Structural Characterization of a New Binding Motif and a Novel Binding Mode in Group 2 WW Domains

doi:10.1016/j.jmb.2007.08.052

J. Mol. Biol. (2007) 373, 1255–1268

Structural Characterization of a New Binding Motif and a Novel Binding Mode in Group 2 WW Domains Ximena Ramirez-Espain 1 , Lidia Ruiz 1 , Pau Martin-Malpartida 1 Hartmut Oschkinat 2 and Maria J. Macias 1 ⁎ 1

Institute for Research in Biomedicine-Protein NMR group, and the Institució catalana de recerca i estudis avançats ICREA, Barcelona Science Park, Josep Samitier 1-5, E-08028 Barcelona, Spain 2

Forschungsinstitut für Molekulare Pharmakologie, Robert-Rössle-Str. 10, 13125 Berlin, Germany Received 26 April 2007; received in revised form 21 August 2007; accepted 23 August 2007 Available online 29 August 2007

Formin homology 1 (FH1), is a long proline-rich region of formins, shown to bind to five WW containing proteins named formin binding proteins (FBPs). FH1 has several potential binding regions but only the PPLPx motif and its interaction with FBP11WW1 has been characterized structurally. To detect whether additional motifs exist in FH1, we synthesized five peptides and investigated their interaction with FBP28WW2, FBP11WW1 and FBP11WW2 domains. Peptides of sequence PTPPPLPP (positive control), PPPLIPPPP and PPLIPPPP (new motifs) interact with the domains with micromolar affinity. We observed that FBP28WW2 and FBP11WW2 behave differently from FBP11WW1 in terms of motif selection and affinity, since they prefer a doubly interrupted proline stretch of sequence PPLIPP. We determined the NMR structure of three complexes involving the FBP28WW2 domain and the three ligands. Depending on the peptide under study, the domain interacts with two proline residues accommodated in either the XP or the XP2 groove. This difference represents a one-turn displacement of the domain along the ligand sequence. To understand what drives this behavior, we performed further structural studies with the FBP11WW1 and a mutant of FBP28WW2 mimicking the XP2 groove of FBP11WW1. Our observations suggest that the nature of the XP2 groove and the balance of flexibility/rigidity around loop 1 of the domain contribute to the selection of the final ligand positioning in fully independent domains. Additionally, we analyzed the binding of a double WW domain region, FBP11WW1-2, to a long stretch of FH1 using fluorescence spectroscopy and NMR titrations. With this we show that the presence of two consecutive WW domains may also influence the selection of the binding mode, particularly if both domains can interact with consecutive motifs in the ligand. Our results represent the first observation of protein–ligand recognition where a pair of WW and two consecutive motifs in a ligand participate simultaneously. © 2007 Elsevier Ltd. All rights reserved.

Edited by M. F. Summers

Keywords: WW domain; NMR; PPLIPP motif; WW domain complex; Group2 WW domain complexes

Introduction Formins are a family of nuclear phosphoproteins shown to participate in a range of processes in*Corresponding author. E-mail address: [email protected]. Abbreviations used: FH1, formin homology region 1; FBP, formin binding protein; TOCSY, total correlated spectroscopy; NOESY, nuclear Overhauser enhancement spectroscopy; PPII, polyproline type II helix.

cluding the formation of actin and microtubule networks during meiosis and mitosis, the maintenance of cell polarity, vesicular trafficking, signaling to the nucleus and embryonic development.1–3 Eight formin binding proteins (FBP) were identified by screening mouse limb bud expression libraries for binders to a conserved ∼120 amino acid long proline-rich region called FH1 or formin homology region 1.4 Of the eight FBPs identified, three proteins contained SH3 domains while the remaining hits contained WW domains, named FBP11, FBP21, FBP23, FBP28, and FBP30.

0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.

1256 The WW domain is a highly abundant functional unit that folds as a triple-stranded antiparallel βsheet and interacts with proline-rich or phosphorylated serine/threonine-proline sites in proteins. 5,6 In all complexes investigated so far, there is a common groove for interaction, named the XP groove, and at least one further groove, which distinguishes the different families of ligands. Many of the WW domains studied belong to group 1. Members of this group recognize ligands containing a polyproline helical turn followed by an aromatic residue, which disrupts the helical structure. Most structures solved so far correspond to this type of WW domains. Group 2-type WW domains form the second largest cluster of WW sequences, with all FH1 WW binders belonging to this group. Group 2 WW domains are characterized by their preference to interact with ligands containing a long polyproline helix. Moreover, aromatic residues are not normally present in these ligands.6 Remarkably, only two complexes, both with FBP11WW1 and PPLPx motifs, illustrate how their domain–ligand interactions occur (Figure 1(a)).7,8 In both complexes the leucine of the motif and the proline following it are recognized by opposite sides of the XP groove (named XPa and XPb sides, respectively; Figure 1(b))

New Binding Preferences for FBP WW Domains

while the proline at position −2 with respect to the leucine is bound in the group 2 specific groove, XP2.7,8 In FBP11WW1, this groove is formed by a histidine located in β1 and a tyrosine located in the middle of β2. A point to note in this is that in both FBP11WW1 complexes the main proline-rich interactions are performed by the group-specific groove of the domain, and not by the common XP groove. However, additional binding experiments performed with peptides using both fluorescence7 and surface plasmon resonance8 have also shown that FBP11WW1 can bind to proline-only sequences with an affinity in the micromolar range.7,9 There is, therefore, experimental evidence to indicate that at least the group 2 WW domain of FBP11WW1 may also recognize its ligands using one alternative binding mode, since the XP groove must also be able to make stabilizing contacts to proline residues for the domain to recognize a proline-only sequence. In the initial work by Chan and co-workers, a qualitative description of the selectivity of the five WW hits was obtained using several constructs of the FH1 region. In this way they found that different domains interacted with different regions of FH1, since a given region showed distinctly lower affinity for FBP21 and FBP28 than for FBP11 but a long

Figure 1. FBP sequences and motifs present in FH1. (a) Cartoon representation of the lowest energy structure of the FBP11WW1 complexes known to date. Interacting residues in the domain are shown in yellow while ligands are shown in orange (left)7 and violet (right).8 Residues in the ligand are labeled. Numbers correspond in both cases to the complex solved by Pires et al.7 (b) Surface representation of the FBP11WW1 domain with the characteristic XP (XPa and XPb) and XP2 grooves labeled in white. Residues that constitute the binding grooves are labeled according to Pires et al.,7 and depicted in black. (c) FBP WW domains identified by Chan et al.4 that bind to the FH1 domain. From top to bottom: FBP11WW1 domain in blue and FBP11WW2 in green (AAD39463), both FBP21 (AAC34810), FBP23 (sequence as described in the original paper), FBP30 (AAF59410) in black and FBP28 (NP_001034563) in orange. Residues shown to interact with the ligand in either FBP11WW17,9 or in FBP11WW2 and FBP28WW2 (this work) are shown in light blue, green and yellow, respectively. Putative interacting residues in the other FBP WW domains are shown in white surrounded by black boxes. Secondary structure elements are boxed in gray and labeled on top as β1, β2 and β3. (d) Ligands selected from the FH1 sequence and the motifs they contain. Peptides used in the structural work are numbered on the top of each sequence.

