Biochemical Journal

Research article

Structural basis and specificity of human otubain 1-mediated deubiquitination

Mariola J. Edelmann, Alexander Iphöfer, Masato Akutsu, Mikael Altun, Katalin di Gleria, Holger B. Kramer, Edda Fiebiger, Sirano Dhe-Paganon, Benedikt M. Kessler

Abstract

OTUB (otubain) 1 is a human deubiquitinating enzyme that is implicated in mediating lymphocyte antigen responsiveness, but whose molecular function is generally not well defined. A structural analysis of OTUB1 shows differences in accessibility to the active site and in surface properties of the substrate-binding regions when compared with its close homologue, OTUB2, suggesting variations in regulatory mechanisms and substrate specificity. Biochemical analysis reveals that OTUB1 has a preference for cleaving Lys48-linked polyubiquitin chains over Lys63-linked polyubiquitin chains, and it is capable of cleaving NEDD8 (neural-precursor-cell-expressed developmentally down-regulated 8), but not SUMO (small ubiquitin-related modifier) 1/2/3 and ISG15 (interferon-stimulated gene 15) conjugates. A functional comparison of OTUB1 and OTUB2 indicated a differential reactivity towards ubiquitin-based active-site probes carrying a vinyl methyl ester, a 2-chloroethyl or a 2-bromoethyl group at the C-terminus. Mutational analysis suggested that a narrow P1′ site, as observed in OTUB1, correlates with its ability to preferentially cleave Lys48-linked ubiquitin chains. Analysis of cellular interaction partners of OTUB1 by co-immunoprecipitation and MS/MS (tandem mass spectrometry) experiments demonstrated that FUS [fusion involved in t(12;6) in malignant liposarcoma; also known as TLS (translocation in liposarcoma) or CHOP (CCAAT/enhancer-binding protein homologous protein)] and RACK1 [receptor for activated kinase 1; also known as GNB2L1 (guanine-nucleotide-binding protein β polypeptide 2-like 1)] are part of OTUB1-containing complexes, pointing towards a molecular function of this deubiquitinating enzyme in RNA processing and cell adhesion/morphology.

  • deubiquitinating enzyme (DUB)
  • interferon-stimulated gene 15 (ISG15)
  • neural-precursor-cell-expressed developmentally down-regulated 8 (NEDD8)
  • otubain 1 (OTUB1)
  • small ubiquitin-related modifier (SUMO)
  • ubiquitin

INTRODUCTION

Ubiquitin is central for many biological processes, including protein turnover, protein targeting, signal transduction and regulation of transcription, and has been implicated in tumorigenesis, neurodegeneration and microbial pathogenesis. The relevance of ubiquitin in these biological processes is reflected by the fact that several hundred genes have so far been linked to ubiquitin conjugation and deconjugation. Primary examples are the E3 ligases and DUBs (deubiquitinating enzymes), the former consisting of approx. 600 genes that are classified into the RING (really interesting new gene) and HECT (homologous with E6-associated protein C-terminus) domain families [1], and the latter are classified into six main families, such as UBPs (ubiquitin-processing proteases), UCHs (ubiquitin C-terminal hydrolases), ataxin-3/Josephin domains, OTUs (ovarian-tumour-domain-containing proteases), pathogen-encoded ubiquitin-processing proteases and JAMM (JAB1/MPN/MOV34 metalloenzyme) proteases [2]. The OTU family has recently attracted attention owing to the presence of conserved sequences found in viruses, bacteria, plants, yeast and humans, and its role in immunity and viral infection [3]. It consists of an approx. 130-amino-acid papain fold with a catalytic triad that is typically found in cysteine proteases [4]. OTUB (otubain) 1 and 2 are close homologues expressed in mammals [5]. Although ubiquitously expressed, OTUB1 is implicated in anergy induction in CD4+ T-lymphocytes, affecting the stability of the lymphocyte-specific E3 ligase GRAIL (gene related to anergy in lymphocytes) [6]. Its immune-related functions were partially ascribed to specific splice variants, including ARF-1 (alternative reading frame 1), found only in these cells [7]. In addition, OTUB1 binds the Yersinia pathogen-encoded virulence factor YpkA [8]. Exemplified by A20 [9,10], OTUB1 also probably acts upon a discrete set of substrates in contrast with Cezanne, which was shown to be a more general DUB [11]. Structural insights obtained for some of the OTU-domain-containing proteins, such as A20 [12,13], OTUB2 [14] and yeast OTU1 in complex with a ubiquitin-Br3 derivative (in which Gly76 was replaced with a bromopropylamine group) [15] confirm that the OTU domain encodes a rudimentary papain-like domain with considerable variations that may reflect functional diversity.

In the present paper, we provide the crystal structure of human OTUB1 and show significant differences when compared with OTUB2. In addition to being selective towards ubiquitin and, to some extent, NEDD8 (neural-precursor-cell-expressed developmentally down-regulated 8), but not ISG15 (interferon-stimulated gene 15) and SUMO (small ubiquitin-related modifier) 1/2/3, OTUB1 prefers cleavage of Lys48-linked ubiquitin over Lys63-linked ubiquitin chains, which is probably attributable to a narrow cavity at the P1′ site. Mass spectrometric identification of cellular interaction partners suggests that OTUB1 may be generally involved in RNA processing and cell adhesion/morphology.

