- Email: [email protected]
ponsin by ataxin-7
FEBS Letters 587 (2013) 2905–2911
journal homepage: www.FEBSLetters.org
Structural basis for recognition of the third SH3 domain of full-length R85 (R85FL)/ponsin by ataxin-7 q Ya-Jun Jiang, Chen-Jie Zhou, Zi-Ren Zhou, Meng Wu, Hong-Yu Hu ⇑ State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China
a r t i c l e
i n f o
Article history: Received 18 April 2013 Revised 1 June 2013 Accepted 4 July 2013 Available online 24 July 2013
Edited by Berend Wieringa Keywords: Ataxin-7 Proline-rich region R85FL/ponsin SH3 domain Polyglutamine Inclusion body
a b s t r a c t Ataxin-7 (Atx7) is a component of the nuclear transcription co-activator complex; its polyglutamine (polyQ) expansion may cause nuclear accumulation and recruit numerous proteins to the intranuclear inclusion bodies. Full-length R85 (R85FL) is such a protein sequestered by polyQ-expanded Atx7. Here, we report that Atx7 specifically interacts with the third SH3 domain (SH3C) of R85FL through its second portion of proline-rich region (PRR). NMR structural analysis of the SH3C domain and its complex with PRR revealed that SH3C contains a large negatively charged surface for binding with the RRTR motif of Atx7. Microscopy imaging demonstrated that sequestration of R85FL by the polyQ-expanded Atx7 in cell is mediated by this specific SH3C–PRR interaction, which is implicated in the pathogenesis of spinocerebellar ataxia 7. Structured summary of protein interactions: Atx7 PP2 and SH3C bind by isothermal titration calorimetry (View Interaction: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12) Atx7 binds to SH3C by pull down (View interaction) Atx7100Q and SH3C colocalize by fluorescence microscopy (View interaction) SH3C and Atx7 bind by nuclear magnetic resonance (View interaction) Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
1. Introduction Spinocerebellar ataxia 7 (SCA7) is an autosomal dominant neurodegenerative disorder characterized by cerebellar ataxia associated with progressive macular dystrophy [1,2]. SCA7 is generally considered to be caused by expansion of a CAG repeat encoding polyglutamine (polyQ) tract in the protein ataxin-7 (Atx7) [3]. Atx7 is a component of mammalian STAGA (SPT3-TAF9-ADAGCN5 acetyltransferase) transcription co-activator complex, and its polyQ expansion inhibits STAGA histone acetyltransferase activity [4,5]. Moreover, its yeast form regulates ubiquitin levels of histone H2B and links histone deubiquitination with gene gating and mRNA export [6]. Thus, impairments of these normal functions by polyQ expansion of Atx7 may underlie the production of SCA7 phenotype. Abbreviations: Atx7, ataxin-7; polyQ, polyglutamine; PRR, proline-rich region; R85FL, full-length R85; SCA7, spinocerebellar ataxia 7; SH3, Src homology 3 q The atomic coordinates and structure factors of free and PP2-bound SHC3 have been deposited in the PDB under codes 2LJ0 and 2LJ1, respectively. ⇑ Corresponding author. Address: Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, 320 Yue-yang Road, Shanghai 200031, China. E-mail address: [email protected] (H.-Y. Hu).
There are nine inherited neurodegenerative disorders related to polyQ expansion, the hallmark of this kind of disease is the formation of nuclear inclusion bodies [7]. The intranuclear inclusions found in human SCA7 brain contain many other proteins, including transcriptional regulators, ubiquitin/proteasome pathway components, cell death associated proteins, and chaperones and their partners [1]. The function of all these proteins can possibly be influenced by polyQ expansion of Atx7. However, whether these proteins are recruited to inclusions by polyQ expanded Atx7 through specific or non-specific protein interactions remains largely unknown. One such protein sequestered by polyQ expanded Atx7 to intranuclear inclusions is full-length R85 (R85FL), a splice variant of ponsin or Cbl-associated protein (CAP). R85FL/ponsin belongs to the SoHo adaptor protein family that regulates cytoskeletal organization and signal transduction [8]. There are three members of this family; all contain three highly conserved Src homology 3 (SH3) domains in the C-termini. As known, SH3 domain specifically recognizes proteins with proline-rich region (PRR) [9–11]. The canonical binding sites for SH3 domains usually include the amino-acid sequences like R/ KxxPxxP (class I) or PxxPxR/K (class II) [10,12]. Like huntingtin (Htt), Atx7 contains a PRR just downstream of the polyQ tract [13].
0014-5793/$36.00 Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.febslet.2013.07.021
2906
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
We herein report identification and structural analysis for the interaction of Atx7 PRR with the third SH3 domain (SH3C) of R85FL. Moreover, R85FL can be sequestered to the nuclear inclusion bodies by polyQ expanded Atx7 through specific SH3C–PRR interactions.
room temperature for 5 h to remove MTSL from the Atx7_PP2 peptide, and then another HSQC spectrum was acquired for reference [15]. The peak intensities with and without MTSL in the HSQC spectra of SH3C were compared for identifying the residues that are close to the N- or C-terminus of the peptide.
2. Materials and methods
2.5. Construction of the complex structure
2.1. Cloning, expression, peptide synthesis, and protein purification
The complex structure was generated by HADDOCK2.1 [16]. The structures of SH3C in the peptide-bound form solved by NMR and Atx7-PP2 generated by homology modeling from the known proline-rich peptide (PDB ID: 2DF6) were used as initials for docking. The residues that have chemical-shift changes larger than the average value were treated as significantly perturbed residues. The active residues included Y13, Y15, N19, D21, D39, W41, P56 and Y59 of SH3C, and all residues of Atx7-PP2. The PRE-derived intermolecular distances were introduced as unambiguous restraints. Only those residues with resonance peaks significantly broadened or almost disappeared by paramagnetic effect were used as inputs for structure calculation. Normally the distance between the amide nitrogen atoms of SH3C and the paramagnetic oxygen of MTSL was set as 10 ± 3 Å. The ambiguous constraints were further modified by the fact that Pro56 of SH3C has intermolecular NOEs with the prolines of Atx7_PP2. The docking protocol included three stages with a final explicit water refinement. At the last iteration, 200 structures were generated and clustered by an RMSD cut-off of 3 Å.
The R85FL gene was cloned from human brain cDNA library (Invitrogen), and inserted into pcDNA3.0 vector with a FLAG tag in N-terminus of the protein. The Atx7 gene with 10 or 100 glutamine repeats in the vector pEGFPN1 was a gift from Dr. A. Brice. The SH3 domains and their mutants were subcloned into pGBTNH [14] and pGEX-4T-3. The peptides were obtained from solid-phase chemical synthesis and analyzed by electrospray mass spectrometry. The plasmids were transformed into Escherichia coli BL21 (DE3) cells to overexpress proteins. The proteins were purified by Ni2+– NTA or GST affinity column, followed by gel filtration chromatography. 15N/13C-labeled proteins were prepared using the M9 minimal medium containing 15NH4Cl and/or 13C6-D-glucose as the sole nitrogen and/or carbon resource, respectively. 2.2. GST pull-down experiments The GST-peptide fusions were added to the glutathione Sepharose 4B beads (Amersham Biosciences) in a PBS buffer (140 mM sodium chloride, 2.7 mM potassium chloride, 10 mM sodium phosphate, 1.8 mM potassium phosphate, pH 7.3), and the suspensions were agitated at 4 °C for 30 min. The beads were washed three times with the same buffer to remove any unbound proteins. An equivalent amount of different SH3 domains were added and incubated at 4 °C for an additional 4 h. Finally, the beads were recovered by centrifugation, and then the samples were resuspended in the sample buffer and subjected to SDS–PAGE (15% gel), followed by Coomassie staining. 2.3. NMR spectroscopy and structure determination The sample containing 15N/13C-labeled SH3C protein (1.0 mM) with or without Atx7_PP2 peptide (4 mM) in a phosphate buffer (20 mM NaH2PO4, 50 mM NaCl, pH6.5) was used for the structural determination by NMR. All data were acquired at 25 °C on a Bruker Avance 600-MHz spectrometer. The spectra including HNCO, HNCACB, CBCA(CO)NH, H(CCO)NH, C(CO)NH and HCCH-TOCSY were obtained for the backbone and side-chain assignments. NOE restraints were obtained from 15N- and 13C-edited NOESY spectra with a mixing time of 100 ms. The NMR data were processed using NMRPipe and analyzed with SPARKY. Backbone dihedral restraints were derived from TALOS program. The structures were calculated using ARIA2.0, assessed with PROCHECK, and displayed with MOLMOL. 2.4. MTSL labeling and paramagnetic relaxation enhancement (PRE) A Cys residue was introduced to the N- or C-terminus of Atx7_PP2, and the peptide fused with GB1 was obtained with pGBTNH expression vector [14]. After purification, MTSL ((1-oxyl2,2,5,5-tetramethyl pyrroline-3-methyl) methanethiosulfonate) (ALEXIS Biochemicals, San Diego) was incubated with Atx7_PP2 at 4 °C overnight. Ten-fold excess of MTSL was used for the reaction and the excess MTSL was removed by desalting [11]. In the PRE experiment, after collecting a HSQC spectrum, dithiothreitol (DTT, 10 mM) was added to the mixture in the NMR tube and stayed at
