis).

Chemical shift assignments Spectra were recorded at 20°C at 500, 600 or 750 MHz on homebuilt spectrometers. Sequential assignment of the backbone N, NH and CαH resonances was achieved using 3D-NOESY-1H–15N-HSQC and 3D-TOCSY-1H–15N-HSQC [44,45] data recorded at 750 MHz on the 15N-labelled sample. Poor TOCSY transfer in the protein sidechains resulted in only partial sidechain 1H assignments being obtained from these data. More complete sidechain 1H assignments

1321

(> 90%) were made from analysis of 13C/15N-labelled protein by 3DCBCA(CO)NH [46], 2D-1H–13C-HSQC [47] and 3D-HCCH-TOCSY [48] spectra recorded at 500 MHz supplemented by 2D-1H-NOESY [49], 2D-1H-TOCSY [50] and 2D-1H-DQF-COSY [51] data recorded at 600 and 750 MHz. During the assignment process, non-degenerate prochiral groups were arbitrarily named such that the lower branch number was assigned to downfield resonance. 1H assignments for the bound peptide were made from 2D X-filtered TOCSY and NOESY data (described below) recorded at 500 MHz using the 13C/15Nlabelled sample. NH and sidechain amide assignments for the free SH2 were made from 2D-1H–15N-HSQC [47] data recorded on a sample containing 0.5 mM peptide and 1.5 mM SH2.

X-filtered experiments High power 1H, 15N, and 13C pulses were applied using field strengths of 27.5 kHz, 5.2 kHz, and 14.7 kHz, respectively. The selective 1H pulse used as part of the DPFGSE water suppression sequence was a 2.0 ms Gaussian 180° pulse [52]. During t1, the 1H–13C couplings were refocused using a 0.5 ms hyperbolic secant 180° pulse [53] applied with a maximum field strength of 12.5 kHz. The 1H, 15N, and 13C transmitters were positioned at 4.7 ppm, 117.5 ppm, and 40.0 ppm respectively. During acquisition, decoupling was achieved using a 1.0 kHz WALTZ-16 modulated field [54] and a 2.5 kHz GARP modulated field [55] for 15N and 13C, respectively. In the X-filtered NOESY experiment, a mixing time of 150 ms was used. DIPSI-2 [56] was applied using a field strength of 7.6 kHz for 22.8 ms in the X-filtered TOCSY experiment. For the X-filtered NOESY, a total of 160 and 1024 (complex) points were acquired in t1 and t2, respectively, using spectral widths of 8000 Hz in both dimensions; 128 transients were recorded for each FID. For the X-filtered TOCSY, a total of 320 (complex) and 1024 (complex) points were acquired in t1 and t2, respectively, using spectral widths of 8000 Hz in both dimensions; 64 transients were recorded for each FID. Prior to Fourier transformation, a mild Lorentz– Gauss transformation was applied in t2 and a cosine bell window function was applied in t1.

Experimental restraints NOEs were categorised as strong, medium and weak, and they were given distance restraint upper bounds of 2.7, 3.5 and 5.0 Å, respectively, based on relative NOE intensities for cross-peaks in residues in well-defined secondary structure elements for which estimation of actual distances was straight forward. The restraint list was based primarily on NOEs observed in the 3D-1H–15N-HSQC-NOESY and 2D1H-NOESY spectra, both recorded at 750 MHz. The values of 3JNα coupling constants were measured from the 2D1H–15N-HMQC-J [57] spectrum recorded at 600 MHz on the 15Nlabelled sample. For values of 3JNα < 5 Hz, φ was restrained to –60 ± 30°; for values of 5 Hz < 3JNα < 5.5 Hz, φ was restrained to –60 ± 40°; and for values of 3JNα > 8 Hz, φ was restrained to –120 ± 40°. Relative rates of backbone NH exchange were estimated from a series of 2D-1H–15N-HSQC spectra recorded at 600 MHz on a sample lyophilised from 90% 1H2O/10% D2O then redissolved in 100% D2O. Hydrogen bond distance restraints were imposed for slowly exchanging NH groups for which a single hydrogen bond acceptor was identified in > 60% of structures in preliminary rounds of structure calculation. The distance restraints used were 1.5–2.3 Å between the hydrogen bond donor and acceptor, and 2.3–3.3 Å between the hydrogen bond donor antecedent and the acceptor.

