Supporting Information

Hanawa-Suetsugu et al. 10.1073/pnas.1113512109 SI Text DHR-2 fragment (residues 1196–1622) was cloned into SI Results. We examined the intra- and intermolecular interfaces the expression vector pDEST10.1 (Invitrogen), as a fusion with of the Src-homology 3 (SH3) domains. To assess all structures N-terminal His and small ubiquitin-like modifier protease affinity containing an SH3 domain, we collected the Data Bank tags and a TEV protease cleavage site. The DOCK2 DHR-2 frag- (PDB) codes in three different ways. The first was a key word ment was expressed in Sf9 cells, using the Bac-to-Bac Baculovirus search for “SH3” in the PDB. The second was a Protein Structure Expression System (Invitrogen). The encoding the human Database Search by DaliLite v. 3, using our dedicator of cytokin- ras-related C3 botulinum toxin substrate 1 (Rac1) (1–177) frag- esis 2 (DOCK2) SH3 domain structure (PDB code 3A98). The ment was cloned into the expression vector pCR2.1 (Invitrogen), last was a homology search in the PDB, using the amino acid as a fusion with an N-terminal His affinity tag and a TEV pro- sequences that are annotated as an SH3 domain in the UniProt tease cleavage site. The T17N mutation was introduced into Rac1 database. Finally, we collected 1,245 PDB codes, and calculated by using a QuikChange site-directed mutagenesis kit (Agilent). the intermolecular and intramolecular interface areas of their The Rac1 (T17N) mutant was synthesized by the E. coli cell-free SH3 domains by PDBePISA (Fig. S5 A and B). system (1, 2). The DOCK2 DHR-2 and Rac1 (T17N) , The intermolecular interface between DOCK2 SH3 and En- after digestion with TEV protease, were separately purified gulfment and cell motility protein 1 (ELMO1) in our structure by ion exchange on a HiTrap Q column and by size-exclusion 2 is approximately 1;450 Å (Fig. S5 A and C), and it is the largest chromatography on a HiLoad 16/60 Superdex 75 pg column. The intermolecular interface (Fig. S5A). The second largest interface DOCK2 DHR-2•Rac1 complex, formed by incubating the pro- 2 (approximately 860 Å ) is that in the p22Phox • p47Phox complex teins together on ice for 1 h, was separated from the uncomplexed structure, which also has the authentic SH3•PXXP interaction proteins by chromatography on a HiLoad 16/60 Superdex 75 pg (Fig. S5D), but it is less than 60% of the size of the DOCK2•- column, preequilibrated with 20 mM Tris•HCl buffer (pH 8.0), ELMO interface. The third is the complex of Cbl-interacting pro- containing 150 mM NaCl, 10% glycerol, and 2 mM DTT. tein of 85 kDa SH3 and ubiquitin. This interface is formed by an 2 approximately 820 Å domain•domain interaction (Fig. S5A). NMR spectroscopy and spectral assignments. All spectra were re- On the other hand, intramolecular interactions of the SH3 corded at 296 K on Bruker Avance 600 and 800 spectrometers domain have been observed with other regions in multidomain equipped with cryoprobes. Resonance assignments were accom- proteins, which may be related to the autoinhibition of the plished using a conventional set of triple-resonance spectra, as protein functions. The SH3 interface in the calcium channel β-2 described previously (4), and were deposited in the Biological 2 subunit is over 1;500 Å (Fig. S5B). The construct lacks the poly- Magnetic Resonance Data Bank (11079). Interproton distance 15 13 proline sequences, and the SH3 domain is buried among the gua- restraints were obtained from N and C edited NOESY spec- nylate kinase domain and two flanking regions. Almost of all the tra, both recorded with a mixing time of 120 ms. All spectra were structures listed in Table S3B include such domain•domain inter- processed using NMRPipe (5), and the programs Kujira (6) and actions. Only cytosol factor 1 has the authentic inter- NMRView (7) were employed for optimal visualization and spec- action of the SH3 domain with a polyproline sequence. However, tral analyses. the interacting motif is not a typical PXXP, but a PXXR motif (Fig. S5E). NMR structure calculations. The DOCK2 SH3-ELMO1 peptide fusion