Autoinhibition and phosphorylation-induced activation mechanisms of cancer and autoimmune disease-related E3 Cbl-b

Yoshihiro Kobashigawaa, Akira Tomitakab, Hiroyuki Kumetaa, Nobuo N. Nodaa,1, Masaya Yamaguchia, and Fuyuhiko Inagakia,2

aDepartment of Structural Biology, Faculty of Advanced Life Science, and bGraduate School of Life Science, Hokkaido University, Sapporo, Hokkaido 001-0021, Japan

Edited by* Joseph Schlessinger, Yale University School of Medicine, New Haven, CT, and approved September 28, 2011 (received for review July 6, 2011)

Cbl-b is a RING-type E3 ligase that functions as a negative reported to be negatively regulated by the TKB domain, showing regulator of T-cell activation and and non- that phosphorylation removes Cbl protein autoinhibi- receptor-type signaling. Cbl-b dysfunction is related tion due to the interaction between the RING and TKB domains. to autoimmune diseases and cancers in . However, the Protease susceptibility analysis has also revealed that the phos- molecular mechanism regulating its E3 activity is largely unknown. phorylation of the Y363 of Cbl-b induces a large conformational NMR and small-angle X-ray scattering analyses revealed that the change that is sensitive to protease digestion (15). However, the unphosphorylated N-terminal region of Cbl-b forms a compact manner in which the molecular mechanism underlying this con- structure by an intramolecular interaction, which masks the inter- formational change leads to the upregulation of E3 activity action surface of the RING domain with an E2 ubiquitin-conjugat- remains elusive. ing enzyme. Phosphorylation of Y363, located in the helix-linker The crystal structure of the c-Cbl N-terminal domain, consist- region between the tyrosine kinase binding and the RING domains, ing of the TKB domain, the helix linker, and RING in complex disrupts the interdomain interaction to expose the E2 binding with UbcH7 (E2) was reported (18). Although this study was

surface of the RING domain. Structural analysis revealed that the the first to report on the complex between RING-type E3 and E2 BIOPHYSICS AND

phosphorylated helix-RING region forms a compact structure in and provided significant insights into the interaction mechanism, COMPUTATIONAL BIOLOGY solution. Moreover, the phosphate group of pY363 is located in the the structural basis for the enhancement of c-Cbl ligation activity vicinity of the interaction surface with UbcH5B to increase affinity through Y371 phosphorylation remains largely unknown. by reducing their electrostatic repulsion. Thus, the phosphorylation Here we report the structural analysis of the Cbl-b N-terminal of Y363 regulates the E3 activity of Cbl-b by two mechanisms: half (Cbl-b39–426; hereafter CBLB-N) both in the unphosphory- one is to remove the masking of the RING domain from the tyro- lated and Y363-phosphorylated states (hereafter pY CBLB-N) sine kinase binding domain and the other is to form a surface to using small-angle X-ray scattering (SAXS) and NMR spectro- enhance binding affinity to E2. scopy, and discuss the structural mechanism of autoinhibition and Y363 phosphorylation induced activation of Cbl-b E3 ligase. he Cbl (c-Cbl, Cbl-b, and Cbl-3) belong to a family Tof RING-type ubiquitin ligases. Like other RING domain Results proteins, the Cbl proteins function as adaptor proteins, simulta- Overall Structural Changes in CBLB-N Due to Y363 Phosphorylation. neously binding to a cognate E2 ubiquitin-conjugating enzyme First, we studied the overall structures of both unphosphorylated and a substrate protein, leading to transfer of ubiquitin to the and phosphorylated CBLB-N in solution using SAXS, as Y363 substrate. This facilitates degradation of the target substrate by was reported to be responsible for phosphorylation-induced liga- proteasomes or, in some cases, . The Cbl proteins func- tion activity (15). The radii of gyration (Rg), as estimated by the tion as a negative regulator of T-cell activation, growth factor Guinier approximation (19), were 24.0 and 26.7 Å for CBLB-N receptor [e.g., epidermal growth factor receptor (EGFR), c-KIT, and pY CBLB-N, respectively (Fig. 1B), suggesting that CBLB-N and platelet-derived growth factor receptor (PDGFR)], and non- is more elongated in the phosphorylated state than in the un- receptor-type tyrosine kinase signaling (e.g., Src family kinases phosphorylated state. Protease susceptibility analysis of Cbl-b and Zap70) (1, 2), and dysfunctional mutations in Cbl proteins revealed that phosphorylated Cbl-b is labile to protease cleavage have been related to human cancer (3–7). Of the three Cbl pro- (15). These results taken together support the idea that pY teins, Cbl-b plays a critical role in the down-regulation of immu- CBLB-N is more extended and more mobile than CBLB-N, nological signaling to induce T-cell anergy (8, 9). Cbl-b knockout indicating that the conformational changes in CBLB-N are mice have been shown to exhibit severe autoimmune diseases induced by phosphorylation at Y363. (10), whereas, in humans, a dysfunctional mutation in Cbl-b was shown to be related to type I diabetes (11) and multiple sclerosis Author contributions: Y.K. and F.I. designed research; Y.K., A.T., H.K., N.N.N., and M.Y. (12). Hence, the Cbl proteins are considered to be a potential performed research; Y.K. and A.T. contributed new reagents/analytic tools; Y.K., A.T., therapeutic target. H.K., and N.N.N. analyzed data; and Y.K. and F.I. wrote the paper. Cbl-b and c-Cbl exhibit high sequence in the The authors declare no conflict of interest. N-terminal region with 86% amino acid identity and share a con- *This Direct Submission article had a prearranged editor. served tyrosine kinase binding (TKB) domain comprised of a Freely available online through the PNAS open access option. four-helix bundle, a Ca2þ-binding EF hand domain and a variant Data deposition: The NMR, atomic coordinates, chemical shifts, and restraints reported in SH2 domain (13, 14), as well as a short helix-linker region and a this paper have been deposited in the BioMagResBank, www.bmrb.wisc.edu (accession RING finger domain that directly associates with E2 proteins no. Q28: BMR17680), and the atomic coordinates have been deposited in the Protein Data (Fig. 1A). The activity of Cbl-b is known to be Bank, www.pdb.org (PDB ID codes 2LDR and 3VGO). up-regulated by the phosphorylation of Y363 (Y371 in c-Cbl), 1Present address: Institute of Microbial Chemistry, Tokyo 141-0021, Japan. which is located in the helix linker (15, 16). The mutation of this 2To whom correspondence should be addressed. E-mail: [email protected]. critical tyrosine residue to phenylalanine abolishes the E3 activity This article contains supporting information online at www.pnas.org/lookup/suppl/ of Cbl (17). Moreover, the E3 activity of the Cbl proteins is doi:10.1073/pnas.1110712108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1110712108 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 27, 2021 A Y371 A D c-Cbl 4H EF SH2 RING PRR UBA 906 G417 I415 Y363 G405 39 343 351 375 426 G367 Cbl-b 982 T418(s) 110 4H EF SH2 RING PRR UBA I421 Helix G407 I375 TKB domain G389 180° 83% 90% T418 86% M392 S368 115 Sequence Identity S403 39 426 CBLB-N T398 39 375 TKB-H Q371 D404 C376 N (ppm) F410 E378 351 426 H-RING L391 C411 I422 R412 Q401 C413 I422 N379 C393 E402 I421 L372 120 T394 S395 K381 * W400 B C E419(s) D380 H390 F370 E I375 C388 E414 L397 100 Y368 V383 C408 D382 A399 C373 F426 125 M132 4H EF SH2 RING 105 E386 E419 I422(s) 13.0 39 426 I415 351 C396 V423 2 I385 14N 15N K374 K384 10 110 A377 V423(s) M53 D424 I422 130 M214 14.0 1 K416 * C (ppm) ln (Intensity) M153 115 N (ppm) M115 0 10.0 9.0 8.0 7.0 M392 M53 15.0 1 0.002 0.003 0.004 0.005 0.006 M365 M214 q2 ( Å -2) HN (ppm) 120 * Intensity 16.0 B R412 C411 R412 C413 T394 M115 0.1 125 V383 I375 M261 M365 17.0