MATERIALS AND METHODS

Cell lines, reagents and antibodies

Chemicals were purchased from Sigma–Aldrich, unless indicated otherwise. The cDNA for human OTUB1 was obtained by carrying out PCR with a spleen cDNA library (Invitrogen) as a template with the following primers: 5′-CTACTAGCTAGCATGGCGGCGGAGGAACC-5′ (forward) and 5′-CGGCGGCTCGAGCTATTTGTAGAGGATATCGTAG-3′ (reverse). The primers included a 5′ NheI and a 3′ XhoI site for directional cloning into a pcDNA3.1 vector containing a C-terminal SBP (streptavidin-binding peptide)–TEV (tobacco etch virus)–HA (haemagglutinin) tag. The OTUB1–SBP–TEV–HA (hereafter referred to as outabain 1–HA) C91S mutant was created using the QuikChange® II site-directed mutagenesis kit (Stratagene). Wild-type and cysteine mutant OTUB1–HA were subcloned into the bicistronic pEF-IRES vector [16], kindly provided by Dr Marek Cebecauer (Section of Molecular and Cellular Medicine, National Lung and Heart Institute, Faculty of Medicine, Imperial College London, London, U.K.), using the 5′ NheI and 3′ XbaI restriction sites. Residues 40–271 were subcloned into pET28a-LIC for structural studies. The cDNA for human OTUB2 was obtained by carrying out PCR with a spleen cDNA library as a template with the following primers: 5′-CTACTAGCTAGCATGAGTGAAACATCTTTCAACC-3′ (forward) and 5′-CGGCGGCTCGAGTCAATGTTTATCGGCTGCATAAAGG-3′ (reverse). The primers contained a 5′ NheI and a 3′ XhoI site for directional cloning into a pcDNA3.1 vector containing a C-terminal SBP–TEV–HA tag as well as a pET28a-LIC vector for bacterial expression. The proteasome inhibitor ZL3VS was kindly provided by Professor Hermen Overkleeft (Department of Bio-organic Synthesis, Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, Leiden, The Netherlands). The antibodies used in the present study are described in the Supplementary Online Data at http://www.BiochemJ.org/bj/418/bj4180379add.htm.

Protein purification, crystallization and structure determination

OTUB1-(40–271) was expressed in Escherichia coli BL21 (DE3) cells and grown in Terrific broth in the presence of 50 μg/ml kanamycin at 37 °C to a D600 of 0.5. Cells were induced overnight at 15 °C with 0.05 mM IPTG (isopropyl β-D-thiogalactoside). The expressed protein was purified from cell pellets using a TALON® metal-affinity resin column (BD Biosciences) at 4 °C, cleaved with thrombin (Sigma T9681), and purified by gel filtration on a HighLoad 16/60 Superdex 200 column (GE Healthcare). Purified protein (68 mg/ml) was crystallized at room temperature (25 °C) using the hanging-drop vapour-diffusion method when mixed with an equal volume of the reservoir solution {30% PEG [poly(ethylene glycol)] 8000, 0.2 M sodium acetate and 0.1 M sodium cacodylate (pH 6.5)}. Crystals were cryoprotected in a 50:50 mixture of Paratone-N and mineral oil and frozen in liquid nitrogen. Diffraction data were obtained using an FR-E generator and RAXIS IV++ detector and processed with HKL2000 [17]. A molecular replacement solution using OTUB2 (PDB code 1TFF) was found with the program MolRep [18]. The refinement procedures were carried out with REFMAC5 [19]. Model fitting to electron-density maps was performed manually using Coot [20]. Final refinement statistics are summarized in Table 1. The generation of recombinant wild-type and P87G mutant OTUB1 and G47P mutant OTUB2 protein is described in detail in the Supplementary Online Data.

View this table:
Table 1 Data collection and refinement statistics

Values in parentheses are for the highest resolution shell. PDB code 2ZFY. RMSD, root mean square deviation.

Preparation of ubiquitin and Ubl (ubiquitin-like protein) substrates and isopeptidase assays

The synthesis and preparation of ubiquitin and Ubl substrate based on the ubiquitin/Ubl orthogonal peptide scaffold containing an N-terminal biotin moiety was performed essentially as described in [21]. Further details can be found in the Supplementary Online Data.

Preparation of ubiquitin-specific active-site probes and activity-based profiling assays

The synthesis and preparation of HA-tagged ubiquitin active-site probes containing a C-terminal Br2 (2-bromoethyl), Cl2 (2-chloroethyl) and VME (vinyl methyl ester) group was performed essentially as reported previously [22]. Further details are described in the Supplementary Online Data.

Linear di-ubiquitin, Lys48-/Lys63-linked tetra-ubiquitin, di-SUMO and ubiquitin-AMC (7-amino-4-methylcoumarin) cleavage assays

Linear di-ubiquitin, Lys48-, Lys63-linked tetra-ubiquitin, di-SUMO2, di-SUMO3, ubiquitin-AMC, UCH-L3 and SENP2 [SUMO1/sentrin/SMT3 (suppressor of mif two 3 homologue 1)-specific peptidase 2] were purchased from Boston Biochem. A 20 μl solution containing reaction buffer (50 mM Tris/HCl, pH 7.4, and 2 mM dithiothreitol), linear di-ubiquitin, Lys48-/Lys63-linked tetra-ubiquitin, di-SUMO2 or di-SUMO3 (250 nM final concentration) was incubated with recombinant OTUB1 or OTUB2 proteins (50 nM final concentration) or 5 μg of cell extract (prepared as described below) in a final reaction volume of 20 μl for the indicated times at 37 °C. The reactions were stopped by adding reducing SDS sample buffer, followed by separation on Tris/Tricine SDS/PAGE [23] and visualization by anti-ubiquitin or anti-SUMO immunoblotting. For the analysis by MS, 1.5 μg of Lys48- or Lys63-linked tetra-ubiquitin was incubated with 1 μg of recombinant OTUB1 for the indicated times and subjected to analysis by infusion electrospray MS (Agilent 1100 series LC/MSD-TOF spectrometer). Ubiquitin-AMC cleavage assays were performed essentially as reported in [21] and further details are described in the Supplementary Online Data.