Structure calculations Structures were generated in X-PLOR [58] using a dynamical simulated annealing protocol based on that of Nilges and co-workers [59], which includes the use of floating stereospecific assignment [60] and ambiguous distance restraints [61]. During preliminary rounds of structure calculations, the energy cut-off for acceptance was set so that the best 20% of structures were retained for analysis of hydrogen bonds

1322

Structure 1997, Vol 5 No 10

and resolution of spectral ambiguity. For the final structures, a family of 200 initial structures were generated from templates with randomised backbone φ and ψ angles by 19.5 ps of dynamics at 2000K followed by annealing to 100K over 10.5 ps. These structures were then subjected to a second round of the high temperature dynamics and simulated annealing, and then were further refined by a final round of simulated annealing where structures were cooled from 2000K to 100K over 30 ps. The 22 sets of coordinates with the lowest overall energy were selected as the final structures. The use of floating stereospecific assignments identified 3 glycine α-methylene groups (33%), 14 β-methylene groups (32%), 3 γ-methylene groups (37%), 4 leucine methyl pairs (40%) and 1 valine methyl pair (25%) that adopted a single orientation consistent with the NOE data. The values in parentheses indicate what proportion of the nondegenerate methylene and methyl pairs this represents in each case.

Accession numbers The coordinates for the family of NMR structures and the energy minimised average have been deposited with the Brookhaven Protein Databank and have accession codes 1AOT and 1AOU, respectively.

Acknowledgements We thank Jonathan Boyd for assistance with the running of the triple resonance experiments, Kristy Downing for useful advice and discussion regarding structure calculation strategies, Tony Willis for amino acid analysis and Carol Robinson for mass spectrometry. Oxford Centre for Molecular Sciences is funded by BBSRC, MRC and EPSRC. TDM and CJM are funded by a LINK grant with Zeneca. GLS is funded by the Wellcome Trust and AJD is supported by the ARC.

References 1. Brown, M.T. & Cooper, J.A. (1996). Regulation, substrates and function of Src. Biochem. Biophys. Acta 1287, 121–149. 2. Xu, W., Harrison, S.C. & Eck, M.J. (1997). Three-dimensional structure of the tyrosine kinase c-Src. Nature 385, 595–602. 3. Sicheri, F. Moarefi, I. & Kuriyan, J. (1997). Crystal structure of the Src family tyrosine kinase Hck. Nature 385, 602–609. 4. Moarefi, I., et al., & Miller, W.T. (1997). Activation of the Src-family tyrosine kinase Hck by SH3 domain displacement. Nature 385, 650–653. 5. Veillette. A. & Davidson, D. (1992). Src-related protein tyrosine kinases and T-cell receptor signalling. Trends Genet. 8, 61–66. 6. Cobb, B.S., Schaller, M.D., Leu, T-H. & Parsons, J.T. (1994). Stable association of pp60src and pp59fyn with the focal adhesionassociated protein tyrosine kinase, pp125FAK. Mol. Cell Biol. 14, 147–155. 7. Shenoy-Scaria, A.M., Gauen, L.K.T., Kwong, J., Shaw, A.S. & Lublin, D.M. (1993). Palmitoylation of an amino-terminal cysteine motif of protein-tyrosine kinases P56Lck and P59Fyn mediates interaction with glycoyl-phosphatidylinositol-anchored proteins. Mol. Cell Biol. 13, 6385–6392. 8. Songyang, Z., et al., & Cantley, L.C. (1993). SH2 domains recognize specific phosphopeptide sequences Cell 72, 767–778. 9. Waksman, G., et al., & Kuriyan, J. (1992). Crystal structure of the phosphotyrosine recognition domain SH2 of v-src complexed with tyrosine-phosphorylated peptides. Nature 358, 646–653. 10. Waksman, G., Shoelson, S.E., Pant, N., Cowburn, D. & Kuriyan, J. (1993). Binding of a high affinity phosphotyrosyl peptide to the src SH2 domain: crystal structures of the complexed and peptide-free forms. Cell 72, 779–790. 11. Gilmer, T., et al., & Berman, J. (1994). Peptide inhibitors of src SH3–SH2-phosphoprotein interactions. J. Biol. Chem. 209, 31711–31719. 12. Xu, R.X., Word, M., Davis, D.G., Rink, M.J., Willard, D.H. & Gampe, R.T. (1995). Solution structure of the human pp60c-src SH2 domain complexed with a phosphorylated tyrosine pentapeptide. Biochemistry 34, 2107–2121. 13. Eck, M.J., Shoelson, S.E. & Harrison, S.C. (1993). Recognition of a high-affinity phosphotyrosyl peptide by the Src homology-2 domain of p56lck. Nature 362, 87–91. 14. Eck, M.J., Atwell, S.K., Shoelson, S.E. & Harrison, S.C. (1994). Structure of the regulatory domains of the Src-family tyrosine kinase Lck. Nature 368, 764–769.