130 E378 180° C376 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.01 0.05 0.1 0.15 0.2 0.25 135 H (ppm) 11.0 10.0 9.0 8.0 7.0 q( -1) K374 F Å F370 E378 HN (ppm) L372 F370 K381 M115 L391 N379 M53 D380 M214 Fig. 1. (A) Domain structures of c-Cbl and Cbl-b. (B) SAXS measurements of C 180° CBLB-N (red) and pY CBLB-N (blue). Inset represents the Guinier plots for 1 15 CBLB-N (red) and pY CBLB-N (blue). (C) Overlay of the H- N HSQC spectra 8.0

between the segmental isotope-labeled CBLB-N (red) and pY CBLB-N (blue) at M132 I385 I422 90° the H-RING moiety. I375 10.0 C (ppm) I415 M365 To obtain further insights into the conformational changes I421 12.0 induced by phosphorylation at Y363, we prepared segmental isotope-labeled CBLB-N using the sortase-mediated protein liga- 14.0 15 M261 tion method (20, 21). Here, uniformly N-labeled helix-RING 1.2 1.0 0.8 0.6 0.4 0.2 (H-RING) was enzymatically attached to the nonlabeled TKB H (ppm) domain. The detailed protocol for the preparation of segmental Fig. 2. (A) Overlay of the 1H-15N HSQC spectra of the segmental isotope-la- isotope-labeled CBLB-N is shown in Fig. S1 A–C and SI Materials beled CBLB-N (red) at the H-RING and the isolated RING domain (blue). Signal and Methods. We confirmed that the E3 activity of segmental assignment of the isolated RING is shown in the spectrum. (B) Lost residues 1 15 isotope-labeled CBLB-N and pY CBLB-N was almost identical (red) in the H- N HSQC spectrum of the segmental isotope-labeled CBLB-N to the authentic proteins (Fig. S1 D and E) though there is an were mapped on the structure of the c-Cbl-N RING domain (1FBV). Red: re- presents lost or shifted residues, Black: removed from the analysis due to insertion sequence of LPETGG prior to H-RING. Fig. 1C shows 1 15 1 15 being missed from the peaks in the H- N HSQC spectrum of the isolated a comparison of the H- N heteronuclear single quantum coher- RING domain or the N-terminal residue in the isolated RING domain, and Yel- ence (HSQC) spectra of the segmental isotope-labeled CBLB-N low: represents nonshifted but observed peaks. The dotted circle represents in the unphosphorylated and phosphorylated states. In the un- the E2 binding region. (C) Overlay of the 1H-13C HMQC spectra of the Ile δ1 phosphorylated state, the H-RING moiety in CBLB-N exhibited methyl selectively labeled CBLB-N (red) and the isolated RING domain (black). broad NMR signals (red). Because CBLB-N was confirmed to Shifted peaks are shown by arrows. Signal assignment of the isolated RING is be monomeric by SAXS at 300 uM (Fig. S1 F and G), signal shown in the spectrum. (D) Shifted (red) and nonshifted (yellow) peaks in C were mapped on the structure of the c-Cbl-N RING domain (1FBV). The broadening was caused by restricted mobility of the H-RING 1 13 dotted circle represents the E2 binding region. (E) Overlay of the H- C moiety through its possible involvement into the core structure of HMQC spectra of the Met methyl signal selectively labeled CBLB-N (red) CBLB-N rather than aggregation. In contrast, in pY CBLB-N, and TKB-H (blue). (F) Methionine residues were mapped on the structure the H-RING moiety exhibited sharp, well-dispersed signals of the TKB-H region of c-CBL-N (1FBV). The helix region is colored pink. (blue), indicating that the H-RING moiety is mobile and inde- pendent from the TKB core. These NMR observations are con- The interface of the RING moiety and TKB-H was also 1 13 sistent with the results from the SAXS and protease susceptibility studied by comparing the H- C heteronuclear multiple quan- analyses, indicating that pY CBLB-N contains a mobile H-RING tum coherence (HMQC) spectra of Ile δ1-methyl-labeled moiety, whereas the H-RING moiety in CBLB-N is incorporated CBLB-N (red) and the isolated RING domain (black) (Fig. 2C). into the core structure. Thus, pY H-RING is considered to be a Spectral overlay revealed that there was an appreciable shift in structural and functional unit possessing ligation activity. the δ1-methyl signals from I375 and I421 between these con- structs, which is consistent with the finding that the I375 main Closed Structure of CBLB-N in the Unphosophorylated State. The chain amide (HN) signal also disappeared in the segmental iso- 1H-15N HSQC spectrum of the segmental isotope-labeled tope-labeled CBLB-N, indicating that I375 and I421 are involved CBLB-N at the H-RING moiety was superimposed onto that of in the interface with TKB-H. The shifted (red) and nonshifted the isolated RING domain (Fig. 2A). Spectral overlay revealed (yellow) Ile residues were mapped on the RING domain in the that the RING moiety in CBLB-N exhibited broad signals (red). crystal structure of c-Cbl complexed with UbcH7 (18) (Fig. 2D). A comparison with the isolated RING domain (blue) showed that It should be noted that Ile375 is located at the binding interface some peaks were absent, possibly due to line broadening by the with E2 (enclosed by a dotted circle). intermediate exchange process. The residues absent in CBLB-N In order to determine the binding interface of TKB-H and were mapped on the RING domain in the crystal structure of the RING moiety, methionine 13C methyl-labeled samples under c-Cbl complexed with UbcH7 (18) (Fig. 2B). Some of the absent a deuterium background were prepared for CBLB-N and TKB- residues were located at the binding interface with E2 (enclosed H, and their 1H-13C HMQC spectra were overlaid (Fig 2E). by a dotted circle), supporting the notion that the RING moiety Resonance assignments of Met methyl signals were obtained by surface required for binding to E2 partially overlaps with that acquiring the spectra of a series of mutants in which the methio- required for the interdomain interaction with the TKB domain nine was replaced by Lys for surface exposed residues and by Ile and helix (hereafter TKB-H). or Leu for buried residues (Fig. S2). Differences in the positions

