doi:10.1006/jmbi.2001.5199 available online at http://www.idealibrary.com on J. Mol. Biol. (2001) 314, 1147±1156

Solution Structures of the YAP65 WW Domain and theVariant L30 K in Complex with the PeptidesGTPPPPYTVG, N-(n-octyl)-GPPPY and PLPPY and theApplication of Peptide Libraries Reveal a MinimalBinding Epitope

Jose Ricardo Pires1*, Fatemeh Taha-Nejad2, Florian Toepert3

Thomas Ast3, Ulrich HoffmuÈ ller3, Jens Schneider-Mergener3

Ronald KuÈ hne1, Maria J. Macias2* and Hartmut Oschkinat1*

1Forschungsinstitut fuÈ rMolekulare PharmakologieRobert-RoÈssle-Str. 1013125 Berlin, Germany2European Molecular BiologyLaboratory, Meyerhofstr. 169117 Heidelberg, Germany3Institut fuÈ r MedizinischeImmunologieUniversitaÈtsklinikum ChariteÂ,Humboldt-UniversitaÈt BerlinSchummmanstr. 20-2110098 Berlin, Germany

E-mail addresses of the corresponoschkinat@fmp-berlin.de; macias@E

Abbreviations used: NOE, nucleaenhancement; NOESY, NOE spectrocorrelated spectroscopy.

0022-2836/01/051147±10 $35.00/0

The single mutation L30 K in the Hu-Yap65 WW domain increased thestability of the complex with the peptide GTPPPPYTVG(Kd � 40(�5) mM). Here we report the re®ned solution structure of thiscomplex by NMR spectroscopy and further derived structure-activityrelationships by using ligand peptide libraries with truncated sequencesand a substituion analysis that yielded acetyl-PPPPY as the smallesthigh-af®nity binding peptide (Kd � 60 mM). The structures of two newcomplexes with weaker binding ligands chosen based on these results(N-(n-octyl)-GPPPYNH2 and Ac-PLPPY) comprising the wild-type WWdomain of Hu-Yap65 were determined. Comparison of the structures ofthe three complexes were useful for identifying the molecular basis ofhigh-af®nity: hydrophobic and speci®c interactions between the side-chains of Y28 and W39 and P50 and P40, respectively, and hydrogenbonds between T37 (donnor) and P50 (acceptor) and between W39(donnor) and T20 (acceptor) stabilize the complex.

The structure of the complex L30 K Hu-Yap65 WW domain/GTPPPPYTVG is compared to the published crystal structure of the dys-trophin WW domain bound to a segment of the b-dystroglycan proteinand to the solution structure of the ®rst Nedd4 WW domain and itsprolin-rich ligand, suggesting that WW sequences bind proline-rich pep-tides in an evolutionary conserved fashion. The position equivalent toT22 in the Hu-Yap65 WW domain sequence is seen as responsible fordifferentiation in the binding mode among the WW domains of group I.

# 2001 Academic Press

Keywords: WW domain; YAP65; NMR structure; polyproline motif;ligands*Corresponding authors

Introduction

The WW domain is a small protein consisting ofabout 34 amino acid residues with two highly con-served tryptophan residues.1,2 To date nearly 100intracellular proteins of great functional diversity

ding authors:MBL-heidelberg.de

r Overhauserscopy; TOCSY, total

are known to contain one or more WW domains,being important in cytoskeletal and regulatory con-text.3,4

WW domains are the smallest natural antiparal-lel b-sheet structures, stable in solution withoutappreciable association and serving frequently as amodel system for understanding the thermodyn-amics and kinetics of the b-sheet fold.5 ± 7 They binddominantly proline-rich sequences as do SH3 andEVH1 domains.2,8 ± 10 According to their ligand pre-dilections the WW domains can be divided in fourgroups.4 The WW domain of the human Yes-kinase

# 2001 Academic Press

1148 Proline-rich Peptides and the WW Domain

associated proteins (Hu-Yap65) recognises a PPxYconsensus, speci®cally the PPPPY sequence.2,11