complex was determined by the conventional triple-reso- SI Materials and Methods. Protein expression and purification for nance technique (8–12). Automated NOE cross-peak assign- structural determination. The encoding the DOCK2 SH3- ments and structure calculations with torsion angle dynamics ELMO1 peptide fusion proteins were obtained by PCR. The were performed using the software package CYANA (13). The DNA fragments, listed in Fig. S1, were cloned into the expression backbone dihedral angle restraints from the TALOS program vector pCR2.1 (Invitrogen), as fusions with an N-terminal His (14) were also included in the calculations, with allowed ranges affinity tag and a tobacco etch virus (TEV) protease cleavage site. of 30°. The final structure calculations with CYANA were The 13C∕15N-labeled fusion proteins were synthesized by the started from 100 conformers with random torsion angle values. Escherichia coli cell-free protein expression system (1–3), and The 20 conformers with the lowest final CYANA target function were purified using a chelating column, as described previously values were selected for the final structure set. The structures (4). The purified proteins were concentrated to 0.7–1.0 mM in were validated using PROCHECK-NMR (15, 16). The structural 20 mM Tris-d11-HCl buffer (pH 7.0), containing 100 mM NaCl, statistics of the DOCK2 SH3-ELMO1 peptide fusion protein are 1 mM dithiothreitol-d10, 10% D2O, and 0.02% NaN3. summarized in Table S1. Figures were generated with the MOL- The expression plasmids for the human DOCK2 (residues 1– MOL (17) and PyMol (18) (http://www.pymol.org) programs. 177) and ELMO1 (residues 532–727) fragments were constructed in a similar manner. The selenomethionine-labeled proteins were Identification of the ELMO1-interacting region of DOCK2 and the synthesized using the large-scale dialysis mode of the E. coli cell- DOCK2-interacting region of ELMO1. In our search for DOCK2 free reaction (1, 2). After digestion with TEV protease, the pro- and ELMO1 regions that are suitable for crystallographic studies, tein complex was purified by ion exchange on a HiTrap Q column we expressed a series of N-terminal fragments of DOCK2 (resi- and by size-exclusion chromatography on a HiLoad 16/60 Super- dues 1–160, 1–177, 1–190, 9–160, 9–177, 9–190, 21–160, 21–177, dex 75 pg column, preequilibrated with 20 mM Tris•HCl buffer and 21–190) and C-terminal fragments of ELMO1 (residues 532– (pH 8.0), containing 150 mM NaCl and 2 mM DTT. All columns 717, 541–717, 550–717, 532–727, 541–727, and 550–727), using were purchased prepacked from GE Healthcare. the small-scale dialysis mode of the E. coli cell-free reaction The boundaries for the DOCK2 DOCK-homology region 2 (1, 2). We found that fragments 1–177 and 1–190 of DOCK2 (DHR-2) domain were determined by expressing various DOCK2 and fragments 532–717, 532–727, and 541–727 of ELMO1 were DHR-2 fragments, using the small-scale dialysis mode of the highly expressed and produced soluble DOCK2•ELMOl com- E. coli cell-free reaction (1, 2). The gene encoding the human plexes. Among them, only the protein complex between frag-

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 1of9 ments 1–177 of DOCK2 and 532–727 of ELMO1 produced dif- fore, the present DOCK2 DHR-2 domain•Rac1 structure is fraction quality crystals. suitable for generating a structural model of the DOCK2•EL- MO1•Rac1 ternary complex (Fig. 5 B and C). Analytical ultracentrifugation. All experiments were performed at 4 °C with an An-50 Ti rotor using Beckman Optima XL-I analy- Calculations of SH3 intermolecular and intramolecular interfaces. To tical ultracentrifuge. The sample buffer was 20 mM Tris•HCl assess all structures containing an SH3 domain, we collected the buffer (pH 8.0), containing 150 mM NaCl and 5 mM β-mercap- PDB codes in three different ways. The first was a key word toethanol. The solvent density and the protein partial specific vo- search for “SH3” in the PDB (http://www.pdbj.org/). The second lume (υ) were estimated using the program Sednterp (http://www. was a Protein