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110712108 Kobashigawa et al. Downloaded by guest on September 27, 2021 of the methyl peaks of M214 and M365 were observed between CBLB-N and TKB-H, suggesting that M214 and M365 are in- volved in the binding with the RING moiety. However, as M365 is located close to the C terminus of TKB-H, the M365 peak shift could be due to a truncation effect. M365 exhibited a sharp peak, whereas that of M214 was appreciably broadened by an inter- mediate chemical exchange process, presumably between the open and closed states. Therefore, M214 is considered to be located at the interface with the RING moiety. The Met residues (M214 is colored red and the others yellow) were mapped on the TKB-H region in the crystal structure of c-Cbl complexed with UbcH7 (18) (Fig. 2F). Although M261 is located in the vicinity of M365, it did not exhibit any peak shift, supporting the idea that the peak shift of M365 is due to a truncation effect. We have also performed a crystallographic analysis of CBLB- N. Unfortunately, due to the perfectly twined nature of the CBLB-N crystal, as well as the dynamic feature of the RING moiety, we could not construct a model of the RING moiety, although a high electron density was observed around M214. This supports the NMR findings that there exists an interdomain interaction between TKB-H and RING in CBLB-N. The interfaces between RING and TKB-H and UbcH7 par- tially overlap with each other (Fig. 2 B and D) so that the associa- tions of RING with TKB-H and E2 are mutually exclusive. The conformational change induced by the phosphorylation of Y363 may be required to disrupt the interaction between RING and 1 15 TKB-H, thereby facilitating the exposure of the RING domain Fig. 3. (A) Overlay of the H- N HSQC spectra of the H-RING moiety in pY CBLB-N (blue) and pY H-RING (red). (B) Ribbon (Left) and surface potential and subsequent association with E2. BIOPHYSICS AND (Right) representations of the solution structure of pY H-RING. pY363 and 1 15 its interacting positively charged residues, K374 and K381, are shown as a COMPUTATIONAL BIOLOGY Solution Structure of the pY H-RING. The H- N HSQC signals of A stick model. (C) The pY H-RING residues shifted upon the addition of 1.5 pY H-RING (Fig. 3 in red) almost entirely overlapped with equivalent molar ratio of UbcH5B were mapped on the structure of pY those of the H-RING moiety in pY CBLB-N (Fig. 3A in blue). H-RING. Red represents chemical-shift changes larger than 0.35 ppm, orange This supports the notion that the H-RING moiety in pY CBLB-N of 0.25, and yellow of 0.15 ppm. Black represents Pro or missing residues. is exposed and independent from the TKB core and that its struc- (D) The UbcH5B residues shifted upon addition of 1.5 equivalent molar of pY ture is similar to that of pY H-RING. We subsequently deter- H-RING were mapped on the structure of UbcH5B. Green represents chemi- mined the solution structure of pY H-RING by NMR (Fig. 3B cal-shift changes larger than 0.05 ppm and magenta represents missing re- and Fig. S3A). Fig. 3B shows the structure of pY H-RING. The sidues. Black represents Pro residues. (E) Docking model between pY H-RING and UbcH5B created by the structural overlay of the crystal structure c-CBL-N RING moiety is composed of two large Zn2þ-binding loops, a β α in complex with the E2 protein UbcH7 (1FBV). (F) UbcH5B is shown as a sur- short three-stranded antiparallel -sheet, and a central -helix, as face charge model, and H-RING (model based on c-Cbl from 1FBV) and pY shown in pink in Fig. 3B. The overall structure of the RING moi- H-RING as a ribbon model. H-RING is colored blue and pY H-RING pink. Side ety is very similar to those of other RING domains (18, 22–30). chains of pY363, K374, and K381 are shown as a stick model. However, the phosphorylation of Y363 induces an additional structure that includes the formation of a parallel β-sheet be- helix linker relative to the RING domain was markedly changed tween the N terminus of the helix linker and the C terminus of by the phosphorylation of Y363, thereby releasing the helix linker the RING domain so that the RING domain has extensive con- and RING from the TKB