Other groups of WW domains bind to a PPLPmotif, like the FE65 WW domain.8,12,13 Recently, atargeting of PGR motifs was proposed for the WWdomain of Npw8 which preferentially recognises ashort proline-rich sequence surrounded by an argi-nine residue.14 Some WW domains bind phos-phorylated ligands. The prolyl-isomerase Pin1 WWdomain recognises speci®cally the phosphorylatedC-terminal repeat of the RNA polymerase II,15

crystal structure.16

The ®rst structure of a WW domain, that of theHu-Yap65 protein (Figure 1) in complex with theproline-rich peptide GTPPPPYTVG, was solved byNMR.17 Although the general orientation of theligand and the speci®city-determing interactionbetween the tyrosine in the peptide and a cavityon the domain surface could already be deter-mined in this ®rst structure, the small number ofintermolecular NOEs has precluded a detaileddescription of the interaction. Mutational studieshave identi®ed crucial amino acids in the WWdomain sequence for protein stability.6 The speci-®city determinants in the proline-rich ligands havebeen analysed by peptide library approaches.17 ± 19

Recently, three more structures of WW domains incomplex with proline-rich peptides have beendetermined by X-ray crystallography and NMRspectroscopy. Two of them, the WW domain ofdystrophin and that of Nedd4, also bind to a pep-tide contaning the PPxY motif.20,21 The othershows the WW domain of Pin1 binding a proline-rich motif including a phosphoserine with the pep-tide bound in the opposite direction.16

In this work, the interactions between proline-rich ligands and the Hu-Yap65 WW domain arefurther analysed by peptide library studies, NMRand UV ¯uorescence titrations, aiming at a moreprecise characterization of individual contributionsto the stability of the complexes. Three new com-

plexes are described by NMR, including theGTPPPPYTVG ligand employed in the originalstudies17 but using the Hu-Yap65 WW mutantL30 K, and two shorter ligands, the n-octyl-GPPPY-NH2 and the acetyl-PLPPY-NH2 togetherwith the wild-type domain. The choice of ligandsresults from the screening of a peptide librarywhich was composed of 190 different proline-richpeptides obtained by single mutations of the pep-tide GTPPPPYTVG, 36 shorter peptide variants ofit, and some peptide-derived compounds. With thestructures obtained by NMR we explain further theobserved structure-activity relationships, anddetermine a high-af®nity peptide of minimallength. This yields a perspective of designing smallnon-peptidic WW ligands.

Results

Library screening and binding constants

A substitutional analysis (Figure 2) was per-formed to investigate the speci®city-determiningpeptide epitope. Single mutations in the peptideGTPPPPYTVG lead to non-detectable bindingwhen P40, P50 or Y70 were substituted by any otheramino acid, with the exceptions that leucine maysubstitute P40, and phenylalanine may replace Y70,both at the expense of af®nity. This result repro-duces the motif PPxY as being essential for bind-ing, in agreement with the literature.2,11 The otherresidues are more tolerant to substitution, but thebinding af®nity of the variants where P60 is substi-tuted is often lower than that of the original pep-tide, indicating that at position 60 proline ispreferable as compared to many other amino acidresidues.22 Examining the binding properties ofshorter peptides we see that for the Hu-Yap65 WWdomain, the PPxY motif is necessary but notenough for the binding as many of the shorter pep-tides (e.g. Ac-PPPYTVG-NH2, Ac-PPPYTV-NH2,Ac-PPPY-NH2), bind much worse than the original

Figure 1. Sequence of the con-structs of the Hu-Yap65 WWdomain and of the ligands investi-gated. Construct 1: wild-type WWdomain from previous work16 withthe numbering adopted throughoutthis paper. Contruct 2: wild-typeWW domain. WWL30;K: L30Kmutant WW domain. Ligandresidues are numbered with anapostrophe (0) to distinguish themfrom domain residues; the number-ing is based on the longestsequence.

Figure 2. (a) Substitutional analysis of the peptide sequence GTPPPPYTVG. All residues of the wild-type (wt)sequence (left column) are substituted by all other L-amino acids (rows) and tested for binding to the Hu-Yap65 WWdomain. (b) Minimization analysis of the peptide sequence GTPPPPYTVG. All possible truncated peptides down to alength of three residues derived from the original sequence were synthesised and tested for binding to the Hu-Yap65WW domain.

Proline-rich Peptides and the WW Domain 1149

peptide inspite of showing the core PPxY motifand we suggest here that Ac-PPPPY is the minimalpeptide required for strong binding.

