Structural Basis for the Recognition of Ubc13 by the Shigella Flexneri Effector Ospi

Featured Communication

Structural Basis for the Recognition of Ubc13 by the Shigella flexneri Effector OspI

Akira Nishide1, Minsoo Kim2, Kenji Takagi1, Ai Himeno2, Takahito Sanada3, Chihiro Sasakawa2,4 and Tsunehiro Mizushima1

1 - Picobiology Institute, Department of Life Science, Graduate School of Life Science, University of Hyogo, 3-2-1 Kouto, Kamigori-cho, Ako-gun, Hyogo 678–1297, Japan 2 - Division of Bacterial Infection Biology, Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108–8639, Japan 3 - Division of Bacteriology, Department of Infectious Disease Control, International Research Center for Infectious Diseases, Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108–8639, Japan 4 - Nippon Institute for Biological Science, 9-2221-1 Shinmachi, Ome, Tokyo 198–0024, Japan

Correspondence to Minsoo Kim and Tsunehiro Mizushima: [email protected]; [email protected] http://dx.doi.org/10.1016/j.jmb.2013.02.037 Edited by Y. Shi

Abstract

Ubc13 is a ubiquitin-conjugating enzyme that plays a key role in the nuclear factor-κB signal transduction pathway in human diseases. The Shigella flexneri effector OspI affects inflammatory responses by catalyzing the deamidation of a specific glutamine residue at position 100 in Ubc13 during infection. This modification prevents the activation of the TNF (tumor necrosis factor) receptor-associated factor 6, leading to modulation of the diacylglycerol–CBM (CARD– Bcl10–Malt1) complex–TNF receptor-associated factor 6–nuclear factor-κB signaling pathway. To eluci- date the structural basis of OspI function, we determined the crystal structures of the catalytically inert OspI C62A mutant and its complex with Ubc13 at resolutions of 3.0 and 2.96 Å, respectively. The structure of the OspI–Ubc13 complex revealed that the interacting surfaces between OspI and Ubc13 are a hydrophobic surface and a complementary charged Legend: A catalytic triad in OspI selectively deamidates surface. Furthermore, we predict that the complemen- Gln100 of Ubc13 to Glu, and this modification suppresses tary charged surface of OspI plays a key role in downstream inflammatory signaling during bacterial infec- substrate specificity determination. tion. The structure of the OspI–Ubc13 complex provides © 2013 Elsevier Ltd. All rights reserved. the mechanism of substrate recognition.