domain and exposing the H-RING tact with the helix linker. Steady-state NOE analysis supported moiety. that RING as well as helix linker and the linker region of N and C terminus of pY H-RING formed rigid structures in solution NMR Examination of the Interaction Between the E2 Protein UbcH5 (Fig. 3B, Fig. S3A, and Table S1). Moreover, we observed long and pY H-RING. To elucidate the biological implications of the range NOEs between K374 Hϵ and the aromatic ring proton structural changes induced by the phosphorylation of Y363 in of pY363, which was further supported by NOEs between sur- H-RING, the interaction of pY H-RING with UbcH5B was rounding residues. The interaction between the RING domain studied by chemical-shift perturbation methods using solution and the helix linker is stabilized by the electrostatic interaction NMR (Fig. S3 G and H). First, unlabeled UbcH5B was titrated 15 between the phosphate group of pY363 on the helix linker and to N-labeled pY H-RING. Upon the addition of aliquots of 15 the positively charged cluster formed by K374 and K381 in the UbcH5B, some of the peaks in N-labeled pY H-RING were RING domain (Fig. 3B). These interactions make pY H-RING gradually shifted by a fast exchange process. From the chemical- a single structural unit presenting the RING domain to E2. This shift change of S403, we estimated that the Kd value for the bind- notion was supported by the fact that Lys Hζ proton signals of ing of pY H-RING to UbcH5B is 1.2 μM (SD of 0.4 μM). both K374 and K381 were observed in the 1H-15N HSQC spectra Next, we mapped the residues with large chemical-shift perturba- (Fig. S3C). The Hζ proton signal from Lys can only be observed in tions on the structure of pY H-RING. Most of the residues cases where chemical exchange with solvent proton is highly re- with large chemical-shift perturbations in the NMR spectra upon stricted due to hydrogen bond formation. We also confirmed the complex formation are located in the two regions: residues 374– interaction of K374 and K381 with the phosphate group of pY363 379 and 408–411 in the loop regions, and residues 397–408 in the by the mutational analysis of H-RING in which K374 and K381 helix of the RING domain (Fig. 3C and Fig. S3G). On the other were replaced by Glu. These mutants did not show significant hand, the helix linker was not perturbed, indicating that this spectral changes upon Y363 phosphorylation (Fig. S3 E and F) region is not directly involved in the interaction with UbcH5B. and gave similar spectra to that of H-RING in contrast to the case The interaction region was similar to the previous report for of the wild type (Fig. S3D). In summary, the arrangement of the c-Cbl H-RING in the unphosphorylated state (24).

Kobashigawa et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on September 27, 2021 15 D226R Next, unlabeled pY H-RING was titrated to N-labeled A R412E C 20 D H K195E/R198E D112R UbcH5B (Fig. 3 and Fig. S3 ). Upon addition of aliquots of 15

pY H-RING, some of the UbcH5B peaks disappeared in inter- 10 E378R mediate exchange process or gradually shifted in fast exchange 5 Relative activity process. Slight precipitation of UbcH5B was observed upon ad- K381E 0

dition of pY H-RING so that the Kd value could not be estimated K129E/R131E WT K381E E378R R412E from this experiment. The largely affected residues in UbcH5B B (1) WT (2) D112R (3) K129E/R131E (4) K195E/R198E D112R D226R

were located in the first helix, the loop between strands β3 and pY CBLB-N K129E/R131E β4, and the loop region connecting the second and third helices K195E/R198E CBLB ~ (Ub)n (Fig. 3D and Fig. S3H). Interaction regions were similar to the

previous report for c-Cbl H-RING in the unphosphorylated state }WB: anti-polyHis

(24). To our knowledge, there are five RING domains for which 0 15 30 60 120 0 15 30 60 120 0 15 30 60 120 0 15 30 60 120 time (min)

the complex structures with E2 have been reported (18, 22, 23, 25, (5) D226R (6) E378R (7) K381E (8) R412E (9) pY CBLB-N 26); CBL RING/UbcH7 [ (PDB) ID: 1FBV], cIAP2 RING/UbcH5B (PDB ID: 3EB6), TRAF6 RING/Ubc13 (PDB ID: 3HCT), Ring1b/UbcH5C (PDB ID: 3RPG), and IDOL RING/UbcH5A (PDB ID: 2YHO). The interaction regions be-

