The NF-κB (nuclear factor κB) regulator A20 antagonises IKK [IκB (inhibitor of κB) kinase] activation by modulating Lys63-linked polyubiquitination of cytokine-receptor-associated factors including TRAF2/6 (tumour-necrosis-factor-receptor-associated factor 2/6) and RIP1 (receptor-interacting protein 1). In the present paper we describe the crystal structure of the N-terminal OTU (ovarian tumour) deubiquitinase domain of A20, which differs from other deubiquitinases but shares the minimal catalytic core with otubain-2. Analysis of conserved surface regions allows prediction of ubiquitin-binding sites for the proximal and distal ubiquitin molecules. Structural and biochemical analysis suggests a novel architecture of the catalytic triad, which might be present in a subset of OTU domains including Cezanne and TRABID (TRAF-binding domain). Biochemical analysis shows a preference of the isolated A20 OTU domain for Lys48-linked tetraubiquitin in vitro suggesting that additional specificity factors might be required for the physiological function of A20 in cells.

INTRODUCTION

TNFα (tumour necrosis factor α) and cytokines such as interleukin-1 activate inflammatory signalling cascades leading to the stimulation of transcription factors such as NF-κB (nuclear factor κB) (reviewed in [1,2]). Among the TNFα-induced genes is A20 [3], a negative regulator of NF-κB, which establishes a negative feedback loop to terminate the response (reviewed in [4,5]). Mice lacking A20 develop severe inflammation and are hyper-responsive to TNFα and lipopolysaccharide, owing to the failure to repress TNFα-induced NF-κB signalling [6,7]. Elegant work by Wertz et al. [8] has elucidated the mechanism of A20 action through dual ubiquitin editing functions (reviewed in [5]).

Ubiquitination has been established as a key signalling event at multiple levels of NF-κB activation (reviewed in [912]). Immediately in response to cytokine stimulation, polyubiquitin chains linked through Lys63 are assembled on receptor-interacting proteins including RIP1, TRAF2 (TNFα receptor interacting protein-2), TRAF6 and NEMO (NF-κB essential modifier) [9]. Such Lys63-linked ubiquitin chains act as scaffolds for the TAK1 [TGFβ (transforming growth factor β)-activated kinase-1] and IKK [IκB (inhibitor of κB) kinase] protein kinase complexes, which incorporate subunits with specific ubiquitin-binding domains (NEMO in the IKK complex, and TAB2 or TAB3 in the TAK1 complex) [9]. Co-localization enables TAK1 autophosphorylation of its activation segment, and subsequently TAK1 phosphorylates and activates downstream kinases including IKKβ. IKKβ then phosphorylates the cytoplasmic IκB–NF-κB complex, which triggers Lys48-linked polyubiquitination and proteasomal degradation of IκB, releasing NF-κB to enter the nucleus and activate transcription of target genes (reviewed in [1,2]). A20 acts as a negative regulator by interfering with NF-κB signalling pathways at the level of TRAF and RIP1 ubiquitination by disassembling Lys63-linked polyubiquitin chains from TRAF2/6 and RIP1, and subsequently replacing them with Lys48-linked polyubiquitin [7,8]. This dual action not only antagonizes downstream signalling, but also targets the substrates for proteasomal degradation [5].

A20 is a 790 amino acid protein with an N-terminal OTU (ovarian tumour) domain (residues 1–370) and seven repeats of A20-like ZnF (zinc finger) domains [8]. The ZnF domains confer E3 ubiquitin ligase activity, and bind to UbcH5-family E2 ubiquitin-conjugating enzymes to generate Lys48-linked polyubiquitin chains [8,13]. Recently, A20-like ZnF domains have also been shown to possess ubiquitin-binding capacity [13,14]. In addition, the ZnF domains are involved in A20 oligomerization [15], and protein–protein interactions with ABIN (A20-binding inhibitor of NF-κB) family members [5,16] and TAX1 binding protein-1 [17]. The N-terminal OTU domain of A20 is a deubiquitinase [8,18]. OTU domains share a cysteine protease catalytic triad common to other deubiquitinase families, and A20 has been shown to hydrolyse Lys48-, Lys63- and branched-ubiquitin chains in vitro and in vivo [18,19]. Structurally, OTU domains belong to a relatively uncharacterized class of deubiquitinating enzymes. The structure of otubain-2, a OTU domain protein with 224 amino acids, has been reported, but sequence identity with A20 is low (15% identical) (PDB accession number 1TFF [20]). Otubain-2 is structurally divergent from other cysteine-dependent deubiquitinases such as USP (ubiquitin-specific protease) or UCH (ubiquitin C-terminal hydrolase) domains [20].