2.6. Isothermal titration calorimetry (ITC) ITC was performed with an ITC200 MicroCalorimeter (MicroCal). Proteins were dissolved at 100 lM in a PBS buffer (140 mM sodium chloride, 2.7 mM potassium chloride, 10 mM sodium phosphate, 1.8 mM potassium phosphate, pH 7.3; or 20 mM sodium phosphate, 50 mM sodium chloride, pH 6.5 or 7.0). Upon reaching the equilibrium temperature of 25 °C, different peptides were titrated from a syringe into a sample chamber holding by 20 injections. Data were obtained using v2 minimization on a model assuming a single set of sites to calculate the binding affinity. 2.7. Cell culture and immunofluorescence microscopy HEK 293T cells were transfected by FuGENE HD transfection reagent (Roche) following the manufacturer’s instructions. AntiFLAG antibody (Sigma) was used for immuno-fluorescence. For confocal microscopy, HEK 293T cells were grown on glass coverslips for 48 h after transfection. Images were obtained on a Leica TCS SP2 confocal microscope (Leica Microsystems). 3. Results 3.1. Atx7 PRR specifically interacts with the third SH3 domain of R85FL By using two-hybrid approach to screen human retina cDNA library, a protein named R85 was identified for binding with Atx7 [13]. In addition, the R85FL was also found to co-localize with Atx7 [13]. R85FL is a splice variant of the SH3P12 gene product. All variants have three repeated SH3 domains located at the C-termini (Fig. 1A). As known, SH3 is a modular domain recognizing proline-rich motifs [9,11], while Atx7 contains a PRR motif located at the N-terminus just behind the polyQ tract (Fig. 1A). To get molecular details of the recognition between Atx7 and R85FL, we examined the interactions of Atx7 PRR with different SH3 domains of R85FL by GST pull-down analysis. The results show that Atx7 PRR can interact with the third SH3 domain (as referred to SH3C)
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
2907
(Fig. 1D), but it neither binds with the first domain (SH3A) (Fig. 1B) nor the second (SH3B) (Fig. 1C). SH3 domain canonically binds with such amino-acid sequence as R/KxxPxxP (class I) or PxxPxR/ K (class II) [10,12]. So we dissected the Atx7 PRR sequence into two parts, namely Atx7_PP1 and Atx7_PP2 (Fig. 1A), each containing a class-II consensus peptide motif. Intriguingly, SH3C only interacts with Atx7_PP2 but not with Atx7_PP1 (Fig. 1E). ITC experiments indicate that Atx7_PP2 has almost the same affinity (KD = 68 lM) to SH3C when compared with the full-length Atx7 PRR sequence (Atx7_PP12) (Fig. S1), suggesting that the PP2 motif in Atx7 is the major binding moiety for SH3C. NMR titration further corroborates our observation from GST pull-down experiments (Fig. S2). The dissociation constant (KD) for SH3C binding with Atx7_PP2 is around a range of 50 lM. However, SH3A is incapable of binding with Atx7_PP1 or Atx7_PP2. Altogether, these results demonstrate that the PP2 motif of Atx7 specifically interacts with the third SH3 domain of R85FL. 3.2. Solution structures of SH3C in free and peptide-bound states
Fig. 1. Atx7 PRR specifically interacts with the third SH3 domain of R85FL. (A) Schematic representation of Atx7, R85FL and the PRR peptide portions of Atx7. PP12, the PRR sequence of Atx7; PP1, the first portion; PP2, the second portion. (B) GST pull-down experiment for testing the interaction of GST-PP12 with SH3A. (C) As in (B), GST-PP12 with SH3B. (D) As in (B), GSP-PP12 with SH3C. (E) GST pulldown experiment for testing the interaction between GST-PP1 or GST-PP2 and SH3C. The GST pull-down experiments were detected by Coomassie blue staining. The extra bands in the gels are due to degradation of GST or GST fused proteins.
The SoHo family adaptor proteins include three members, vinexin, Arg-binding protein 2 (ArgBP2) and R85FL (ponsin or CAP). Each protein has one SoHo domain in N-terminus and three SH3 domains in C-terminus. The SH3 domain is highly conserved among the family members with an identity up to 60% (Fig. S3), whereas only 17% identity among the three SH3 domains of R85FL (Fig. 2A). The structures of SH3A (2DL3) and SH3B (2O9S) are available in the Protein Data Bank. By using heteronuclear NMR approach, we solved the solution structures of the SH3C domain of R85FL in free and PP2bound states. The best 15 final structures, as judged from energetic and geometrical considerations, have a backbone RMSD of 0.44 ± 0.08 Å for SH3C alone or 0.61 ± 0.14 Å for it binding with
Fig. 2. Solution structures of SH3C in free and peptide-bound states. (A) Sequence alignment of the three SH3 domains from R85FL. The conserved residues are in blue box, while the negatively charged residues in SH3C that are not present in SH3A or SH3B are indicated with arrows. Positions of the b-strands, 310 helix, RT loop, N-Src loop and distal loop are indicated at the top of the sequence. (B) A cluster of 15 best structures of SH3C in free state (PDB ID: 2LJ0). (C) A cluster of 15 best structures of SH3C in Atx7_PP2 bound state (PDB ID: 2LJ1). (D) Comparison of the average structures of SH3C in free (blue) and peptide bound (green) states. (E) Surface representation for the three SH3 domains from R85FL. The positive and negative charges are shown in blue and red, respectively, while the hydrophobic regions are shown in gray. PDB ID: SH3A (2DL3), SH3B (2O9S) and SH3C (2LJ0).
2908
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
Fig. 3. Interaction between Atx7_PP2 and SH3C of R85FL. (A) The binding affinity of SH3C with Atx7_PP2 is dependent on the salt concentration in solution. The dissociation constant (KD) values were obtained from ITC experiments in pH 6.5 or pH 7.0. (B) Chemical-shift perturbation of SH3C upon addition of Atx7_PP2 at the PP2/SH3C ratio of 6:1. The chemical-shift change was calculated by a combination of hydrogen and nitrogen ðDdav ¼ SQRTðDd2H þ Dd2N =25ÞÞ. The residues with chemical-shift changes larger than the average (mean) were treated as significantly perturbed residues, including Y15, Q18, N19, D20, D21, I30, V31, W41, V43, T54, F55 and Y59. (C) Intermolecular NOEs between Atx7_PP2 and SH3C. The three columns represent 13C-edited NOE signals for the side-chain of Pro56 in SH3C (left), 13C-edited NOEs for Pro56 saturated with Atx7_PP2 (middle), and the 13C-filtered/edited intermolecular NOEs for Pro56 with Atx7_PP2 (right). The arrows indicate the intermolecular NOEs for Pro56 of SH3C with prolines of Atx7_PP2. (D) Ratios of the peak intensities in the HSQC spectra with or without titration of N-terminally MTSL-labeled Atx7_PP2. The region of residues with strong paramagnetic effect is indicated by an arrow. (E) As in (D), with C-terminally labeled Atx7_PP2 peptide. Table 1 Binding affinities (KD) of SH3C or its mutants with Atx7_PP2 peptide or its mutants as determined by isothermal titration calorimetry. Atx7 PRR Peptide
PP12
PP2
PP2-R6A
PP2-R7A
PP2-T8A
PP2-R9A
68.0 ± 4.6
980 ± 19
253 ± 45
862 ± 44
588 ± 96
D20A 139 ± 24
D21A >1000
D39A 571 ± 41
T54A 78.1 ± 10.9
Y59A >1000
R85L_SH3C KD (lM) SH3C Mutant KD (lM)
46.5 ± 7.1 Atx7_PP2 Y13A >1000
the PP2 peptide (Table S1). Both structures of the free and PP2bound forms display a b-barrel fold of five b-strands and a 310 helix (Fig. 2B and C), showing a typical SH3 domain architecture. In comparison of the two average structures in free and PP2-bound states, it appears that binding of the PP2 peptide does not cause any significant conformational change (Fig. 2D), because the backbone RMSD of the two structures is only 1.22 Å. There is only a small structural difference in the RT-loop and n-Src loop regions, which are the most flexible regions of SH3 domains but engaged in the peptide binding. The binding surface for PRR on a typical SH3 domain can be divided into two areas, one is the PxxP binding pocket and the other is located in between the RT-loop and n-Src loop [17]. The second region is contributable to the ligand specificity of SH3 domain. The significant difference among the three SH3 domains of R85FL is that SH3C contains more negatively charged residues (e.g. D20, D21 and D39) on its surface (Fig. 2E). This negatively charged area
lies in between the RT-loop and n-Src loop. In comparison, the corresponding surface of SH3A is roughly hydrophobic almost without negative charge, while the surface of SH3B contains less negative charge than that of SH3C. This charge difference might provide molecular basis for the specificities of the SH3 domains binding with partners. Actually, SH3B of R85FL binds with non-canonical proline-rich motif [18], while SH3A of vinexin, a highly conserved homolog of SH3A of R85FL, can also bind with a non-canonical proline-rich motif [19]. 3.3. Structural insights into the specific interaction between SH3C and Atx7_PP2 Since SH3C has a negatively charged surface and Atx7_PP2 contains a RRTR motif with positive charges, the electric charges on the surface may play important roles in mediating the electrostatic
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
2909
Fig. 4. Complex structure of SH3C with Atx7_PP2. (A) Surface representation of SH3C showing the grooves for Atx7_PP2 binding. The positive charges, negative charges, and the hydrophobic regions are shown in blue, red and gray, respectively. The R6, R7, T8 and R9 residues of Atx7_PP2 are illustrated in blue, while other residues are shown in cyan. The hydrophobic interaction (left) and electrostatic interaction (right) are also highlighted in the graphs. (B) Detailed view of the interfaces for interaction of R85FL_SH3C (ribbon) with Atx7_PP2 (stick). (C) As in (B), for interaction of R85FL_SH3B (ribbon) and the PRR peptide of paxillin (stick) (PDB ID, 2O9V).