Structure Database Search by DaliLite v. 3 (http:// jphilo.mailway.com/). Sedimentation equilibrium experiments ekhidna.biocenter.helsinki.fi/dali_server/), using our DOCK2 were performed with protein concentrations of 0.80, 0.40, and SH3 domain structure (PDB code 3A98). The last was a homol- 0.20 mg∕mL. Data were obtained at 10,000, 12,000, and ogy search in the PDB, using the amino acid sequences that are 14,000 rpm. The equilibrium data were fitted using the manufac- annotated as an SH3 domain in the UniProt database (http:// turer’s software. www.uniprot.org/uniprot/). Finally, we collected 1,245 PDB codes, and calculated the intermolecular and intramolecular in- Crystallization and data collection. The DOCK2•ELMOl complex terface areas of their SH3 domains by PDBePISA (http:// crystals for structure analysis were grown by the hanging-drop www.ebi.ac.uk/msd-srv/prot_int/pistart.html)(Fig.S5A and B). vapor-diffusion method, by mixing the protein solution with an We listed the PDB codes of the proteins with intermolecular in- 2 equal volume of reservoir solution, containing 100 mM diammo- terfaces greater than 800 Å and intramolecular interfaces larger 2 nium hydrogen citrate and 12% PEG 3350. The crystals were than 1;000 Å . Although 18 structures are included in the list, transferred to reservoir solution containing 25% PEG 400 as a there are essentially five different types shown in Fig. S5B. cryoprotectant and then flash-cooled in liquid nitrogen. The sin- gle-wavelength anomalous dispersion data were collected at Plasmids for biological experiments. The pcDNA-DOCK2-HA, 100 K at BL-17A at the Photon Factory (Table S2). The diffrac- pBJ1-DOCK2-HA, and pBJ1-ELMO1-HA plasmids were de- tion data were processed with the HKL2000 program (19). scribed previously (31). For bacterial expression of the GST- The DOCK2 DHR-2•Racl complex crystals for structure ana- ELMO1 fusion protein, full-length ELMO1 was cloned into lysis were grown by the sitting-drop vapor-diffusion method, pGEX-6P-1 (GE Healthcare). The genes encoding the N- and by mixing the protein solution with an equal volume of reservoir C-terminal fragments of DOCK2 with different tags were created solution, containing 100 mM sodium citrate and 15% PEG 6000. by PCR, and cloned into pENTR-3C (Invitrogen) and pBJ1-neo, The crystals were transferred to reservoir solution containing respectively. 25% glycerol as a cryoprotectant and then flash-cooled in liquid nitrogen. The X-ray diffraction data were collected at 100 K at Pull-down assay and immunoprecipitation. Rac activation was ana- BL-41XU at SPring-8 (Table S3) and were processed with the lyzed as described previously (32). To assess ELMO1 binding, HKL2000 program (19). HEK293T cell lysates expressing DOCK2 or its mutants were in- cubated with the GST–ELMO1 fusion protein for 1 h at 4 °C. The X-ray structure determination and refinement. For the structure precipitated DOCK2 was detected by immunoblotting with an • determination of the DOCK2 ELMOl complex, the program anti-HA antibody (3F10; Roche Diagnostics). The interaction be- autoSHARP (20) was used to locate the selenium atoms and to tween the N- and C-terminal fragments of DOCK2 was analyzed calculate the initial phases. RESOLVE (21) was used for the den- by transiently expressing differentially tagged proteins in sity modification, and it placed 66% of the amino acid residues. HEK293T cells, followed by immunoprecipitation using antibo- Models were manually adjusted to the electron density by using dies against the tags. the Coot program (22). The refinement was performed with the CNS program (23). The final refinement statistics are summar- Preparation of stromal cells. Peripheral lymph nodes were cut into ized in Table S2. The quality of the model was inspected by the 1 ∕ program PROCHECK (24). The solvent accessible surface areas pieces and digested in medium containing mg mL collagenase D (Roche Diagnostics), 0.1 mg∕mL liberase blendzyme 2 (Roche were calculated with the program AREAIMOL (25), and the 0 1 ∕ structure comparisons were accomplished