Ligand-Induced Strain in Hydrogen Bonds of the c-Src SH3 Domain Detected by NMR

doi:10.1006/jmbi.2000.4274 available online at http://www.idealibrary.com on

J. Mol. Biol. (2000) 304, 497±505

COMMUNICATION

Ligand-Induced Strain in Hydrogen Bonds of the c-Src SH3 Domain Detected by NMR Florence Cordier1, Chunyu Wang2, Stephan Grzesiek1* and Linda K. Nicholson2* 1

Biozentrum der UniversitaÈt Basel, Klingelbergstrasse 70 Basel, Switzerland CH-4058 2

Department of Molecular Biology and Genetics, Cornell University, Ithaca, NY 14853, USA

Changes in the molecular conformation of proteins can result from a variety of perturbations, and can play crucial roles in the regulation of biological activity. A new solution NMR method has been applied to monitor ligand-induced changes in hydrogen bond geometry in the chicken c-Src SH3 domain. The structural response of this domain to ligand binding has been investigated by measuring trans-hydrogen bond 15 N-13C0 scalar couplings in the free state and when bound to the high af®nity class I ligand RLP2, containing residues RALPPLPRY. A comparison between hydrogen bonds in high resolution X-ray structures of this domain and those observed via h3JNC0 couplings in solution shows remarkable agreement. Two backbone-to-side-chain hydrogen bonds are observed in solution, and each appears to play a role in stabilization of loop structure. Reproducible ligand-induced changes in trans-hydrogen bond scalar couplings are observed across the domain that translate into Ê . The changes in hydrogen bond length ranging between 0.02 to 0.12 A observed changes can be rationalized by an induced ®t mechanism in which hydrogen bonds across the protein participate in a compensatory response to forces imparted at the protein-ligand interface. Upon ligand binding, mutual intercalation of the two Leu-Pro segments of the ligand between three aromatic side-chains protruding from the SH3 surface wedges apart secondary structural elements within the SH3 domain. This disruption is transmitted in a domino-like effect across the domain through networks of hydrogen bonded peptide planes. The unprecedented resolution obtained demonstrates the ability to characterize subtle structural rearrangements within a protein upon perturbation, and represents a new step in the endeavor to understand how hydrogen bonds contribute to the stabilization and function of biological macromolecules. # 2000 Academic Press

*Corresponding authors

Keywords: hydrogen bond; NMR; SH3; ligand binding; trans-hydrogen bond J coupling

Abbreviations used: SH3, Src homology domain 3; SH2, Src homology domain 2; RLP2, proline-rich peptide ligand for SH3, with the sequence RALPPLPRY; H-bond, hydrogen bond; h3JNC0 , three bond transhydrogen bond scalar coupling constant between amide nitrogen and carbonyl carbon nuclei; HNCO, NMR experiment which correlates chemical shifts of amide protons and nitrogen atoms with scalar coupled carbonyl carbon atoms; LINK32, the linker region between the SH3 and SH2 domains in c-Src and Src family members; GST, glutathione S-transferase. E-mail addresses of the corresponding authors: [email protected]; [email protected] 0022-2836/00/040497±9 $35.00/0

Characterization of conformational changes in atomic detail has primarily been accomplished through comparison of high resolution X-ray structures for proteins crystallized in different forms, such as with and without ligand. Such comparisons are limited by the atomic coordinate Ê precision, typically on the order of 0.1 to 0.3 A Ê resolution structure.1 The higher resfor a 1.5 A olution of X-ray structures versus standard solution NMR structures comes with the caveat that interpretation of detected changes is complicated by in¯uences of crystal packing forces, particularly in loops and turns at the protein surface. Hence, a solution-based technique with # 2000 Academic Press

498

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

Ê would offer unpreceprecision exceeding 0.1 A dented resolution for detecting perturbationinduced conformational changes in an environment free of crystal packing artifacts. The direct observation of hydrogen bonds (Hbonds) in proteins is now possible through a recently developed solution NMR technique which uniquely identi®es H-bond partners by trans-Hbond 15N-13C0 scalar couplings (h3JNC0 ).2 ± 4 These scalar couplings arise from electron-electron interactions across the hydrogen bridge and re¯ect a correlation of electronic wave functions of the donor and acceptor groups. Although these electronic interactions undoubtedly depend on a complete geometric description of the H-bond including H-bond angles,5 the N


O or H


O distance is recognized as a predominant factor determining the size of h3JNC0 -couplings in regular protein secondary structure elements.2,6 An exponential dependence of h3JNC0 on the N