interaction between the two peptide chains, thus ion strength may also influence the complex. Indeed, ITC experiments show that the dissociation constant (KD) for SH3C binding with PP2 is significantly increased with an increase of salt concentration in the buffer (Fig. 3A), suggesting that the electrostatic interactions are contributable to the formation of the SH3C–PP2 complex. NMR chemicalshift perturbation exhibits that some residues like Y15, D21, D39, W41 and Y59, experience significant chemical-shift changes upon PP2 binding (Fig. 3B). These residues are also contributable to constructing the xP groove and the specificity pocket of SH3 domain [9,10,12]. Thus, both the proline-mediated hydrophobic interaction and the arginine-conducted electrostatic interaction are crucial to forming the complex between Atx7_PP2 and SH3C. More evidence comes from the NOE signals between Atx7_PP2 and SH3C (Fig. 3C). The b-methylene group of Pro56 of SH3C has only intraresidue NOE signals in the free state, whereas some intermolecular NOE signals appear in the complex, though they are still unsigned due to 5 continuous prolines in Atx7_PP2. The intermolecular NOE data from 13C filtered/edited NOESY spectra are confirmable for this assumption (Fig. 3C, right panel). According to the chemical-shift pattern, these intermolecular NOE signals may come from one of the proline residues of Atx7_PP2. We further characterized the interaction between SH3C and Atx7_PP2 by PRE experiments. The cross peaks of Y13, S14, Y15 and I16 of SH3C are significantly broadened when it binds with the N-terminally MTSL-labeled peptide (Fig. 3D), indicating that the N-terminus of Atx7_PP2 is seated around the xP groove of SH3C formed by Y13 and Y15. Meanwhile, C-terminally labeled Atx7_PP2 causes large peak broadening of D38, D39 and G40 (Fig. 3E), suggesting that the C-terminus of
Atx7_PP2 interacts with the acidic residues in the N-Src loop of SH3C. The ITC experiments on various mutants of both Atx7_PP2 and SH3C further confirm the viewpoint that the extended RRTR motif of Atx7_PP2 and the key residues forming the xP groove and specificity pocket of SH3C are important for the specific interaction (Table 1). The dramatic decrease of the binding affinities in the D21A and D39A mutants can account for the different recognition modes for polyproline peptides among the three SH3 domains of R85FL (Fig. 1 and Fig. S2). Taken together, these results suggest that both the xP groove and specificity pocket of SH3C are engaged in the interaction with Atx7_PP2, while both the PPPPP and RRTR motifs in Atx7_PP2 are also contributable to the specific interaction. 3.4. Structure model for the complex of SH3C and Atx7_PP2 Due to the relatively weak interaction and difficulty in NMR assignment of the Atx7_PP2 peptide, we have not obtained enough unambiguous intermolecular NOEs for direct computation of the complex structure, so we generated a rational complex structure by HADDOCK method [16], using the information from chemical shift perturbation (Fig. 3B), intermolecular NOEs (Fig. 3C), PRE restraints (Fig. 3D and E) and binding affinities of various mutants (Table 1). With the experimental structure constraints as an input, a major cluster of the initial complex structures were generated, and a refined structure was finally obtained (Fig. 4). In the structure model, the PPPPP motif is docked on the hydrophobic xP groove of SH3C and the RRTR motif is directed to the negatively charged specificity pocket (Fig. 4A). The prolines of Atx7_PP2 are contacting with the xP groove of SH3C formed by Y13, Y15, W41,
2910
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
to dysfunction [7]. Using confocal microscopy, we investigated the c0-localization of Atx7 with R85FL in cells. Normal Atx7 (Atx710Q) is a nucleus localized protein, while R85FL is distributed mainly in cytosol (Fig. 5, 1st row). The polyQ expanded Atx7 (Atx7100Q) forms nuclear inclusion bodies, which can sequester R85FL into the nucleus (Fig. 5, 2nd row). As a control, double mutation (I16A/Q18A) in the SH3C moiety of R85FL, which does not alter the interaction with Atx7_PP2, can also be sequestered into the nucleus by Atx7100Q (Fig. 5, 3rd row). However, the D39A mutant in the SH3C moiety of R85FL that significantly reduces the binding affinity to the PP2 motif of Atx7 cannot be sequestered by Atx7100Q (Fig. 5, 4th row). Moreover, deletion of SH3C in R85FL significantly abolishes sequestration of R85FL into the nucleus by Atx7100Q (Fig. 5, 5th row). This observation indicates that the SH3C–PP2 interaction mediates the association of Atx7 and R85FL in cells. The polyQ expanded Atx7 is prone to formation of inclusion bodies and may significantly alter the localization of R85FL from cytosol to nucleus. This kind of abnormal interactions of the polyQ expanded Atx7 with other partners may be implicated in the pathological process of SCA7.
4. Discussion
Fig. 5. PolyQ expanded Atx7 (Atx7100Q) can sequester R85FL in cells. Normal Atx710Q-GFP and polyQ expanded Atx7100Q-GFP were co-transfected with either wild-type R85FL or its mutants in the SH3C moiety into HEK 293T cells, and performed imaging with confocal fluorescence microscopy. Nuclei were stained with DAPI. I16A/Q18A is a double mutant of R85FL in the SH3C moiety, while D39A is a point mutant of R85FL in the SH3C domain. DSH3C stands for the SH3C deletion mutant of R85FL. Scalar bar = 10 lM.
P56 and Y59 (Fig. 4B). The positively charged RRTR motif is located in the negatively charged specificity pocket formed by D21, E22, D38 and D39. On the Atx7_PP2 side, R6 is pointed to the acidic pocket of SH3C formed by D21 and E22, while both R7 and R9 located at the pocket formed by D38 and D39. This is different from the complex structure of SH3B of R85FL, in which the paxillin peptide forms a type-II polyproline helix conformation [18]. The specificity pocket of SH3B is comprised of Val, Asp, Asn and Ile residues, corresponding to the residues of C37, D38, G40 and T54 in SH3C (Fig. 2A), but there is no any charge involved in the interaction. In essence, this structure model can explain the binding specificity for Atx7_PP2 and the unique property of SH3C among the SH3 domains of R85FL. 3.5. PolyQ expanded Atx7 (Atx7100Q) can sequester R85FL through specific interaction with SH3C Formation of inclusion bodies is the hallmark of polyQ expanded neurodegenerative diseases, and the cellular inclusions can sequester various kinds of proteins and probably lead them
SH3 is a modular domain commonly found in all eukaryotic genomes [10]. R85FL/ponsin is an adaptor protein that harbors three SH3 domains in the C-terminus. Adaptor proteins are usually composed of multiple modular domains for linking different parts of enzymatic cascade or signal transduction pathway [20,21]. So the three SH3 domains of R85FL must have evolved unique properties to accommodate different partners. Comparing the structures of these three SH3 domains from R85FL reveals that the major difference for binding diverse PRR partners comes from the negatively charged region between the RT loop and n-Src loop (Fig. 2E). This structural difference among these three SH3 domains may provide a possibility for R85FL to function both specifically and versatilely in living cells. Most SH3 domains recognize linear polyproline motif through two pockets; one binds to the canonical PxxP core motif and the other binds to the extended motif beyond the PxxP core [17]. Both the canonical and non-canonical peptides associate with SH3 domains through these two pockets. Although lots of SH3 domain structures are available, it is still very hard to predict the interaction partners of a SH3 domain, largely because of the high versatility of SH3 domains. Previous studies showed that the SH3C domain of R85FL can interact with c-Cbl [22], 1-Afadin [23] and c-Abl [24]. The SH3C domain interacts with the PRR-containing partners with a novel consensus motif of PxPPxRxSSL pattern at their binding sites [25]. Besides the PxxP motif, the extended motif also plays a crucial role in SH3 domain recognition [26–29]. However, these SH3 domains bind with different polyproline peptides with the dissociation constants around 1 lM, whereas the binding affinity for R85FL_SH3C with Atx7_PP2 is much weaker, till to an extent of about 50 lM. The RxxK motif of Gab2 was reported to recognize the SH3C domain of Grb2 [30]; both side-chains of Arg and Lys residues point to the negatively charged region of the RT loop (Fig. S4A). The interaction between Atx7 and R85FL is a little different from that between Gab2 and Grb2. The negatively charged region of R85FL_SH3C includes both RT loop and n-Src loop, and each residue of the RRTR motif in Atx7 contributes to the interaction (Table 1). Another example is from the structural study of Mona/ Gads SH3C [27], which recognizes a PxxPxR/KxxK motif. The PxxP motif binds to the two xP grooves of SH3C, while the R/KxxK motif points to the negatively charged region of the RT loop (Fig. S4B). To our knowledge, the interaction between R85FL_SH3C and Atx7_PP2 represents a novel pattern. The diversity of the binding