– Introduction to crystallize OspI Ubc13 complexes. To compar- atively assess the binding of OspI to Ubc13, we Numerous Gram-negative pathogenic bacteria generated catalytic cysteine-inactive mutants deliver virulence factors, called effectors, into host (C62A, C62S and C65S) and performed in vitro cells through the type III secretion system. To pull-down analysis. As shown in Fig. 1a, the OspI establish infection, some effectors mimic or hijack C62A mutant showed increased binding to Ubc13 the host signaling pathway, whereas others interfere compared with wild-type OspI. Therefore, we used with the host's innate immune system.1 an OspI C62A mutant for crystallization with Ubc13. Ubiquitination is one of the most prevalent post- We obtained crystals of OspI C62A in complex with translational modifications. Protein ubiquitination is Ubc13 and determined the structure by the molecular catalyzed by a sophisticated cascade system consist- replacement method (Fig. 1b and Table 1). The ing of ubiquitin-activating (E1), ubiquitin-conjugating refined structure model contains residues 21–212 of (E2), and ubiquitin-ligating (E3) enzymes. Among OspI C62A, residues 1–152 of Ubc13, and four iodine these enzymes, E3 enzymes are responsible for the atoms. The structure of OspI has an α/β fold, with four selection of target proteins. E3 enzymes are a diverse β-strands (β1–β4), seven α-helices (α1–α7), and a family of proteins [RING (really interesting new gene) 310 helix (310-1). The electron densities of the N- or HECT (homologous to E6-associated protein terminal domain (residues 1–21) of OspI are not C terminus type)] and multiple protein complexes.2 visible, suggesting that this region is flexible. Ubc13 TNF (tumor necrosis factor) receptor-associated adopts a canonical E2 fold, consisting of a four- factor 6 (TRAF6) is a RING-type E3 ligase. TRAF6 stranded antiparallel β-sheet (S1–S4), four α-helices – regulates the nuclear factor-κB activation signaling (H1 H4), and a 310 helix (310-1). The folds of the OspI pathway to induce immune and inflammatory re- C62A and Ubc13 in the complex were almost identical with those of each free form (Fig. 1c). The sponses and confer protection against apoptosis. α TRAF6 functions with a heterodimer E2 that consists root-mean-square deviation (r.m.s.d.) of the C atoms between free and Ubc13-bound OspI was of Ubc13 and a ubiquitin E2 variant (Uev1A) to α synthesize K63-linked polyubiquitin chains on target 1.2 Å for 191 residues, whereas that of the C atoms proteins and TRAF6 itself.3 This K63-linked poly- between free and OspI-bound Ubc13 was 1.7 Å for ubiquitination leads to the activation of the down- 148 residues. However, the H1 helix of Ubc13 had a stream signaling pathway. Unlike K48-linked different conformation when Ubc13 was bound to polyubiquitin chains that are hallmarks of proteaso- OspI. The N-terminal end of the H1 helix had shifted mal degradation, K63-linked polyubiquitin is nonde- about 1.8 Å closer to OspI (Fig. S1). gradative and has been shown to function as a signaling moiety in DNA damage repair processes Interaction between OspI and Ubc13 and innate immunity pathways.4,5 OspI, a Shigella flexneri effector, is a glutamine – deamidase that selectively deamidates the glutamine The structure of the OspI Ubc13 complex residue at position 100 in Ubc13 to a glutamic acid revealed that OspI and Ubc13 are bound by interactions of 310-1 and loops α2–α3, 310-1–α4, residue. This modification inhibits the E2 ubiquitin- β –β β –β conjugating activity, which is required for TRAF6 1 2, and 3 4 of OspI with H1, H2, and loops S3–S4 and 310-1–H2 of Ubc13 (Fig. 2a and Fig. activation, and leads to the suppression of TRAF6- 2 dependent inflammatory signaling during bacterial S2). A total of 970 Å of the accessible surface area is buried at the interface between OspI and Ubc13. infection. We recently determined the OspI crystal – – structure, which has a cysteine protease-like fold and Ubc13 is bound to OspI by its S3 S4 and 310-1 H2 a conserved active-site catalytic triad (Cys62, His145, loops against the concave surface of OspI, and and Asp160).6 Mutation of the catalytic triad caused Ubc13 extends its Gln100 to the active site of OspI. OspI to lose its Ubc13 deamidation activity. To better The residues involved in intermolecular formation understand the mechanistic details underlying OspI- are Asp59, Gly60, Ser63, Ile87, Thr92, Phe95, mediated TRAF6 inhibition, we determined the crystal Asp103, Glu141, Gln142, Ala143, Thr144, Tyr170, structures of the OspI C62A–Ubc13 complex and the and Asn184 of OspI and Gly3, Leu4, Arg6, Arg7, OspI C62A mutant. Arg14, Met64, Ser96, Pro97, Ala98, Leu99, Gln100, Thr103, and Leu106 of Ubc13. Gly60, Glu141, Gln142, Thr144, and Asn184 of OspI form a Results and Discussion potential salt bridge and hydrogen bond with Ala98 and Arg6 of Ubc13, respectively (Fig. S2). The interacting surfaces of these two proteins are a Overall structure of the OspI–Ubc13 complex hydrophobic surface and a complementary charged surface (Fig. 2b). To investigate the roles of To define how OspI engages its substrate and residues involved in complex formation, we intro- catalyzes the deamidation of Ubc13, we attempted duced point mutations into OspI and Ubc13. We Recognition of Ubc13 by OspI 2625