Identification of OTUB1 interactors by MS/MS (tandem MS)

HEK (human embryonic kidney)-293T cells were grown to confluence in DMEM (Dulbecco's modified Eagle's medium) containing 10% (v/v) FBS (fetal bovine serum) and 1% streptomycin/penicillin. Cells were then transferred to 150-mm-diameter, 100-mm-diameter, 50-mm-diameter or six-well tissue culture dishes at a concentration of 0.4×106/ml and grown overnight at 37 °C. The cells were then washed and the transfection was performed using SuperFect reagent (Qiagen), according to the manufacturer's protocol, followed by incubation overnight at 37 °C. Approx. 109 cells per sample were lysed using glass beads (Sigma) in 150 mM NaCl, 5 mM CaCl2, 50 mM Tris/HCl (pH 7.4) and 250 mM sucrose containing protease inhibitor cocktail (Roche Applied Science) as described in [22]. Alternatively, cells were lysed in 0.1% NP-40 (Nonidet P40), 150 mM NaCl, 20 mM CaCl2 and 50 mM Tris/HCl (pH 7.4). For immunoprecipitation, protein lysates (30 mg per sample) were first diluted in NET buffer (50 mM Tris/HCl, 5 mM EDTA, 150 mM NaCl and 0.5% NP-40, pH 7.4) to a protein concentration of 1 mg/ml, and pre-cleared with Protein A–agarose beads for 1 h at 4 °C. Immunoprecipitation was then carried out either for 2 h or overnight at 4 °C. Material was eluted from the beads using the TEV protease (Invitrogen) for 2 h at 4 °C and separated by SDS/PAGE and visualized using silver staining. Gel bands that were unique to lanes containing OTUB1 as well as the corresponding areas in the control lane were excised and subjected to in-gel digestion with trypsin and analysis by nano-LC (liquid chromatography)–MS/MS as described in [24].

Gel filtration and immunoblotting

Fractionation of crude cell extracts prepared from HEK-293T cells expressing control vector, wild-type or C91S mutant OTUB1 was performed using size-exclusion chromatography with an HPLC system (Agilent HP1200). Sample protein concentration was determined using a Lowry assay (Bio-Rad Laboratories). Samples of 300 μl (2.5 mg/ml) were loaded and separation was achieved using a Zorbax G450 gel-filtration column (9.6 mm×250 mm; Agilent) at a flow rate of 0.7 ml/min in lysis buffer (0.5% NP-40, 150 mM NaCl, 5 mM CaCl2 and 50 mM Tris/HCl, pH 7.4). Fractions of 0.5 or 0.7 ml were collected, and protein material was precipitated using chloroform/methanol [25], followed by SDS/PAGE-based separation and immunoblotting using anti-FUS [fusion involved in t(12;6) in malignant liposarcoma; also known as TLS (translocation in liposarcoma) or CHOP (CCAAT/enhancer-binding protein homologous protein)] (Santa Cruz Biotechnology) or anti-OTUB1 antibodies.

Molecular graphics

All structural Figures were generated using PyMOL v.99vc6 (DeLano Scientific; http://pymol.sourceforge.net/).

RESULTS

Overall structure

OTUB1 is an OTU family member with a conserved ovarian tumour domain between residues 85 and 271. Removal of the N-terminal sequence allowed OTUB1 (residues 40–271) to readily crystallize and diffract to 1.7 Å (1 Å=0.1 nm) resolution. A molecular replacement strategy using co-ordinates from the OTUB2 structure was employed to determine the OTUB1 structure, revealing all residues in the construct except for the first five and Asp238 in the β3–β4 loop. As with other papain-like proteases, the active site is formed at the interface of an α-helical lobe (α3–α10) and a β-sheet lobe (β1–β5), centred on the catalytic cysteine (Cys91), which is situated at the C-terminal pole of the α3 helix (Figure 1A, left-hand panel). The P-side of the active site runs along one side of the interface, and the P′-side runs along the opposing side (Figures 1A and 1C). Interestingly, the conformation of the catalytic triad histidine residue (His265) and its distance from the catalytic cysteine (Cys91, 5.5 Å) and catalytic aspartate (Asp267, 4.6 Å) residues is incompatible with catalysis (Figure 1B, top panel). Instead, His265 is sandwiched between the guanidinium and prolyl side chains of Arg262 and Pro87 respectively, interacting with the catalytic Asp267 through a pair of water molecules. In addition, the catalytic Cys91 is tied-up through hydrogen bonds with the backbone carbonyl oxygen of Arg86 and a water molecule that also interacts with Glu214. The conserved Glu214 from the α9–α10 loop is inserted into the P1 pocket, forming hydrogen bonds with the C-terminal pole of the α3 helix and the backbone of the catalytic Cys91, thereby blocking access to the active site (Figure 1B, top panel). This unique structure corresponds to a non-productive conformation of the catalytic centre not observed in any other cysteine proteases (Figure 1B). Our structure of the ligand-free OTUB1 might therefore represent an autoinhibited state, which is probably involved in a ligand-dependent induced-fit mechanism.

Figure 1 OTUB1 crystal structure and a comparison between OTUB2 and yeast OTU1 bound to ubiquitin-Br3

(A) Molecular surface and overlaying ribbon model of OTUB1 (residues 40–271, left-hand panel), OTUB2 (middle panel) and yeast OTU1–ubiquitin-Br3 (Ub) (right-hand panel, ubiquitin in dark grey, OTU1 in blue). The conserved amino acids expected to be part of the catalytic centre are indicated in red, blue, green and purple respectively. (B) A stereo view of the catalytic site regions of OTUB1 (top panel), OTUB2 (middle panel) and yeast OTU1–ubiquitin (lower panel) illustrates hydrogen-bonding networks between residues within the catalytic site. Ubiq., ubquitin. (C) Display of OTUB1 (pink) and OTUB2 (yellow) as ribbon models, in which the α-helices and β-sheets are numbered. (D) Overlaid ribbon models of OTUB1 (pink) and yeast OTU1–ubiquitin-Br3 (grey) showing the catalytic centre. Catalytic site residues are indicated as in (B), and Trp175 (yeast OTU1, grey) and Glu214 (OTUB1, pink) are shown.