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
partners for an SH3 domain is expanded by combination of the PxxP motif and an extended motif. PolyQ disorders share several common pathological features, including the nuclear accumulation and aggregation of the disease proteins [7,31]. Numerous proteins can be sequestered by the nuclear inclusions, so that their functions must be abolished, which can also be recognized as a pathogen for the polyQ diseases. Our previous work has shown that polyQ expanded huntingtin (Htt) can sequester Htt two-hybrid protein A (HYPA) through its PRR interaction, and thus impairs the function of HYPA in pre-mRNA splicing [32]. It is quite possible to be a similar case in R85FL. The R85FL splicing variant ponsin/CAP regulates cytoskeleton organization [33,34], insulin signal pathway [22,35,36], and receptor-mediated endocytosis [37]. Actually, R85FL colocalizes with Atx7 in the brain of SCA patients [13]. The sequestration of R85FL by polyQ expanded Atx7 possibly leads to impairment of the normal function of R85FL, which may partially contribute to the pathology of the SCA7 disease. Acknowledgements The authors would thank Dr. A. Brice for providing the plasmid coding for ataxin-7, Dr. C.W. Jin at Peking University for technical assistance in NMR data acquisition, and Dr. Alexander M.J.J. Bonvin for the help of generating the complex structure using HADDOCK software. This work was supported by the grants from the National Basic Research Program of China (2011CB911104, 2012CB911003), and the National Natural Science Foundation of China (10979070). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.febslet.2013. 07.021. References [1] Lebre, A.S. and Brice, A. (2003) Spinocerebellar ataxia 7 (SCA7). Cytogenet. Genome Res. 100, 154–163. [2] Garden, G.A. and La Spada, A.R. (2008) Molecular pathogenesis and cellular pathology of spinocerebellar ataxia type 7 neurodegeneration. Cerebellum 7, 138–149. [3] David, G. et al. (1997) Cloning of the SCA7 gene reveals a highly unstable CAG repeat expansion. Nat. Genet. 17, 65–70. [4] Palhan, V.B. et al. (2005) Polyglutamine-expanded ataxin-7 inhibits STAGA histone acetyltransferase activity to produce retinal degeneration. Proc. Natl. Acad. Sci. USA 102, 8472–8477. [5] McMahon, S.J., Pray-Grant, M.G., Schieltz, D., Yates 3rd, J.R. and Grant, P.A. (2005) Polyglutamine-expanded spinocerebellar ataxia-7 protein disrupts normal SAGA and SLIK histone acetyltransferase activity. Proc. Natl. Acad. Sci. USA 102, 8478–8482. [6] Kohler, A., Schneider, M., Cabal, G.G., Nehrbass, U. and Hurt, E. (2008) Yeast ataxin-7 links histone deubiquitination with gene gating and mRNA export. Nat. Cell Biol. 10, 707–715. [7] Havel, L.S., Li, S. and Li, X.J. (2009) Nuclear accumulation of polyglutamine disease proteins and neuropathology. Mol. Brain 2, 21. [8] Kioka, N., Ueda, K. and Amachi, T. (2002) Vinexin, CAP/ponsin, ArgBP2: a novel adaptor protein family regulating cytoskeletal organization and signal transduction. Cell Struct. Funct. 27, 1–7. [9] Li, S.S. (2005) Specificity and versatility of SH3 and other proline-recognition domains: structural basis and implications for cellular signal transduction. Biochem. J. 390, 641–653. [10] Zarrinpar, A., Bhattacharyya, R.P. and Lim, W.A. (2003) The structure and function of proline recognition domains. Sci. STKE, RE8. [11] Gao, Y.G., Yan, X.Z., Song, A.X., Chang, Y.G., Gao, X.C., Jiang, N., Zhang, Q. and Hu, H.Y. (2006) Structural insights into the specific binding of huntingtin proline-rich region with the SH3 and WW domains. Structure 14, 1755–1765. [12] Cesareni, G. (2005) Modular Protein Domains, Wiley-VCH, Weinheim.
2911
[13] Lebre, A.S. et al. (2001) Ataxin-7 interacts with a Cbl-associated protein that it recruits into neuronal intranuclear inclusions. Hum. Mol. Genet. 10, 1201– 1213. [14] Bao, W.J., Gao, Y.G., Chang, Y.G., Zhang, T.Y., Lin, X.J., Yan, X.Z. and Hu, H.Y. (2006) Highly efficient expression and purification system of small-size protein domains in Escherichia coli for biochemical characterization. Protein Expr. Purif. 47, 599–606. [15] Berliner, L.J., Grunwald, J., Hankovszky, H.O. and Hideg, K. (1982) A novel reversible thiol-specific spin label: papain active site labeling and inhibition. Anal. Biochem. 119, 450–455. [16] de Vries, S.J., van Dijk, A.D., Krzeminski, M., van Dijk, M., Thureau, A., Hsu, V., Wassenaar, T. and Bonvin, A.M. (2007) HADDOCK versus HADDOCK: new features and performance of HADDOCK2.0 on the CAPRI targets. Proteins 69, 726–733. [17] Kim, J., Lee, C.D., Rath, A. and Davidson, A.R. (2008) Recognition of noncanonical peptides by the yeast Fus1p SH3 domain: elucidation of a common mechanism for diverse SH3 domain specificities. J. Mol. Biol. 377, 889–901. [18] Gehmlich, K. et al. (2007) Paxillin and ponsin interact in nascent costameres of muscle cells. J. Mol. Biol. 369, 665–682. [19] Zhang, J., Li, X., Yao, B., Shen, W., Sun, H., Xu, C., Wu, J. and Shi, Y. (2007) Solution structure of the first SH3 domain of human vinexin and its interaction with vinculin peptides. Biochem. Biophys. Res. Commun. 357, 931–937. [20] Good, M.C., Zalatan, J.G. and Lim, W.A. (2011) Scaffold proteins: hubs for controlling the flow of cellular information. Science 332, 680–686. [21] Buday, L. and Tompa, P. (2010) Functional classification of scaffold proteins and related molecules. FEBS J. 277, 4348–4355. [22] Ribon, V., Printen, J.A., Hoffman, N.G., Kay, B.K. and Saltiel, A.R. (1998) A novel, multifuntional c-Cbl binding protein in insulin receptor signaling in 3T3-L1 adipocytes. Mol. Cell. Biol. 18, 872–879. [23] Mandai, K., Nakanishi, H., Satoh, A., Takahashi, K., Satoh, K., Nishioka, H., Mizoguchi, A. and Takai, Y. (1999) Ponsin/SH3P12: an L-afadin- and vinculinbinding protein localized at cell–cell and cell-matrix adherens junctions. J. Cell Biol. 144, 1001–1017. [24] Lin, W.H., Huang, C.J., Liu, M.W., Chang, H.M., Chen, Y.J., Tai, T.Y. and Chuang, L.M. (2001) Cloning, mapping, and characterization of the human sorbin and SH3 domain containing 1 (SORBS1) gene: a protein associated with c-Abl during insulin signaling in the hepatoma cell line Hep3B. Genomics 74, 12–20. [25] Kurakin, A., Hoffman, N.G. and Kay, B.K. (1998) Molecular recognition properties of the C-terminal Sh3 domain of the Cbl associated protein, Cap. J. Pept. Res. 52, 331–337. [26] Bauer, F., Schweimer, K., Meiselbach, H., Hoffmann, S., Rosch, P. and Sticht, H. (2005) Structural characterization of Lyn-SH3 domain in complex with a herpesviral protein reveals an extended recognition motif that enhances binding affinity. Protein Sci. 14, 2487–2498. [27] Lewitzky, M., Harkiolaki, M., Domart, M.C., Jones, E.Y. and Feller, S.M. (2004) Mona/Gads SH3C binding to hematopoietic progenitor kinase 1 (HPK1) combines an atypical SH3 binding motif, R/KXXK, with a classical PXXP motif embedded in a polyproline type II (PPII) helix. J. Biol. Chem. 279, 28724– 28732. [28] Hoelz, A., Janz, J.M., Lawrie, S.D., Corwin, B., Lee, A. and Sakmar, T.P. (2006) Crystal structure of the SH3 domain of betaPIX in complex with a high affinity peptide from PAK2. J. Mol. Biol. 358, 509–522. [29] Stollar, E.J., Garcia, B., Chong, P.A., Rath, A., Lin, H., Forman-Kay, J.D. and Davidson, A.R. (2009) Structural, functional, and bioinformatic studies demonstrate the crucial role of an extended peptide binding site for the SH3 domain of yeast Abp1p. J. Biol. Chem. 284, 26918–26927. [30] Harkiolaki, M. et al. (2009) Distinct binding modes of two epitopes in Gab2 that interact with the SH3C domain of Grb2. Structure 17, 809–822. [31] Orr, H.T. and Zoghbi, H.Y. (2007) Trinucleotide repeat disorders. Annu. Rev. Neurosci. 30, 575–621. [32] Jiang, Y.J., Che, M.X., Yuan, J.Q., Xie, Y.Y., Yan, X.Z. and Hu, H.Y. (2011) Interaction with polyglutamine-expanded huntingtin alters cellular distribution and RNA processing of huntingtin yeast two-hybrid protein A (HYPA). J. Biol. Chem. 286, 25236–25245. [33] Ribon, V., Herrera, R., Kay, B.K. and Saltiel, A.R. (1998) A role for CAP, a novel, multifunctional Src homology 3 domain-containing protein in formation of actin stress fibers and focal adhesions. J. Biol. Chem. 273, 4073–4080. [34] Langhorst, M.F., Jaeger, F.A., Mueller, S., Sven Hartmann, L., Luxenhofer, G. and Stuermer, C.A. (2008) Reggies/flotillins regulate cytoskeletal remodeling during neuronal differentiation via CAP/ponsin and Rho GTPases. Eur. J. Cell Biol. 87, 921–931. [35] Baumann, C.A. et al. (2000) CAP defines a second signalling pathway required for insulin-stimulated glucose transport. Nature 407, 202–207. [36] Lesniewski, L.A. et al. (2007) Bone marrow-specific Cap gene deletion protects against high-fat diet-induced insulin resistance. Nat. Med. 13, 455–462. [37] Tosoni, D. and Cestra, G. (2009) CAP (Cbl associated protein) regulates receptor-mediated endocytosis. FEBS Lett. 583, 293–300.