O distance has recently been established.6 Hence, measurement of H-bond scalar coupling constants not only provides de®nitive identi®cation of paired H-bond donors and acceptors, but should also enable characterization of changes in H-bond lengths independent of X-ray structure data. Here, changes in h3JNC0 for the chicken c-Src Src homology domain 3 (SH3) upon binding the class I proline-rich ligand RLP2 (RALPPLPRY)7 have been investigated. The small b-sandwich SH3 fold (Figure 1(a)) has been used extensively for studying protein stability and folding,8 ± 10 and for probing the structural determinants for ligand speci®city.11 ± 13 This domain is employed herein to investigate the structural response of a protein to ligand binding. Reproducible changes in measured h3 JNC0 values of up to 0.3 Hz were observed that translate into changes in H-bond length of up to Ê . This ultra-high resolution allows evaluation 0.12 A of subtle ligand-induced structural changes, providing insight on the propagation of binding interactions across the domain. This is the ®rst application of this NMR technique to examine such ligand-induced structural changes at picometer resolution. Figure 1. Observed H-bonds for Src SH3 in solution. (a) Ribbon diagram of the Src SH3 domain (constructed using PDB accession code 2src). A high degree of twist of the b-meander (b1-b2-b3-b4) is facilitated by the diverging turn. The two faces of the resulting barrel-like bsandwich are composed of a small sheet formed by b5b1-b2, and a larger sheet formed by b2-b3-b4. (b) Selected region of the long-range TROSY-HNCO spectrum obtained for the unliganded c-Src SH3 domain. Crosspeaks marked as Resi/Resj are due to h3JNiC0 j H-bond scalar couplings between the 15N nucleus of residue i and 13C0 nucleus of residue j. Resonances marked by the superscript ``s'' denote incompletely suppressed sequential correlations between the 15N nucleus of residue i and 13C0 nucleus of residue i 1, whereas resonances marked by the superscript ``i'' denote intraresidue 2JNC0

correlations. The NMR sample for unliganded SH3 (250 ml in a Shigemi microtube) contained 3.9 mM 2H, 13 C, 15N-labelled SH3, 50 mM NaH2PO4 (pH 6.6), 200 mM NaSO2, 10 mM DTT, 5 mg/ml pepstatin, 50 mM EDTA, 50 mM chloramphenicol, 10 mg/ml benzamidine, and 95 % H2O/5 % 2H2O). The SH3/RLP2 complex sample was generated by dissolving excess RLP2 ligand (4.4 mole equivalents of lyophilized peptide) in the unliganded sample and readjusting the pH to 6.6. (c) Topology diagram showing H-bonds observed in solution. Arrows point from donor to acceptor. Backbonebackbone H-bonds are implied unless an explicit sidechain is shown.

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

499

Comparison with previously determined structures The hydrogen bond partners identi®ed by measured h3JNC0 values in both free and RLP2bound SH3 are in general agreement with the backbone topology of ®ve antiparallel b-strands previously established by NMR7,14 (PDB-Code 1rlp, Figure 1(b) and (c)). Due to the limited precision of conventional NMR structures with respect to H-bonds, it is not surprising that 10 H-bonds observed by the long-range HNCO Ê in the exhibit H


O distances larger than 2.5 A 1rlp structure. A more precise comparison is Ê X-ray crystal strucpossible with the two 1.5 A tures of repressed human Src (PDB-Codes 1fmk and 2src).15,16 Similar to the SH3/RLP2 complex, the SH3 part of each crystal structure is bound through intramolecular interactions to a polyproline type II helical linker connecting the SH2 and kinase domains. The major difference between the two structures is in the activation loop of the kinase. In the 1fmk structure, the kinase activation loop is not ordered, while in the 2src structure, obtained using different crystallization conditions and in the presence of an ATP analog, this loop is ordered. It is thought that the ordered activation loop re¯ects the conformational state of Src in solution, and that the crystallization conditions for the latter structure As minimized crystal packing artifacts.16 described below, H-bonds observed in both free and ligand-bound SH3 in fact agree best with the 2src structure. A comparison between H-bonds in the 2src X-ray structure and those observed via h3JNC0 couplings reveals remarkable precision and agreement. Omitting the ¯exible N and C termini of the solution structure, residues 7 to 61 in the 2src structure contain 27 backbone-backbone H-bonds Ê . Of these with H