1257

New Binding Preferences for FBP WW Domains

region bound equally well to all constructs (FBPWW aligned sequences are shown in Figure 1(c)). The second idea derived from these studies was that a certain level of cooperativity of binding occurred in FBP11, since the two consecutive FBP11 WW domains bound a peptide of sequence APPTPPPLPP with reduced affinity compared to longer FH1 motifs. All these results raise questions as to whether other motifs different from PPLPx can be recognized by group 2 domains and as to whether the FBP11 WW domains might be able to work as a tandem in their binding to the FH1 region of the formins. To test these hypotheses we set up a structural analysis by using a series of peptides derived from the FH1 region and several FBP WW domains to understand if different FBP WW domains may be fine-tuned for different regions in the FH1 sequence (Figure 1(d)). From the peptides we chose, all were proline-rich peptides containing at least an xxPLxx sequence. One of the peptides studied in detail was chosen to be similar to the one used in the existing FBP11WW1 and APPTPPPLPP complex, to facilitate comparison to previous structural results. Three ligands interacted with an affinity in the micromolar range. To analyze how these three ligands interact we solved their NMR structures in complex with FBP28WW2. Two of the complexes, one of them the complex with the peptide PTPPPLPP, show similar sets of contacts despite a difference in the sequence motif of the peptide. As expected, the PTPPPLPP complex of FBP28WW2 also re-sembled the complex with FBP11WW1 previously determined. However, a complex with a ligand of sequence PPLIPPPP shows that the ligand is bound with an orientation shifted by a helical turn with respect to the other peptides, describing a new form of interaction. To explain why both binding modes occur, we made further studies with FBP11WW1 and with FBP28WW2_Y11H, an FPB28 mutant which mimics the XP2 groove of FBP11WW1. We found that the selection of the binding mode is dependent on the XP2 residue composition and the balance of flexibility/rigidity around loop 1. Our binding experiments with these domains allow us to suggest that the alternative binding mode shown here could be present in other group 2 domains. To investigate the suggestion of cooperativity between WW domains of a single FBP protein, and noticing that the PPLP and PPLIPP motifs are consecutive in FH1 sequence, we prepared a peptide of sequence TPPPLPPPLIPPPP and studied its interaction with the FBP11WW1-2 stretch by fluorescence spectroscopy and NMR titrations. Titration experiments performed with the tandem show substantial chemical shift changes at residues belonging to both domains. The changes are less intense when independent domains are titrated with short ligands. We believe that these discoveries expand the present understanding of group II WW domain binding and will open the search towards discovering new interacting motifs for individual domains

and tandems. For the latter, in particular, our results explore a new mode for binding that WW domains may have at their disposal when working together.

Results and Discussion Selection of peptide motifs and FBP domains The protein sequence of the conserved FH1 region in mouse, used to identify the FBPs by P. Leder and co-workers, contains several proline-rich regions where different motifs can be distinguished. In a simplified manner, these regions can be divided into either polyproline-only stretches of up to 12 consecutive proline residues or those containing interrupted proline regions. Indeed, members of group 2 WW domains have been shown to bind to both types of motifs with similar affinities.7,9 However, characterization of a complex with purely polyproline stretches where the peptide orientation may not be unique (several possible frames could be chosen by the protein) would provide un-interpretable results by NMR. We thus investigated the interaction of FBPWW domains to regions containing no more than four consecutive proline residues separated by one or two hydrophobic amino acids. In this way, a region spanning 27 residues of the FH1 domain (872– 899 in formin 1) was divided into five peptides containing one of three types of possible target motifs, PPxPP, PPxxPP or PxPP. We also prepared a 14-residue ligand containing two consecutive motifs. Motifs and ligands are shown in Figure 1(d). Peptide assignment strategy All the investigated peptides have a high proline content, responsible for the formation of the polyproline type II helical structure (PPII). The preference for this structure is supported by the pattern of nuclear Overhauser enhancements (NOEs) that define them as being in the trans-proline conformation. The PPII helix is rigid and extended. In the studies performed with individual domains and with short ligands this results in four or five peptide residues interacting with the domain. The result of binding is a set of chemical shift changes observed at the interacting residues. Since the chemical shift changes affect most of the proline residues of the peptide, their changes contribute to facilitate their intra-residual assignment. To further facilitate their assignment, we also synthesized several peptides of each type, varying the number of proline residues at their termini. In all cases we employed solid-phase synthesis using Rink-amide resin for the synthesis. This resin leaves an NH2 group bound to the C-terminal residue after cleavage. These signals are straightforward to assign and in the NOE spectroscopy (NOESY) experiment they provide NOEs to the last proline, contributing to facilitate the proline identification. Since most of the resonances were assignable, and particularly those corresponding to

1258 the interacting residues, we did not consider production of labeled peptides using recombinant techniques. Chemical shift assignments of FBPWW domains To obtain information about new binding motifs in FH1 as well as to determine if the previously identified FBP11WW1 binding grooves are common to other FBPWW domains, we decided to follow the potential interactions by measuring heteronuclear single quantum coherence (HSQC) spectra of 15Nlabeled FBPWW domains in the presence of increasing amounts of unlabeled peptides. However, in order to compare the affected residues, knowledge of the amide assignments is required. While assignments of FBP28WW2 and FBP11WW1 domains have been reported,7,10 the FBP11WW2 domain has been suggested to be marginally folded in solution.11 Since both FBP11 WW domains are connected by a short loop, we also prepared a sample including both domains, which could be used to investigate the possibility of simultaneously binding. We prepared the four samples using the same experimental conditions described for FBP28WW2,6 thus avoiding denaturant conditions, and acquired the NMR data at 285K. Sets of 15N- NOESY and TOCSY and 2D 1H-1H NOESY and TOCSY run in 2 H2O were acquired with each sample, allowing us to assign both independently produced FBP11WW1 and WW2 domains. With respect to the tandem and in order to identify the residues taking part in the interaction we tried to use the assignment reported by Gao et al.11 However, since in their experimental conditions the second domain was partially unfolded, we could not fit their assignments to our data. Thus, we prepared a double-labeled sample of the tandem and assigned it using standard triple resonance experiments. To further facilitate the assignment and to resolve some ambiguities (both domains share the YYYN motif) we combined the backbone triple-resonance experiments with a 15N-NOESY experiment measured with a threefold ligand excess. With this experiment residues across the strands can be connected and we could use this additional information to distinguish if they are in the first domain or in the second. NMR titrations to identify new motifs and interacting regions of FBP WW domains Proline-rich peptides are highly hygroscopic and therefore peptide concentration determined by gravity can suffer from an overestimation of the actual concentration of peptide used. To overcome this, peptide concentrations were determined by elemental analysis of the respective peptide's solutions. HSQC spectra of 15N-labeled FBP28WW2 in the presence of increasing amounts of unlabeled peptides yield amide chemical shift changes upon addition of peptides containing either PPxPP or PPxxPP motifs. The average value of these changes calculated at the last titration point is shown in