journal homepage: www.FEBSLetters.org
Structural basis for recognition of the third SH3 domain of full-length R85 (R85FL)/ponsin by ataxin-7 q Ya-Jun Jiang, Chen-Jie Zhou, Zi-Ren Zhou, Meng Wu, Hong-Yu Hu ⇑ State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China
a r t i c l e
i n f o
Article history: Received 18 April 2013 Revised 1 June 2013 Accepted 4 July 2013 Available online 24 July 2013
Edited by Berend Wieringa Keywords: Ataxin-7 Proline-rich region R85FL/ponsin SH3 domain Polyglutamine Inclusion body
a b s t r a c t Ataxin-7 (Atx7) is a component of the nuclear transcription co-activator complex; its polyglutamine (polyQ) expansion may cause nuclear accumulation and recruit numerous proteins to the intranuclear inclusion bodies. Full-length R85 (R85FL) is such a protein sequestered by polyQ-expanded Atx7. Here, we report that Atx7 specifically interacts with the third SH3 domain (SH3C) of R85FL through its second portion of proline-rich region (PRR). NMR structural analysis of the SH3C domain and its complex with PRR revealed that SH3C contains a large negatively charged surface for binding with the RRTR motif of Atx7. Microscopy imaging demonstrated that sequestration of R85FL by the polyQ-expanded Atx7 in cell is mediated by this specific SH3C–PRR interaction, which is implicated in the pathogenesis of spinocerebellar ataxia 7. Structured summary of protein interactions: Atx7 PP2 and SH3C bind by isothermal titration calorimetry (View Interaction: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12) Atx7 binds to SH3C by pull down (View interaction) Atx7100Q and SH3C colocalize by fluorescence microscopy (View interaction) SH3C and Atx7 bind by nuclear magnetic resonance (View interaction) Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
1. Introduction Spinocerebellar ataxia 7 (SCA7) is an autosomal dominant neurodegenerative disorder characterized by cerebellar ataxia associated with progressive macular dystrophy [1,2]. SCA7 is generally considered to be caused by expansion of a CAG repeat encoding polyglutamine (polyQ) tract in the protein ataxin-7 (Atx7) [3]. Atx7 is a component of mammalian STAGA (SPT3-TAF9-ADAGCN5 acetyltransferase) transcription co-activator complex, and its polyQ expansion inhibits STAGA histone acetyltransferase activity [4,5]. Moreover, its yeast form regulates ubiquitin levels of histone H2B and links histone deubiquitination with gene gating and mRNA export [6]. Thus, impairments of these normal functions by polyQ expansion of Atx7 may underlie the production of SCA7 phenotype. Abbreviations: Atx7, ataxin-7; polyQ, polyglutamine; PRR, proline-rich region; R85FL, full-length R85; SCA7, spinocerebellar ataxia 7; SH3, Src homology 3 q The atomic coordinates and structure factors of free and PP2-bound SHC3 have been deposited in the PDB under codes 2LJ0 and 2LJ1, respectively. ⇑ Corresponding author. Address: Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, 320 Yue-yang Road, Shanghai 200031, China. E-mail address: [email protected] (H.-Y. Hu).
There are nine inherited neurodegenerative disorders related to polyQ expansion, the hallmark of this kind of disease is the formation of nuclear inclusion bodies [7]. The intranuclear inclusions found in human SCA7 brain contain many other proteins, including transcriptional regulators, ubiquitin/proteasome pathway components, cell death associated proteins, and chaperones and their partners [1]. The function of all these proteins can possibly be influenced by polyQ expansion of Atx7. However, whether these proteins are recruited to inclusions by polyQ expanded Atx7 through specific or non-specific protein interactions remains largely unknown. One such protein sequestered by polyQ expanded Atx7 to intranuclear inclusions is full-length R85 (R85FL), a splice variant of ponsin or Cbl-associated protein (CAP). R85FL/ponsin belongs to the SoHo adaptor protein family that regulates cytoskeletal organization and signal transduction [8]. There are three members of this family; all contain three highly conserved Src homology 3 (SH3) domains in the C-termini. As known, SH3 domain specifically recognizes proteins with proline-rich region (PRR) [9–11]. The canonical binding sites for SH3 domains usually include the amino-acid sequences like R/ KxxPxxP (class I) or PxxPxR/K (class II) [10,12]. Like huntingtin (Htt), Atx7 contains a PRR just downstream of the polyQ tract [13].
0014-5793/$36.00 Ó 2013 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.febslet.2013.07.021
2906
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
We herein report identification and structural analysis for the interaction of Atx7 PRR with the third SH3 domain (SH3C) of R85FL. Moreover, R85FL can be sequestered to the nuclear inclusion bodies by polyQ expanded Atx7 through specific SH3C–PRR interactions.
room temperature for 5 h to remove MTSL from the Atx7_PP2 peptide, and then another HSQC spectrum was acquired for reference [15]. The peak intensities with and without MTSL in the HSQC spectra of SH3C were compared for identifying the residues that are close to the N- or C-terminus of the peptide.
2. Materials and methods
2.5. Construction of the complex structure
2.1. Cloning, expression, peptide synthesis, and protein purification
The complex structure was generated by HADDOCK2.1 [16]. The structures of SH3C in the peptide-bound form solved by NMR and Atx7-PP2 generated by homology modeling from the known proline-rich peptide (PDB ID: 2DF6) were used as initials for docking. The residues that have chemical-shift changes larger than the average value were treated as significantly perturbed residues. The active residues included Y13, Y15, N19, D21, D39, W41, P56 and Y59 of SH3C, and all residues of Atx7-PP2. The PRE-derived intermolecular distances were introduced as unambiguous restraints. Only those residues with resonance peaks significantly broadened or almost disappeared by paramagnetic effect were used as inputs for structure calculation. Normally the distance between the amide nitrogen atoms of SH3C and the paramagnetic oxygen of MTSL was set as 10 ± 3 Å. The ambiguous constraints were further modified by the fact that Pro56 of SH3C has intermolecular NOEs with the prolines of Atx7_PP2. The docking protocol included three stages with a final explicit water refinement. At the last iteration, 200 structures were generated and clustered by an RMSD cut-off of 3 Å.
The R85FL gene was cloned from human brain cDNA library (Invitrogen), and inserted into pcDNA3.0 vector with a FLAG tag in N-terminus of the protein. The Atx7 gene with 10 or 100 glutamine repeats in the vector pEGFPN1 was a gift from Dr. A. Brice. The SH3 domains and their mutants were subcloned into pGBTNH [14] and pGEX-4T-3. The peptides were obtained from solid-phase chemical synthesis and analyzed by electrospray mass spectrometry. The plasmids were transformed into Escherichia coli BL21 (DE3) cells to overexpress proteins. The proteins were purified by Ni2+– NTA or GST affinity column, followed by gel filtration chromatography. 15N/13C-labeled proteins were prepared using the M9 minimal medium containing 15NH4Cl and/or 13C6-D-glucose as the sole nitrogen and/or carbon resource, respectively. 2.2. GST pull-down experiments The GST-peptide fusions were added to the glutathione Sepharose 4B beads (Amersham Biosciences) in a PBS buffer (140 mM sodium chloride, 2.7 mM potassium chloride, 10 mM sodium phosphate, 1.8 mM potassium phosphate, pH 7.3), and the suspensions were agitated at 4 °C for 30 min. The beads were washed three times with the same buffer to remove any unbound proteins. An equivalent amount of different SH3 domains were added and incubated at 4 °C for an additional 4 h. Finally, the beads were recovered by centrifugation, and then the samples were resuspended in the sample buffer and subjected to SDS–PAGE (15% gel), followed by Coomassie staining. 2.3. NMR spectroscopy and structure determination The sample containing 15N/13C-labeled SH3C protein (1.0 mM) with or without Atx7_PP2 peptide (4 mM) in a phosphate buffer (20 mM NaH2PO4, 50 mM NaCl, pH6.5) was used for the structural determination by NMR. All data were acquired at 25 °C on a Bruker Avance 600-MHz spectrometer. The spectra including HNCO, HNCACB, CBCA(CO)NH, H(CCO)NH, C(CO)NH and HCCH-TOCSY were obtained for the backbone and side-chain assignments. NOE restraints were obtained from 15N- and 13C-edited NOESY spectra with a mixing time of 100 ms. The NMR data were processed using NMRPipe and analyzed with SPARKY. Backbone dihedral restraints were derived from TALOS program. The structures were calculated using ARIA2.0, assessed with PROCHECK, and displayed with MOLMOL. 2.4. MTSL labeling and paramagnetic relaxation enhancement (PRE) A Cys residue was introduced to the N- or C-terminus of Atx7_PP2, and the peptide fused with GB1 was obtained with pGBTNH expression vector [14]. After purification, MTSL ((1-oxyl2,2,5,5-tetramethyl pyrroline-3-methyl) methanethiosulfonate) (ALEXIS Biochemicals, San Diego) was incubated with Atx7_PP2 at 4 °C overnight. Ten-fold excess of MTSL was used for the reaction and the excess MTSL was removed by desalting [11]. In the PRE experiment, after collecting a HSQC spectrum, dithiothreitol (DTT, 10 mM) was added to the mixture in the NMR tube and stayed at