O distances smaller than 2.5 A 27, 22 are clearly observed in the HNCO experiment for either free or RLP2-bound SH3. Resonance overlap for four of the remaining ®ve Hbonds precludes a de®nitive distinction between presence and absence. Only the H-bond across the single turn of the 310 helix (NHY57


OP54) is clearly missing, with an upper limit for the absolute value of the h3JNC0 coupling constant of 0.21 and 0.12 Hz for free and complexed SH3, respectively. Therefore, the backbone H-bonds of the SH3/RLP2 complex in solution are essentially identical to the Hbonds observed in the 2src crystal structure with a minor difference in the 310 helical turn. Inspection of the 1fmk structure reveals several differences located close to the nSrc loop. For one Ê H-bond (NHI31


OT5), the H


O distance is 2.0 A Ê in the 2src structure. This Hin 1fmk, but 4.4 A bond is also observed by an h3JNC0 correlation in the HNCO experiment. In contrast, two backbone H-bonds across the nSrc loop (NHV32


OL41 and NHL41


ON33) and an H-bond between the amide hydrogen of T35 and the side-chain (Od1) of N33

Figure 2. Location of the two backbone-to-side-chain H-bonds observed, and conservation of features in the Hck SH3 domain. The observed NHT35-OdN33 and NHK25OdD12 H-bonds both occur on the solvent-exposed faces of loops and are accompanied by backbone H-bonds across each loop (top). Comparable loop-stabilizing interactions are present in the SH3 domain from Src family member Hck (PDB accession code 1qcf), where the RT loop H-bonds are identical but the nSrc loop backbone-to-side-chain H-bond is replaced by hydrophobic packing and electrostatic interactions between the N33- and L41-equivalent side-chains (bottom).

(Figure 2) are observed both in the HNCO experiment and in the 2src structure, but are absent in the 1fmk structure. The corresponding distances in Ê . Therefore, a the 1fmk structure all exceed 3.8 A different conformation of the nSrc loop is trapped in each of the two X-ray crystal forms, whereas the solution structure shows all four H-bonds associated with these two conformations. Examination of additional SH3 crystal structures with conserved nSrc loop sequences reveals three general cases: the H-bonds involving T32, N33, or L41 are either all present, are all absent, or the two backbone Hbonds are present and the N33- and L41-equivalent side-chains exhibit non-H-bonded electrostatic or hydrophobic interactions across the n-Src loop (Figure 2). Hence, formation of the two backbone H-bonds across the n-Src loop appears to require an additional interaction that stabilizes the confor-

500

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

Figure 3 (legend opposite)

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

mation of the N33-equivalent side-chain. Stabilization of this loop may be important for ligand binding since it contains several residues that interact directly with bound ligand.7 An additional backbone-to-side-chain H-bond is observed via a h3JNC0 correlation (NHK25


Od2 D12), which is present in both the 1fmk and 2src crystal structures. Analogous to the NHT35


Od1 N33 H-bond in the n-Src loop, this H-bond appears to stabilize the RT loop, as two backbone H-bonds (NHY13


OF23 and NHF23


OY13) are located in similar relative positions (Figure 2). It is interesting to note that D12 is a highly conserved residue in SH3 domains and makes a signi®cant energetic contribution to the ensemble of folding transition states for both Src and a-spectrin SH3 domains.10 Induced fit at the ligand binding interface The SH3/RLP2 binding interaction induces changes in h3JNC0 values for a number of H-bonds (Figure 3(a)) which clearly exceed the reproducibility of the HNCO experiment. Estimates of this reproducibility were either based on the standard deviations of three independent measurements (RLP2-bound SH3), or on error propagation from the spectral noise (free SH3). In both cases, the errors were typically between 2-5 % of the measured values, although weaker couplings generally yielded larger fractional errors (Table 1). Using the previously established exponential relationship,6 the changes observed in h3JNC0 translate into changes in H-bond length on the order of

501 2-12 pm (Figure 3(a)). The observed changes are highly correlated with regard to their location in the structure (Figure 3(b)). For example, the changes in h3JNC0 values which exceed the reproducibility error indicate that several H-bonds between strands b3 and b4 and between strands b1 and b5 lengthen upon ligand binding. In a similar way H-bonds involving residues L29 and Q30 in the N-terminal region of strand b2 become longer. In contrast, H-bonds involving residues I31 and N33 located towards the C terminus of strand b2 become shorter. These observations can be rationalized by an induced ®t mechanism for RLP2 binding (Figure 3(b) and (c)). Upon ligand binding, a key H-bond at the binding surface, NHS55