Structural comparison of OTUB1 and OTUB2

Despite being highly homologous with OTUB2 {RMSD (root mean square deviation) value of 1.68, Figure 1 and [14]}, OTUB1 displays a number of different structural features (Figures 1C and 1D), particularly in the vicinity of the catalytic triad and the P1/P1′ pockets. Importantly, in contrast with OTUB1, the structure of OTUB2 reveals a more canonical active site, where the catalytic His224 is within hydrogen-bonding distance of the catalytic cysteine and the pKa of His224 is enhanced by hydrogen bonds with the side chains of adjacent residues, including those of Asn226 and Thr45 [14] (Figure 1B, middle panel). But, unlike OTUB1, the P1/P2 pocket of OTUB2 is partially encumbered by the presence of Ser223 from the β3–β4 loop. Differences in the OTUB1 β4–β5 and OTUB2 β3–β4 loop suggest that it may undergo conformational changes upon ubiquitin binding that then allow rearrangement of His265 for catalytic cysteine activation.

The P′-side of OTUB1, where the leaving lysyl group from a target substrate or ubiquityl group from a chained substrate would be expected to bind, is significantly different from OTUB2. In addition to a dissimilar surface charge property in this region, the α2 helix of OTUB2 is kinked, a feature that is not observed in the OTUB1 structure (Figure 1C). An additional structural restriction of the catalytic cleft in OTUB1 is the presence of Pro87 that is located in close proximity to Cys91 (Figure 1B). In the OTUB2 structure, this position is occupied by Gly47. The presence of a more bulky side chain could potentially contribute to OTUB1's cleavage preference for isopeptide bonds over ubiquitin C-terminal fusions, a trait that is not observed for OTUB2 [14]. A similar argument was made for explaining the specificity of the SUMO-specific protease SENP2 towards lysine deconjugation [26].

Structural comparison of OTUB1 and OTUB2 with yeast OTU1 bound to ubiquitin indicates conformational changes upon ligand binding

Yeast OTU1 has been crystallized recently in a covalent complex with a ubiquitin C-terminal Br3 derivative [15]. Although there is minimal primary sequence homology between the OTU domains of yeast OTU1 and human OTUB1 (<15% identity), they share similar structural features, and a comparison may provide interesting insights into the conformational changes that occur upon ubiquitin binding (Figures 1A and 1B). In addition to the ubiquitin tail, which makes typical residue-specific peptide interactions with the active site (classical P1–P4 substrate residues and the correponding S1–S4 protease pockets), one face of the core domain of ubiquitin is also recognized by OTU1. This area of the enzyme, herein referred to as the SD site [the subsite of the enzyme that docks with the core globular domain of ubiquitin (residues 1–70)], mediates numerous interactions, including those centred on His192 and Ile221, and is likely to contribute to the induced-fit mechanism of the enzyme, the Km of the ubiquityl substrate, or the Kd of the ubiquityl inhibitor. OTUB2 contains a discordant loop (residues 198–204) that would probably clash with ubiquitin binding. This region of OTUB1 forms a well-defined β-strand (β4), extending the β-sheet lobe, but not interfering with the putative ubiquitin-binding region (Figure 1C). The flexible nature of a single residue in the β3–β4 loop (Arg238) in OTUB1 hints at the possibility that OTUB1 might adopt a conformation similar to that of OTUB2 during a regulatory cycle, and that the OTUB1 and OTUB2 structures may represent two conformational states in a regulatory pathway. Alternatively, these structural differences may suggest diverse ubiquitin-binding modes.

Differential proteolytic fine specificities between OTUB1 and OTUB2

To profile potential differences in substrate specificity between OTUB1 and OTUB2, we subjected these enzymes to reaction with various ubiquitin-based active-site probes that have subtle differences in their reactive electrophiles. Probes with a C-terminal Cl2, Br2 or VME group were incubated with enzyme and analysed by electrophoresis and immunoblotting (Figure 2). OTUB2 reacted with all three probes equally well, whereas OTUB1 reacted with the alkyl halides, but not the VME probe (Figures 2B and 2C). These results hint that OTUB1 and OTUB2 may have slightly different specificities, probably expressed by structural differences proximal to their catalytic centres. Furthermore, this supports the notion that OTUB1 is autoinhibited and that specific substrates, or interactors, are required for its activation, probably through an induced-fit mechanism.

Figure 2 OTUB1 and OTUB2 reactivity towards ubiquitin probes with different C-terminal chemical groups

(A) HA-tagged ubiquitin equipped either with a C-terminal Br2 (HA–ubiquitin-Br2), Cl2 (HA–ubiquitin-Cl2) or VME moiety (HA–ubiquitin-VME) was used to test the fine specificity of OTUB1 and OTUB2. (B) Crude extracts prepared from HEK-293T cells overexpressing OTUB1–HA or OTUB2–HA (C) were incubated with HA–ubiquitin-Br2, HA–ubiquitin-Cl2 or HA–ubiquitin-VME, analysed by SDS/PAGE and anti-HA immunoblotting. Ub, ubiquitin.