2.6. Isothermal titration calorimetry (ITC) ITC was performed with an ITC200 MicroCalorimeter (MicroCal). Proteins were dissolved at 100 lM in a PBS buffer (140 mM sodium chloride, 2.7 mM potassium chloride, 10 mM sodium phosphate, 1.8 mM potassium phosphate, pH 7.3; or 20 mM sodium phosphate, 50 mM sodium chloride, pH 6.5 or 7.0). Upon reaching the equilibrium temperature of 25 °C, different peptides were titrated from a syringe into a sample chamber holding by 20 injections. Data were obtained using v2 minimization on a model assuming a single set of sites to calculate the binding affinity. 2.7. Cell culture and immunofluorescence microscopy HEK 293T cells were transfected by FuGENE HD transfection reagent (Roche) following the manufacturer’s instructions. AntiFLAG antibody (Sigma) was used for immuno-fluorescence. For confocal microscopy, HEK 293T cells were grown on glass coverslips for 48 h after transfection. Images were obtained on a Leica TCS SP2 confocal microscope (Leica Microsystems). 3. Results 3.1. Atx7 PRR specifically interacts with the third SH3 domain of R85FL By using two-hybrid approach to screen human retina cDNA library, a protein named R85 was identified for binding with Atx7 [13]. In addition, the R85FL was also found to co-localize with Atx7 [13]. R85FL is a splice variant of the SH3P12 gene product. All variants have three repeated SH3 domains located at the C-termini (Fig. 1A). As known, SH3 is a modular domain recognizing proline-rich motifs [9,11], while Atx7 contains a PRR motif located at the N-terminus just behind the polyQ tract (Fig. 1A). To get molecular details of the recognition between Atx7 and R85FL, we examined the interactions of Atx7 PRR with different SH3 domains of R85FL by GST pull-down analysis. The results show that Atx7 PRR can interact with the third SH3 domain (as referred to SH3C)
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
2907
(Fig. 1D), but it neither binds with the first domain (SH3A) (Fig. 1B) nor the second (SH3B) (Fig. 1C). SH3 domain canonically binds with such amino-acid sequence as R/KxxPxxP (class I) or PxxPxR/ K (class II) [10,12]. So we dissected the Atx7 PRR sequence into two parts, namely Atx7_PP1 and Atx7_PP2 (Fig. 1A), each containing a class-II consensus peptide motif. Intriguingly, SH3C only interacts with Atx7_PP2 but not with Atx7_PP1 (Fig. 1E). ITC experiments indicate that Atx7_PP2 has almost the same affinity (KD = 68 lM) to SH3C when compared with the full-length Atx7 PRR sequence (Atx7_PP12) (Fig. S1), suggesting that the PP2 motif in Atx7 is the major binding moiety for SH3C. NMR titration further corroborates our observation from GST pull-down experiments (Fig. S2). The dissociation constant (KD) for SH3C binding with Atx7_PP2 is around a range of 50 lM. However, SH3A is incapable of binding with Atx7_PP1 or Atx7_PP2. Altogether, these results demonstrate that the PP2 motif of Atx7 specifically interacts with the third SH3 domain of R85FL. 3.2. Solution structures of SH3C in free and peptide-bound states
Fig. 1. Atx7 PRR specifically interacts with the third SH3 domain of R85FL. (A) Schematic representation of Atx7, R85FL and the PRR peptide portions of Atx7. PP12, the PRR sequence of Atx7; PP1, the first portion; PP2, the second portion. (B) GST pull-down experiment for testing the interaction of GST-PP12 with SH3A. (C) As in (B), GST-PP12 with SH3B. (D) As in (B), GSP-PP12 with SH3C. (E) GST pulldown experiment for testing the interaction between GST-PP1 or GST-PP2 and SH3C. The GST pull-down experiments were detected by Coomassie blue staining. The extra bands in the gels are due to degradation of GST or GST fused proteins.
The SoHo family adaptor proteins include three members, vinexin, Arg-binding protein 2 (ArgBP2) and R85FL (ponsin or CAP). Each protein has one SoHo domain in N-terminus and three SH3 domains in C-terminus. The SH3 domain is highly conserved among the family members with an identity up to 60% (Fig. S3), whereas only 17% identity among the three SH3 domains of R85FL (Fig. 2A). The structures of SH3A (2DL3) and SH3B (2O9S) are available in the Protein Data Bank. By using heteronuclear NMR approach, we solved the solution structures of the SH3C domain of R85FL in free and PP2bound states. The best 15 final structures, as judged from energetic and geometrical considerations, have a backbone RMSD of 0.44 ± 0.08 Å for SH3C alone or 0.61 ± 0.14 Å for it binding with
Fig. 2. Solution structures of SH3C in free and peptide-bound states. (A) Sequence alignment of the three SH3 domains from R85FL. The conserved residues are in blue box, while the negatively charged residues in SH3C that are not present in SH3A or SH3B are indicated with arrows. Positions of the b-strands, 310 helix, RT loop, N-Src loop and distal loop are indicated at the top of the sequence. (B) A cluster of 15 best structures of SH3C in free state (PDB ID: 2LJ0). (C) A cluster of 15 best structures of SH3C in Atx7_PP2 bound state (PDB ID: 2LJ1). (D) Comparison of the average structures of SH3C in free (blue) and peptide bound (green) states. (E) Surface representation for the three SH3 domains from R85FL. The positive and negative charges are shown in blue and red, respectively, while the hydrophobic regions are shown in gray. PDB ID: SH3A (2DL3), SH3B (2O9S) and SH3C (2LJ0).
2908
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
Fig. 3. Interaction between Atx7_PP2 and SH3C of R85FL. (A) The binding affinity of SH3C with Atx7_PP2 is dependent on the salt concentration in solution. The dissociation constant (KD) values were obtained from ITC experiments in pH 6.5 or pH 7.0. (B) Chemical-shift perturbation of SH3C upon addition of Atx7_PP2 at the PP2/SH3C ratio of 6:1. The chemical-shift change was calculated by a combination of hydrogen and nitrogen ðDdav ¼ SQRTðDd2H þ Dd2N =25ÞÞ. The residues with chemical-shift changes larger than the average (mean) were treated as significantly perturbed residues, including Y15, Q18, N19, D20, D21, I30, V31, W41, V43, T54, F55 and Y59. (C) Intermolecular NOEs between Atx7_PP2 and SH3C. The three columns represent 13C-edited NOE signals for the side-chain of Pro56 in SH3C (left), 13C-edited NOEs for Pro56 saturated with Atx7_PP2 (middle), and the 13C-filtered/edited intermolecular NOEs for Pro56 with Atx7_PP2 (right). The arrows indicate the intermolecular NOEs for Pro56 of SH3C with prolines of Atx7_PP2. (D) Ratios of the peak intensities in the HSQC spectra with or without titration of N-terminally MTSL-labeled Atx7_PP2. The region of residues with strong paramagnetic effect is indicated by an arrow. (E) As in (D), with C-terminally labeled Atx7_PP2 peptide. Table 1 Binding affinities (KD) of SH3C or its mutants with Atx7_PP2 peptide or its mutants as determined by isothermal titration calorimetry. Atx7 PRR Peptide
PP12
PP2
PP2-R6A
PP2-R7A
PP2-T8A
PP2-R9A
68.0 ± 4.6
980 ± 19
253 ± 45
862 ± 44
588 ± 96
D20A 139 ± 24
D21A >1000
D39A 571 ± 41
T54A 78.1 ± 10.9
Y59A >1000
R85L_SH3C KD (lM) SH3C Mutant KD (lM)
46.5 ± 7.1 Atx7_PP2 Y13A >1000
the PP2 peptide (Table S1). Both structures of the free and PP2bound forms display a b-barrel fold of five b-strands and a 310 helix (Fig. 2B and C), showing a typical SH3 domain architecture. In comparison of the two average structures in free and PP2-bound states, it appears that binding of the PP2 peptide does not cause any significant conformational change (Fig. 2D), because the backbone RMSD of the two structures is only 1.22 Å. There is only a small structural difference in the RT-loop and n-Src loop regions, which are the most flexible regions of SH3 domains but engaged in the peptide binding. The binding surface for PRR on a typical SH3 domain can be divided into two areas, one is the PxxP binding pocket and the other is located in between the RT-loop and n-Src loop [17]. The second region is contributable to the ligand specificity of SH3 domain. The significant difference among the three SH3 domains of R85FL is that SH3C contains more negatively charged residues (e.g. D20, D21 and D39) on its surface (Fig. 2E). This negatively charged area
lies in between the RT-loop and n-Src loop. In comparison, the corresponding surface of SH3A is roughly hydrophobic almost without negative charge, while the surface of SH3B contains less negative charge than that of SH3C. This charge difference might provide molecular basis for the specificities of the SH3 domains binding with partners. Actually, SH3B of R85FL binds with non-canonical proline-rich motif [18], while SH3A of vinexin, a highly conserved homolog of SH3A of R85FL, can also bind with a non-canonical proline-rich motif [19]. 3.3. Structural insights into the specific interaction between SH3C and Atx7_PP2 Since SH3C has a negatively charged surface and Atx7_PP2 contains a RRTR motif with positive charges, the electric charges on the surface may play important roles in mediating the electrostatic
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
2909
Fig. 4. Complex structure of SH3C with Atx7_PP2. (A) Surface representation of SH3C showing the grooves for Atx7_PP2 binding. The positive charges, negative charges, and the hydrophobic regions are shown in blue, red and gray, respectively. The R6, R7, T8 and R9 residues of Atx7_PP2 are illustrated in blue, while other residues are shown in cyan. The hydrophobic interaction (left) and electrostatic interaction (right) are also highlighted in the graphs. (B) Detailed view of the interfaces for interaction of R85FL_SH3C (ribbon) with Atx7_PP2 (stick). (C) As in (B), for interaction of R85FL_SH3B (ribbon) and the PRR peptide of paxillin (stick) (PDB ID, 2O9V).