OD38, is lengthened beyond the point of detection. This Hbond connects the 310 helix to strand b3 in the free SH3 domain. A similar, somewhat smaller reduction in h3JNC0 is observed for a second H-bond at the SH3/RLP2 interface, NHL10


OY57, which connects the 310 helix to strand b1. In the SH3/ RLP2 solution structure,7 the two Leu-Pro segments of RLP2 intercalate between three bulky aromatic side-chains (Y11, Y57 and W39) which protrude like ®ngers from the SH3 surface (Figure 3(c)). These three residues are each located on or near one of the three secondary structure elements, b1, 310, and b3, that converge at the binding surface. It is apparent that the mutual intercalation between these three aromatic sidechains and the two Leu-Pro segments of the ligand wedges apart the NHS55


OD38 and NHL10


OY57

Figure 3. Changes in H-bond length induced by ligand binding. (a) Measured h3JNC0 values and corresponding NO distances. The h3JNC0 values measured for free (open circles) and RLP2-bound (®lled circles) SH3 are plotted as a function of donor residue. Corresponding acceptor groups are shown in Figure 1(c). Error bars re¯ect uncertainty in Ê , a scale re¯ecting the correpeated measurements. Using the exponential relationship RNO ˆ 2.75 0.25 ln( h3JNC0 ) A responding distance between the N and O atoms in the H-bond is shown on the right side of the plot. Secondary structure elements are mapped across the top. Data for residues V8, Y13, F23, and A42 are missing for the free form due to spectral overlap. A trans-H-bond scalar coupling was not observed for residue 55 in the RLP2-bound form of SH3. The gray bar at this residue re¯ects an upper limit for the possible value of jh3JNC0 j. This limit is derived from the signal to noise ratio of a reference spectrum as described previously by Dingley et al.30 (b) Propagation of strain across the 5-stranded b-sheet is illustrated by a two-dimensional schematic diagram of changes in H-bond length. In order to visualize the H-bond network and protein ligand-interactions more clearly, the area within the broken box is repeated in the upper part of the ®gure. Double-sided arrows indicate an increase, inward-pointing arrows a decrease, and dashed lines no change in H-bond length (within the error limits of Figure 3(a)). Propagation of changes in H-bond length occurs via a domino-like mechanism through H-bond networks, where change of one peptide plane orientation alters the orientations of other peptide planes within the same network. For example, the Hbond network starting at peptide plane Q49/T50 shows correlated changes across its entirety, spanning strands b4, b3, b2, b1, and b5. In a similar way, a correlation of changes in h3JNC0 values is found for the sheet-spanning H-bond network starting at peptide plane T50/G51. The H-bond network that involves the Y52/I53 and W40/L41 peptide planes shows shortening of the H-bonds on each end but no change in the central H-bond, which could result from coordinated changes in orientation of both peptide planes. These three networks are highlighted in yellow. (c) Intercalation of the L3-P4 and L6-P7 segments of the RLP2 ligand into the comb-like SH3 surface, and location of the NHL10


OY57 and NHS55


OD38 H-bonds (dotted lines) that lengthen upon ligand binding. In all ligand-bound SH3 structures determined to date, the ligand adopts a poly-proline type-II helix, a prism-like structure that packs like a zipper into the shallow binding groove on the side of the SH3 domain. The three aromatic residues that protrude from the SH3 surface (green) and residues of the RLP2 ligand (R1 blue, L3 and L6 orange, P4 and P7 teal and noninteracting A2 and P5 gray) are shown as van der Waals surfaces. Non-interacting ligand residues R8 and Y9 are not shown.

502

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

Table 1. Measured

h3

Donor

Acceptor

Phe7 Val8 Ala9 Tyr11 Leu10 Tyr13 Arg16 Thr17 Leu21 Phe23 Lys24 Lys25 Gly26 Glu27 Leu29 Gln30 Ile31 Ile31 Val32 Thr35 Glu36 Trp40 Leu41 Ala42 His43 Ser44 Leu45 Thr46 Gly48 Gly51 Tyr52 Ile53 Ser55 Tyr57 Val58 Ala59 Ser61 Ser63 Ala66 Glu67