Fig. 1. Crystal structure of the OspI C62A–Ubc13 complex. (a) Ubc13 binding activity of OspI mutants. Pull-down analysis using GST–OspI mutants with Ubc13. GST–OspI wild type and its mutants were expressed in BL21. The cell extracts were incubated with glutathione–Sepharose 4B, and the immobilized proteins were washed extensively with a pull-down buffer [25 mM Tris–HCl (pH 7.5), 20 mM 2-mercaptoethanol, and 150 mM NaCl]. His–Ubc13 and its mutants were expressed in BL21, purified by Ni-NTA affinity chromatography and anion-exchange chromatography, and dialyzed in the pull-down buffer. The resulting immobilized glutathione–Sepharose 4B containing GST–OspI wild type or its mutants were incubated with 40 μM Ubc13 and its mutants in 200 μl of the pull-down buffer on ice for 1 h and then washed two times. GST-bound proteins were suspended in SDS-PAGE sample buffer and subjected to SDS-PAGE, followed by Coomassie Brilliant Blue staining. (b) A ribbon diagram of the OspI C62A–Ubc13 complex. OspI C62A and Ubc13 are colored blue and orange, respectively. Secondary structural elements are labeled. The side chains of the active-site residues (A62–H145–D160) of OspI and Q100 of Ubc13 are shown in stick models. (c) Structural comparison of OspI (PDB code 3B21; magenta) and Ubc13 (PDB code 1JBB; green) in their free forms and in the OspI C62A–Ubc13 complex. The molecular orientation is the same as that in (b). mutated selected residues of the OspI C62A–Ubc13 deamidation. The R14A mutation did not affect interface to Ala (R6A, R14A, M64A, and L99A) in binding to OspI and was deamidated by OspI. Next, Ubc13 and conducted a glutathione S-transferase we created a series of OspI mutants to determine (GST) pull-down experiment to monitor complex the residues important for OspI–Ubc13 complex formation (Fig. 2c, left). The R6A and L99A mutants formation. We performed an in vitro pull-down assay showed very weak binding to OspI. Mutation M64A with OspI mutants and Ubc13 (Fig. 2c, right). We also reduced binding. However, the R14A mutant found that OspI I87A, F95A, and Q142A mutants had no effect. This result suggested that R6, M64, showed a decreased ability to bind to Ubc13 and L99 of Ubc13 participate in OspI binding. We compared with the C62A mutant. In addition, we investigated substrate deamidation with a series of investigated the substrate deamidation activity with Ubc13 mutants using native PAGE (Fig. S3). a series of OspI mutants using native PAGE. As Consistent with their very weak binding, the R6A shown in Fig. 2d, OspI mutants I87A, F95A, and and L99A mutants were not deamidated by OspI. Y170A have relatively low deamidation activity. Although the M64A mutant showed reduced binding These results suggest that the hydrophobic in- to OspI, the reduced binding was sufficient for teractions are crucial for OspI–Ubc13 binding. 2626 Recognition of Ubc13 by OspI

Table 1. Crystallographic data collection and refinement statistics

OspI C62A–Ubc13 complex OspI C62A mutant Data collection Wavelength (Å) 0.9 0.9 Space group P321 P212121 Unit cell parameters (Å) a 119.2 67.5 b 119.2 67.8 c 70.0 85.0 Resolution (Å) 50.0–2.96 (3.00–2.96) 50.0–3.00 (3.05–3.00) No. of measured reflections 56,970 49,134 No. of unique reflections 12,193 8194 Completeness (%) 99.8 (100.0) 99.9 (100.0) Redundancy 3.4 (3.5) 6.0 (5.7) Mean I/σ(I) 13.3 (4.2) 12.5 (5.3) Rmerge (%) 9.4 (30.0) 16.0 (29.4)