The dissociation constants of the complexesformed by the Hu-Yap65 WW domain and someof the peptides of the library were determinedquantitatively by following the changes of theintrinsic ¯uorescence of the trypthophan residuesin the domain upon addition of increasingamounts of the diferent peptides. As a positivecontrol of our binding experiments we have usedthe peptide KbAbAEYPPYPPPPYPSG for whichwe observed the strongest ¯uorescence changes, asdescribed.19 These changes include a shift in themaximum wavelength of emission of about 5 nm(from 345 nm to 340 nm) and an about twofoldincrease in the intensity of the emission, yielding aKd value of 6 mM. The titration of the WW domainwith GTPPPPYTVG increases the ¯uorescence by20 % , Kd � 52(�5) mM,17 and in a similar mannerdoes the addition of the shorter versions of it, Ac-PPPPYTV-NH2, Ac-PPPPYT-NH2 and Ac-PPPPY-NH2. The addition of Ac-PLPPY-NH2 or N-(n-octyl)-GPPPY-NH2 only produces a slight increasein the ¯uorescence intensity of ca. 8 % and 1 %,respectively, and it requires about ten times morepeptide than in the case of GTPPPPYTVG. By ®t-

Table 1. Dissociation constants for the complexes formed by

Ligand Kd (mM)

KbAbAEYPPYPPPPYPSG 6GTPPPPYTVG 52Ac-PPPPYTV-NH2 60Ac-PPPPYT-NH2 70-80Ac-PPPPY-NH2 60Ac-PLPPY-NH2 500n-Octhyl-GPPPY-NH2 700

n.d., not determined.

ting the changes of the ¯uorescence intensity at340 nm as a function of the added ligand, andassuming the formation of a 1:1 complex, weobtain the binding constants Kd � 70 mM for Ac-PPPPYTV-NH2 and Ac-PPPPY-NH2, 700 mM for N-(n-octyl)-GPPPY-NH2, and 500 mM for Ac-PLPPY,respectively (Table 1). We also followed the chemi-cal shifts changes of the indole HN signals of W39detecting shifts of 0.65, 0.19, 0.05 and ÿ0.04 ppmfor the above-mentioned peptides, respectively.Shift changes observed by addition ofKbAbAEYPPYPPPPYPSG and GTPPPPYTVG werestronger than for acetyl-PLPPY or N-(n-octyl)-GPPPY-NH2 (Table 1). Although the latter twoligands display lower af®nity than the wild-typepeptide, our NMR studies performed on the com-plexes con®rmed that all peptides speci®cally bindat the same position and with the same orientationas the original GTPPPPYTVG peptide on thedomain surface.

Structures

The 2D-NOESY and 2D-TOCSY spectra of theL30 K WW domain in complex with the peptideGTPPPPYTVG and of the wild-type domain incomplex with either acetyl-PLPPY or N-(n-octhyl)-

the wild-type WW domain and poly-proline ligands

Increase in fluorescence(%)

Change in the chemical shift ofthe indole NH of W39 (ppm)

100 0.6520 0.1920 0.1220 n.d.20 0.218 0.051 ÿ0.04

1150 Proline-rich Peptides and the WW Domain

GPPPY-NH2 were assigned, using the previousassignment of the Hu-Yap65 WW/GTPPPPYTVGcomplex as a starting point. All NMR structureswere calculated with a simulated annealing proto-col using distance restraints derived from NOEvalues taken from 2D-NOESY spectra obtained inH2O and/or 2H2O. For complex 1 (L30 K WWdomain and GTPPPPYTVG ligand),Kd � 40(�5) mM,17 a total of 590 distance restraintswere used in the calculations, classi®ed in 188sequential, 63 medium-range, 198 long-range and50 intermolecular (WW domain-ligand inter-actions), and included also ten hydrogen bondsbetween the strands and 71 intra-residualrestraints. For complex 2 (wild-type WW domainand N-(n-octyl)-GPPPY-NH2 ligand) and complex3 (wild-type WW domain and acetyl-PLPPYligand) a total of 422 and 330 restraints were usedin the calculation, respectively (Table 2).

For each complex we have calculated 200 struc-tures. Ensembles of the 20 lowest energy structuresare shown in Figure 3(a), (b) and (c), respectively.The structures were superimposed using the back-bone residues 16 to 39 of the WW domain. Theaverage root-mean-square difference (RMSD) forthe three structures, including all non-hydrogenatoms from residues G16 to W39 of the proteinand all non-hydrogen atoms for the ligand areabout 1 AÊ . The resolution achieved for the WWdomains alone is similar in the three complexes,with RMSD values of about 0.8 AÊ for the residues13 to 42. Residues located at both termini displayconformational ¯exibility, since neither mediumnor long-range NOEs were observed for them.Regarding the ligand conformation, complex 1,

Table 2. Structure statistics for the complexes 1, 2 and 3

Comple

A. Number of restraintsIntra-residual 71Sequential 188Medium-range 63Long-range 198Intermolecular 50Hydrogen-bond mimics (two restraints each) 10Total 590