15. Mikol, V., Baumann, G., Keller, T.H., Manning, U. & Zurini, M.G.M. (1995). The crystal structure of the SH2 domain of p56lck complexed with two phosphonopeptides suggest a gated peptide binding site. J. Mol. Biol. 246, 344–355. 16. Tong, L., Warren, T.C., King, J., Betageri, R., Rose, J. & Jakes, S. (1996). Crystal structure of the human p56lck SH2 domain in complex with two short phosphotyrosyl peptides at 1.0 Å and 1.8 Å resolution. J. Mol. Biol. 256, 601–610. 17. Metzler, W.J., Leitings, B., Pryor, K., Mueller, L. & Bennett, T.F. (1996). The three-dimensional solution structure of the SH2 domain from P56blk kinase. Biochemistry, 35, 6201–6211. 18. Booker, G.W., et al., & Campbell, I.D. (1992). Structure of the SH2 domain of the p85α subunit of the phosphatidylinosityl-3-OH kinase. Nature 358, 684–686. 19. Pascal, S.M., et al., & Forman-Kay, J.D. (1994). Nuclear magnetic resonance structure of an SH2 domain of phospholipase C-γ1 complexed with a high affinity binding peptide. Cell 77, 461–472. 20. Overduin, M., Rios, C.B., Mayer, B.J., Baltimore, D. & Cowburn, D. (1992). 3-dimensional solution structure of the src homology-2 domain of c-Abl. Cell 70, 697–704. 21. Maignan, S., Guilloteau, J-P., Fromage, N., Arnoux, B., Becquart, J. & Ducruix, A. (1995). Crystal structure of the mammalian Grb2 adaptor. Science 268, 291–292. 22. Hatada, M.H., et al., & Karas, J.L. (1995). Molecular basis for interaction of the protein tyrosine kinase ZAP-70 with the T-cell receptor. Nature 377, 32–38. 23. Narula, S.S., et al., & Dalgarno, D.C. (1995). Solution structure of the C-terminal SH2 domain of the human tyrosine kinase SykC complexed with a phosphotyrosine pentapeptide. Structure 3, 1061–1073. 24. Lee, C-H., et al., & Kuriyan, J. (1994). Crystal structures of peptide complexes of the amino-terminal SH2 domain of the Syp tyrosine phosphatase. Structure 2, 423–438. 25. Zhou, M-M., et al., & Fesik, S.W. (1995). Solution structure of the Shc SH2 domain complexed with a tyrosine-phosphorylated peptide from the T-cell receptor. Proc. Natl. Acad. Sci. USA 92, 7784–7788. 26. Courtneidge, S.A., Goutbroze, L., Cartwright, A., Heber, A., Scherneck, S. & Feuteun, J. (1991). Identification and characterization of the hamster polyomavirus middle T antigen. J. Virol. 