Structure refinement Resolution range (Å) 33.14–2.96 41.63–3.00 No. of reflections 12,150 7750 Rcryst (%) 24.5 24.8 Rfree (%) 28.6 26.9 r.m.s.d. bond lengths (Å) 0.009 0.003 r.m.s.d. bond angles (°) 1.381 0.404 No. of protein atoms 2699 2980 Others (iodine) 4 0 Average B-factors (Å2) Protein 72.2 12.9 Iodine 98.5 — Ramachandran plot (%) Most favored region 77.0 84.1 Additional allowed 21.0 15.9 Generously allowed 2.0 0 Values in parentheses are for the highest-resolution shell. OspI and its mutants were cloned into the pGEX6P-1 vector and expressed in Escherichia coli BL21. OspI was purified in a stepwise process using glutathione–Sepharose 4B, anion-exchange chromatography, and gel-filtration chromatography. The GST moiety was proteolytically removed by PreScission protease. For the OspI–Ubc13 complex, we used the OspI C62A mutant and His–Ubc13. Ubc13 was cloned into the pET28 vector and expressed in E. coli BL21. Ubc13 was purified by Ni-affinity chromatography, and the His tag was cleaved from Ubc13 using thrombin. The OspI C62A mutant and Ubc13 were combined and purified by gel-filtration chromatography. Fractions containing the OspI C62A–Ubc13 complex were then concentrated to 51 mg/ml and used for crystallization. OspI C62A was concentrated to 40 mg/ml by ultrafiltration in 25 mM Tris–HCl (pH 7.5) and 1 mM DTT. Crystallization of the OspI C62A mutant was performed using the sitting-drop vapor-diffusion method at 293 K in drops containing a mixture of 1 μl of protein solution and 1 μlof reservoir solution, which consisted of 0.2 M magnesium acetate, 0.1 M Tris–HCl (pH 6.5), and 25% (w/v) polyethylene glycol 8000. The OspI C62A–Ubc13 crystals were prepared using the hanging-drop vapor-diffusion method at 293 K in drops containing a mixture of 1 μlof protein solution and 1 μl of reservoir solution, which consisted of 0.4 M potassium iodide, 0.1 M 4-morpholineethanesulfonic acid (pH 6.5), 4 mM DTT, and 15% polyethylene glycol 8000. X-ray diffraction data sets for OspI C62A and the OspI C62A–Ubc13 complex were collected at 100 K on BL44XU (SPring-8). Data processing and reduction were conducted with HKL2000.7 The structure of the OspI C62A mutant was determined by molecular replacement in MOLREP8,9 using as a search model the wild-type OspI (PDB ID 3B21) structure6 lacking residues 59–77. The structure of the OspI C62A–Ubc13 complex was determined using the molecular replacement technique PHENIX10 and the wild-type OspI and Ubc1311 (PDB ID 1JBB). The models were subsequently improved through alternate cycles of manual rebuilding using Coot12 and refinement with the program REFMAC513 and CNS.14,15 There were no residues in the disallowed regions of the Ramachandran plot drawn by program PROCHECK.16 Structure figures were generated using PyMOL17 and CCP4MG.18

Catalytic site in OspI by 2.7 Å, and this acts to extend the α3 helix, which displaces Cys62 toward the active-site triad. We found that the active site of OspI showed We next examined whether the conformational significant conformational changes in the structure of change in the active-site loop can occur in the the complex6,19 (Fig. 3a). In the active-site Cys62 of uncomplexed OspI C62A mutant. The OspI C62A γ wild-typeOspI,theS position was distantly mutant protein was crystallized, and its structure was positioned compared with that of the cysteine solved at a resolution of 3.0 Å (Fig. S4). The OspI protease family. However, marked structural C62A crystal structure comprises two molecules in changes were found in the complex in the α2–α3 the asymmetric unit. The overall structures of these loop compared with wild-type OspI; Ala62 is shifted two proteins are similar (r.m.s.d. of 0.3 Å for 192 Cα Recognition of Ubc13 by OspI 2627 atoms). We will focus on chain A in the discussion active-site structure of the C62A mutant is similar to below. The overall structure of OspI C62A corre- that of the OspI C62A–Ubc13 complex. The side sponds well with that of the wild type (Fig. S4). The chain of Ala62 corresponding to Cys62 of OspI is