with the LSQKAB pro- Diagnostics), and . mg mL DNase I at 37 °C for 45 min. Stro- gram (25). The graphic figures were created with the PyMol pro- mal cells were enriched by eliminating leukocytes, using anti- gram (18). CD45 antibody-conjugated microbeads (Miltenyi). Cells were For the structure determination of the DOCK2 DHR-2 do- then suspended in DMEM/10% FBS and cultured on a fibronec- – main•Rac1 complex, the program PHASER (26) was used to cal- tin-coated, glass bottom dish for approximately 10 14 d. Cells α 10 ∕ culate the initial phases by molecular replacement. The were stimulated with TNF- ( ng mL) the day before the mi- DHR-2 coordinates from the DOCK9 DHR-2•Cdc42 structure gration assay. (PDB code 2WM9) (27) and the Rac1 coordinates from the − − Dbl homology/pleckstrin homology cassette of Trio•Rac1 (PDB Migration assay. BW5147α β cells stably expressing the wild-type code 2NZ8) (28) were used as search models. The models were or mutant DOCK2 were suspended in RPMI medium 1640/10% manually adjusted to the electron density by using the Coot pro- FBS, and loaded on a monolayer of stromal cells. After incuba- gram (22). The refinement was performed with the PHENIX pro- tion for 4 h, the cells were placed on a microscope (model IX81; gram (29). The quality of the model was inspected by the program Olympus) equipped with an incubation chamber (Chamlide TC; PROCHECK (24). The final refinement statistics are summar- Live Cell Instrument). Images were captured at 20-s intervals for ized in Table S3. The final model included all the residues of 20 min with a CCD camera (CoolSNAP HQ/OL; Photometrics). the DOCK2 DHR-2 domain and Rac1 with well-defined struc- Migration of individual cells was tracked using the Metamorph tures, whereas a 14-residue loop region (residues 1257–1270) imaging software (Molecular Devices), and cell speed was calcu- was missing, due to disorder, in the recently reported DOCK2 lated by dividing the total path length by the total assay time. DHR-2 domain•Rac1 structure (PDB code 2YIN) (30). There- Data were analyzed with ANOVA followed by Dunnett’s test.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 2of9 1. Kigawa T, Matsuda T, Yabuki T, Yokoyama S (2007) Bacterial cell-free system for highly 17. Koradi R, Billeter M, Wuthrich K (1996) MOLMOL: A program for display and analysis efficient protein synthesis. Cell-Free Protein Synthesis, eds Spirin AS, Swartz JR (Wiley, of macromolecular structures. J Mol Graph 14:51–55. New York), pp 83–97. 18. DeLano WL (2002) Pymol (computer program) (DeLano Scientific, San Carlos, CA). 2. Kigawa T, et al. (2004) Preparation of Escherichia coli cell extract for highly productive 19. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscilla- – cell-free protein expression. J Struct Funct Genomics 5:63 68. tion mode. Methods Enzymol 207:307–326. 3. Matsuda T, et al. (2007) Improving cell-free protein synthesis for stable-isotope label- 20. Vonrhein C, Blanc E, Roversi P, Bricogne G (2007) Automated structure solution with ing. J Biomol NMR 37:225–229. autoSHARP. Methods Mol Biol 364:215–230. 4. Tochio N, et al. (2006) Solution structure of the SWIRM domain of human histone de- 21. Terwilliger TC (2000) Maximum-likelihood density modification. Acta Crystallogr methylase LSD1. Structure 14:457–468. 56:965–972. 5. Delaglio F, et al. (1995) NMRPipe: A multidimensional spectral processing system based 22. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. Acta on UNIX pipes. J Biomol NMR 6:277–293. Crystallogr 60:2126–2132. 6. Kobayashi N, et al. (2007) KUJIRA, a package of integrated modules for systematic and interactive analysis of NMR data directed to high-throughput NMR structure studies. J 23. Brunger AT, et al. (1998) Crystallography & NMR system: A new software suite for – Biomol NMR 39:31–52. macromolecular structure determination. Acta Crystallogr 54:905 921. 