B. RMSD (AÊ ) from experimental restraints0.026 � 0

C. X-PLOR potential energy (kcal molÿ1)Etotal 143 �Ebonds 7.4 � 0Eangles 94 �Eimpropers 11.8 �EvdW 10.5 �Enoe 20 �D. RMSD (AÊ ) between average structure and selected set of structuresWW domain resid. 16-39 Backbone 0.32

All non-H 0.80WW domain resid. 16-39 Backbone 0.37� ligand resid. 30-70 All non-H 0.78

that is the most stable of the three, yields a slightlybetter de®ned structure supported by 50 unam-biguously assigned intermolecular NOEs (Table 2).The spectra of this complex show contacts betweenthe two central proline residues, P40 and P50, withthe aromatic residues Y28 and W39 of the WWdomain. An analysis of the resulting structure ofcomplex 1 shows the presence of hydrogen bondsbetween T37 and P50 and between the indol NH ofW39 and the carbonyl group of the amino acid T20.The ligand conformation at the binding siteresembles that of a poly-proline helix type II(average dihedral angles observed for the 20 struc-tures with lowest energy are for P30:Phi � ÿ 72.4(�0.2) � Psi � 158(�15) �; for P40: Phi �ÿ 69(�1) � Psi � 171(�3) �; for P50: Phi � ÿ 71(�1) �

Psi � 168(�6) � and for P60: Phi � ÿ 72(�4) �). Adetailed representation of the peptide conformationis shown in Figure 4(a), and the interaction withthe domain0s surface in Figure 4(b).

For complexes 2 and 3 only about 25 intermole-cular NOEs were found, respectively, and theseNOEs are also weaker than those observed in thespectra of complex 1. Therefore both structures areslightly less well de®ned than that of complex 1(Figure 3(b) and (c)). Nevertheless, a similar pat-tern of intermolecular NOEs between the peptidesand the domain were observed (Table 3), placingresidues Y70, P60 and P50 at almost the same pos-ition in the binding site of the WW domain. More-over, in complex 3 several NOEs were observedbetween L40 and W39 mimicking those of complex1, between P40 and W39.

x 1 Complex 2 Complex 3

52 40134 12142 33148 9126 2510 10422 330

.001 0.025 � 0.001 0.042 � 0.002

4 125 � 4 162 � 2.5 5.7 � 0.4 8.6 � 0.4

2 87 � 2 99 � 38 10 � 1 11 � 18 9 � 1 13 � 1

2 14 � 1 30 � 3

0.31 0.350.77 0.840.45 0.471.08 0.92

Figure 3. Superposition of the 20structures of lowest energy calcu-lated by restrained simulatedannealing for the complex formedby: (a) the L30K mutantWW domain and the ligandGTPPPPYTVG (complex 1); (b) bythe wild-type WW domain andthe ligand N-(n-octyl)-GPPPYNH2

(complex 2); and (c) the wild-typeWW domain an the ligand acetyl-PLPPY (complex 3). The structureswere superimposed by the back-bone of the WW domain usingresidues 16 to 39.

Proline-rich Peptides and the WW Domain 1151

Figure 5(a) shows the superposition of the threecomplexes using the backbone of the WW domain,residues 16 to 39.

Discussion

The solution structures reported here in combi-nation with the peptide library studies allowed usto obtain a re®ned picture of the interactions of theHu-Yap65 WW domain and the peptideGTPPPPYTVG, although we did not investigate itdirectly. An important role was played by themutant L30K, which showed improved stabilityand solubility re¯ected also in a slightly higherstability for the complex. As a result, better NMRdata could be acquired and we could assign 50intermolecular NOEs (domain-ligand interactions).

These new data de®ned very well the confor-mation of the ligand and its location at the bindingsite of the domain without using modelling tools.A detailed analysis of the binding sites in the threecomplexes, involving either mutations in the pro-tein or in the ligand peptide, reveals general mech-anisms of peptide binding and suggests that thenative complex has the same structure. These struc-tures, together with the peptide binding studies,allow us to understand further the principles thatgovern these interactions.

From the structures and the peptide library stu-dies it can be deduced that the interaction of theGTPPPPYTVG ligand and the WW domain is dri-ven by the speci®c hydrophobic interactionbetween the two central and essential proline resi-dues P40 and P50 with aromatic residues Y28 andW39 of the WW domain and by the formation of

Figure 4. (a) Cartoon representation of the structure of the the complex L30K Hu-Yap WW domain/GTPPPPYTVG.(b) Poly-proline peptide bound to the surface of the L30K Hu-Yap65 WW domain; the surface is coloured by lipophi-licity, in brown the more lipophilic regions and in green more hydrophilic. Hydrogen bond donnors are colouredred.