Fig. 2 (legend on next page) 2628 Recognition of Ubc13 by OspI

turned out toward His145 (Fig. 3a). We previously covers the H1 helix and S3–S4 and 310-1–H2 reported that the crystal structure of OspI is the loops.25,26 TRAF6, CHIP, and OspI recognize Ubc13 closest to that of AvrPphB, a papain-like cysteine through similar interfaces (Fig. 3b and Fig. S7). The protease.19 In the superimposed structures, the binding of TRAF6 and CHIP to Ubc13 mainly involves catalytic His and main-chain atoms of Cys (Ala for hydrophobic interactions. The interacting surfaces OspI C62A) and aspartic acids align well in OspI C62A between OspI and Ubc13 are a hydrophobic surface and AvrPphB (Fig. 3a). At the active site of AvrPphB, and a complementary charged surface. These proteins Asn93 functions to form an oxyanion hole. In OspI, a have different overall structures but present a very Gln162 residue is located at a position corresponding to similar hydrophobic binding surface (Figs. 2band3d). that of Asn93 in AvrPphB (Fig. 3a). In addition, Gln100 In the crystal structure of Ubc13 complexed with of Ubc13 in the complex protrudes into the active-site the deubiquitinating enzyme OTUB1 (OTU-domain- cleft. The distance between Gln100 of Ubc13 and containing ubiquitin aldehyde-binding protein 1), it Ala62 of OspI in the complex suggests that substrate has been recently reported that OTUB1 suppresses residue Gln100 can occupy an appropriate position in K63-linked polyubiquitination by binding Ubc13.27,28 wild-type OspI for deamidation (Fig. S5). Recent OTUB1 interacts with Ubc13 through the H1 helix studies have identified several bacterial toxins and and S3–S4 and 310-1–H2 loops, which are the same type III secreted effectors that act as glutamine regions involved in OspI binding (Fig. S7). Further- deamidases.20,21 Among these deamidases, Cif family more, OTUB1 and Ubc13 interactions are mainly proteins deamidate Gln40 of ubiquitin and NEDD8. The hydrophobic, as are OpsI–Ubc13 interactions. Al- crystal structures of Cif family proteins complexed with though OspI and OTUB1 interact with Ubc13 their substrates have been reported.22,23 Although the similarly, OspI inhibits Ubc13 through a different overall structures of OspI and Cifs [Protein Data Bank mechanism. OTUB1 inhibits Ubc13–E3 interaction (PDB) IDs 4F8C, 4HCN, and 4HCP] and their active by occupying the Ubc13 interface with E3s.27,28 sites are similar (r.m.s.d. range, 4.3–4.4 Å), the However, the wild-type OspI–Ubc13 complex is not glutamine substrates of Cifs are located about 3.0 Å stable. The deamidation of Gln100 may decrease closer to the active site than that of OspI (Fig. S6a). The the affinity of OspI and TRAF6 for Ubc13. structures of the interfaces between deamidases and Next, we compared the structures of free Ubc13 their substrates have different shape complementarity and OspI-bound Ubc13 with the known structures of and surface charge distribution (Fig. S6b). three complexes (Ubc13–TRAF6, Ubc13–CHIP, and Ubc13–OTUB1). The overall structure of Ubc13 is The interaction surface on Ubc13 almost identical among these complexes (r.m.s.d. range for 146–150 Cα atoms, 1.4–1.7 Å) (Fig. S8). Ubc13 forms a heterodimer with its E2 variant, Their overlapping binding sites in Ubc13 suggest Uev1a, and this heterodimer interacts with specific that a convergence exists in E2 recognition. TRAF6, ubiquitin ligases such as TRAF6 and CHIP (carboxy CHIP, and OTUB1 interact with not only Ubc13 but terminus of Hsp70-interacting protein). The structures also UbcH5.29–31 However, OspI does not specifi- of Ubc13–Uev1a,11,24 Ubc13–TRAF6,25 and Ubc13– cally bind with UbcH5 (Fig. 3c). The complementary CHIP26 have been previously determined. The struc- charged interactions between OspI and Ubc13 may ture of the Ubc13–Uev1a complex showed that the S2, play a key role in specificity determination for Ubc13 S3, and S4 strands and loops S2–S3 and S4–310-1 of (Figs. 2b and 3d and Fig. S9). The Shigella effector Ubc13 interact with Uev1a. We generated a model of protein IpaH family is a ubiquitin ligase. The IpaH the Ubc13–Uev1a–OspI complex by superimposing family was shown to have ubiquitin ligase activity in Ubc13 from the Ubc13–Uev1a and Ubc13–OspI the presence of UbcH5 in vitro.32 This substrate structures. This model showed that Uev1a is located selectivity between these effectors may be an at one end of Ubc13, whereas OspI is located at the advantage at different stages of infection. other end (Fig. 3b). Previous studies have indicated In summary, we have solved the crystal structures that the binding region of TRAF6 and CHIP in Ubc13 of the OspI C62A mutant and the OspI C62A–Ubc13