New Binding Preferences for FBP WW Domains

Figure 2(a). Of the four peptides investigated, the one containing the PxPP motif can be regarded as a negative control, since no changes were observed, even in the presence of a tenfold excess of peptide. In the three positive cases binding occurs in fast exchange, with a 1:1 stoichiometry according to the titrations performed using NMR spectroscopy. The HSQC values with the complete titration steps are shown as Supplementary Data, Figure 1(a)–(d)). We investigated the binding of 15 N-labeled FBP11WW1 with either the PPxPP peptide previously studied by Pires et al. (positive control, data not shown), or with the PPxxPP motif (Supplementary Data, Figure 1(e)) as well as the binding of 15 N-labeled FBP11WW2 with the PPxxPP motif. In this case, the chemical shift changes obtained at the last step of the titration (2.7-fold excess of ligand) are shown in Figure 2(b). The HSQC showing both the good dispersion of the amides characteristic of a folded sample and the chemical shift changes upon binding are shown in Supplementary Data, Figure 1(f). All three domains investigated suffer chemical shift changes in residues localized in both XP and XP2 grooves. Residues 10–13 in the β1 strand, 18–21 in β2 and 28–31 in the β3 strand, manifest average changes well above the threshold in all complexes (numbers are maintained as by Macias et al.10 for FBP28WW2, equivalent residues are also affected for FBP11WW1 and WW2 domains, respectively). Changes are also observed at the side-chain level, particularly in the aromatic rings of Y11, Y19, Y21 and W30 (data not shown). Thus, based on the chemical shift changes observed in both XP and XP2 grooves, we can conclude that all interacting peptides occupy the binding sites described in the FBP11WW1 complexes. Additionally, a second conclusion is that FBP11WW1 binds more efficiently to the PPLPx motif previously described while both FBP11WW2 and FBP28WW2 seem to prefer the PPxxPP motifs. With respect to the tandem construct, the addition of increasing amounts of the 14 residue ligand induced above average chemical shift changes in 33 residues. In titrations with each independently prepared domain, about 17 residues are normally affected upon binding, suggesting that in this case both domains are involved in ligand interaction. Interestingly, we observed that some residues bind in slow exchange, an exceptional feature not observed when titrations were performed with FBP28 or FBP11 independently expressed domains. The bar representation corresponding to the last titration point is shown in Figure 2(c). The assigned HSQC together with the changes observed upon addition of increasing amounts of the peptide are shown in Figure 2(d). Affinity values obtained for the FBP complexes using fluorescence spectroscopy Affinity values obtained with standard far-UV fluorescence measuring tryptophane quenching at

New Binding Preferences for FBP WW Domains

1259

Figure 2. Interaction between FBP28WW2 and four peptides selected from FH1. (a) Binding of FBP28WW2 to four peptides (PPPLIPPPP, PTPPPLPP, PPLIPPPP and PPGLGPLPP) shown as a bar representation calculated as the difference of the average changes observed for each residue at saturation with respect to the free domain. Amino acid labels refer to FBP28WW with numbering maintained. (b) Binding of FBP11WW2 to the PPPLIPPPP peptide shown as a bar representation, generated as described above. (c) Bar representation of the chemical shift changes calculated as described for (a). (d) HSQC titration corresponding to the construct containing the pair of FBP11WW1-2 domains. Different colors represent chemical shift changes induced upon addition of increasing amounts of the ligand. The molar equivalents of ligand used in the titration are shown in the inset of the Figure. Most changes are obtained at a 2.6-fold excess of peptide. This value is boxed. To facilitate the identification of residues in the Figure we used blue for FBP11WW1 amides and green for FBP11WW2. (e) Table of dissociation constants obtained following tryptophan quenching fluorescence upon addition of the ligands. Steady-state fluorescence intensity data were fitted to the equation: F = 1−((Qmax * [peptide])/Kd + [peptide])), where F is the fluorescence intensity, Qmax is the maximum fluorescence that can be quenched, and Kd is the dissociation constant. Curve fitting and statistics were generated with Kaleidagraph 4 (Synergy software, Reading PA). Data are expressed as means ± S.D. (standard deviation). Data points are the averages of three experiments.

343 nm upon addition of increasing amounts of peptide solutions12 are given in Figure 2(e). The HSQC values corresponding to the changes induced by the different motifs and domains are shown in Supplementary Data, Figure 1. According to the affinity constants determined, the FBP28WW2 domain binds the PPPLIPPPP peptide with the highest affinity (94(±11) μM), while the loss of one N-terminal proline as in PPLIPPPP decreases the affinity to 156(±12) μM. The PPxPP motif interacts more weakly with a binding con-

stant of 255(±21) μM. The FBP11WW2 domain also interacts with a similar affinity with PPPLIPPPP (89(±14) μM). In fact, the PPPLIPPPP peptide binds to both FBP11WW2 and FBP28WW2 with the highest micromolar affinity constant ever determined for a group 2 WW domain.7,13 The pair of FBP11WW1-2 domains bind to the 14 residue peptide with a slightly better value of 53(±20) μM. To better understand the features responsible for the affinity displayed by both FBP28WW2 and FBP11WW2 domains to the PPxxPP motifs, we set

1260 up to determine the complexes between FBP28WW2 and three peptides containing either the PPxPP or PPxxPP motifs. Structure of FBP28WW2 in complex with the PTPPPLPP peptide: L8′ contacts Y19 and W30 We prepared two related sequences, the APPTPPPLPP previously used in the structural study of FBP11WW1 by Pires et al.,7 and also a shorter version of it, PTPPPLPP, taking into account the structural information available from the FBP11WW1 complexes. To facilitate the comparison between our data and that of FBP11WW1, we maintained the numbering used for the ligand in the FBP11WW1 complex with APPTPPPLPP. We observed that upon addition of the FBP28WW2 domain to each of the two ligands, equivalent residues were affected (namely the PPLPP region). The affinities we determined for both peptides are also similar (Figure 2(e)). Thus, for the determination of the complex we used the shortest version to simplify the peptide assignment. Figure 3(a) shows the NMR ensemble of the PTPPPLPP-FBP28WW2 complex. A surface and a cartoon representation to ease interpretation are shown in Figure 3(b). The calculation of this complex is supported by 52 manually assigned intermolecular NOEs (some shown in Figure 3(a) in Supplementary Data), which define the peptide