Leu29 Ala59 Glu27 Tyr57 Tyr57 Phe23 Asp20 Asp20 Tyr52 Tyr13 Glu27 Asp12 Ala9 Lys24 Phe7 His43 Thr5 Leu29 Leu41 Asn33 Asn34 Ile53 Asn33 Gly51 Gln30 Gln49 Arg28 Ser44 Ser44 Ala42 Thr19 Trp40 Asp38 Pro54 Ser55 Val8 Thr6 Pro60 Ser63 Ile64

JNC0 -coupling constants for the c-Src SH3 domain in the free and RLP2-complexed state

RO


H O O O O O O OD2 OD1 O O OE1 OD1 O O O O O O O OD1 O O O O O O O O O O O O O O O O O O O O

a

h3

1.80286 1.89895 2.0715 2.34563 (1rlp) 1.71302 1.78808 1.93124 2.32488 1.83566 2.11924 2.15925 1.67051 1.90185 2.10474 1.94728 1.94197 1.9956 (1fmk) 2.35883 (1rlp) 2.44312 2.28104 2.12581 2.13211 2.21166 2.04218 2.09642 1.86486 2.24474 2.49821 (1rlp) 2.0265 2.09906 2.08188 2.00492 1.73533 2.16125 2.09585 (1rlp) 1.98342 2.02885 2.24658 2.38712 1.98189

JNC0 free SH3b 0.7192 ovlpe ovlp <0.2206 0.6958 ovlp NAf <0.3325 0.4688 ovlp <0.2336 0.5700 0.4988 ovlp 0.6853 0.6838 0.3103 <0.1431 0.4060 0.2280 ovlp 0.6278 0.3350 ovlp 0.7642 0.8883 0.4061 <0.2152 ovlp 0.7596 0.3895 0.7590 0.5301 <0.2141 <0.2158 0.4383 0.2969 <0.1776 <0.1418 <0.0747

Errorc 0.0142 0.0117 0.0108 0.0077 0.0097 0.0073 0.0084 0.0120 0.0168 0.0309 0.0191 0.0146 0.0097 0.0088 0.0155 0.0096 0.0349 0.0390 0.0639 0.0148 0.0186

h3 JNC0 SH3-RLP2 complex

0.6687 0.9128 ovlp <0.1412 0.6183 0.6370 <0.2098 <0.1005 0.5253 0.4525 NA 0.5731 0.5155 ovlp 0.6932 0.6185 0.4080 <0.0803 0.3960 0.2376 <0.0983 0.5210 0.3999 0.4905 0.6970 0.8180 0.3965 <0.1175 ovlp 0.4755 0.3264 0.7131 <0.2702 <0.1280 ovlp 0.4519 0.2353 <0.1019

Errord 0.02441 0.02137 0.01831 0.01239 0.01340 0.00821 0.01527 0.01551 0.00828 0.01797 0.01360 0.01874 0.04544 0.02185 0.01171 0.01265 0.01733 0.03097 0.01240 0.01923 0.02750 0.01410 0.00670 0.01579 <0.0856 <0.0598

The vector pGEX-SH3 (gift from David Shalloway), encoding a GST-c-Src-SH3 fusion protein was transformed into E. coli BL21(DE3) cells which were grown under conditions of 95 % 2H2O and with 15NH4Cl and 13C6-glucose as the sole nitrogen and carbon source, respectively.28 Puri®cation proceeded as described elsewhere.20 Besides the native Src residues 77 to 146 (numbered 2 to 67 herein), this protein contains the additional non-c-Src residues GS and GIHRQ at its N and C terminus, repectively. The RLP2 peptide was synthesized and puri®ed (98 % homogeneity) by the Cornell BioResources Center. Long-range TROSY-HNCO experiments were carried out and h3JNC0 -values were derived from the spectra as described previously by Cordier & Grzesiek2 and Wang et al.29 Typical data sets were acquired as two-dimensional H(N)CO spectra with a total measuring time of 64 hours on a Bruker DMX600 instrument at 25  C. a Ê for the 2src structure. Hydrogen atom Unless noted otherwise, RH


O refers to amide proton oxygen acceptor distances in A Ê proton-nitrogen distance. positions were built into the X-ray structures by using the program XPLOR31 and assuming a 1.01 A Ê . For certain H-bonds where RH


O was larger than 2.5 A Ê in the Hydrogen bonds are listed for which RH


O was smaller than 2.5 A Ê. 2src structure, the corresponding values in the 1rlp or 1fmk structures are listed when they were smaller than 2.5 A b J-values are absolute values given in Hz. c Error estimates were derived from the noise of the spectra. d Error estimates are standard deviations derived from the results of three independent experiments. e Spectral overlap. f Not assigned.