Fig. 2. Interactions between OspI and Ubc13. (a) Stereo view of the OspI–Ubc13 interface showing amino acids of OspI (blue) and Ubc13 (orange). Broken lines indicate hydrogen bonds. (b) Surface potential representation of OspI and Ubc13. The complementary surface patches (A and B) responsible for complex formation are depicted by a circle. Red, blue, and white represent acidic, basic, and neutral, respectively. (c) Characterization of the crucial residues in the OspI–Ubc13 interface. A pull-down assay was performed as described in Fig. 1a using Ubc13 mutants with GST–OspI C62A (left). A pull-down assay using GST–OspI mutants with wild-type Ubc13 (right). GST–OspI and its mutants were purified as described in Fig. 1a. GST-bound proteins were pulled down from lysates of bacteria expressing Ubc13–S tag. The bound proteins were immunoblotted with anti-S tag and anti-GST antibodies. (d) Deamidation activities of mutant OspI proteins. GST–OspI and its mutants were expressed in BL21; purified stepwise using affinity chromatography, anion-exchange chromatography, and gel-filtration chromatography; and dialyzed in reaction buffer. Ubc13 was incubated with purified GST–OspI or its mutants at 30 °С for 10 min in the reaction buffer (20 mM Tris–HCl at pH 7.5, 100 mM NaCl, and 0.1 mM DTT). The reaction samples were separated by native PAGE and stained with Coomassie Brilliant Blue. Recognition of Ubc13 by OspI 2629