New Binding Preferences for FBP WW Domains

orientation on the surface of the domain. Seventeen of these NOEs are observed between the side-chain of L8′ to the aromatic rings of Y19 and W30 in the XP groove, which were unambiguously assigned. Regarding the peptide's proline residues, P6′ is accommodated between Y11 and Y21, with its ring perpendicular to the β-sheet. P9′ on the other hand is situated between Y21 and W30 and is perpendicular with respect to the β-sheet plane. In addition W30 shows NOEs to the delta protons of P19′ (which has the lowest chemical shift for delta protons of all peptidic proline residues) and Y11 contacts P5′. In the initial calculations we used only the gamma protons of P5′ as restraints, since the β-sheets were overlapped with P7′. As expected, due to the sequence similarity of the domain and ligands, this complex is very similar to those obtained for FBP11WW1.7,8 In three cases the peptide contacts the domain via equivalent residues and in a similar manner. Besides the equivalent NOEs, two hydrogen bonds between the peptide and the domain are observed in the FBP28WW2 complex. One involves the carbonyl group of P6′, which is close to the hydroxyl group of Y21 and was also observed in the FBP11WW1 complex. The second involves W30 Hε1 and the carbonyl group of P9′. This potential hydrogen bond is absent in the FBP11WW1 complex. Statistics of the calculation including the analysis with Procheck-NMR are given in Table 1.

Figure 3. Solution structure of FBP28WW in complex with the PTPPPLPP peptide. (a) Stereo view of the best-fit backbone (N, Cα, C′) superposition of the ten lowest energy structures after water refinement with the PTPPPLPP ligand (domain in blue and ligand in red). (b) Left: Surface representation of the lowest energy structure (ligand in dark blue). Right: Cartoon representation with the secondary structural elements of the domain shown in green, residues participating in contacts in dark yellow and ligand in dark blue.

1261

New Binding Preferences for FBP WW Domains

Table 1. Summary of structural calculations and restraints PTPPLPP

PPPLIPPPP

PPLIPPPP

547 109 29 74 327 52 3 68 24

528 109 26 67 326 40 1 69 24

587 109 25 75 326 44 2 69 24

2.9 × 10−3±9 × 10−4 2.3 × 10−3±5 × 10−4 0.19 ± 0.02

2.2 × 10−3±9 × 10−4 2.1 × 10−3±1 × 10−4 0.17 ± 0.04

1.5 × 10−3±6 × 10−4 2.7 × 10−3±4 × 10−4 0.21 ± 0.04

(6–34, 6′–10′) 0.34 ± 4 × 10−2 1.01 ± 0.12 0.60 ± 0.12 1.15 ± 0.2

(6–34, 2′–7′) 0.43 ± 6 × 10−2 0.98 ± 0.14 0.72 ± 0.15 1.17 ± 0.2

(6–34, 3′–9′) 0.50 ± 6 × 10−2 1.05 ± 8 × 10−2 0.58 ± 7 × 10−2 1.32 ± 5 × 10−2

−1580.15 ± 65 89.3

−1620.15 ± 44 87.6

−1550.17 ± 38 92.9

10.7

12.4

7.1

Number of structural restraints All Sequential (|i–j| = 1) Medium range (2 ≤ |i–j| ≤ 4) Long range (|i–j| N 4) Intraresidual Intermolecular Total ambiguous restraints Dihedral angles Hydrogen bonds r.m.s. deviation from restraintsa bSANb All NOE distance restraints (Å) Hydrogen bonds (Å) Dihedral angles (°) Average atomic r.m.s. deviation from the mean structure (Å) Residues backbone 2nd structure Residues heavy atoms 2nd structure All structured residues (backbone) All structured residue (heavy atoms) Structural quality EL.-J.c Residues in most favored (5–34 and 2′–9′)(%) regions of Ramachandran plotd Residues in additionally allowed region a

No dihedral angle restraint was violated by more than 5° and no distance restraint was violated after the water refinement step. bSAN refers to the ensemble of the ten structures with the lowest energy. EL-J is the Lennard-Jones energy calculated using the CHARMM PARMALLH6 parameters. EL-J was not included in the target function during the structure calculation. d Excluding glycine and proline residues. b c

The few discrepancies observed in the three structures (this paper, Pires et al.7 and Kato et al.9) are probably due to the differences observed at the domain sequence level. In particular, the key differences occur in the XP2 groove and in loop 1. In FBP11WW1, the position equivalent to Y11 in the first strand is a histidine (the corresponding residue number in FBP11WW1 is either H15 or H20,7,8. In loop 1 FBP11WW1 has a proline (P18/P23), which participates in ligand binding. The FBP28WW2 equivalent residue is an alanine (A14), which does not contribute to the interactions. These changes may also explain the different affinities observed in the three complexes.7,8 FBP28WW2 binds PPLIPP peptides using two binding modes To facilitate the assignment of this motif we prepared three peptides, PPPLIPPPP, PPLIPPPP and PPLIPPP. In all three cases, chemical shift changes occurred in both domain and ligand residues, which we could follow up to saturation with HSQCs or up to a ratio of 1:3 protein-to-peptide with 2D-NOESYs. We observed, however, that the changes in the aromatic residues of the domain were different depending on the peptide under investigation. PPPLIPPPP (the best binder accordingly to the affinity values) and PTPPPLPP (the worse binder) induced similar changes to Y11, Y19 and W30. However, the ad-

dition of PPLIPPP or PPLIPPPP (the second best binder) shifts the delta protons of Y19 by 0.2 ppm and W30 Hη2 and Hζ3 by 0.12 ppm with respect to the other complexes. Regarding the three PPLIPP ligands and in contrast to the changes observed in the PPLPP peptide, we observed that the non-proline residue whose chemical shift changes in the presence of the domain is the isoleucine, and not the leucine. Indeed, some isoleucine protons change up to 0.15 ppm (for instance, the QG2 methyl and the HG13 proton). With respect to the proline residues we observed that in both PPLIPPP and PPLIPPPP peptides, the chemical shift of proline residues preceding the “LI” residues are more similar to that of the free peptides than in the case of PPPLIPPPP, while the three peptides manifest similar changes in the C-terminal proline residues upon binding. These differences, the differences in affinities measured by fluorescence spectroscopy and the ones detected at the aromatic residues in the domain suggested that at least the N-terminal part of the peptides might be accommodated differently in PPPLIPPPP compared to PPLIPPPP and PPLIPPP. Structure of FBP28WW2 in complex with the PPPLIPPPP and PPLIPPPP peptides The NOEs observed between the domain and the different peptides confirmed two different binding