interaction between the two peptide chains, thus ion strength may also influence the complex. Indeed, ITC experiments show that the dissociation constant (KD) for SH3C binding with PP2 is significantly increased with an increase of salt concentration in the buffer (Fig. 3A), suggesting that the electrostatic interactions are contributable to the formation of the SH3C–PP2 complex. NMR chemicalshift perturbation exhibits that some residues like Y15, D21, D39, W41 and Y59, experience significant chemical-shift changes upon PP2 binding (Fig. 3B). These residues are also contributable to constructing the xP groove and the specificity pocket of SH3 domain [9,10,12]. Thus, both the proline-mediated hydrophobic interaction and the arginine-conducted electrostatic interaction are crucial to forming the complex between Atx7_PP2 and SH3C. More evidence comes from the NOE signals between Atx7_PP2 and SH3C (Fig. 3C). The b-methylene group of Pro56 of SH3C has only intraresidue NOE signals in the free state, whereas some intermolecular NOE signals appear in the complex, though they are still unsigned due to 5 continuous prolines in Atx7_PP2. The intermolecular NOE data from 13C filtered/edited NOESY spectra are confirmable for this assumption (Fig. 3C, right panel). According to the chemical-shift pattern, these intermolecular NOE signals may come from one of the proline residues of Atx7_PP2. We further characterized the interaction between SH3C and Atx7_PP2 by PRE experiments. The cross peaks of Y13, S14, Y15 and I16 of SH3C are significantly broadened when it binds with the N-terminally MTSL-labeled peptide (Fig. 3D), indicating that the N-terminus of Atx7_PP2 is seated around the xP groove of SH3C formed by Y13 and Y15. Meanwhile, C-terminally labeled Atx7_PP2 causes large peak broadening of D38, D39 and G40 (Fig. 3E), suggesting that the C-terminus of
Atx7_PP2 interacts with the acidic residues in the N-Src loop of SH3C. The ITC experiments on various mutants of both Atx7_PP2 and SH3C further confirm the viewpoint that the extended RRTR motif of Atx7_PP2 and the key residues forming the xP groove and specificity pocket of SH3C are important for the specific interaction (Table 1). The dramatic decrease of the binding affinities in the D21A and D39A mutants can account for the different recognition modes for polyproline peptides among the three SH3 domains of R85FL (Fig. 1 and Fig. S2). Taken together, these results suggest that both the xP groove and specificity pocket of SH3C are engaged in the interaction with Atx7_PP2, while both the PPPPP and RRTR motifs in Atx7_PP2 are also contributable to the specific interaction. 3.4. Structure model for the complex of SH3C and Atx7_PP2 Due to the relatively weak interaction and difficulty in NMR assignment of the Atx7_PP2 peptide, we have not obtained enough unambiguous intermolecular NOEs for direct computation of the complex structure, so we generated a rational complex structure by HADDOCK method [16], using the information from chemical shift perturbation (Fig. 3B), intermolecular NOEs (Fig. 3C), PRE restraints (Fig. 3D and E) and binding affinities of various mutants (Table 1). With the experimental structure constraints as an input, a major cluster of the initial complex structures were generated, and a refined structure was finally obtained (Fig. 4). In the structure model, the PPPPP motif is docked on the hydrophobic xP groove of SH3C and the RRTR motif is directed to the negatively charged specificity pocket (Fig. 4A). The prolines of Atx7_PP2 are contacting with the xP groove of SH3C formed by Y13, Y15, W41,
2910
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
to dysfunction [7]. Using confocal microscopy, we investigated the c0-localization of Atx7 with R85FL in cells. Normal Atx7 (Atx710Q) is a nucleus localized protein, while R85FL is distributed mainly in cytosol (Fig. 5, 1st row). The polyQ expanded Atx7 (Atx7100Q) forms nuclear inclusion bodies, which can sequester R85FL into the nucleus (Fig. 5, 2nd row). As a control, double mutation (I16A/Q18A) in the SH3C moiety of R85FL, which does not alter the interaction with Atx7_PP2, can also be sequestered into the nucleus by Atx7100Q (Fig. 5, 3rd row). However, the D39A mutant in the SH3C moiety of R85FL that significantly reduces the binding affinity to the PP2 motif of Atx7 cannot be sequestered by Atx7100Q (Fig. 5, 4th row). Moreover, deletion of SH3C in R85FL significantly abolishes sequestration of R85FL into the nucleus by Atx7100Q (Fig. 5, 5th row). This observation indicates that the SH3C–PP2 interaction mediates the association of Atx7 and R85FL in cells. The polyQ expanded Atx7 is prone to formation of inclusion bodies and may significantly alter the localization of R85FL from cytosol to nucleus. This kind of abnormal interactions of the polyQ expanded Atx7 with other partners may be implicated in the pathological process of SCA7.
4. Discussion
Fig. 5. PolyQ expanded Atx7 (Atx7100Q) can sequester R85FL in cells. Normal Atx710Q-GFP and polyQ expanded Atx7100Q-GFP were co-transfected with either wild-type R85FL or its mutants in the SH3C moiety into HEK 293T cells, and performed imaging with confocal fluorescence microscopy. Nuclei were stained with DAPI. I16A/Q18A is a double mutant of R85FL in the SH3C moiety, while D39A is a point mutant of R85FL in the SH3C domain. DSH3C stands for the SH3C deletion mutant of R85FL. Scalar bar = 10 lM.
P56 and Y59 (Fig. 4B). The positively charged RRTR motif is located in the negatively charged specificity pocket formed by D21, E22, D38 and D39. On the Atx7_PP2 side, R6 is pointed to the acidic pocket of SH3C formed by D21 and E22, while both R7 and R9 located at the pocket formed by D38 and D39. This is different from the complex structure of SH3B of R85FL, in which the paxillin peptide forms a type-II polyproline helix conformation [18]. The specificity pocket of SH3B is comprised of Val, Asp, Asn and Ile residues, corresponding to the residues of C37, D38, G40 and T54 in SH3C (Fig. 2A), but there is no any charge involved in the interaction. In essence, this structure model can explain the binding specificity for Atx7_PP2 and the unique property of SH3C among the SH3 domains of R85FL. 3.5. PolyQ expanded Atx7 (Atx7100Q) can sequester R85FL through specific interaction with SH3C Formation of inclusion bodies is the hallmark of polyQ expanded neurodegenerative diseases, and the cellular inclusions can sequester various kinds of proteins and probably lead them
SH3 is a modular domain commonly found in all eukaryotic genomes [10]. R85FL/ponsin is an adaptor protein that harbors three SH3 domains in the C-terminus. Adaptor proteins are usually composed of multiple modular domains for linking different parts of enzymatic cascade or signal transduction pathway [20,21]. So the three SH3 domains of R85FL must have evolved unique properties to accommodate different partners. Comparing the structures of these three SH3 domains from R85FL reveals that the major difference for binding diverse PRR partners comes from the negatively charged region between the RT loop and n-Src loop (Fig. 2E). This structural difference among these three SH3 domains may provide a possibility for R85FL to function both specifically and versatilely in living cells. Most SH3 domains recognize linear polyproline motif through two pockets; one binds to the canonical PxxP core motif and the other binds to the extended motif beyond the PxxP core [17]. Both the canonical and non-canonical peptides associate with SH3 domains through these two pockets. Although lots of SH3 domain structures are available, it is still very hard to predict the interaction partners of a SH3 domain, largely because of the high versatility of SH3 domains. Previous studies showed that the SH3C domain of R85FL can interact with c-Cbl [22], 1-Afadin [23] and c-Abl [24]. The SH3C domain interacts with the PRR-containing partners with a novel consensus motif of PxPPxRxSSL pattern at their binding sites [25]. Besides the PxxP motif, the extended motif also plays a crucial role in SH3 domain recognition [26–29]. However, these SH3 domains bind with different polyproline peptides with the dissociation constants around 1 lM, whereas the binding affinity for R85FL_SH3C with Atx7_PP2 is much weaker, till to an extent of about 50 lM. The RxxK motif of Gab2 was reported to recognize the SH3C domain of Grb2 [30]; both side-chains of Arg and Lys residues point to the negatively charged region of the RT loop (Fig. S4A). The interaction between Atx7 and R85FL is a little different from that between Gab2 and Grb2. The negatively charged region of R85FL_SH3C includes both RT loop and n-Src loop, and each residue of the RRTR motif in Atx7 contributes to the interaction (Table 1). Another example is from the structural study of Mona/ Gads SH3C [27], which recognizes a PxxPxR/KxxK motif. The PxxP motif binds to the two xP grooves of SH3C, while the R/KxxK motif points to the negatively charged region of the RT loop (Fig. S4B). To our knowledge, the interaction between R85FL_SH3C and Atx7_PP2 represents a novel pattern. The diversity of the binding