Fig. 3. Structural features of the activity of OspI. (a) Stereo view of comparison of the catalytic sites among AvrPphB (green), OspI WT (purple), OspI C62A (yellow), and Ubc13-bound OspI C62A (cyan). All atoms of His145 and Asp160 were superimposed on the corresponding residues of the counterpart proteins by LSQKAB in the CCP4 program suite. (b) Comparison of the Uev1a, TRAF6, CHIP, OTUB1, and OspI binding sites on Ubc13. Ubc13 is shown as a ribbon model, and Uev1a, TRAF6, CHIP, and OTUB1 are shown as Cα trace representations and are colored orange (Ubc13), blue (OspI), violet (Uev1a; PDB code 2C2V), dark green (CHIP; PDB code 2C2V), purple (TRAF6; PDB code 3HCT), gray (OTUB1; PDB code 4DHZ), and light yellow (OTUB1; PDB code 3VON). (c) UbcH5a binding activity of OspI. A pull-down assay was performed as described in Fig. 1a using UbcH5a with GST–OspI C62A. (d) Surface potential representation of the Ubc13 binding site of TRAF6, CHIP, and OTUB1. The complementary surface patches responsible for complex formation are depicted by a circle. 2630 Recognition of Ubc13 by OspI complex. The refined structure in conjunction with References mutational data provided further insight into the interaction between OspI and Ubc13. Although the 1. Kim, M., Ashida, H., Ogawa, M., Yoshikawa, Y., mechanisms through which OspI regulates TRAF6 Mimuro, H. & Sasakawa, C. (2010). Bacterial in- activity remain unclear, our results have important teractions with the host epithelium. Cell Host Microbe, consequences for understanding the molecular 8,20–35. 2. Deshaies, R. J. & Joazeiro, C. A. (2009). RING domain basis of the inhibition of TRAF6-dependent inflam- – matory signaling in response to bacterial infection. E3 ubiquitin ligases. Annu. Rev. Biochem. 78, 399 434. 3. Chen, Z. J. (2005). Ubiquitin signalling in the NF-κB pathway. Nat. Cell Biol. 7, 758–765. κ Accession numbers 4. Iwai, K. (2012). Diverse ubiquitin signaling in NF- B activation. Trends Cell Biol. 22, 355–364. 5. Panier, S. & Durocher, D. (2009). Regulatory ubiqui- Atomic coordinates and structure factors of OspI – tylation in response to DNA double-strand breaks. C62A and OspI C62A Ubc13 complex have been DNA Repair (Amst), 8, 436–443. deposited in the Protein Data Bank with accession 6. Sanada, T., Kim, M., Mimuro, H., Suzuki, M., Ogawa, numbers 3W30 and 3W31, respectively. M., Oyama, A. et al. (2012). The Shigella flexneri Supplementary data to this article can be found effector OspI deamidates UBC13 to dampen the online at http://dx.doi.org/10.1016/j.jmb.2013.02.037 inflammatory response. Nature, 483, 623–626. 7. Otwinowski, Z. & Minor, W. (1997). Processing of X- ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. 8. Collaborative Computational Project, Number 4. (1994). The CCP4 suite: programs for protein crystal- Acknowledgements lography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50, 760–763. We thank all members of beamline BL44XU for help 9. Vagin, A. & Teplyakov, A. (2009). Molecular replace- in data collection at SPring-8. Diffraction data were ment with MOLREP. Acta Crystallogr., Sect. D: Biol. collected at Osaka University using beamline BL44XU Crystallogr. 66,22–25. at SPring-8 (Proposal Nos. 2011B6645, 2012A6744, 10. Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. and 2012B6744) equipped with an MX225-HE detec- B., Davis, I. W., Echols, N. et al. (2010). PHENIX: a comprehensive Python-based system for macromo- tor (Rayonix), financially supported by Academia lecular structure solution. Acta Crystallogr., Sect. D: Sinica and National Synchrotron Radiation Research Biol. Crystallogr. 66, 213–221. Center (Taiwan). We are grateful to R. Whittier and M. 11. VanDemark, A. P., Hofmann, R. M., Tsui, C., Pickart, Fukumatsu for their advice. This work was supported C. M. & Wolberger, C. (2001). Molecular insights into in part by the following grants from the Ministry of polyubiquitin chain assembly: crystal structure of the Education, Culture, Sports, Science and Technology Mms2/Ubc13 heterodimer. Cell, 105, 711–720. of Japan: a Grant-in-Aid for Specially Promoted 12. Emsley, P. & Cowtan, K. (2004). Coot: model-building Research [23000012 (C.S.)], a Grant-in-Aid for tools for molecular graphics. Acta Crystallogr., Sect. – Scientific Research on Innovative Areas [24112009 D: Biol. Crystallogr. 60, 2126 2132. (T.M.)], a Grant-in-Aid for Scientific Research (B) 13. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by [24370049 (T.M.)], and a Grant-in Aid for Young the maximum-likelihood method. Acta Crystallogr., Scientists (A) [23689027 (M.K.)]. Part of this work was Sect. D: Biol. Crystallogr. 53, 240–255. supported by grants from the Naito Foundation (M.K.). 14. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography & NMR system: a new software suite Received 11 December 2012; for macromolecular structure determination. Acta Received in revised form 23 February 2013; Crystallogr., Sect. D: Biol. Crystallogr. 54, 905–921. Accepted 27 February 2013 15. Brunger, A. T. (2007). Version 1.2 of the Crystallog- Available online 27 March 2013 raphy and NMR system. Nat. Protoc. 2, 2728–2733. 16. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Keywords: Thornton, J. M. (1993). PROCHECK: a program to Shigella flexneri; check the stereochemical quality of protein structures. – Ubc13; J. Appl. Crystallogr. 26, 283 291. deamidase; 17. Delano, W. L. (2002). The PyMOL molecular graphics, system. v. 0. 97. ubiquitination; 18. McNicholas, S., Potterton, E., Wilson, K. S. & Noble, crystal structure M. E. (2011). Presenting your structures: the CCP4mg molecular-graphics software. Acta Crystallogr., Sect. Abbreviations used: D: Biol. Crystallogr. 67, 386–394. TRAF6, TNF receptor-associated factor 6; GST, 19. Zhu, M., Shao, F., Innes, R. W., Dixon, J. E. & Xu, Z. glutathione S-transferase; PDB, Protein Data Bank. (2004). The crystal structure of Pseudomonas Recognition of Ubc13 by OspI 2631