1152 Proline-rich Peptides and the WW Domain

hydrogen bonds between T37 and P50 and betweenW39 and the carbonyl group of the amino acid pre-ceding P30.

The segment of the peptide GTPPPPYTVG thatis really contacting the surface of the domain(Figure 4(a) and (b)), comprises the carbonyl group

Figure 5. Comparison of thestructures. (a) Superposition of thestructures of the complexes of theHu-Yap65 WW domain: L30 KWW domain and the peptideGTPPPPYTVG (black), wild-typeWW domain and the petide N-(n-octyl)-GPPPY-NH2 (green) andwild-type WW domain and thepeptide acetyl-PLPPY (magenta).(b) L30K WW domain of theHu-Yap65 and the peptideGTPPPPYTVG (complex 1, black)superimposed to the complexformed by the WW domain of dys-trophin and a poly-proline contain-ing fragment of dystroglican (blue).(c) L30K WW domain of theHu-Yap65 and the peptideGTPPPPYTVG (complex 1, black)superimposed to the complexformed by the WW domain ofNedd4 and its proline-rich ligand(red). The structures were superim-posed by the a-carbon atoms of theresidues that compose the bindingpocket (T22, Y28, L30 or K30, H32,Q35, T37 and W39) and are theonly residues displayed togetherwith the ligands.

Table 3. Intermolecular interacitons observed in thecomplexes formed by the WW domain of Hu-Yap65and poly-proline ligands

Proline-rich Peptides and the WW Domain 1153

of T20 until residue Y70. This segment contributesto most of the binding energy, because screeningof shortened peptides yielded the high af®nitybinding minimal peptide Ac-PPPPY-NH2.

The substitution analysis (Figure 2) shows alsothat the two other proline residues, P60 and P30, arenot essential for binding, but are still required forincreasing its af®nity. They probably play a role instabilising a poly-proline helix type II structure,orienting the proline side-chains and the carbonylgroups towards the appropriated hydrogen bonddonors.

To date, besides the structure of the Yap65 WWdomain complexed to the peptide GTPPPPYTVG,another two well de®ned structures of WWdomains belonging to group I complexed to theircorresponding proline-rich peptides are describedin the literature.20,21 All these WW domains bindpeptides with the core motif PPxY, but in all cases¯anking residues seem to be important for increas-ing the af®nity and speci®city to the WW domains.Thus, in the complex of the Nedd4 WW domainwith a fragment of b-ENaC(TLPIPGTPPPNYDSL),21 it is observed that the Cterminus of the ligand is folded such that the C-terminal leucine is involved in further contactswith the surface of the WW domain, increasing thestability of the complex. Also, in the complex ofthe dystrophin WW domain with a fragment of b-dystroglycan (KNMTPYRSPPPYVPP)20 contactsbetween the domain and residues in the N-term-inal half ofthe ligand are observed. They thus contribute tothe total strengh and selectivity of the interactionof dystrophin and b-dystroglycan. Supportingeffects by EF hand domains neighbouring the WWdomain are proposed to play a role in further stabi-lizing the complex.

Nevertheless, by comparison of the three com-plex structures involving WW domains of group Iwe can identify a very similar mechanism of recog-nition of the core motif PPxY. In summary, theessential proline residues stack against the con-served residues W39 and Y28, and the tyrosine isaccomodated in a shallow hydrophobic pocketformed by residues K30, H32 and Q35.

This form of poly-proline recognition shown byWW domains is very similar to the mechanismused by SH3 domains.10 However, in the SH3domain case, exposed aromatic residues form twodifferent grooves. Peptides that bind to SH3domain have the core motif PxxP. One of the pro-line residues of this motif stacks against a tryptho-phan forming the ®rst hydrophobic groove, whilethe second proline stacks against one aromatic resi-due in the second.