1262 modes. While in the PPLIPPP and PPLIPPPP peptides we detected unambiguous NOEs from Y11 to the peptide's isoleucine, in the case of the PPPLIPPPP the isoleucine showed NOEs to W30. Intrigued by these differences we set out to determine the structures of both complexes. The PPPLIPPPP complex: I5 ′ binds in the XP groove The calculation of the PPPLIPPPP-FBP28WW2 complex is based on 40 manually assigned NOEs between the peptide and the domain. Statistics of the calculation including the analysis with ProcheckNMR are given in Table 1. The superposition of the ten best structures after water refinement is shown in Figure 4(a). As in the previous complex, the XP groove is involved in recognition of a non-proline/ proline peptide residue (Figure 4(b)). However, in this case the isoleucine I5′ interacts with the domain, and not the leucine as in the PPLPP complex. Indeed, we observe ten NOEs between I5′ and Y19 and W30 in the domain (some NOEs are shown in Supplementary Data, Figure 3(b)). This is not surprising if one takes into account that the interaction occurs with the hydrophobic residue in position −1 with respect to the interacting proline, which in this case is P6′. To define the orientation of the peptide, we used unambiguously assigned NOEs from proline P3′ to both Y11 and Y21, from P6′ gammas to Y21 and

New Binding Preferences for FBP WW Domains

W30 and from P7′ deltas to W30. P2′ shows NOEs from the delta and gamma protons to Y11 as well as to the methyl group of T13. S28 beta protons show NOEs with P6′ gammas and deltas and W30 with P8′. As displayed in the Figures, both P3′ and P6′ proline residues are oriented perpendicularly to the β-sheet plane, similarly to the orientation of the equivalent proline residues in the previous complex (P6′and P9′, respectively). Furthermore, two potential hydrogen bonds are also observed between the peptide and the domain, one of them involving the carbonyl group of P3′, which is close to the hydroxyl group of Y21. The second involving the hydroxyl group of S28 and the carbonyl group of L4′. Phosphorylation of these hydroxyl groups may have implications in ligand binding regulation, as we have recently shown in the case of the interaction of Itch WW3 domain and its ligand.14 Since these hydrogen bonds were not used as restraints, they are not depicted in the Ligplot representation of Supplementary Data, Figure 2(b). The PPLIPPPP complex: I5′ binds in the XP2 groove To facilitate comparisons, we maintained the same numbering in the ligand as in the preceding complex. Thus, the first proline of this peptide is P2′. In the bound state, this peptide has all alpha and delta proline protons resolved and only some of the

Figure 4. Solution structure of FBP28WW in complex with the PPPLIPPPP peptide. (a) Stereo view of the best-fit backbone (N, Cα, C′) superposition of the ten lowest energy structures after water refinement with the ligand (domain in blue and ligand in red). (b) Left: Surface representation of the lowest energy structure (ligand in brown). Right: Cartoon representation with the secondary structural elements of the domain shown in green, residues participating in contacts in dark yellow and ligand in brown.

New Binding Preferences for FBP WW Domains

beta or gamma protons are overlapped, allowing the unambiguous assignment of most of the observed NOEs with the domain, and in particular of those that define the peptide orientation. For the complex we used a ratio 1:3 protein– peptide. In the NOESY experiments we observed that the side-chain of I5′ packs in the cavity formed by the aromatic rings of Y11 and Y19, while its alpha proton is close to Y21. No contacts between this isoleucine and W30 were observed, unlike in the preceding complex. Once again, L4′ did not contact the domain, and only NOEs that connect it sequentially to P3′ and to I5′ were observed. We found that P3′ contacts to Y11 and that P6′ contacts to both Y11 and Y21 aromatic rings, P8′ contacts Y19 and W30 and P9′ W30 and Y21 rings. Two hydrogen bonds between the ligand and the protein, which were not included in the calculation, are observed in the complex structure. They result from the hydroxyl group of the Y19 with the carbonyl of P6′ and the hydroxyl group of S28 with the carbonyl of P7′, respectively. As with the other two complexes, the ensemble of ten lowest energy structures is shown, with a surface and cartoon representation of the main features of the complex highlighted (Figure 5(a) and (b)). Most of the 44 manually assigned intermolecular NOEs

1263 are represented with the Ligplot projection (Supplementary Data, Figure 3(c)). Statistics of the calculation including the analysis with Procheck-NMR are given in Table 1. The main difference between this and the previous complexes can be thus described as a shift along the polyproline helix equivalent of one turn. Regarding the domain, the main change involves small rotamer accommodation of both Y19 and Y21 aromatic rings. The PPLIPPP peptide also binds in this way, since there are NOEs observed between the isoleucine and both tyrosine residues, Y11 and Y19 while P'8 is close to the W30 (data not shown). In all, the “PLIP” complexes with FBP28WW2 show that although both peptides are bound using the same interacting pockets on the FBP28WW2 surface, the peptide is accommodated differently, translated by one full turn of the polyproline helix. FBP11WW1 binds the PPPLIPPPP peptide: I5′ binds in the XP2 groove To understand whether other group 2 WW domains show this dual behavior we chose to investigate the FBP11 domains, since they were previously shown to bind the FH1 region used in this work.4 We selected FBP11WW1 because of

Figure 5. Solution structure of FBP28WW in complex with the PPLIPPPP peptide. (a) Stereo view of the best-fit backbone (N, Cα, C′) superposition of the ten lowest energy structures after water refinement with the ligand (domain in blue and ligand in red). (b) Left: Corresponding surface representation of the lowest energy structure (ligand in dark green). Right: Cartoon representation with the secondary structural elements of the domain shown in green, residues participating in contacts in dark yellow and ligand in dark green.

1264 the different residue found at position 11 (histidine instead of tyrosine present in FBP11WW2 and FBP28WW2), shown to be involved in binding the PPLPx peptides (Figure 1(a) and (b)). Upon addition of increasing amounts of the PPPLIPPPP peptide followed by HSQC experiments (up to an excess of 1:10 protein–ligand ratio) we observed chemical shift changes in residues concentrated in both XP and XP2 grooves. These changes, however, are smaller than the ones previously described in the literature for the PPLPx type of peptides, and also for FBP28WW and the same ligand (Supplementary Data, Figure 1(f)) 8,9 indicating that the domain binds the PPPLIPPPP peptide with lower affinity than that of PPLPx and purely proline stretches. Although the affinity may not be high, and this implies that we have to use a considerable peptide excess for NOESY experiments (1:4 protein–peptide), we looked for the signature pattern of intermolecular NOEs that could define the interaction, as before. In the 2D-NOESY experiment, we observed NOEs from I5′ towards H11 and Y19 (Supplementary Data, Figure 2(c)). Weak NOEs to proline residues were observed, and some of them could be assigned from W30 and Y19 to P8′. The intermolecular NOE volumes, however, are smaller than the equivalent ones observed in the FBP28WW2 complexes, precluding the calculation of a structure. Nonetheless, these observations are compatible with the orientation observed in the case of FBP28WW2 complexed with the PPLIPPPP peptide and are in contrast with the orientation previously described for FBP11WW1 and the PPLPx ligands. Y11H-FBP28WW2 mutant Both FBP11WW1 and FBP28WW2 sequences show a highly conserved XP groove, a single change in the XP2 groove (H to Y) and one in the first loop (P to A). To clarify if the different behavior displayed by both domains towards the PPPLIPPPP ligand is related to the histidine/tyrosine change, we prepared a FBP28WW2 mutant, in which Y11 is mutated to H, to resemble the XP2 groove of FBP11WW1. We then tested the interaction of this mutant with either PPLIPPPP or PPPLIPPPP peptides. Again, upon addition of each peptide to the mutant we observed chemical shift changes in both XP and XP2 grooves. However, regardless of the peptide under investigation, I5′ shows NOEs with residues located in both grooves, indicating that the Y11H change has a critical role in peptide accommodation. We suggest that this is probably due to the size increase of the XP2 cavity as the result of the change of a tyrosine by a smaller histidine, an increase which is not compensated by any means, such as the interaction of P14 in loop1 of FBP11WW1 with the ligand. Thus, in the mutant the peptide cannot find a unique orientation. It therefore seems that the additional N-terminal proline in the ligand is not the unique driving force to dictate the ligand orientation and that the