7. Johnson BA, Blevins RA (1994) NMR View: A computer program for the visualization 24. Laskowski RA, MacArthur MW, Moss DS, Thornton JM (1993) PROCHECK: A program to and analysis of NMR data. J Biomol NMR 4:603–614. check the stereochemical quality of protein structures. J Appl Crystallogr 26:283–291 8. Bax A, et al. (1994) Measurement of homo- and heteronuclear J couplings from quan- 25. CCP4, Collaborative Computational Project No. 4. (1994) The CCP4 suite: Programs for titative J correlation. Methods Enzymol 239:79–105. protein crystallography. Acta Crystallogr 50:760–763. 9. Cavanagh J, Fairbrother WJ, Palmer AG, Skelton NJ (1996) Protein NMR Spectroscopy, 26. Storoni LC, McCoy AJ, Read RJ (2004) Likelihood-enhanced fast rotation functions. Principles and Practice (Academic, New York), pp 411–531. Acta Crystallogr 60:432–438. 1 13 10. Ikura M, Kay LE, Bax A (1990) A novel approach for sequential assignment of H, C, 27. Yang J, Zhang Z, Roe SM, Marshall CJ, Barford D (2009) Activation of Rho GTPases by 15 and N spectra of proteins: Heteronuclear triple-resonance three-dimensional NMR DOCK exchange factors is mediated by a nucleotide sensor. Science 325:1398–1402. – spectroscopy. Application to calmodulin. Biochemistry 29:4659 4667. 28. Chhatriwala MK, Betts L, Worthylake DK, Sondek J (2007) The DH and PH domains of 11. Kay LE (1997) NMR methods for the study of protein structure and dynamics. Biochem Trio coordinately engage Rho GTPases for their efficient activation. J Mol Biol Cell Biol 75:1–15. 368:1307–1320. 12. Wüthrich K (1986) NMR of Proteins and Nucleic Acids (Wiley, New York), pp 117–199. 29. Adams PD, et al. (2002) PHENIX: Building new software for automated crystallographic 13. Guntert P (2004) Automated NMR structure calculation with CYANA. Methods Mol structure determination. Acta Crystallogr 58:1948–1954. Biol 278:353–378. 30. Kulkarni K, Yang J, Zhang Z, Barford D (2011) Multiple factors confer specific Cdc42 14. Cornilescu G, Delaglio F, Bax A (1999) Protein backbone angle restraints from search- ing a database for chemical shift and . J Biomol NMR 13:289–302. and Rac protein activation by dedicator of cytokinesis (DOCK) nucleotide exchange – 15. Hooft RW, Vriend G (1996) Improved coordinate reconstruction from stereo diagrams. factors. J Biol Chem 286:25341 25351. J Mol Graph 14:168–172. 31. Sanui T, et al. (2003) DOCK2 regulates Rac activation and cytoskeletal reorganization 16. Laskowski RA, Rullmannn JA, MacArthur MW, Kaptein R, Thornton JM (1996) AQUA through interaction with ELMO1. Blood 102:2948–2950. and PROCHECK-NMR: Programs for checking the quality of protein structures solved 32. Fukui Y, et al. (2001) Haematopoietic cell-specific CDM family protein DOCK2 is essen- by NMR. J Biomol NMR 8:477–486. tial for migration. Nature 412:826–831.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 3of9 Fig. S1. List of the DOCK2 SH3-ELMO1 peptide fusion protein constructs. The underlined residues are the new amino acids added to the C terminus of the human ELMO1 sequence, and the italicized characters represent the seven extra N-terminal residues derived from the expression vector and the linker con- necting the DOCK2 SH3 domain with C-terminal region of ELMO1. The residue numbers are shown at the top. The construct used for the structure determina- tion is colored red.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 4of9 Fig. S2. NMR structure of the DOCK2 SH3-ELMO1 peptide fusion protein. (A) Schematic representation of the fusion construct used for solution structure determination by NMR. (B) Chemical shift changes for the 15N and 1HN atoms of the DOCK2 SH3 domain, upon complex formation with the ELMO1 peptide. The weighted-averaged chemicalqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi shift changes between the DOCK2 SH3 domain and the complex formation with the ELMO1 peptide ð Þ¼ δ ð Þ2 þ 0 1 × δ ð Þ2 (averaged chemical shift ppm 1 HN ppm . 