tween the RING domain and E2 in these protein complexes were 0 15 30 60 120 0 15 30 60 120 0 15 30 60 120 0 15 30 60 120 0 15 30 60 120 found to be essentially the same (Fig. S3I), which was also sup- ported by NMR and/or mutational analysis for other RING do- Fig. 4. Autoubiquitination assay of CBLB-N variants. (A) The mutation sites mains of (27), CNOT4 (28, 29), and BRCA1 (30). Thus, in the TKB-H region were mapped and colored yellow. The helix region is we constructed the complex model between pY H-RING and colored pink (Left). Mutation sites in the RING moiety were mapped and UbcH5B by overlaying the RING domain of pY H-RING and colored yellow (Right). (B) Time course of the autoubiquitination of CBLB- N, pY CBLB-N, and CBLB-N mutants. Ubiquitination was monitored by the UbcH5B to the RING region and UbcH7 of the c-Cbl/UbcH7 E appearance of a ladder pattern in the Western blotting. For all autoubiqui- complex structure (1FBV) (18), respectively (Fig. 3 ). The com- tination experiments, N terminus hexahistidine-tag-attached ubiquitin was plex model is consistent with and supported by the results of used and detected using anti-histidine-tag antibody. (C) Autoubiquitination the NMR titration analysis, despite the fact NMR data were not activity of pY CBLB-N and CBLB-N mutants relative to that of the wild type. considered at all for the model construction. Samples from the autoubiquitination assay at a time course of 60 min were applied for Western blotting in single SDS-PAGE gel, and the band density Affinity of CBLB-N and CBLB-N Variants Toward UbcH5B. Fluorescence was quantified using ImageJ software. Reaction reached a plateau within polarization spectroscopy (hereafter FP) was next used to eval- 15 min for pY CBLB-B, thus underestimate the relative activity obtained at 60 min. Error bar indicates standard deviation. uate the affinity of CBLB-N and CBLB-N variants for UbcH5B (Fig. S4). Titration measurements of the affinity of CBLB-N We, therefore, designed several TKB domain mutants in which toward Alexa 488-labeled UbcH5B revealed the estimated Kd the interaction between TKB-H and RING was disrupted. Prior μ 3 3 μ value of 97.5 M (SD of . M). CBLB-N has a much lower to mutant design, the CBLB-N structure was modeled by rigid affinity to UbcH5B than isolated H-RING (22.7 μM: SD of 1 0 μ body docking between TKB-H and RING using the HADDOCK . M), indicating the masking of the RING moiety in program (31), incorporating the NMR chemical-shift perturba- CBLB-N. The affinity of pY CBLB-N toward UbcH5B was mea- μ tion data (Fig. 2). A detailed description of the docking calcula- sured, and the Kd value was estimated to be 1.1 M (SD of 0 1 μ tions between TKB-H and RING is given in Table S2 and the . M), which is similar to that of isolated pY H-RING docking structure is shown in Fig. S5 A and B and statistics (1.2 μM: SD of 0.4 μM) obtained by NMR, indicating that the are shown in Table S2. In the crystallographic analysis, we could masking was released in pY CBLB-N. This is consistent with the observe a low but extensive electron density around the RING results of the SAXS and NMR measurements. It should be noted region in the HADDOCK model (Fig. S5C and Table S3), which that a comparison of the K values between H-RING and pY d supports the NMR- and HADDOCK-derived model. The RING H-RING revealed that the phosphorylation of Y363 increases moiety fitted into a shallow groove on TKB-H, and the interac- the affinity toward UbcH5B by about 20-fold. It can be concluded tion between TKB-H and the RING moiety seemed to be mainly that the phosphorylation of Y363 in CBLB-N facilitates the B association of the E2 protein, not only through unmasking but electrostatic in nature (Fig. S5 ). We, therefore, focused on the also through formation of a proper E2 binding surface. Indeed, exposed charged residues on the TKB-H surface. Four mutants in the TKB region, D112R, K129E/R131E, K195E/R198E, and the phosphate group of pY363 in pY H-RING interacts with A K374 and K381 to fix the structure of the H-RING moiety and D226R (Fig. 4 ), were prepared and applied to the autoubiqui- to reduce its basic surface potential (Fig. 3 B and F). This surface tination assay. These mutants exhibited two- to sevenfold higher is partially involved in the binding to the conserved, positively E3 activity than the wild type. This supported the idea that the charged surface of the α1 helix in UbcH5B so that neutralization E3 activity of CBLB-N is autoinhibited by the interdomain inter- of the positively charged surface by the phosphate group of Y363 action with the TKB region. increases the binding affinity between RING and UbcH5B We next designed RING domain mutants that would disrupt (Fig. 3F). the masked structure. From the docking study and the data for Next, the E3 activity of the Cbl-b variants was studied by mea- the TKB mutants, the masking was expected to be maintained suring their autoubiquitination activity. Consistent with previous mainly by electrostatic interaction. Hence, we focused on the data (15, 16), the ubiquitination activity of CBLB-N was mark- charged residues, the signals of which disappeared in the segmen- edly enhanced by Y363 phosphorylation (Fig. 4 B and C), which tal-labeled CBLB-N (Fig. 2B). We then performed autoubiquiti- also supports our structural data for phosphorylation-induced nation assay of three RING mutants, E378R, K381E, and R412E. CBLB-N unmasking and activation. Among these mutants, K381E exhibited markedly higher activity, whereas E378R and R412E exhibited lower activity compared The Closed Structure of CBLB-N Based on a Docking Study, the Design to CBLB-N. As E378 and R412 are located on the interface of Mutants, and Ubiquitination Assay. According to the NMR and between RING and E2 (18), the lower activity of these mutants FP data, the RING domain appears to be masked by TKB-H, is thought to be disruption of association with UbcH5B. Intrigu- thereby reducing the binding affinity of the RING moiety to E2. ingly, R420 in c-CBL (R412 in Cbl-b) is a mutational hot spot in