The structures of complex 1 described here arevery similar to that of the WW domain of dystro-phin in complex with the b-dystroglycan peptide.A superimposition of both structures is presentedin Figure 5(b). The RMSD value between both com-plexes (considering the residues displayed inFigure 5(b)) is 1.10 AÊ . However, slightly differentpositions were observed for P40, P50 and T37.Although in both complexes the hydroxyl of T37points towards the carbonyl of P50, they are closerto each other in the L30K complex than in the

1154 Proline-rich Peptides and the WW Domain

X-ray structure. The formation of this hydrogenbond is in agreement with mutational studies per-formed on the wild-type WW domain that showsthat T37 can only be mutated to serine, all othernatural amino acid residues weaken the binding ofthe poly-proline ligand, demonstrating the require-ment of a hydroxyl group for binding.23 Thisobservation is further supported by the fact thatnearly all WW domain sequences known so fardisplay S/T at this position. Dystrophin and theHu-Yap65 bind proline-rich peptides in a verysimilar way.20 However, some of the differencespointed out can be more related to speci®c differ-ences between both binding sites rather than onlyto experimental limitations. For instance, P50 ismore perpendicular to the Y28 ring in the Yap65L30 K complex than it is in the Dystrophin WWcomplex. However, the enviroment of Y28 is alsoquite different in both domains, while the last resi-due of the ®rst strand in the dystrophin WWdomain is a serine the equivalent position in Yap65is a threonine. The latter has a methyl group extrathat it is located between both aromatic rings ofthe binding site, modifying the space available forthe central proline of the ligand (Figure 5(b)). Sincethe mutation T22S leads to weaker binding of theligand,23 we believe that this methyl plays a posi-tive role by increasing the hydrophobic interactionsbetween P50 and the domain; this is also supportedby NOEs between the proline and the threonine. Inthe Nedd4 complex,21 there is a histidine in theequivalent position of T22 and as a consequence ofits bigger size, the proline ring of the ligand isshifted further away from the phenylalanine ringthan in dystrophin and Yap65 (Figure 5(c)). Webelive that this position can also play a role in the®nal orientation of the peptide ligand, and itshould be taken into account when de®ning theamino acid residues required to de®ne the speci-®city of the binding.

Our peptide library screening has also identi®edsome weaker ligands that have also been studiedby NMR in complex to the wild-type WW domain(Figure 3(b) and (c)). These structures have beencompared to the complex with the strongest bind-ing ligand GTPPPPYTVG, allowing to pinpointcrucial interactions. The main difference betweenthe three complexes is in the interaction betweenresidues 40 and 30 of the peptides and W39 whichmay account for the different stability. The ligandN-(n-octhyl)-GPPPY-NH2 lacks the carbonyl groupnecessary for the hydrogen bond with the indoleNH of W39, and this seems to be the cause of thesmaller af®nity (Figure 3(b)). This should also bethe explanation for the small af®nity of other pep-tides of the library (e.g. Ac-PPPYTVG-NH2, Ac-PPPYTV-NH2, Ac-PPPY-NH2). The replacement ofP40 by leucine in complex 3 (Figure 3(c)), seems todestabilise the interaction, probably due to adecrease in the hydrophobic contact between L40and W39 and to a decrease of the preformed poly-proline type II helical population. This is, however,the only substitution allowed at this position by

mutational studies, probably because of the theleucine can still keep the helical character of apoly-proline sequence.

Conclusion

We have determined by NMR a solution struc-ture of the L30 K Hu-Yap65 WW domain in com-plex with the GTPPPPYTVG ligand and thusdescribed in more detail the mode of binding ofpoly-proline ligands to the WW domain of Hu-Yap65. The interaction is very similar tothat recently described for the b-dystroglycan-dystrophin WW domain complex, a WW sequenceclosely related to the Hu-Yap65. With the high-res-olution structures now available we can outline theprinciples that govern the interactions betweengroup I and IV of peptides but additional struc-tures from other representative and divergentsequences are required before we can fully under-stand the mechanisms controling the speci®city ofa particular WW sequence. The NMR structurestogether with the peptide library screening yieldthe core Ac-PPPPY as necessary and suf®cient forthe af®nity in the micromolar range to theHu-Yap65 WW domain, beside identifying severalchemical groups of fundamental importance forthe interaction in both the ligand and the domain.These results should now be of use for the designof high af®nity small WW ligands.

Materials and Methods

Constructs of the WW domain and ligands

Synthesis of the cellulose-bound peptide library. Thecellulose-bound peptide library was produced by semi-automatic SPOT Synthesis24 (Abimed, Langenfeld,Germany; software LISA (Jerini GmbH, Berlin,Germany)) on Whatman 50 cellulose membranes (What-man, Maidstone, UK), as described.25

The construct of the WW L30 K mutant was preparedby recombinant technology following proceduresdescribed in the literature.17

Determination of Kd by fluorescence

The dissociation constants of the different complexeswere determined by ¯uorescence spectroscopy asdescribed,19 by titration of a 10 mM solution of the WWdomain with a 50 mM solution of the binding peptideboth in 10 mM potassium phosphate buffer (pH 7.4),100 mM KCl. The temperature was set to 25 �C, the exci-tation wavelength was 298 nm and the changes in theintensity of the ¯uorescence at 340 nm by the addition ofthe ligand was used for the calculation.