65, 3301–3308. 27. Otting, G. & Wüthrich, K. (1989). Extended heteronuclear editing of 2D 1H NMR spectra of isotope-labelled proteins, using the X(ω1, ω2) double half filter. J. Magn. Reson. 85, 586–594. 28. Rutger, H.A., Hilbers, C.W., Konings, R.N.H. & Hallenga, K. (1995). A 13C double-filtered NOESY with strongly reduced artefacts and improved sensitivity. J. Biomol. NMR 5, 427–432. 29. Gemmecker, G., Olejniczak, E.T. & Fesik, S.W. (1992). An improved method for selectively observing protons attached to 12C in the presence of 1H–13C spin pairs. J. Magn. Reson. 96, 199–204. 30. Hwang, T. & Shaka, A.J. (1995). Water suppression that works. Excitation sculpting using arbitrary waveforms and pulsed field gradients. J. Magn. Reson. A112, 275–279. 31. Levitt, M.H. (1986). Composite Pulses. Prog. NMR Spectrosc. 18, 61–122. 32. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Cryst. 24, 946–950. 33. Pintar, A., Hensmann, M., Jumel, K., Pitkeathly, M., Harding, S.E. & Campbell, I. D. (1996). Solution studies of the SH2 domain from the Fyn tyrosine kinase: secondary structure, backbone dynamics and protein association. Eur. Biophys. J. 24, 371–380. 34. Hyberts, S.G., Goldberg, M.S., Havel, T.F. & Wagner, G. (1992). The solution structure of eglin C based on the measurement of many NOEs and coupling constants and its comparison with X-ray structures. Protein Sci. 1, 736–751. 35. Morris, A.L., MacArthur, M.W., Hutchinson, E.G. & Thornton J.M. (1992). Stereochemical quality of protein structure coordinates. Proteins 12, 345–364. 36. Harper, E.T. & Rose, G.D. (1993). Helix stop signals in proteins and peptides: the capping box. Biochemistry 32, 7605–7609. 37. Dasgupta, S. & Bell, J.A. (1993). Design of helix ends: amino acid preferences, hydrogen bonding and electrostatic interactions. Int. J. Peptide Protein Res. 41, 499–511. 38. Gronenborn, A.M. & Clore, G. M. (1994). Identification of N-terminal helix capping boxes by means of 13C chemical shifts. J. Biomol. NMR 4, 455–458. 39. Serrano, L., Sancho, J., Hirshberg, M. & Fersht, A.R. (1992). α-Helix stability in proteins. I. Empirical correlations concerning substitution of side-chains at the N and C-caps and the replacement of alanine by glycine at solvent exposed surfaces. J. Mol. Biol. 227, 544–559.