OTUB1 specificity towards ubiquitin and Ubls

In order to test the deubiquitination properties of OTUB1, we examined its ability to cleave linear di-ubiquitin as well as Lys48- and Lys63-linked tetra-ubiquitin by immunoblotting and MS. OTUB1 did not cleave di-ubiquitin, consistent with previous observations [5] (Figure 3A). At an enzyme/substrate ratio of 1:5, OTUB1 clearly showed a preference for cleaving Lys48-linked tetra-ubiquitin when assessed by anti-ubiquitin immunoblotting (Figure 3B). This preference was confirmed by MS-based analysis (Figure 3C). The use of a higher enzyme/substrate ratio (1:1.5) in the latter analysis indicated further that OTUB1 may also slowly cleave Lys63-linked ubiquitin chains (Figure 3C). Ubiquitin-cleavage assays performed with a truncated form of OTUB1 lacking the first 40 amino acids did not affect the proteolytic properties of OTUB1 (results not shown), indicating that the N-terminal region of OTUB1 does not influence cleavage preference in vitro.

Figure 3 OTUB1 has a preference for cleaving Lys48-linked tetra-ubiquitin over Lys63-linked tetra-ubiquitin

(A) Recombinant OTUB1 or cell extract was incubated with linear di-ubiquitin for 1 or 15 h at 37 °C, separated by SDS/PAGE and analysed by anti-ubiquitin immunoblotting. (B) Recombinant OTUB1 was incubated with either Lys48-linked tetra-ubiquitin (K48 Ub4; upper panels) or Lys63-linked tetra-ubiquitin (K63 Ub4; lower panels), and cleavage was monitored by SDS/PAGE and anti-ubiquitin immunoblotting. As a control, the tetra-ubiquitin substrates were incubated either with crude cell extract or no enzyme. Molecular masses are indicated in kDa. (C) Ub4 K48 (upper panels) or Ub4 K63 (lower panels) were incubated with recombinant OTUB1 for either 2 or 18 h followed by TOF (time-of-flight)–MS analysis. (D) Effect of OTUB1 on polyubiquitination. HEK-293T control cells or transfected with wild-type OTUB1 or C91S mutant were treated or not with the proteasome inhibitor ZL3VS (50 μM final concentration) for 4 h at 37 °C, and crude extracts were analysed by SDS/PAGE and anti-ubiquitin immunoblotting (top panel). Alternatively, crude cell extracts prepared from HEK-293T control or cells expressing OTUB1 or C91S mutant were incubated with ZL3VS (50 μM final concentration) for 30 min at 37 °C and analysed by SDS/PAGE and anti-ubiquitin immunoblotting (α-Ub IB). As a loading control, anti-tubulin immunoblotting was performed. Ctrl, control; wt, wild-type.

We examined further the ability of OTUB1 to catalyse general deubiquitination reactions by examining crude extracts prepared from HEK-293T cells overexpressing either wild-type or C91S inactive mutant OTUB1. We observed only minor effects on the level of polyubiquitinated material in vitro and in living cells (Figure 3D). This observation is consistent with previous findings [5,6] and suggests that OTUB1 may act only on a discrete set of substrates.

In contrast with OTUB1, OTUB2 appears to readily cleave Lys63-linked tetra-ubiquitin in preference to Lys48-linked tetra-ubiquitin (Figure 4A). We therefore evaluated structural differences between OTUB1 and OTUB2 that might explain their differential reactivity towards the active-site probe ubiquitin–VME and the different ubiquitin chains. In particular, we noted that the P1′ site of OTUB1 was sterically restricted by the presence of a proline residue at position 87, which is a glycine residue (Gly47) in the corresponding position in the OTUB2 structure (Figure 1B). OTUB1 P87G and OTUB2 G47P mutant proteins were therefore generated and tested for their ability to cleave different ubiquitin chains. Interestingly, we observed that the steric restriction imposed by a proline residue in this area of the P1′ site in OTUB2 affected the ability of cleaving Lys63-linked tetra-ubiquitin, whereas cleavage slightly improved upon the introduction of a glycine residue at this position in OTUB1 (Figure 4A). Consistent with this, we observed that cleavage of ubiquitin-AMC was also altered, although to a greater extent by the G47P mutation in OTUB2 compared with the inverse modification in OTUB1 (Figure 4B). Combined, these results support the notion that OTUB1 has a more restricted specificity, mainly towards Lys48-linked ubiquitin chains, owing to a narrower P1′ site, and that OTUB1 and OTUB2 may act on a different set of substrates, although the overall structure of these enzymes is similar.

Figure 4 Modification of the P1′ site affects cleavage of Lys63- and Lys48-linked tetra-ubiquitin by OTUB1 and OTUB2

(A) Modification of the P1′ site affects cleavage of Lys63- and Lys48- linked tetra-ubiquitin by OTUB1 and OTUB2. Recombinant wild-type OTUB1, P87G mutant OTUB1, wild-type and G47P mutant OTUB2 were incubated with either Lys48- or Lys63-linked tetra-ubiquitin (Ub4-K48 and Ub4-K63 respectively) for the indicated time at 37 °C. Samples were separated by SDS/PAGE and analysed by anti-ubiquitin immunoblotting. A longer exposure of the experiment performed with wild-type and P87G mutant OTUB1 is shown in the lower-left-hand panel. Anti-histidine immunoblotting (α-HIS) indicates equal loading of added enzyme. Molecular masses are indicated in kDa. (B) Effect of P1′ modifications on ubiquitin-AMC cleavage by OTUB1 and OTUB2. Ubiquitin-AMC (100 nM) was incubated with wild-type or mutant G47P OTUB2, UCH-L3 or OTUB1 at the indicated concentrations (left-hand panel). Cleavage of ubiquitin-AMC with wild-type or mutant P87G OTUB1 was observed at higher concentrations only (right-hand panel). AMC cleavage was measured as a function of time by fluorescence (380 nm excitation/460 nm emission). a.u., arbitrary units; Ub, ubiquitin.