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110712108 Kobashigawa et al. Downloaded by guest on September 27, 2021 human cancer malignancies (3–7). On the other hand, CBLB-N CBLB-N at the H-RING moiety with the isolated pY H-RING (K381E) exhibited the highest activity among the mutants we domain clearly supports the idea that the H-RING moiety in studied. From these observations, K381 may be located in the pY CBLB-N is exposed and mobile, in a manner similar to the interface region between RING and TKB-H domains. Thus, the isolated pY H-RING. Thus, the helix region and the RING do- results of E3 activity, as well as the FP and SAXS analyses, main work together as a structural and functional unit to recruit support our view that the dynamic structural change induced by E2 proteins. The structure of pY CBLB-N was subsequently mod- Y363 phosphorylation enhances the E3 activity of Cbl-b. eled based on the structure of TKB and pY H-RING. Because of the β-sheet formation between the N and C termini of the Discussion H-RING moiety, the location of the H-RING moiety relative to The present structural analyses of CBLB-N revealed that the the TKB domain may be significantly different from that ob- interdomain interaction between RING and TKB-H leads to the served in the crystal structure of c-Cbl complexed with UbcH7, formation of a compact structure in the unphosphorylated state and close to the Zap70 phosphotyrosine-containing peptide bind- in which the binding site between the RING domain and E2 is ing site, particularly as the linker between TKB and H-RING masked by TKB-H. In contrast, pY CBLB-N has an extended is comprised of four residues (Fig. 5A and Fig. S6). structure in which the phosphorylated H-RING moiety is freely From the present structural and functional analyses, the fol- accessible to the E2 proteins. Moreover, the H-RING moiety lowing model for Cbl-b signaling can be proposed (Fig. 5B). markedly changes its structure upon phosphorylation, thereby Upon activation of the cell surface receptors, including the T-cell increasing its affinity to the E2 protein. This was confirmed by receptor, EGFR, PDGFR, c-KIT, and so on, the tyrosine phos- biophysical and biochemical analyses together with mutational phorylation of the receptors by protein tyrosine kinases creates analysis. However, this process is not merely a closed-to-open a binding site for the Cbl-b TKB domain, thus recruiting Cbl-b. conformation change induced by Y363 phosphorylation. Because of the intramolecular interaction of the RING domain First, we considered the structure of CBLB-N in the closed with TKB-H, Cbl-b is in a closed state that masks the binding site state. We modeled a closed CBLB-N structure by docking the for the E2 protein. However, when Cbl-b is phosphorylated, the RING domain with the crystal structure of TKB-H (modeled H-RING becomes exposed, thereby enhancing the affinity for from c-Cbl: 1FBV) on the basis of chemical-shift perturbation the E2 protein. Thus, activated Cbl-b catalyzes ubiquitination of studies (Fig. 2) using the HADDOCK program. Most of the RING residues on the TKB-H binding interface disappeared, the receptor- and nonreceptor-type tyrosine kinases to down reg- ulate signaling. In conclusion, our structural and biochemical

indicating that these residues are in an intermediate exchange BIOPHYSICS AND studies have revealed a regulatory mechanism of the E3 activity

process and showing that CBLB-N in the unphosphorylated state COMPUTATIONAL BIOLOGY is in an equilibrium between a closed and a partially open state of Cbl-b, which is closely related to cancer as well as autoimmune (Fig. 5A). This view is consistent with the basal E3 activity of diseases in humans. Cbl-b in the unphosphorylated state as well as the crystal struc- Materials and Methods ture of CBLB-N, where the electron density of the RING domain See detail also for SI Materials and Methods. is diffusive. This suggests that the RING domain is not fixed but is in a dynamic equilibrium between several forms. The partially Protein Expression and Purification. All the proteins were expressed as hexa- open state may be considered to be an ensemble of multiple histidine tag fusion proteins using Escherichia coli strain Rossetta (DE3) at states, one of which may be similar to the crystal structure of 25 °C, purified using Ni2þ-affinity column chromatography, followed by c-Cbl in which the RING moiety is released from the TKB region HRV3C protease digestion to remove the tag, and further purified by gel while the helix region is fixed on the TKB domain (18). In some of filtration chromatography using Superdex 75 (GE Healthcare). the partially open states, both the helix and the RING regions may be released from the TKB domain so that the Y363 of Fluorescence Labeling of the Protein and Fluorescence Polarization Measure- Cbl-b or the Y371 of c-Cbl becomes susceptible to phosphoryla- ment. Alexa 488-conjugated UbcH5B was prepared as described previously (32). Fluorescence polarization was measured at 22 °C with a buffer contain- tion. Once phosphorylated, the helix region is detached from the ing 20 mM MES (pH 7.0) and 150 mM NaCl with 1 μM of Alexa 488-labeled TKB domain and interacts with the RING domain through ionic UbcH5B using an RF-5300PC (Shimadu) fluorescence spectrometer. Aliquots interactions between the phosphate group of pY363 and the of the solution containing wild-type or mutant CBLB-N, pY CBLB-N, pY positively charged cluster in the RING domain. A comparison of H-RING, or H-RING were added to this sample and the fluorescence anisotro- 1 15 the H- N HSQC spectra of the segmental isotope-labeled pY py was measured. A