H-bonds which hold the b1, 310, and b3 secondary structure elements together (Figure 3(c)). The response of the protein to this separating force is not restricted to the direct interaction site, but is

then transmitted further along strands b1 and b3 such that lengthening is observed for the majority of H-bonds forming the antiparallel b-sheet connections b1/b5 and b3/b4 (Figure 3(b)).

503

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

Transmission of strain through the SH3 backbone Changes in h3JNC0 values reveal a net lengthening of H-bonds upon binding RLP2. In mechanical terms, the amount of elongation along a given direction is de®ned as strain, while stress is de®ned as force per unit area.17 Therefore, mapping the changes in H-bond length onto the SH3 structure elucidates the propagation of strain that occurs through H-bond networks. The observed changes are not only correlated along the direction of b-strands, but also along H-bond networks connecting peptide planes across the entire b-sheet as described in detail in Figure 3(b). This observation seems not surprising, since the repositioning of one peptide plane would necessarily alter the orientation of other peptide planes within such an H-bond network. Hence, structural rearrangement at the protein-ligand interface is propagated in a domino-like effect through networks formed by H-bonded peptide planes. For each of the three networks highlighted in Figure 3(b), one H-bond is in the b3/b4 interface. Therefore, all changes in H-bond geometry in these networks can be traced back to a location that is close to the site of interaction with the RLP2 peptide. The net lengthening of H-bonds accompanying ligand binding shows no correlation with the large increase in hydrogen-deuterium (H/2H) exchange protection factors (Pex) induced by ligand binding (Wang et al., unpublished results). While the h3JNC0 coupling constant directly re¯ects on the electronelectron interactions across the hydrogen bridge, Pex depends on many factors including solvent accessibility and the change in free energy of the whole system due to the unfolding process (
GH/2H). The lack of correlation between h3JNC0 and Pex cautions against a direct interpretation of Pex in terms of H-bond strength, and against a direct interpretation of the h3JNC0 coupling constant in terms of protein stability. Correlation between changes in h3JNiCj0 and changes in amide proton chemical shift A correlation between the isotropic chemical shift of the amide proton (dHN) and the H-bond length has long been recognized in proteins.18,19 Likewise, a linear relationship between the measured h3JNC0 values and dHN was observed,2 also implying a correlation between changes in these two parameters (
h3JNC0 and
dHN). For the c-Src SH3 domain, this relationship is observed for several but not all residues (data not shown), indicating that additional effects signi®cantly in¯uence
dHN. The most likely source of these changes is { Indeed, a correction for ring current effects deduced from high resolution coordinates yields considerable improvements in the correlation between dHN and h3JNC0 values for proteins containing a large number of aromatic side-chains (F.C., unpublished results).

variations in ring current effects caused by the repositioning of aromatic side-chains upon ligand binding.{ Changes in dHN per se cannot be taken as quantitative predictors of changes in H-bond length. In contrast, h3JNC0 values are not affected by such secondary effects and present a more reliable indication for changes in the H-bond geometry. Biological Implications The detection of minute geometric changes in Hbond geometry in this study is made possible by the very high spatial resolution derived from the precise measurement of H-bond coupling constants. Ligand binding to the SH3 domain induces strain in H-bonds that buttress the SH3/RLP2 binding surface, and small accompanying changes in H-bond length are propagated throughout the domain. This illustrates how major changes at the protein-ligand interface are compensated for by minor readjustments that span the whole domain. A key question is whether the observed changes in H-bond geometry are of functional signi®cance and whether the propagation of these minute changes in H-bond geometry could facilitate the transmission of larger structural changes throughout the molecule. Recent solution NMR evidence suggests that the event of RLP2 binding to the SH3 ligand surface causes structural changes in the interface to the highly conserved linker (LINK32) between the SH3 and Src homology domain 2 (SH2) domains. This interface is formed by residues in Ê ) from b1, b2 and b5, which are all remote (>5 A the bound ligand. In a construct consisting of the contiguous SH3-LINK32-SH2 domains, this linker region is ordered20 and is partly helical (J. Young et al., unpublished results), similar to the structure of the repressed state of the nearly intact Src kinase.15,16 Binding of RLP2 to the SH3 domain of this construct alters the structure of LINK32 and changes the relative orientation of the SH3 and SH2 domains (J. Reinking et al., unpublished results). This transmission of structural changes upon RLP2 binding is also re¯ected by changes in chemical shifts that not only occur throughout the SH3 domain, but also in the LINK32 and SH2 domains of the SH3LINK32-SH2 construct.21 In a similar way, RLP2 binding to the isolated SH3 domain induces changes in backbone (Wang et al., unpublished results) and side-chain (C. Wang et al., unpublished results) dynamics in regions remote from the ligand binding surface. Therefore, the changes in H-bond geometry which are observed throughout the SH3 domain are consistent with other NMR observations indicating a change in dynamics and transmission of structural rearrangements to the LINK32 interface and other remote regions upon RLP2 binding. The relative orientations and structural states of the SH3 and SH2 domains and of the LINK32 region connecting these two domains are not