NMR spectroscopy

A series of 2D NOESY26 with 130, 150, 180 ms mixing-time and 2D TOCSY27 with 20, 35 and 70 ms spinlockfor the constructs of WW domain alone (1.2 mM) for theligands alone (2.4 mM) and for the complexes (1:2) wereacquired at 600 or 750 MHz on DRX Bruker spec-trometers. Experiments were carried out in 10 mM pot-

Proline-rich Peptides and the WW Domain 1155

assium phosphate buffer (pH 6), 100 mM NaCl, 0.1 mMDTT, 0.1 mM EDTA, in H2O and in 2H2O at 15 �C.Spectra were processed with XWINNMR (Bruker) andanalysed with ANSIG.28

Structure calculations

NOE restraints derived from the NOESY experimentwere analysed as described using a lower boundary of1.8 AÊ and an upper boundary of 6 AÊ . Structures werecalculated by simulated annealing from random coor-dinates29 at 2000 K by using X-PLOR 3.130 and ¯oatingstereospeci®c assignment. Force constants for NOE were50 kcal molÿ1 AÊ ÿ2 for bond lengths, bond angles; andimproper angles were 1000 kcal molÿ1 AÊ ÿ2, 500 kcalmolÿ1 radÿ2 and 500 kcal molÿ1 radÿ2, respectively.

Atomic coordinates and NMR restraints

The Atomic coordinates and NMR restraints for com-plexes 1, 2 and 3 (20 structures each) were deposited inthe RCSB Protein Data Bank, entry numbers respectively1JMQ, 1K9Q and 1K9R.

References

1. Bork, P. & Sudol, M. (1994). The WW domain: a sig-naling site in dystrophin? Trends Biochem. Sci. 19,531-533.

2. Chen, H. I. & Sudol, M. (1995). The WW domain ofYes-associated protein binds a proline-rich ligandthat differs from the consensus established for Srchomology 3-binding modules. Proc. Natl Acad. Sci.USA, 92, 7819-7823.

3. Sudol, M. (1996). Structure and function of the WWdomain. Prog. Biophys. Mol. Biol. 65, 113-132.

4. Sudol, M. & Hunter, T. (2000). New wrinkles for anold domain. Cell, 103, 1001-1004.

5. Koepf, E. K., Petrassi, H. M., Sudol, M. & Kelly, J. W.(1999). WW: an isolated three-stranded antiparallelb-sheet domain that unfolds and refolds reversibly;evidence for a structured hydrophobic cluster inurea and GdnHCl and a disordered thermalunfolded state. Protein Sci. 8, 841-853.

6. Macias, M. J., Gervais, V., Civera, C. & Oschkinat,H. (2000). Structural analysis of WW domains anddesign of a WW prototype. Nature Struct. Biol. 7,375-379.

7. Ibragimova, G. T. & Wade, R. C. (1999). Stability ofthe b-sheet of the WW domain: a moleculardynamics simulation study. Bioph. J. 77, 2191-2198.

8. Chan, D. C., Bedford, M. T. & Leder, P. (1996).Forming binding proteins bear WWP/WW domainsthat bind proline-rich peptides and functionallyresemble SH3 domains. EMBO J. 15, 1045-1054.

9. Carl, U. D., Pollmann, M., Orr, E., Gertlere, F. B.,Chakraborty, T. & Wehland, J. (1999). Aromatic andbasic residues within the EVH1 domain of VASPspecify ist interaction with proline-rich ligands.Curr. Biol. 9, 715-718.

10. Zarrinpar, A. & Lim, W. A. (2000). Converging onproline: the mechanism of WW domain peptide rec-ognition. Nature Struct. Biol. 7, 611-613.

11. Chen, H. I., Einbond, A., Kwak, S. J., Linn, H.,Koepf, E., Peterson, S., Kelly, J. W. & Sudol, M.(1997). Characterization of the WW domain ofhuman Yes-associated protein and its polyproline-containing ligands. J. Biol. Chem. 272, 17070-17077.

12. Bedford, M. T., Chan, D. C. & Leder, P. (1997). FBPWW domains and the Abl SH3 domain bind to aspeci®c class of proline-rich ligands. EMBO J. 16,2376-2383.