New Binding Preferences for FBP WW Domains

domain sequence plays a more important role. By comparing the results obtained from looking at FBP28, FBP11WW1 and Y11H-FBP28WW1 it appears that the size of the XP2 groove (Y or H) and the flexibility of loop 1 (P or A) may share responsibility in the fine-tuning of binding. The results from the mutant also support the idea that both binding modes are bona fide observations and not the result of an artifact derived from the use of short synthetic ligands. The role of dual binding modes: cracking the tandem code? FBP28 (also termed CA150) contains two additional group 2 WW domains. The three domains, however, are separated from one another by long loops. The interaction of FBP28WW2 sequence and the FH1 ligand is probably controlled by ligand accessibility, with the contribution of the remaining WW domains in the process of motif selection being negligible. Nevertheless it cannot be discarded that for a given domain, having two binding modes with “PLIP” motifs may facilitate the formation of dynamic complexes, as there is no need to search for a single optimal orientation. In this line of thought is a molecular dynamic simulation study of the GYF domain using wild-type and mutant sequences. In this theoretical work it is suggested that binding variability may be justified by a decrease of the entropic penalty of binding while maintaining the degree of specificity required for function.15 This explanation may also apply for FBP28WW2 and some of its complexes. A different scenario may occur if we consider the interaction of the FBP11 pair of WW domains and FH1. So far, there is little information regarding specific interactions of WW tandems and long ligands. Only the pair of consecutive WW domains present in the yeast splicing factor Prp40 have been shown to constitute a structured tandem.16 FBP11WW1-2 and Prp40WW1-2 sequences are very similar. Moreover, in both cases the WW domains are separated by a short linker of about ten residues. Therefore, it may be that, as described for Prp40, both domains in FBP11 can also form a unique platform and interact at the same time with a long enough region of FH1 containing both PPxPP and PPxxPP motifs. This hypothesis is sustained by the affinity value obtained by fluorescence spectroscopy with the 14 residue peptide (the best affinity constant measured so far for group 2 WW complexes) and by the strong chemical shift changes observed in the FBP11WW1-2 upon titration with this peptide. During the NMR titration (Figure 2(c) and (d)), the most affected residues are located in the XP and XP2 grooves of each domain. For instance, Y20 and Y61 (equivalent residues in WW1 and WW2, respectively), which were overlapped in the free tandem, shift away from each other upon addition of the ligand. The equivalent pairs S28 and S64, Y19 and Y60, T9 and K50 or H11 and Y52 behave in a similar

1265

New Binding Preferences for FBP WW Domains

manner. Interestingly, in all these cases the observed changes are of similar magnitude. Only the residues connecting the first to the second strand behave differently. In fact, the changes corresponding to the first domain are smaller than the ones induced in the second domain. In particular residues S54 and D55 change by 0.5 ppm, which are the biggest changes observed in all binding experiments performed so far with FBP WW domains. In the first domain the equivalent residue of D55 is P14, which is absent in the HSQC experiment, but S13, which corresponds to S54, does not change as much as S54. In addition, several residues in the linker connecting the domains are also affected. For instance, K32, D34, D35, L36, E41, L43, L44 and S45 change significantly. In all, these features support the hypothesis that both domains may participate in the interaction as an entire unit. Considering the structures of the FBP11WW1 and FBP28WW2 complexes, we could try to visualize how the tandem of FBP11WW1-2 domains may interact with the 14 residue ligand. For this we need to consider the key positions for selectivity, the first leucine (L8′ in PTPPLPP peptide) and the isoleucine (I5′ in the PPLIPP peptides). These residues in the 14 residue ligand are separated by four residues, L5′ and I10′. Both the distance and the regular polyproline II conformation of this peptide are significant if, as deduced from the binding experiments, we are to consider cooperative binding. If the PPLIPP region occurs as in the PPPLIPPPP–FBP28WW2 complex, the isoleucine I19′ will be sandwiched by Y60 and W71 in the XP groove of FBP11WW2. Thus, the four amino acid separation appears not to be enough to cover the distance between the XP of the FBP11WW1, where the L5′ should fit, and the XP groove of FBP11WW2. There is therefore a case for thinking that the binding between FBP11WW2 and the PPxxPP peptide might occur as for PPLIPPPP when binding

to the double motif ligand. Using this binding mode, there is no constraint to binding between the first WW domain and its PPLPP ligand. A model of the possible mode of this interaction is shown in Figure 6(a) and (b). The model shows that the domains are oriented in such a way that their sidechains are adequately accommodated, suggesting that both domains can interact at once. In this way they would be able to chaperone each other's function, working cooperatively. This hypothesis was put forward recently, suggesting that binding of tandem WW domains should be even considered independently from single WW domains, since one domain may interfere with the ligand selection of the second.17 Although it seems clear that both domains participate in binding as a unit, and that a global improvement of affinity is observed, an atomic description of the interaction is required to understand which binding mode is used by each domain in the complex.

Conclusions Group 2 WW domains can bind proline-rich ligands ranging from purely poly-proline sequences to others interrupted by either one non-proline residue such us previously identified PPLPx, or two as the PLIP sequences described here. The structures reported here show that the ligand can be accommodated using two binding modes. The selection of the binding mode does not require electrostatic interactions, as observed in SH3 complexes. In this case it seems that the composition and size of the XP2 groove, or even the presence of an additional WW domain, could play a role deciding where to accommodate the non-proline residue, whether in the XP2 or in the XP groove. The results presented here might help understand the mystery of why so

Figure 6. Cartoon representation of a model interaction of a tandem of WW domains and two proline-rich motifs. (a) Surface and (b) cartoon representation for the hypothetical binding of the tandem assuming that the tandem will use both binding modes. Across the surface and on the side, elements of secondary structure and some side-chains are highlighted. The backbone of the peptide is shown in yellow.