Fig. 5. (A) Schematic representation of the regulation of E3 activity of Cbl-b by phosphorylation. CBLB-N is in equi- librium between the closed inactive and open partially ac- tive states. In the closed state, the E2 binding region is masked by the TKB region, but in the open partially active state, the E2 binding region is open. Upon phosphoryla- tion of Y363, the H-RING region shows a marked change in its structure to completely expose the E2 binding sur- B face and moves to an open fully active state. (B) Schematic representation of the regulation of E3 activity of Cbl-b by phosphorylation. Cbl-b is in a closed state and the E2 bind- ing region of RING is masked by the TKB region. Upon stimulation, the receptor is phosphorylated and subse- quently recruits Cbl-b via association with the TKB do- main. The Y363 of Cbl-b is phosphorylated by kinases that are also recruited to the receptor site, and Cbl-b be- comes open. Phosphorylation-induced unmasking and conformation change enhances the affinity of H-RING to the E2 protein, facilitating the ubiquitination of the re- ceptor to down regulate the signals.

Kobashigawa et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on September 27, 2021 13 Protein Ligation and Purification of the Product. Protein ligation was carried Met-labeled CBLB-N, TKB-H, and RING; and Ile δ1-methyl CH3-labeled out at room temperature as described previously (20). Ligation product was TKB-H, RING, and CBLB-N were dissolved in 20 mM MES (pH 6.3), 1 mM 2 digested by HRV3C protease, followed by pass through the Ni-nitrilotriace- CaCl2, 2 mM DTT, 150 mM NaCl in 90% H2O∕10% H2O or 20 mM deuterated tate, and further purified by gel filtration using Superdex 75 (GE Healthcare) MES (pH meter direct read of 6.3), 1 mM CaCl2, 2 mM deuterated DTT, and 13 (Fig. S1). 150 mM NaCl in 100% D2O. The CH3 resonances of the Met residues were assigned using a series of mutants of CBLB-N and TKB-H (Fig. S2) and Phosphorylation of CBLB-N and H-RING. The phosphorylation reaction was the isolated RING domain. carried out using fusion protein between c-Src kinase domain and Zap70 Titration measurements were carried out at 25 °C. Small aliquots of (SDGYTPEP) fragment to enhance phosphorylation efficiency (33). The phos- nonlabeled protein (pY H-RING or UbcH5B) were added to the 15N-labeled

phorylation reaction was monitored by SDS-PAGE and immunoblotting using protein solution. The dissociation constant (Kd) of pY H-RING for UbcH5B PY20 (Zymed), a mouse monoclonal antibody against phosphotyrosine. was estimated from the amide proton chemical-shift changes. Specificity of the phosphorylation at Y363 was confirmed using the Y363F mutant protein (Fig. S7). Small-Angle X-ray Scattering Measurements. All samples were dissolved in 20 mM Tris·HCl buffer (pH 8.0) and 150 mM NaCl. A protein concentration In Vitro Ubiquitination Assay. Ubiquitination reaction was carried out at of 6 mg∕mL was used for all SAXS measurements. The SAXS data were 30 °C in 25 μL of reaction solution containing 20 mM Hepes-KOH (pH 7.5), collected at 25 °C using the Nano-viewer (RIGAKU) at the Open Facility, 50 mM KCl, 5 mM MgCl2, 1 mM DTT, 1 mM ATP, 10 mM creatine phosphate, Hokkaido University Sousei Hall. Solvent scattering was corrected for the use 0.25 μg E1 (Sigma), 0.5 μg UbcH5B, 5.0 μg CBLB-N variant, 0.5 μg N terminus of buffer solutions identical to that used for the sample. Scattering data were hexahistidine tag attached ubiquitin, and 10 μg creatine kinase. Reaction was analyzed using the Guinier approximation (19). Ið0Þ, intensity at zero scatter- terminated at 0, 15, 30, 60, and 120 min for Western blotting analysis. Ubi- ing angle, and Rg, the radius of gyration, were calculated using the AutoRg quitination was monitored by immunoblotting using peroxidase conjugated software (34). antipolyhistidine antibody (Sigma). Samples from autoubiquitination time course at 60 min were applied for Western blotting in the single SDS-PAGE ACKNOWLEDGMENTS. This work was supported by the Targeted Proteins gel and the band density was quantified by using ImageJ software. Research Program, the matching Program for Innovations in Future Drug Discovery and Medical Care, the Funding Program for World-Leading Inno- NMR Spectroscopy. All NMR experiments were carried out at 25 °C on a Varian vative R&D on Science and Technology, and a Grant-in-Aid for Scientific 15 Inova 500, 600, or 800 MHz NMR spectrometer. The segmental N-labeled Research on Innovative Areas from the Ministry of Education, Science, and 15 13 15 13 CBLB-N and pY CBLB-N; Nor C∕ N-labeled pY H-RING and H-RING; CH3- Culture, Japan.