Y.-J. Jiang et al. / FEBS Letters 587 (2013) 2905–2911
partners for an SH3 domain is expanded by combination of the PxxP motif and an extended motif. PolyQ disorders share several common pathological features, including the nuclear accumulation and aggregation of the disease proteins [7,31]. Numerous proteins can be sequestered by the nuclear inclusions, so that their functions must be abolished, which can also be recognized as a pathogen for the polyQ diseases. Our previous work has shown that polyQ expanded huntingtin (Htt) can sequester Htt two-hybrid protein A (HYPA) through its PRR interaction, and thus impairs the function of HYPA in pre-mRNA splicing [32]. It is quite possible to be a similar case in R85FL. The R85FL splicing variant ponsin/CAP regulates cytoskeleton organization [33,34], insulin signal pathway [22,35,36], and receptor-mediated endocytosis [37]. Actually, R85FL colocalizes with Atx7 in the brain of SCA patients [13]. The sequestration of R85FL by polyQ expanded Atx7 possibly leads to impairment of the normal function of R85FL, which may partially contribute to the pathology of the SCA7 disease. Acknowledgements The authors would thank Dr. A. Brice for providing the plasmid coding for ataxin-7, Dr. C.W. Jin at Peking University for technical assistance in NMR data acquisition, and Dr. Alexander M.J.J. Bonvin for the help of generating the complex structure using HADDOCK software. This work was supported by the grants from the National Basic Research Program of China (2011CB911104, 2012CB911003), and the National Natural Science Foundation of China (10979070). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.febslet.2013. 07.021. References [1] Lebre, A.S. and Brice, A. (2003) Spinocerebellar ataxia 7 (SCA7). Cytogenet. Genome Res. 100, 154–163. [2] Garden, G.A. and La Spada, A.R. (2008) Molecular pathogenesis and cellular pathology of spinocerebellar ataxia type 7 neurodegeneration. Cerebellum 7, 138–149. [3] David, G. et al. (1997) Cloning of the SCA7 gene reveals a highly unstable CAG repeat expansion. Nat. Genet. 17, 65–70. [4] Palhan, V.B. et al. (2005) Polyglutamine-expanded ataxin-7 inhibits STAGA histone acetyltransferase activity to produce retinal degeneration. Proc. Natl. Acad. Sci. USA 102, 8472–8477. [5] McMahon, S.J., Pray-Grant, M.G., Schieltz, D., Yates 3rd, J.R. and Grant, P.A. (2005) Polyglutamine-expanded spinocerebellar ataxia-7 protein disrupts normal SAGA and SLIK histone acetyltransferase activity. Proc. Natl. Acad. Sci. USA 102, 8478–8482. [6] Kohler, A., Schneider, M., Cabal, G.G., Nehrbass, U. and Hurt, E. (2008) Yeast ataxin-7 links histone deubiquitination with gene gating and mRNA export. Nat. Cell Biol. 10, 707–715. [7] Havel, L.S., Li, S. and Li, X.J. (2009) Nuclear accumulation of polyglutamine disease proteins and neuropathology. Mol. Brain 2, 21. [8] Kioka, N., Ueda, K. and Amachi, T. (2002) Vinexin, CAP/ponsin, ArgBP2: a novel adaptor protein family regulating cytoskeletal organization and signal transduction. Cell Struct. Funct. 27, 1–7. [9] Li, S.S. (2005) Specificity and versatility of SH3 and other proline-recognition domains: structural basis and implications for cellular signal transduction. Biochem. J. 390, 641–653. [10] Zarrinpar, A., Bhattacharyya, R.P. and Lim, W.A. (2003) The structure and function of proline recognition domains. Sci. STKE, RE8. [11] Gao, Y.G., Yan, X.Z., Song, A.X., Chang, Y.G., Gao, X.C., Jiang, N., Zhang, Q. and Hu, H.Y. (2006) Structural insights into the specific binding of huntingtin proline-rich region with the SH3 and WW domains. Structure 14, 1755–1765. [12] Cesareni, G. (2005) Modular Protein Domains, Wiley-VCH, Weinheim.
2911
[13] Lebre, A.S. et al. (2001) Ataxin-7 interacts with a Cbl-associated protein that it recruits into neuronal intranuclear inclusions. Hum. Mol. Genet. 10, 1201– 1213. [14] Bao, W.J., Gao, Y.G., Chang, Y.G., Zhang, T.Y., Lin, X.J., Yan, X.Z. and Hu, H.Y. (2006) Highly efficient expression and purification system of small-size protein domains in Escherichia coli for biochemical characterization. Protein Expr. Purif. 47, 599–606. [15] Berliner, L.J., Grunwald, J., Hankovszky, H.O. and Hideg, K. (1982) A novel reversible thiol-specific spin label: papain active site labeling and inhibition. Anal. Biochem. 119, 450–455. [16] de Vries, S.J., van Dijk, A.D., Krzeminski, M., van Dijk, M., Thureau, A., Hsu, V., Wassenaar, T. and Bonvin, A.M. (2007) HADDOCK versus HADDOCK: new features and performance of HADDOCK2.0 on the CAPRI targets. Proteins 69, 726–733. [17] Kim, J., Lee, C.D., Rath, A. and Davidson, A.R. (2008) Recognition of noncanonical peptides by the yeast Fus1p SH3 domain: elucidation of a common mechanism for diverse SH3 domain specificities. J. Mol. Biol. 377, 889–901. [18] Gehmlich, K. et al. (2007) Paxillin and ponsin interact in nascent costameres of muscle cells. J. Mol. Biol. 369, 665–682. [19] Zhang, J., Li, X., Yao, B., Shen, W., Sun, H., Xu, C., Wu, J. and Shi, Y. (2007) Solution structure of the first SH3 domain of human vinexin and its interaction with vinculin peptides. Biochem. Biophys. Res. Commun. 357, 931–937. [20] Good, M.C., Zalatan, J.G. and Lim, W.A. (2011) Scaffold proteins: hubs for controlling the flow of cellular information. Science 332, 680–686. [21] Buday, L. and Tompa, P. (2010) Functional classification of scaffold proteins and related molecules. FEBS J. 277, 4348–4355. [22] Ribon, V., Printen, J.A., Hoffman, N.G., Kay, B.K. and Saltiel, A.R. (1998) A novel, multifuntional c-Cbl binding protein in insulin receptor signaling in 3T3-L1 adipocytes. Mol. Cell. Biol. 18, 872–879. [23] Mandai, K., Nakanishi, H., Satoh, A., Takahashi, K., Satoh, K., Nishioka, H., Mizoguchi, A. and Takai, Y. (1999) Ponsin/SH3P12: an L-afadin- and vinculinbinding protein localized at cell–cell and cell-matrix adherens junctions. J. Cell Biol. 144, 1001–1017. [24] Lin, W.H., Huang, C.J., Liu, M.W., Chang, H.M., Chen, Y.J., Tai, T.Y. and Chuang, L.M. (2001) Cloning, mapping, and characterization of the human sorbin and SH3 domain containing 1 (SORBS1) gene: a protein associated with c-Abl during insulin signaling in the hepatoma cell line Hep3B. Genomics 74, 12–20. [25] Kurakin, A., Hoffman, N.G. and Kay, B.K. (1998) Molecular recognition properties of the C-terminal Sh3 domain of the Cbl associated protein, Cap. J. Pept. Res. 52, 331–337. [26] Bauer, F., Schweimer, K., Meiselbach, H., Hoffmann, S., Rosch, P. and Sticht, H. (2005) Structural characterization of Lyn-SH3 domain in complex with a herpesviral protein reveals an extended recognition motif that enhances binding affinity. Protein Sci. 14, 2487–2498. [27] Lewitzky, M., Harkiolaki, M., Domart, M.C., Jones, E.Y. and Feller, S.M. (2004) Mona/Gads SH3C binding to hematopoietic progenitor kinase 1 (HPK1) combines an atypical SH3 binding motif, R/KXXK, with a classical PXXP motif embedded in a polyproline type II (PPII) helix. J. Biol. Chem. 279, 28724– 28732. [28] Hoelz, A., Janz, J.M., Lawrie, S.D., Corwin, B., Lee, A. and Sakmar, T.P. (2006) Crystal structure of the SH3 domain of betaPIX in complex with a high affinity peptide from PAK2. J. Mol. Biol. 358, 509–522. [29] Stollar, E.J., Garcia, B., Chong, P.A., Rath, A., Lin, H., Forman-Kay, J.D. and Davidson, A.R. (2009) Structural, functional, and bioinformatic studies demonstrate the crucial role of an extended peptide binding site for the SH3 domain of yeast Abp1p. J. Biol. Chem. 284, 26918–26927. [30] Harkiolaki, M. et al. (2009) Distinct binding modes of two epitopes in Gab2 that interact with the SH3C domain of Grb2. Structure 17, 809–822. [31] Orr, H.T. and Zoghbi, H.Y. (2007) Trinucleotide repeat disorders. Annu. Rev. Neurosci. 30, 575–621. [32] Jiang, Y.J., Che, M.X., Yuan, J.Q., Xie, Y.Y., Yan, X.Z. and Hu, H.Y. (2011) Interaction with polyglutamine-expanded huntingtin alters cellular distribution and RNA processing of huntingtin yeast two-hybrid protein A (HYPA). J. Biol. Chem. 286, 25236–25245. [33] Ribon, V., Herrera, R., Kay, B.K. and Saltiel, A.R. (1998) A role for CAP, a novel, multifunctional Src homology 3 domain-containing protein in formation of actin stress fibers and focal adhesions. J. Biol. Chem. 273, 4073–4080. [34] Langhorst, M.F., Jaeger, F.A., Mueller, S., Sven Hartmann, L., Luxenhofer, G. and Stuermer, C.A. (2008) Reggies/flotillins regulate cytoskeletal remodeling during neuronal differentiation via CAP/ponsin and Rho GTPases. Eur. J. Cell Biol. 87, 921–931. [35] Baumann, C.A. et al. (2000) CAP defines a second signalling pathway required for insulin-stimulated glucose transport. Nature 407, 202–207. [36] Lesniewski, L.A. et al. (2007) Bone marrow-specific Cap gene deletion protects against high-fat diet-induced insulin resistance. Nat. Med. 13, 455–462. [37] Tosoni, D. and Cestra, G. (2009) CAP (Cbl associated protein) regulates receptor-mediated endocytosis. FEBS Lett. 583, 293–300.