1266

New Binding Preferences for FBP WW Domains

Experimental Procedures

meters equipped with either a triple- resonance z-gradient probe or with a z-gradient cryo-probe. 2D 1H-TOCSY and NOESY20 and 3D 15 N-edited NOESY experiments 21 together with the WW assignment corresponding to the free FBP28WW domain10 were used to assign the different complexes. Peptides were assigned in both free and bound forms using 2D 1H-TOCSY, NOESY and ROESY experiments with different mixing times. Similar experiments were run to assign the FBP11WW1-PPPLIPPPP complex. For the tandem construct, the following triple resonance experiments CBCA(CO)NH, CBCANH and a (H)CCC (CO)NH were used. To resolve some ambiguities in the assignment a 15N-NOESY experiment in the presence of three equivalents of ligand was also employed. All these experiments were run at 285 K and on the Bruker DRX600 equiped with a cryo-probe.

Sample preparation

Peptide binding and affinity measurements by NMR

The mouse construct of FBP28WW2 was prepared essentially as described.10 The FBP28H11YWW2 mutant was obtained using the QuickChange site-directed mutagenesis kit (Stratagene). This mutant and both mouse FBP11WW1 and FBP11WW2 as well as the contruct containing both domains (FBP11WW1-2) were cloned into the pETM-11 vector. This modified pET24d expression vector (Novagen) was developed and kindly provided by Gunter Stier (EMBL, Heidelberg). The plasmid codes for Nterminal 6× His tagged linked to the protein of interest with a sequence that contains a TEV protease restriction site. After Nickel affinity purification the protein was cut with the protease and further purified using a by gel filtration on a HiLoad™ Superdex™ 30 prepgrade (GE healthcare Life Sciences). All samples were dissolved in 100 mM sodium phosphate buffer (pH5.8), 50 mM NaCl, 0.02% (w/v) NaN3 in 90% H2O/10% 2H2O or 100% 2H2O. All sequences were confirmed by DNA sequencing and the purified proteins by mass spectrometry.

Addition of each of the four peptides shown in Figure 1 to the 15N- labeled FBP28WW was done until saturation (protein–ligand ratio of up to 1:21). The sum of the geometrical average chemical shift differences (ACSD) were calculated for all of these residues using the equation:

many WW domains appear in two, three or four copies in the same protein. Certainly more structural information of other WW sequences belonging to this group including tandem sequences are required before we can start to outline the rules governing the selection of binding modes. With respect to tandem complexes, so far ignored from both functional and structural points of view, the model of interaction here presented may open a window towards understanding how and why WW domains may operate either solo or as a tandem in regulating cellular mechanisms.

Peptide synthesis From the FH1 sequence APPTPPPLPPPLIPPPPPLPPGLGPLPP, six peptides were synthesized using Fmoc solid phase peptide synthesis:18 APPTPPPLPP, P T P P P L P P, P P P L I P P P P, P P L I P P P P, P P L I P P P, TPPPLPPPLIPPPPP and PPGLGPLPP. Colorimetric tests to ensure amino acid incorporation were done as described19 with either ninhydrin to detect the presence of free primary amines or, when proline residues were incorporated, chloranyl to detect the presence of free secondary amines. In all cases the N-terminal amino group was acetylated while the C terminus contains an amide from the cleavage reaction of the Rink amide support (Novabiochem). All peptides were purified by reverse phase chromatography in a RESOURCE column (Amersham). With this column there is no need to use acidic conditions in the purification protocol, and thus, addition of base to neutralize the peptides is not required. This avoids the presence of a high salt concentration in the ulterior titrations with the protein. Each peptide was identified by mass spectrometry. The concentrations of the respective solutions were quantified by chemical analysis.

ðACSDÞ ¼ ½ðDdN Þ2 þ ð5DdH Þ2 0:5

where Δ δX are the changes in chemical shift in the 1H and 15N dimensions for each point in the titration with respect to the reference spectrum. To automate data treatment of such binding experiments by NMR 1H-15N HSQC titrations, the in-house software AffFE was developed†. Affinity measurements by fluorescence spectroscopy The dissociation constants of the different complexes were determined by fluorescence spectroscopy at 295 K. Fluorescence was measured in an Aminco Bowman series 2 luminescence spectrofluorometer, titrating a 2–5 μM solution of WW domains with 5 mM solutions of each of the peptides. The buffer conditions were the same as for the NMR experiments. Changes in the domain fluorescence emission upon addition of the ligand were used for calculating the dissociation constant as described in legend to Figure 2(e).22 Structure calculation Inter- and intramolecular proton distance restraints were generated from fully assigned peaks obtained from a set of 2D NOESY experiments measured in 90–10 H2O/ 2 H2O and 100% 2H2O using 120, 160 and 180 ms mixing times. NOESY spectra were analyzed and integrated using CARA/XEASY.23 All data were processed using NMRPipe/NMRDraw and/or with Xwinnmr.24 For structure calculation, only experimentally obtained 3J (HN,Hα) coupling constants, hydrogen bond restraints and manually assigned peaks were used. Due to the high proline content of the peptides and the fact that they were all in trans conformation, the peptide was restricted to adopt a polyproline type II helical (PPII) conformation during the calculation. The structures of the complexes

NMR data assignment NMR experiments were carried out at 285 and at 295 K on either a Bruker DRX-500, 600 or DRX-800 spectro-

ð1Þ

† Available at www.maciasnmr.net

1267

New Binding Preferences for FBP WW Domains were calculated using CNS25 and ARIA 1.226 program with in-house modifications. Since only unambiguously assigned restraints were used the protocol was reduced to two iterations of 1 and 60 structures, respectively, with an additional step of water refinement. Cooling steps were increased, as recently recommended27 to 16,000 even though the great majority of peaks used in this calculation were unambiguously assigned. We observed that water refinement using the default protocol values produced structures with better Ramachandran values but with more NOE violations than the input structures used to perform the refinement, in agreement with Nilges' and coworker's observations.26 As these authors claimed, introduction of modifications in the run.cns protocol, which only affect the water refinement protocol, and not the protocols used for the previous iterations are difficult to define. Thus, we instead modified the water refinement protocol itself, by weighing the value of hydrogen bonds, NOEs and dihedrals restraints by a factor of 10. In this way all experimental restraints are used during the refinement process and we thus obtain structures in better agreement with our data while retaining good Ramachandran values. The modified protocol will be provided upon request. The quality of the final structures was evaluated using the program PROCHECK-NMR.28 Models Generated with ARIA1.2 program, with the same protocol used to calculate the complexes. The full-length peptide was restricted to adopt a polyproline type II helix due to the high proline content and also due to the experimental evidence obtained from the NMR analysis of the peptide. All remaining restraints (hydrogen bonds, dihedral and NOEs) were taken from the three isolated complexes and combined into a single file. Data Bank accession codes Deposition of assignments and coordinates- BioMagResBank database‡ accession numbers 11007,11008 15453 and Protein Data Bank§ accession number 2rly, 2rmO, 2jup.