Research Article Structure of a Fyn SH2–peptide complex Mulhern et al.

40. Sibanda, B.L., Blundell, T.L. & Thornton, J.M. (1989). Conformation of beta-hairpins in protein structures. A systematic classification with applications to modelling by homology, electron density fitting and protein engineering. J. Mol. Biol. 206, 759–777. 41. Semba, K., et al., & Toyashima, K. (1986), yes-related protooncogene, syn, belongs to the protein-tyrosine kinase family. Proc. Natl. Acad. Sci. USA 83, 5459–5463. 42. McLeod, M., Stein, M. & Beach, D. (1987). The product of the mei3+ gene, expressed under the control of the mating-type locus, induces meiosis and sporulation in fission yeast. EMBO J. 6, 729–736. 43. Studier, F.W., Rosenberg, A.H., Dunn, J.J. & Dubendorff, J.W. (1990). Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60–89. 44. Marion, D., et al., & Clore, G.M. (1989). Overcoming the overlap problem in the assignment of 1H-NMR spectra of larger proteins by use of 3-dimensional heteronuclear 1H–15N hartman hahn multiple quantum coherence and nuclear overhauser multiple quantum coherence spectroscopy — application to interleukin-1β. Biochemistry 28, 6150–6156. 45. Zuiderweg, M., McIntosh, L.P., Dahlquist, F.W. & Fesik, S.W. (1990). 3-dimensional 13C-resolved protein NOE spectroscopy of uniformly 13C-labelled proteins for the NMR assignment and structure determination of larger molecules. J. Magn. Reson. 86, 210–216. 46. Grzeisk, S. & Bax, A. (1992). Correlating backbone amide and sidechain resonances in larger proteins by multiple relayed triple resonance NMR. J. Am. Chem. Soc. 114, 6291–6293. 47. Brühwiler, D. & Wagner, G. (1986). Selective excitation of 1H resonances coupled to 13C — hetero COSY and relay experiments with 1H detection for a protein. J. Magn. Reson. 69, 546–551. 48. Fesik, S.W., Eaton, H.L., Olejniczak, E.T., Zuiderweg, E.R.P., McIntosh, L.P. & Dahlquist, F.W. (1990). 2D and 3D NMR spectroscopy employing 13C–13C magnetisation transfer by isotropic mixing – spin system identification in large proteins. J. Am. Chem. Soc. 112, 886–888. 49. Jeener, J., Meier, B.H., Bachmann, P. & Ernst, R.R. (1979). Investigation of exchange processes by two-dimensional NMR spectroscopy. J. Chem. Phys. 71, 4546–4553. 50. Braunschweiler, L. & Ernst, R.R. (1983). Coherence transfer by isotropic mixing: application to proton correlation spectroscopy. J. Magn. Reson. 53, 521–528. 51. Rance, M., Sørensen, O.W., Bodenhausen, G., Wagner, G., Ernst, R.R. & Wüthrich, K. (1983). Improved spectral resolution in COSY 1H-NMR spectra of proteins via double quantum filtering. Biochem. Biophys. Res. Commun. 117, 479–485. 52. Bauer, C., Freeman, R., Frenkiel, T., Keeler, J. & Shaka, A.J. (1984). Gaussian Pulses. J. Magn. Reson. 58, 442–457. 53. Silver, M.S., Joseph, R.I. & Holt, D.I. (1984). Highly selective π/2 and π pulse generation. J. Magn. Reson. 59, 347–351. 54. Shaka, A.J., Keeler, J., Frenkiel, T. & Freeman, R. (1983). An improved sequence for broadband decoupling: WALTZ-16. J. Magn. Reson. 52, 335–338. 55. Shaka, A.J., Baker, P.B. & Freeman, R. (1985). Computer-optimised decoupling scheme for wideband applications and low-level operation. J. Magn. Reson. 64, 547–552. 56. Shaka, A.J., Lee, C.J. & Pines, A. (1988). Iterative schemes for bilinear operators; application to spin decoupling. J. Magn. Reson. 77, 274–293. 57. Kay, L.E. & Bax, A. (1990). New methods for the measurement of NH–CαH coupling constants in 15N-labelled proteins. J. Magn. Reson. 86, 110–126. 58. Brünger, A.T. (1992). X-PLOR Version 3.1. A system for X-ray crystallography and NMR. Yale University Press, New Haven, CT. 59. Nilges, M., Gronenborn, A.M., Brünger, A.T. & Clore, G.M. (1991). Determination of three-dimensional structures of proteins by simulated annealing with interproton distance restraints: application to crambin, potato carboxypeptidease inhibitor and barley serine proteinase inhibitor 2. Protein Eng. 2, 27–38. 60. Holak, T., Nilges, M. & Oschkinat, H. (1989). Improved strategies for the determination of protein structures from NMR data: the solution structure of acyl carrier protein. FEBS Lett. 242, 649–654. 61. Nilges, M. (1995). Calculation of protein structures with ambiguous distance restraints: automated assignment of ambiguous NOE crosspeaks and disulphide connectivities. J. Mol. Biol. 245, 645–660. 62. Merrit, E.A. & Murphy, M. (1994). Raster3D Version 2.0: a program for photorealistic and schematic plots of protein structures. Acta Cryst. D 50, 869–873. 63. Bacon, D.J. & Anderson, W.F. (1988). A fast algorithm for rendering space-filling molecule pictures. J. Mol. Graphics 6, 219–220.

1323