avirulence protein AvrPphB: a papain-like fold with a 26. Zhang, M., Windheim, M., Roe, S. M., Peggie, M., distinct substrate-binding site. Proc. Natl Acad. Sci. Cohen, P., Prodromou, C. & Pearl, L. H. (2005). USA, 101, 302–307. Chaperoned ubiquitylation—crystal structures of the 20. Cruz-Migoni, A., Hautbergue, G. M., Artymiuk, P. J., CHIP U box E3 ubiquitin ligase and a CHIP-Ubc13– Baker, P. J., Bokori-Brown, M., Chang, C. T. et al. Uev1a complex. Mol. Cell, 20, 525–538. (2011). A Burkholderia pseudomallei toxin inhibits 27. Wiener, R., Zhang, X., Wang, T. & Wolberger, C. helicase activity of translation factor eIF4A. Science, (2012). The mechanism of OTUB1-mediated inhibition 334, 821–824. of ubiquitination. Nature, 483, 618–622. 21. Taieb, F., Nougayrede, J. P. & Oswald, E. (2011). 28. Sato, Y., Yamagata, A., Goto-Ito, S., Kubota, K., Cycle inhibiting factors (cifs): cyclomodulins that usurp Miyamoto, R., Nakada, S. & Fukai, S. (2012). the ubiquitin-dependent degradation pathway of host Molecular basis of Lys-63-linked polyubiquitination cells. Toxins (Basel), 3, 356–368. inhibition by the interaction between human deubiqui- 22. Crow, A., Hughes, R. K., Taieb, F., Oswald, E. & tinating enzyme OTUB1 and ubiquitin-conjugating Banfield, M. J. (2012). The molecular basis of enzyme UBC13. J. Biol. Chem. 287, 25860–25868. ubiquitin-like protein NEDD8 deamidation by the 29. Windheim, M., Peggie, M. & Cohen, P. (2008). Two bacterial effector protein Cif. Proc. Natl Acad. Sci. different classes of E2 ubiquitin-conjugating enzymes USA, 109, E1830–E1838. are required for the mono-ubiquitination of proteins 23. Yao, Q., Cui, J., Wang, J., Li, T., Wan, X., Luo, T. et al. and elongation by polyubiquitin chains with a specific (2012). Structural mechanism of ubiquitin and NEDD8 topology. Biochem. J. 409, 723–729. deamidation catalyzed by bacterial effectors that 30. Xu, Z., Kohli, E., Devlin, K. I., Bold, M., Nix, J. C. & induce macrophage-specific apoptosis. Proc. Natl Misra, S. (2008). Interactions between the quality Acad. Sci. USA, 109, 20395–20400. control ubiquitin ligase CHIP and ubiquitin conjugating 24. Moraes, T. F., Edwards, R. A., McKenna, S., enzymes. BMC Struct. Biol. 8, 26. Pastushok, L., Xiao, W., Glover, J. N. & Ellison, M. 31. Juang, Y. C., Landry, M. C., Sanches, M., Vittal, V., J. (2001). Crystal structure of the human ubiquitin Leung, C. C., Ceccarelli, D. F. et al. (2012). OTUB1 conjugating enzyme complex, hMms2-hUbc13. Nat. co-opts Lys48-linked ubiquitin recognition to suppress Struct. Biol. 8, 669–673. E2 enzyme function. Mol. Cell, 45, 384–397. 25. Yin, Q., Lin, S. C., Lamothe, B., Lu, M., Lo, Y. C., Hura, 32. Rohde, J. R., Breitkreutz, A., Chenal, A., Sansonetti, G. et al. (2009). E2 interaction and dimerization in the P. J. & Parsot, C. (2007). Type III secretion effectors of crystal structure of TRAF6. Nat. Struct. Mol. Biol. 16, the IpaH family are E3 ubiquitin ligases. Cell Host 658–666. Microbe, 1,77–83.