Acknowledgements We acknowledge the IRB for financial support and to the NMR facility of the University of Barcelona for measurement time and support. We are indebted to J. Ashurst, N. Görner and S. Moulton for some sample preparations, to B. Morales for assistance with the fluorescence measurements and to H.-Y. Hu for the amide assignment of a FBP11WW1-2 construct. X.R.E. and P.M.M. acknowledge EMBL and the IRB for pre-doctoral fellowships. This work was supported by the Spanish Ministerio de Educación y Ciencia, grants: GEN2003-20642-C09-04 and BFU2005-06276 (to M.J.M.) and also for financial support to access to the NMR facility. ‡ http://www.bmrb.wisc.edu § http://www.rcsb.org

Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2007.08.052

References 1. Wallar, B. J. & Alberts, A. S. (2003). The formins: active scaffolds that remodel the cytoskeleton. Trends Cell Biol. 13, 435–446. 2. Romero, S., Le Clainche, C., Didry, D., Egile, C., Pantaloni, D. & Carlier, M. F. (2004). Formin is a processive motor that requires profilin to accelerate actin assembly and associated ATP hydrolysis. Cell, 119, 419–429. 3. Rivero, F., Muramoto, T., Meyer, A. K., Urushihara, H., Uyeda, T. Q. & Kitayama, C. (2005). A comparative sequence analysis reveals a common GBD/FH3-FH1FH2-DAD architecture in formins from Dictyostelium, fungi and metazoa. BMC Genom. 6, 28. 4. Chan, D. C., Bedford, M. T. & Leder, P. (1996). Formin binding proteins bear WWP/WW domains that bind proline-rich peptides and functionally resemble SH3 domains. EMBO J. 15, 1045–1054. 5. Bork, P. & Sudol, M. (1994). The WW domain: a signalling site in dystrophin? Trends Biochem. Sci. 19, 531–533. 6. Macias, M. J., Wiesner, S. & Sudol, M. (2002). WW and SH3 domains, two different scaffolds to recognize proline-rich ligands. FEBS Letters, 513, 30–37. 7. Pires, J. R., Parthier, C., Aido-Machado, R., Wiedemann, U., Otte, L., Bohm, G. et al. (2005). Structural basis for APPTPPPLPP peptide recognition by the FBP11WW1 domain. J. Mol. Biol. 348, 399–408. 8. Kato, Y., Miyakawa, T., Kurita, J. I. & Tanokura, M. (2006). Structure of a FBP11 WW1-PL ligand complex reveals the mechanism of proline-rich ligand recognition by group-II/III WW domains. J. Biol. Chem. 281, 40321–40329. 9. Kato, Y., Hino, Y., Nagata, K. & Tanokura, M. (2006). Solution structure and binding specificity of FBP11/ HYPA WW domain as Group-II/III. Proteins:Struct. Funct. Genet. 63, 227–234. 10. Macias, M. J., Gervais, V., Civera, C. & Oschkinat, H. (2000). Towards an understanding of beta-sheet structures: design of a prototype WW domain. Nature Struct. Biol. 7, 375–379. 11. Gao, Y. G., Yan, X. Z., Song, A. X., Chang, Y. G., Gao, X. C., Jiang, N. et al. (2006). Structural insights into the specific binding of huntingtin proline-rich region with the SH3 and WW domains. Structure, 14, 1755–1765. 12. Hays, J. L. & Watowich, S. J. (2003). Oligomerizationinduced modulation of TPR-MET tyrosine kinase activity. J. Biol. Chem. 278, 27456–27463. 13. Kato, Y., Nagata, K., Takahashi, M., Lian, L., Herrero, J. J., Sudol, M. & Tanokura, M. (2004). Common mechanism of ligand recognition by group II/III WW domains: redefining their functional classification. J. Biol. Chem. 279, 31833–31841. 14. Morales, B., Ramirez-Espain, X., Shaw, A. Z., MartinMalpartida, P., Yraola, F., Sanchez-Tillo, E. et al. (2007). NMR structural studies of the ItchWW3 domain reveal that phosphorylation at T30 inhibits the interaction with PPxY-containing ligands. Structure, 15, 473–483. 15. Gu, W., Kofler, M., Antes, I., Freund, C. & Helms, V.

1268

16.

17. 18. 19. 20. 21.

22.

(2005). Alternative binding modes of proline-rich peptides binding to the GYF domain. Biochemistry, 44, 6404–6415. Wiesner, S., Stier, G., Sattler, M. & Macias, M. J. (2002). Solution structure and ligand recognition of the WW domain pair of the yeast splicing factor Prp40. J. Mol. Biol. 324, 807–822. Sudol, M., Recinos, C. C., Abraczinskas, J., Humbert, J. & Farooq, A. (2005). WW or WoW: the WW domains in a union of bliss. IUBMB Life, 57, 773–778. Wellings, D. A. & Atherton, E. (1997). Standard Fmoc protocols. Methods Enzymol, 289, 44–67. Vazquez, J., Qushair, G. & Albericio, F. (2003). Qualitative colorimetric tests for solid phase synthesis. Methods Enzymol. 369, 21–35. Wüthrich, K. (1986). NMR of Proteins and Nucleic Acids. John Wiley & Sons, Inc., New York. Sattler, M., Schleucher, J. & Griesinger, C. (1999). Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Prog. NMR Spectrosc. 34, 93–158. Macias, M. J., Hyvonen, M., Baraldi, E., Schultz, J., Sudol, M., Saraste, M. & Oschkinat, H. (1996). Structure of the WW domain of a kinase-associated protein complexed with a proline-rich peptide. Nature, 382, 646–649.

New Binding Preferences for FBP WW Domains

23. Bartels, C., Xia, T.-H., Billeter, M., Güntert, P. & Wüthrich, K. (1995). The program XEASY for computer-supported NMR spectral analysis of biological macromolecules. J. Biomol. NMR, 5, 1–10. 24. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J. & Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR, 6, 277–293. 25. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. 26. Linge, J. P., Williams, M. A., Spronk, C. A., Bonvin, A. M. & Nilges, M. (2003). Refinement of protein structures in explicit solvent. Proteins:Struct. Funct. Genet. 50, 496–506. 27. Fossi, M., Oschkinat, H., Nilges, M. & Ball, L. J. (2005). Quantitative study of the effects of chemical shift tolerances and rates of SA cooling on structure calculation from automatically assigned NOE data. J. Magn. Reson. 175, 92–102. 28. Laskowski, R. A., Rullmannn, J. A., MacArthur, M. W., Kaptein, R. & Thornton, J. M. (1996). AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J. Biomol. NMR, 8, 477–486.