504 only crucial in the repression of Src kinase activity15,16,22 but also in managing the assembly of multiprotein complexes in the activated state of the kinase.23 ± 27 Therefore, the observed changes in H-bond geometry upon RLP2 binding together with the concomitant changes in dynamics and structure in the region of the LINK32 interface and other remote regions could indicate a functional role in the regulation of the entire Src tyrosine kinase.

Ligand-Induced Hydrogen Bond Strain in c-Src SH3

12.

13.

14.

Acknowledgements We thank Norma Pawley for very helpful comments and discussion, and Frank Delaglio and Dan Garrett (NIH/NIDDK) for use of their software tools. This research was funded by a National Cancer Institute grant to L.K.N. (1 R55 CA77478-01A1), and by SNF grant 31-610 757.00 to S.G. Support for F.C. and C.W. was provided by A. v. Humboldt and Olin Foundation fellowships, respectively.

References 1. Kuriyan, J., Karplus, M. & Petsko, G. A. (1987). Estimation of uncertainties in X-ray re®nement results by use of perturbed structures. Proteins: Struct. Funct. Genet. 2, 1-12. 2. Cordier, F. & Grzesiek, S. (1999). Direct observation of hydrogen bonds in proteins by interresidue h3JNC0 Scalar Couplings. J. Am. Chem. Soc. 121, 1601-1602. 3. Cornilescu, G., Hu, J.-S. & Bax, A. (1999). Identi®cation of the hydrogen bonding network in a protein by scalar couplings. J. Am. Chem. Soc. 121, 2949-2950. 4. Meissner, A. & Sùrensen, O. (2000). Title. J. Magn. Reson. 143, 387-390. 5. Scheurer, C. & BruÈschweiler, R. (1999). Quantumchemical characterization of nuclear spin-spin couplings across hydrogen bonds. J. Am. Chem. Soc. 121, 8661-8662. 6. Cornilescu, G., Ramirez, B. E., Frank, M. K., Clore, G. M., Gronenborn, A. M. & Bax, A. (1999). Correlation between h3JNC0 and hydrogen bond length in proteins. J. Am. Chem. Soc. 121, 6275-6279. 7. Feng, S., Chen, J. K., Yu, H., Simon, J. A. & Schreiber, S. L. (1994). Two binding orientations for peptides to the Src SH3 domain: development of a general model for SH3-ligand interactions. Science, 266, 1241-1247. 8. Farrow, N. A., Zhang, O., Forman-Kay, J. D. & Kay, L. E. (1997). Characterization of the backbone dynamics of folded and denatured states of an SH3 domain. Biochemistry, 36, 2390-2402. 9. Riddle, D. S., Grantcharova, V. P., Santiago, J. V., Alm, E., Ruczinski, I. & Baker, D. (1999). Experiment and theory highlight role of native state topology in SH3 folding. Nature Struct. Biol. 6, 1016-1024. 10. Martinez, J. C. & Serrano, L. (1999). The folding transition state between SH3 domains is conformationally restricted and evolutionarily conserved. Nature Struct. Biol. 6, 1010-1016. 11. Yu, H., Chen, J. K., Feng, S., Dalgarno, D. C., Brauer, A. W. & Schreiber, S. L. (1994). Structural

15. 16.

17.

18. 19.

20.

21.

22. 23.

24. 25. 26.