We next examined the possibility that OTUB1 recognizes other ubiquitin-like modifiers. For this purpose, an N-terminally biotinylated seven-amino-acid peptide was conjugated to ubiquitin, NEDD8, ISG15 or SUMO1 in order to create substrates with an isopeptide bond. These were incubated with either OTUB1 or a positive control (Figure 5A). Our results indicate that OTUB1 has a clear preference for ubiquitin and, at a lower rate, for NEDD8 conjugates. In addition, OTUB1 was unable to cleave di-SUMO2 and di-SUMO3 substrates (Figure 5B). Since these experiments are based on the use of a mono- or di-ubiquitin/Ubl-based substrate, the specificity may be different for polyubiquitin/Ubls, particularly in living cells where OTUB1 may be present in multiprotein complexes in which co-factors could influence its cleavage specificity.

Figure 5 OTUB1 cleaves ubiquitin and NEDD8 conjugates, but not SUMO1/2/3 or ISG15

(A) Ubiquitin, NEDD8, ISG15 and SUMO1 were conjugated to a biotinylated 7-mer peptide and incubated with either recombinant OTUB1 or controls including UCH-L3, crude lysate or SENP2. Cleavage was monitored by SDS/PAGE and streptavidin–HRP (horseradish peroxidase) immunoblotting. (B) Di-SUMO2 and di-SUMO3 conjugates were incubated with recombinant OTUB1 or cell lysate as a control, and the cleavage was monitored at the indicated times, followed by SDS/PAGE and anti-SUMO immunoblotting. The position of the 20 kDa band is indicated.

Initial analysis of intracellular OTUB1-interacting proteins

As shown previously, OTUB1 was found to be in complex with GRAIL and USP (ubiquitin-specific peptidase) 8 in lymphocytes [6] and can interact with the enteropathogen Yersinia encoded virulence factor YpkA (YopO) [8]. In order to identify potential OTUB1-interacting proteins in a non-haemopoietic context, we performed proteomics-based studies using co-immunoprecipitation and identification by MS/MS. HEK-293T cells were transiently transfected with either wild-type or catalytically inactive (C91S) OTUB1–HA. The inactive mutant was used as a ‘substrate trap’, preventing potential substrates from being turned over and released [27]. Cell extracts were prepared after 24 h, either with or without detergent solubilization, and subjected to anti-HA immunoprecipitation, followed by cleavage of bound material using TEV protease. This approach was used previously to reduce background arising from non-specific protein binding to beads [28]. Analysis by gel electrophoresis, silver staining and LC MS/MS revealed a total of >80 proteins, of which 26 were uniquely present when OTUB1–HA was expressed, but not detected when OTUB1–HA was absent (Figure 6 and Supplementary Table S1 at http://www.BiochemJ.org/bj/418/bj4180379add.htm). The catalytically inactive OTUB1 mutant (C91S) showed a slightly faster migration profile, a feature commonly observed with mutant proteins. We also detected OTUB1 material that was of higher molecular mass, indicating the presence of post-translational modifications.

Figure 6 MS/MS-based screen of OTUB1-containing protein complexes

(A) HEK-293T control cells or HEK-293T cells transfected with OTUB1 or C91S mutant were lysed either using glass beads or NP-40 detergent-containing buffer, followed by an anti-HA immunoprecipitation. Isolated material was eluted from beads using TEV protease and analysed by SDS/PAGE and silver staining. Protein bands present in OTUB1 or C91S lanes and the corresponding regions in the control were subjected to LC–MS/MS analysis. A total of 26 out of >80 proteins were found to be present only in OTUB1/C91S-containing lanes. Molecular masses are indicated in kDa. Ctrl, control; DDX, DEAD box polypeptide; EBP2, EBNA1 (Epstein–Barr virus nuclear antigen 1)-binding protein 2; hc, heavy chain; hnRNP, heterogeneous nuclear ribonucleoprotein; lc, light chain; NPM, nucleophosmin; PCNA, proliferating-cell nuclear antigen; WT, wild-type. (B) MS/MS spectra of a tryptic peptide derived from RACK1 (upper panel) and a tryptic peptide derived from FUS/TLS (lower panel).

Among the hits, two proteins, FUS/TLS and RACK1 [receptor for activated kinase 1; also known as GNB2L1 (guaninenucleotide-binding protein β polypeptide 2-like 1)] were evaluated further with specific pull-down and follow-up biochemical experiments. Immunoprecipitation of OTUB1–HA followed by anti-FUS/TLS and anti-RACK1 immunoblotting confirmed their presence in OTUB1 complexes, but not in controls (Figure 7A). Since FUS/TLS can exist in different forms [29] and also as fusion proteins in the context of oncogenic transformations [30,31], we examined the effect of OTUB1 on FUS/TLS stability by size-exclusion chromatography and immunoblotting using an anti-FUS polyclonal antibody that detects multiple molecular-mass forms of this protein (Figure 7B, left-hand panel). FUS/TLS appeared predominantly as high-molecular-mass species, but upon overexpression of OTUB1, a FUS/TLS polypeptide of the expected molecular mass of ∼60 kDa was detected (Figure 7B, left-hand panel). Using a monoclonal antibody that recognizes a C-terminal epitope of FUS/TLS, we observed a similar ∼60 kDa form that appears to be stabilized upon overexpression of OTUB1, but not in the presence of the mutant C91S (Figure 7B, right-hand panel). We noted that only a small fraction of high-molecular-mass FUS/TLS was affected by OTUB1, and therefore examined the extent of polyubiquitinated FUS/TLS present in cells. Transfection experiments with HA-tagged ubiquitin, subsequent anti-HA immunoprecipitation and anti-FUS immunoblotting indicated that a small fraction of FUS is indeed ubiquitinated (Figure 7C). Combined, these data suggest that the catalytic activity of OTUB1 may affect the ubiquitination state of a small subset of FUS/TLS protein.