15 N ppm ). The residue numbers and the secondary structures are shown at the bottom. RT loop, arginine and threonine pair loop. (C) A stereoview of the backbone superimposition of the final 20 structures by NMR. Residues Asp8–Thr70 of the DOCK2 SH3 domain (blue), residues Arg697–Phe722 of the ELMO1 peptide (red), and the N-terminal sequence derived from the expression vector and the linker residues (gray) are shown. (D) The solution structure of the DOCK2 SH3–ELMO1 peptide fusion protein. The structure of the DOCK2 SH3 domain is shown as a ribbon (blue), with the side chains of the seven residues, Tyr17, Phe19, Trp44, Arg46, Ile58, Pro60, and Phe63, that exhibit NOE connectivities with the ELMO1 peptide shown as stick models (red). The ELMO1 peptide is shown as a stick model. The labels of the two Pro residues forming the PxxP motif (Pro714 and Pro717) are underlined.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 5of9 Fig. S3. Sequence alignment of the CDM protein family (Caenorhabditis elegans CED-5, mammalian , and Myoblast city) and the ELMO proteins. (A) Sequence alignment of the N-terminal region of DOCK2 and the CDM family proteins. The secondary structure of DOCK2 in the DOCK2/ELMO1 complex is shown above the sequence. Asterisks indicate the contact residues of DOCK2 in the DOCK2/ELMO1 complex (black and red asterisks indicate hydrophobic bonds and hydrogen bonds, respectively). (B) Sequence alignment of the C-terminal regions of the ELMO proteins. The secondary struc- ture of ELMO1 in the DOCK2•ELMOl complex is shown above the sequence. Asterisks indicate the contact residues of ELMO1 in the DOCK2•ELMO1 complex. VL1—3 indicate variable loops 1–3, respectively.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 6of9 Fig. S4. The crystal structure of DOCK2•ELMOl complex. (A) Sedimentation equilibrium analysis of the DOCK2•ELMOl complex. The sedimentation equili- brium data are plotted with the residuals from the best fit to a single ideal species. This plot shows the data with the DOCK2•ELMOl complex protein at 0.4 mg∕mL and a speed of 14,000 rpm. The estimated partial specific volume of the protein was 0.428, and the calculated solvent density was 1.011 g∕mL. (B) Cartoon representation of the three-helix bundle domain of DOCK2, showing the interactions between the helices in DOCK2. Hydrophobic Leu, Ile, Val, and Trp residues are shown. (C) Internal helical domain structure of ELMO1, showing the interactions between the hydrophobic Leu, Ile, and Pro residues from the Eα1 and Eα3 helices. (D) Interface between the pleckstrin homology domain (Eα2) and the flanking helices (Eα1 and 3) of ELMO1. Conserved Leu and Trp residues around Gly674 are shown. (B–D) The DOCK2 and ELMO1 residues are labeled in black and red, respectively. (E) Superimposition of the ELMO1 structure in the DOCK2•ELMOl complex and the reported DOCK2-unbound ELMO1 structure (blue) (PDB code 2VSZ). The Pro542 residue and the Pro residues in the Pro-rich tail are colored red.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 7of9 (A) SH3 intermolecular interface PDB ID interface in Å2 SH3 interaction Molecule name 1 3a98 1,454.1 polyproline DOCK2, ELMO1 hox hox 2 1wlp 855.0 polyproline P47P , P22P 3 2k6d 816.5 domain CIN85, Ubiquitin

(B) SH3 intramolecular interface PDB ID interface in Å2 SH3 interaction Molecule name 1 1t0j 1,714.9 domain Calcium channel beta-2 subunit 2 1vyu 1,613.8 domain Calcium channel beta-3 subunit 3 1vyt 1,613.4 domain Calcium channel beta-3 subunit 4 1t3s 1,595.9 domain Calcium channel beta-2 subunit 5 1t3l 1,524.6 domain Calcium channel beta-2 subunit 6 1vyv 1,516.7 domain Calcium channel beta-4 subunit 7 3shw 1,433.8 domain Tight junction protein ZO-1 8 1uec 1,379.8 polyproline Neutrophil cytosol factor 1 9 1ng2 1,371.4 polyproline Neutrophil cytosolic factor 1 10 2dx1 1,365.5 domain Rho guanine nucleotide exchange factor 4 11 2eyz 1,359.8 domain CT10-Regulated Kinase isoform II 12 3tsz 1,356.7 domain Tight junction protein ZO-10 13 2pz1 1,355.1 domain Rho guanine nucleotide exchange factor 4 14 3tsw 1,198.2 domain Tight junction protein ZO-1 15 2dvj 1,074.5 domain CT10-Regulated Kinase isoform II 16 3lh5 1,074.5 domain Tight junction protein ZO-1 17 2fo0 1,052.6 domain Proto-oncogene tyrosine-protein kinase ABL1 18 3kfv 1,033.6 domain Tight junction protein ZO-3 Entries with the same color represent essentially same protein.