13. Ermekova, K. S., Zambrano, N., Linn, H., Minopoli,G., Gertler, F., Russo, T. & Sudol, M. (1997). TheWW domain of neural protein FE65 interacts withproline-rich motifs in Mena, the mammalian homo-log of Drosophila enabled. J. Biol. Chem. 272, 32869-32877.

14. Komuro, A., Saeki, M. & Kato, S. (1999). Associationof two nuclear proteins, Npw38 and NpwBP, via theinteraction between the WW domain and a novelproline-rich motif containing, glycine and arginine.J. Biol. Chem. 274, 36513-36519.

15. Lu, P.-J., Zhou, X. Z., Shen, M. & Lu, K. P. (1999).Function of WW domains as phosphoserine- orphosphothreonine-binding modules. Science, 283,1325-1328.

16. Verdecia, M. A., Bowman, M. E., Lu, K. P., Hunter,T. & Noel, P. (2000). Structural basis for phosphoser-ine-proline recognition by group IV WW domains.Nature Struct. Biol. 7, 639-643.

17. Macias, M. J., HyvoÈnen, M., Baraldi, E., Schultz, J.,Sudol, M., Saraste, M. & Oschkinat, H. (1996). Struc-ture of the WW domain of a kinase-associated pro-tein complexed with a proline-rich peptide. Nature,382, 646-649.

18. Espanel, X. & Sudol, M. (1999). A single pointmutation in a group I WW domain shifts its speci-®city to that of group II WW domains. J. Biol. Chem.274, 17284-17289.

19. Koepf, E. K., Petrassi, H. M., Ratnaswamy, G., Huff,M. E., Sudol, M. & Kelly, J. W. (1999). Characteriz-ation of the structure and function of W! F WWdomain variants: identi®cation of a nativelyunfolded protein that folds upon ligand binding.Biochemistry, 38, 14338-14351.

20. Huang, X., Poy, F., Zhang, R., Joachimiak, A., Sudol,M. & Eck, M. J. (2000). Structure of a WW domaincontaining segment of dystrophin in complex withb-dystroglycan. Nature Struct. Biol. 7, 634-638.

21. Kanelis, V., Rotin, D. & Forman-Kay, J. D. (2001).Solution structure of a Nedd4 WW domain-EnaCpeptide complex. Nature Struct. Biol. 8, 407-412.

22. Linn, H., Ermekova, K. S., Rentschler, S., Sparks,A. B., Kay, B. K. & Sudol, M. (1997). Using molecu-lar repertoires to identify high-af®nity peptideligands of the WW domain of human and mouseYAP. Biol. Chem. 378, 531-537.

23. ToÈpert, F., Pires, J. R., Landgraf, C., Oschkinat, H. &Schneider-Mergener, J. (2001). Synthesis of an arraycomprising 837 variants of the hYAP WW proteindomain. Angew. Chem. Int. Ed. Engl. 40, 897-900.

24. Kramer, A. & Schneider-Mergener, J. (1998).Synthesis and screening of peptide libraries oncontinuous cellulose membrane supports. MethodsMol. Biol. 87, 25-39.

25. Kramer, A., Reineke, U., Dong, L., Hoffmann, B.,Hoffmuller, U., Winkler, D. et al. (1999). Spot syn-thesis: observations and optimizations. J. Pept. Res.54, 319-327.

26. Jeener, J., Meier, B. H., Bachmann, P. & Ernst, R. R.(1979). Investigation of exchange processes by two-dimensional n.m.r. spectroscopy. J. Chem. Phys. 71,4546-4553.

27. Braunschweiler, L. & Ernst, R. R. (1983). Coherencetransfer by isotropic mixing: application to protoncorrelation spectroscopy. J. Magn. Reson. 53, 521-529.

1156 Proline-rich Peptides and the WW Domain

28. Kraulis, P. J. (1989). ANSIG: a program for theassignment of protein 1H 2D NMR spectra by inter-active graphics. J. Magn. Reson. 24, 627-633.

29. Kuszewski, J., Nilges, M. & Brunger, A. T. (1992).Sampling and ef®ciency of metric matrix distancegeometry: a novel partial metrization algorithm.J. Biomol. NMR, 2, 33-56.

30. BruÈ nger, A. T. (1993). X-PLOR, Version 3.1: A Systemfor X-ray Crystallography and NMR, Yale UniversityPress, New Haven, CT.

Edited by M. F. Summers

(Received 23 July 2001; received in revised form 22October 2001; accepted 22 October 2001)

http://www.academicpress.com/jmb

Supplementary Material for this paper is availableon IDEAL