1. Ryan PE, Davies GC, Nau MM, Lipkowitz S (2006) Regulating the regulator: Negative 19. Guinier A, Fournet G (1955) Small-Angle Scattering of X-ray. (Wiley, New York). regulation of Cbl ubiquitin ligases. Trends Biochem Sci 31:79–88. 20. Kobashigawa Y, Kumeta H, Ogura K, Inagaki F (2009) Attachment of an NMR-invisible 2. Swaminathan G, Tsygankov AY (2006) The Cbl family proteins: Ring leaders in regula- solubility enhancement tag using a sortase-mediated protein ligation method. tion of . J Cell Physiol 209:21–43. J Biomol NMR 43:145–150. 3. Niemeyer CM, et al. (2010) Germline CBL mutations cause developmental abnormal- 21. Mao H, Hart SA, Schink A, Pollok BA (2004) Sortase-mediated protein ligation: a new – ities and predispose to juvenile myelomonocytic leukemia. Nat Genet 42:794 800. method for protein engineering. J Am Chem Soc 126:2670–2671. 4. Fernandes MS, et al. (2010) Novel oncogenic mutations of CBL in human acute myeloid 22. Yin Q, et al. (2009) E2 interaction and dimerization in the crystal structure of TRAF6. leukemia that activate growth and survival pathways depend on increased metabo- Nat Struct Mol Biol 16:658–666. lism. J Biol Chem 285:32596–32605. 23. Mace PD, et al. (2008) Structures of the cIAP2 RING domain reveal conformational 5. Adélaïde J, et al. (2010) Gain of CBL-interacting protein, a possible alternative to CBL changes associated with ubiquitin-conjugating enzyme (E2) recruitment. J Biol Chem mutations in myeloid malignancies. Leukemia 24:1539–1541. 283:31633–31640. 6. Sanada M, et al. (2009) Gain-of-function of mutated C-CBL tumour suppressor in 24. Huang A, et al. (2009) E2-c-Cbl recognition is necessary but not sufficient for ubiqui- myeloid neoplasms. Nature 460:904–908. – 7. Kales SC, Ryan PE, Nau MM, Lipkowitz S (2010) Cbl and human myeloid neoplasms: tination activity. J Mol Biol 385:507 519. The Cbl comes of age. Cancer Res 70:4789–4794. 25. Bentley ML, et al. (2011) Recognition of UbcH5c and the nucleosome by the Bmi1/ 8. Mueller DL (2004) E3 ubiquitin ligases as T cell anergy factors. Nat Immunol 5:883–890. Ring1b ubiquitin ligase complex. EMBO J 30:3285–3297. 9. Chiang YJ, et al. (2000) Cblb regulates the CD28 dependence of T-cell activation. 26. Zhang L, et al. (2011) The IDOL-UBE2D complex mediates sterol-dependent degrada- Nature 403:216–220. tion of the LDL receptor. Dev 25:1262–1274. 10. Bachmaier K, et al. (2000) Negative regulation of lymphocyte activation and autoim- 27. Linke K, et al. (2008) Structure of the MDM2/MDMX RING domain heterodimer reveals munity by the molecular adaptor Cbl-b. Nature 403:211–216. dimerization is required for their ubiquitylation in trans. Cell Death Differ 15:841–848. 11. Yokoi N, et al. (2008) Identification and functional analysis of CBLB mutations in type 1 28. Albert TK, et al. (2002) Identification of a ubiquitin-protein ligase subunit within the diabetes. Biochem Biophys Res Commun 368:37–42. CCR4-NOT repressor complex. EMBO J 21:355–564. 12. Sanna S, et al. (2010) Variants within the immunoregulatory CBLB are associated 29. Dominguez C, et al. (2004) Structural model of the UbcH5B/CNOT4 complex revealed – with multiple sclerosis. Nat Genet 42:495 407. by combining NMR, mutagenesis, and docking approaches. Structure 12:633–644. 13. Meng W, Sawadikosol S, Burakoff SJ, Eck MJ (1999) Structure of the amino-terminal 30. Brzovic PS, et al. (2003) Binding and recognition in the assembly of an active BRCA1/ – domain of Cbl complexed to its binding site on ZAP-70 kinase. Nature 398:84 90. BARD1 ubiquitin-ligase complex. Proc Natl Acad Sci USA 100:5646–5651. 14. Keane MM, et al. (1995) Cloning and characterization of -b: A SH3 binding protein 31. Dominguez C, Boelens R, Bonvin AMJJ (2003) HADDOCK: A protein–protein docking with homology to the c-cbl proto-oncogene. Oncogene 10:2367–2377. approach based on biochemical or biophysical information. J Am Chem Soc 15. Kassenbrock CK, Anderson SM (2004) Regulation of ubiquitin protein ligase activity in 125:1731–1737. c-Cbl by phosphorylation-induced conformational change and constitutive activation 32. Kobashigawa Y, et al. (2011) Phosphoinositide-incorporated lipid-protein nanodiscs: A by tyrosine to glutamate point mutations. J Biol Chem 279:28017–28027. – 16. Thein CBF, Walker F, Langdon WY (2001) RING finger mutations that abolish c-Cbl- tool for studying protein-lipid interactions. Anal Biochem 410:77 83. directed polyubiquitination and downregulation of the EGF receptor are insufficient 33. Kobashigawa Y, Naito M, Inagaki F (2007) An efficient method for protein phosphor- for cell transformation. Mol Cell 7:355–366. ylation using the artificially introduced of cognate-binding modules into kinases and 17. Levkowitz G, et al. (1999) Ubiquitin ligase activity and tyrosine phosphorylation substrates. J Biotechnol 131:458–465. underlie suppression of growth factor signaling by c-Cbl/Sli-1. Mol Cell 4:1029–1040. 34. Petoukhov MV, Konarev PV, Kikhney AG, Svergun DI (2007) ATSAS 2.1—towards 18. Zheng N, Wang P, Jeffrey PD, Pavletich NP (2000) Structure of a c-Cbl-UbcH7 complex: automated and web-supported small-angle scattering data analysis. J Appl Crystallogr RING domain function in ubiquitin-protein ligases. Cell 102:533–539. 40:s223–s228.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110712108 Kobashigawa et al. Downloaded by guest on September 27, 2021