Figure 7 OTUB1 is in complex with RACK1 and FUS/TLS

(A) OTUB1 was immunoprecipitated from HEK-293T control cells or cells transfected with OTUB1, followed by anti-RACK1 and anti-FUS/TLS immunoblotting. (B) OTUB1 affects the molecular-mass distribution of FUS/TLS. Equal amounts of cell extracts derived from control or HEK-293T cells expressing wild-type or C91S mutant OTUB1 were separated by size-exclusion chromatography. Individual fractions were analysed by SDS/PAGE and immunoblotting using an anti-OTUB1 and a polyclonal anti-FUS/TLS antibody that recognizes multiple forms (left-hand panel), or a monoclonal anti-FUS/TLS antibody that recognizes a ∼60 kDa form (right-hand panel). Note that the fraction numbers are not comparable between the left- and right-hand panels. The arrow indicates the observed ∼60 kDa form of FUS/TLS. One out of three independent experiments is shown. Molecular masses are indicated in kDa. (C) HEK-293T cells were transfected with either empty vector or HA–ubiquitin. Cell extracts were prepared after 24 h and subjected to anti-HA immunoprecipitation, followed by SDS/PAGE separation and analysis by anti-HA and anti-FUS immunoblotting. Ctrl, control; MW, molecular mass (indicated in kDa).

DISCUSSION

The structure of OTUB1 reveals a number of properties that address activation and substrate specificity. In analogy to autoinhibition observed in HAUSP (herpesvirus-associated ubiquitinspecific peptidase), USP8 and USP14 [3234], where the unliganded active site is blocked by surface loops, the apo form of OTUB1 adopts a conformation of the active site that appears to be closed and not competent for catalysis. A comparison between OTUB1 and the yeast OTU1–ubiquitin-Br3 structure suggests further that a conformational change would activate their respective catalytic triads (Figure 1) [15]. We propose that the tight OTUB1 β4–β5 loop immediately before His265 (Arg-Pro-Gly264) and the α9–α10 loop, which influences the catalytic cysteine residue through Glu214, could play a role by mediating a conformational change upon ubiquitin binding, allowing His265 to participate in the catalytic triad. However, analysis of the geometry of the active sites revealed subtle variations between OTUB1 and OTUB2, which were reflected by a differential reactivity with alkyl halide and VME-based ubiquitin probes (Figure 2) [22], but also ubiquitin-AMC (Figure 4B). This may be explained by subtle structural differences between the P1′ sides of those enzymes. In particular, steric hindrance by the presence of Pro87 in OTUB1's catalytic centre may prevent the accommodation of the AMC fluorescent group and the VME side chain of the probe, whereas the alkyl halides are able to react with the catalytic cysteine residue to undergo a nucleophilic substitution reaction. Consistent with this notion, the OTUB2 structure contains Gly47 at the equivalent position (Figure 1). Mutational analysis using recombinant OTUB1 P87G and OTUB2 G47P mutants confirmed that steric hindrance at this position compromises the ability to cleave ubiquitin-AMC and to react with the probes, but substitution of glycine for Pro87 in OTUB1 did not improve ubiquitin-AMC cleavage, neither did it restore reactivity towards the ubiquitin–VME probe (Figure 4B and results not shown). This suggests that other structural constrains must exist, or that the introduction of a mutation in this position may also lead to more complex rearrangements of the P1′ site. Reactivity of OTUB1 towards the ubiquitin-Cl2 probe was observed in the present study, but was not detected in a previous study [22]. This might be due to overexpression of OTUB1 in the labelling assay (Figures 2B and 2C), whereas, at physiological levels, other DUBs present in cell extracts may effectively compete for binding to the ubiquitin–chloride probe. Different structural features and discrepancies in the reactivity towards the active site ubiquitin probes between OTUB1 and OTUB2 may hint towards a divergence in their cleavage or substrate specificities. This hypothesis was confirmed by the finding that OTUB1 has a preference for cleaving Lys48-linked tetra-ubiquitin over Lys63-linked tetra-ubiquitin, whereas the opposite appears to be the case for OTUB2. We observed that full-length OTUB2 retained cleavage activity which was unchanged when deleting the C-terminal amino acids 229–234 compared with a previous study ([5] and M. Akutsu and S. Dhe-Paganon, unpublished work). Modification of the P1′ site in OTUB1 by a P87G and in OTUB2 by a G47P substitution affected the ability of these enzymes to cleave Lys63-linked tetra-ubiquitin in the opposite way (Figure 4). This result is fairly surprising, since Lys63-linked polyubiquitin chains are in an extended conformation compared with Lys48-linked chains and therefore would be expected to be less susceptible to structural constraints in the P1′ pocket [35]. Recent structural information as well as predictions suggest that the P1′ site harbours the isopeptide bond lysine side chain, and that the nature of the binding pocket for distal ubiquitin is partially responsible for the affinity towards different ubiquitin chains [36]. Our results provide insights into the role of non-catalytic side chains that can influence proteolysis of different ubiquitin chains. However, other structural elements, such as a kinked α2 helix present in OTUB2, but not OTUB1, may be necessary to determine linkage preference.

The oxyanion hole in OTUB1 appears to be formed by the backbone amide of the catalytic Cys91 and the backbone amide of Asp88. This is very similar to the oxyanion hole geometry observed in OTUB2, A20 and yeast OTU1, thereby confirming that this may be an element specifically observed in OTUs, but not other cysteine protease families [12,14,15].

With the exception of virus-encoded OTU variants that have acquired an additional specificity for ISG15 [3], ovarian-tumour-domain-containing DUBs have a preference for ubiquitinyl derivatives. Our data confirm this finding and indicate further that OTUB1 may have some reactivity towards NEDD8. This apparent dual specificity has also been observed for other DUBs, such as UCH-L1, UCH-L3 [37,38], USP21 and the COP9 signalosome [39,40]. As observed with the OTU domains of A20 and yeast OTU1, OTUB1 has a preference for Lys48-linked polyubiquitin over Lys63-linked polyubiquitin (Figure 3) [15], whereas OTUB2 prefers the latter (Figure 4). However, as observed for A20, the in vitro result may not necessarily reflect biological behaviour in living cells, a trait that remains to be examined for OTUB1-specific substrates.