(C) PDB ID: 3a98 (D) PDB ID:1wlp (E) PDB ID:1uec

Fig. S5. (A) SH3 intermolecular interface. (B) SH3 intramolecular interface. (C–E) The structures of DOCK2 (green) of ELMO1 (yellow) (PDB code 3A98), p47phox (blue), and p22Phox (pink) (PDB code 1WLP) and Neutrophil cytosol factor 1 (orange) (PDB code 1UEC) are shown, respectively. The conserved Trp residue of the SH3 domain and its interacting residues are shown by stick models.

Table S1. Structural statistics for the DOCK2 SH3-ELMO1 peptide fusion complex (20 structures)

NOE upper distance restraints Intraresidual, ji − jj¼0 486 Medium range, 1 ≤ ji − jj ≤ 4 689 Long range, ji − jj > 4 735 Total 1,910 Dihedral angle restraints, ϕ and φ 64 2 CYANA target function value, Å 0.13 ± 0.05 No. of violations Distance violations, >0.30 Å0 Dihedral angle violations, >5.0°0 Rmsd deviation from the averaged coordinates,* Å Backbone atoms 0.78 ± 0.17 Heavy atom 1.19 ± 0.15 Ramachandran plot,* % Residues in most favored regions 76.6 Residues in additional allowed regions 23.4 Residues in generously allowed regions 0.0 Residues in disallowed regions 0.0 Rmsd Bond length, Å 0.003 Bond angle, ° 0.458 *Residues from I704–F722 of ELMO1 peptide and K9–E68 of DOCK2 SH3.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 8of9 Table S2. X-ray data collection, phasing, and refinement statistics of the DOCK2•ELMO1 complex

Data collection Space group P212121 Cell dimensions a, b, c, Å 63.9, 104.0, 124.7 α, β, γ, ° 90, 90, 90 Wavelength, Å 0.97888 Resolution, Å 50–2.1 (2.18–2.10) Unique reflections 49,140 Redundancy 7.2 (7.4) Completeness, % 100.0 (100.0) I∕σðIÞ 28.7 (2.6) R sym, % 8.8 (78.5) SAD analysis No. of Se sites 18 Phasing power 1.27 FOMSAD* 0.39 † FOMRESOLVE 0.59 Refinement Resolution, Å 49.9-2.1 (2.23-2.10) No. of reflections 48,976 No. of protein atoms 5,417 No. of solvent atoms 133 R work, % 22.6 R ‡ free, %26.8 2 Avg B factor, Å 52.9 Rmsd bond length, Å 0.007 Rmsd bond angle, ° 1.2 Ramachandran plot 89.9, 9.9, 0.2, 0.0 All numbers in parentheses represent last outer shell statistics. SAD, single-wavelength anomalous dispersion. *Figure of merit after autoSHARP. †Figure of merit after RESOLVE. ‡Free R factor is calculated for 5% of randomly selected reflections excluded from refinement.

Table S3. X-ray data collection and refinement statistics of the DOCK2-DHR2•Rac1 complex

Data collection Space group P65 Cell dimensions a, b, c, Å 168.7 168.7 129.7 α, β, γ, ° 90, 90, 120 Wavelength, Å 1 Resolution, Å 50–3.0 (3.11–3.00) Unique reflections 41,625 Redundancy 3.2 (1.9) Completeness, % 99.7 (100.0) I∕σðIÞ 9.08 (1.95) R sym, % 12.9 (59.4) Refinement Resolution, Å 48.7-3.01 (3.08-3.01) No. of reflections 41,565 No. of protein 9,898 atoms No. of solvent 10 atoms R work, % 18.3 R free,* % 23.4 2 Avg B factor, Å 46 Rmsd bond 0.009 length, Å Rmsd bond angle, 1.2 ° Ramachandran 94.5, 5.5, 0.0, 0.0 plot All numbers in parentheses represent last outer shell statistics. *Free R factor is calculated for 5% of randomly selected reflections excluded from refinement.

Hanawa-Suetsugu et al. www.pnas.org/cgi/doi/10.1073/pnas.1113512109 9of9