27.

basis for the binding of proline-rich peptides to SH3 domains. Cell, 76, 933-945. Weng, Z., Rickles, R. J., Feng, S., Richard, S., Shaw, A. S., Schreiber, S. L. & Brugge, J. S. (1995). Structure-function analysis of SH3 domains: SH3 binding speci®city altered by single amino acid substitutions. Mol. Cell. Biol. 15, 5627-5634. Viguera, A. R., Arrondo, J. L., Musacchio, A., Saraste, M. & Serrano, L. (1994). Characterization of the interaction of natural proline-rich peptides with ®ve different SH3 domains. Biochemistry, 33, 1092510933. Yu, H., Rosen, M. K. & Schreiber, S. L. (1993). 1H and 15N assignments and secondary structure of the Src SH3 domain. FEBS Letters, 324, 87-92. Xu, W., Harrison, S. C. & Eck, M. J. (1997). Threedimensional structure of the tyrosine kinase c-Src. Nature, 385, 595-602. Xu, W., Doshi, A., Lei, M., Eck, M. J. & Harrison, S. C. (1999). Crystal structures of c-Src reveal features of its autoinhibitory mechanism. Mol. Cell, 3, 629-638. Shigley, J. E. (1977). Mechanical Engineering Design, (Holman, J. P., ed.), 3rd edit., McGraw-Hill Series in Mechanical Engineering McGraw-Hill, New York. Wagner, G., Pardi, A. & Wuthrich, K. (1983). Hydrogen bond length and proton NMR chemical shifts in proteins. J. Am. Chem. Soc. 105, 5948-5949. Pardi, A., Wagner, G. & Wuethrich, K. (1983). Protein conformation and proton nuclear magnetic resonance chemical shifts. Eur. J. Biochem. 137, 445454. Tessari, M., Gentile, L. N., Taylor, S. J., Shalloway, D. I., Nicholson, L. K. & Vuister, G. W. (1997). Heteronuclear NMR studies of the combined Src homology domains 2 and 3 of pp60 c-Src: effects of phosphopeptide binding. Biochemistry, 36, 1456114571. Reinking, J. L., Gentile, L. N., Tessari, M., Vuister, G. & Nicholson, L. K. (1998). Mechanisms of regulation in the regulatory apparatus of pp60 c-Src. In Structure, Motion, Interaction and Expression of Biological Macromolecules. Proceedings of the Tenth Conversation, State University of New York, Albany, NY 1997 (Sarma, R. H. & Sarma, M. H., eds), Adenine Press. Williams, J. C., Wierenga, R. K. & Saraste, M. (1998). Insights into Src kinase functions: structural comparisons. Trends Biochem Sci. 23, 179-184. Nakamoto, T., Sakai, R., Ozawa, K., Yazaki, Y. & Hirai, H. (1996). Direct binding of C-terminal region of p130Cas to SH2 and SH3 domains of Src kinase. J. Biol. Chem. 271, 8959-8965. Taylor, S. J. & Shalloway, D. (1994). An RNA-binding protein associated with Src through its SH2 and SH3 domains in mitosis. Nature, 368, 867-871. Fumagalli, S., Totty, N. F., Hsuan, J. J. & Courtneidge, S. A. (1994). A target for Src in mitosis. Nature, 368, 871-874. Alexandropoulos, K. & Baltimore, D. (1996). Coordinate activation of c-Src by SH3- and SH2-binding sites on a novel p130Cas-related protein, Sin. Genes Dev. 10, 1341-1355. Thomas, J. W., Ellis, B., Boerner, R. J., Knight, W. B., White, G. C., II & Schaller, M. D. (1998). SH2- and SH3-mediated interactions between focal adhesion kinase and Src. J. Biol. Chem. 273, 577-583.

Ligand-Induced Hydrogen Bond Strain in c-Src SH3 28. Guo, W., Sutcliffe, M. J., Cerione, R. A. & Oswald, R. E. (1998). Identi®cation of the binding surface on Cdc42Hs for p21-activated kinase. Biochemistry, 37, 14030-14037. 29. Wang, Y. X., Jacob, J., Cordier, F., Wing®eld, P., Stahl, S. J. & Lee-Huang, S., et al. (1999). Measurement of h3JNC0 connectivities across hydrogen bonds in a 30 kDa protein. J. Biomol. NMR, 14, 181-184.

505 30. Dingley, A. J., Masse, J. E., Peterson, R. D., Bar®eld, M., Feigon, J. & Grzesiek, S. (1999). Internucleotide scalar couplings across hydrogen bonds in watson-crick and hoogsteen base pairs of a DNA triplex. J. Am. Chem. Soc. 121, 6019-6027. 31. BruÈnger, A. T. (1992). X-PLOR (version 3.1) A System for X-ray Crystallography and NMR, Yale University Press, New Haven, CT.

Edited by M. F. Summers (Received 5 September 2000; received in revised form 27 October 2000; accepted 27 October 2000)