As a step towards addressing this issue, proteomics- and MS-based screens for OTUB1 interactors and substrates identified a number of candidates (see Supplementary Table S1). We noted a considerable amount of co-immunoprecipitating nuclear proteins, such as cell growth nucleolar proteins, DNA topoisomerases and several RNA helicases sharing a DEAD (Asp-Glu-Ala-Asp) box motif (Figure 6 and Supplementary Table S1). OTUB1 does contain a predicted NLS (nuclear localization sequence) (residues 69–85), and initial microscopy studies suggest that OTUB1 subsets are present in the nucleus ([41] and A. Iphöfer, M. J. Edelmann and B. M. Kessler, unpublished work). On the other hand, RNA-binding proteins are abundant and commonly found as contaminants in MS-based pull-down assays [22], although all proteins listed were not detectable in controls without the bait protein (Figure 6). Among the hits, FUS/TLS and RACK1 were validated by co-immunoprecipitation experiments. FUS/TLS has been described as an RNA-splicing factor involved in chromosomal aberrations observed in Ewing tumours, erythroleukaemia, acute myelocytic/lymphoblastic leukaemia and fibrosarcoma [4244], and it was shown to be present in protein networks containing CIP29 (cytokine-induced protein of 29 kDa) and DDX39 (DEAD box polypeptide 39) [45]. The major form of ∼60 kDa can be recognized with an anti-FUS/TLS monoclonal antibody (Figures 7A and 7B) [46]. Additional high-molecular-mass forms can also be detected using a polyclonal anti-FUS/TLS antibody (Figure 7B). OTUB1, when overexpressed in HEK-293T cells, has an effect on the molecular size of a subset of FUS/TLS-containing material (Figure 7B). This appears to be dependent on the catalytic activity of OTUB1, suggesting that a functional OTUB1 enzyme is required for this process. Using co-transfection experiments with HA-tagged ubiquitin, we were able to observe that a small fraction of the FUS/TLS-containing high-molecular-mass material is ubiquitinated (Figure 7C). Alteration of a small subset did not lead to any significant changes in total FUS/TLS protein levels, although the ∼60 kDa form of FUS/TLS appears to be present in more chromatography fractions when wild-type OTUB1 was overexpressed (Figure 7B, left- and right-hand panels). Attempts to address whether FUS/TLS may be a direct substrate of OTUB1 turned out to be challenging, in part due to the fact that FUS/TLS may exist in multiple forms, and that a major fraction of high-molecular-mass material reactive to polyclonal anti-FUS antibody is not ubiquitinated [47]. OTUB1 may act as a DUB in this case, or could function as a scaffold protein to mediate activity of another DUB, such as USP8 [6], although its effect on FUS appears to be dependent on intact enzymatic activity (Figure 7B). FUS/TLS, as well as RACK1, was also suggested to play a role in the initiation of cell spreading by modulating the formation of SICs (spreading initiation centres) [48]. Taken together, our results suggest that OTUB1 may be linked to these biological processes, and provide the framework for more detailed studies on the functional contribution of OTUB1 within these pathways.

FUNDING

The Structural Genomics Consortium is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canadian Foundation for Innovation, Genome Canada through the Ontario Genomics Institute, GlaxoSmithKline, Karolinska Institutet, the Knut and Alice Wallenberg Foundation, the Ontario Innovation Trust, the Ontario Ministry for Research and Innovation, Merck & Co., the Novartis Research Foundation, the Swedish Agency for Innovation Systems, the Swedish Foundation for Strategic Research and the Wellcome Trust. B. M. K. is supported by a Medical Research Council New Investigator Award [grant number 75743]. M. A. is supported by the Swedish Brain Foundation, the Loo and Hans Ostermans Foundation for Geriatric Research and the Foundation for Geriatric Diseases at the Karolinska Institutet, Stockholm, Sweden.

Acknowledgments

We are indebted to Dr Mark Cebecauer (Imperial College London, London, U.K.) for providing us with the pEF-IRES expression constructs. We also thank Dr Herman Overkleeft (Leiden University, Leiden, The Netherlands) for the gift of proteasome inhibitors, Dr Nicola Ternette for expert assistance with the generation of constructs, Dr Norma Masson for the ubiquitin–HA construct and Dr Kerry R. Love and Hidde L. Ploegh (Whitehead Institute, MIT, Boston, MA, U.S.A.) for providing us with the pTYB2 plasmid. We also acknowledge the Computational Biology Research Group at the University of Oxford for use of their services in this project.

Abbreviations: AMC, 7-amino-4-methylcoumarin; Br2, 2-bromoethyl; Cl2, 2-chloroethyl; DUB, deubiquitinating enzyme; FUS, fusion involved in t(12;16) in malignant liposarcoma; GRAIL, gene related to anergy in lymphocytes; HA, haemagglutinin; HEK, human embryonic kidney; ISG15, interferon-stimulated gene 15; LC, liquid chromatography; MS/MS, tandem MS; NEDD8, neural-precursor-cell-expressed developmentally down-regulated 8; NP-40, Nonidet P40; OTU, ovarian-tumour-domain-containing protease; OTUB, otubain; RACK1, receptor for activated kinase 1; SBP, streptavidin-binding peptide; SUMO, small ubiquitin-related modifier; SENP2, SUMO1/sentrin/SMT3 (suppressor of mif two 3 homologue 1)-specific peptidase 2; TEV, tobacco etch virus; TLS, translocation in liposarcoma; Ubl, ubiquitin-like protein; UCH, ubiquitin C-terminal hydrolase; USP, ubiquitin-specific peptidase; VME, vinyl methyl ester

References

View Abstract