In the present study we have analysed the A20 OTU domain crystal structure, and provide biochemical insight into the catalytic mechanism and in vitro activity.

MATERIALS AND METHODS

Cloning

Full-length A20 was cloned from human cDNA, and shorter constructs were PCR-amplified using standard methods. The A20 OTU domain (residues 1–366; A201–366) was cloned into the pGEX6P1 vector [GE Healthcare; N-terminal GST (glutathione transferase)-tag with a PreScission cleavage site] using BamHI and NotI. The pOPIN-E (C-terminal His6 tag) construct was prepared with the Infusion 2.0 system (Clontech). Mutagenesis was performed using the QuikChange® site-directed mutagenesis kit (Stratagene). All constructs were verified by sequencing.

Protein purification and crystallization

GST-tagged A201–366 was expressed in Escherichia coli BL21 (DE3) cells, at 25 °C for 16 h after induction with 200 μM IPTG (isopropyl β-D-thiogalactoside) at a D600 of 0.8. Cells from 2 litres of culture were lysed by sonication in 50 ml of lysis buffer [270 mM sucrose, 50 mM Tris (pH 8.0), 1 mM EDTA, 1 mM EGTA, 10 mM sodium β-glycerophosphate, 50 mM sodium fluoride, 10 mM 2-mercaptoethanol, 1 mM benzamidine, 0.1 mg/ml DNaseI and 1 mg/ml lysozyme], and cleared by centrifugation (20000g for 45 min at 4 °C). The cleared lysate was incubated with 6 ml of equilibrated glutathione S-sepharose 4B resin (GE Healthcare) for 1 h, and was subsequently washed with 50 ml of lysis buffer, 500 ml of buffer A [25 mM Tris (pH 8.5), 1 mM EDTA and 5 mM DTT (dithiothreitol)] plus 500 mM NaCl, and 500 ml of buffer A plus 200 mM NaCl. The GST-tag was cleaved on the resin with 200 μg of GST-tagged PreScission protease overnight. The cleaved A20 OTU domain was subjected to anion-exchange chromatography (HiTrap Q FF) where it eluted as a single peak in a NaCl gradient from 50 to 500 mM. The peak fractions were concentrated to 5 ml and subjected to gel filtration (Superdex 75) in buffer A plus 200 mM NaCl. The protein was concentrated to 13 mg/ml using a VivaSpin (10 kDa molecular mass cut-off) concentrator and was used in crystallization. For structure determination by anomalous phasing techniques, the protein was produced in the E. coli B834 strain and grown in minimal medium with SeMet (seleno-methionine) substituting for methionine, using standard protocols, and purified with increasing DTT concentration (10 mM) in all buffers. All protein purifications were performed at 4 °C.

The A20 OTU domain structure was determined from crystals grown from 1.3 to 1.6 M magnesium sulfate and 0.1 M Mes (pH 6.5–6.9) after 7 days at 14 °C. For synchrotron data collection, crystals were soaked in mother liquor containing 15% ethylene glycol, and were frozen in a nitrogen cryostream.

Data collection, phasing and refinement

Diffraction data on A20 OTU domain crystals were collected at the ESRF (European Synchrotron Radiation Facility), beamlines ID29 and ID14-2. A three wavelength MAD dataset was collected to 3.7 Å (1 Å=0.1 nm) resolution from SeMet crystals and used for phasing. An initial set of sites was determined with the SHELX/hkl2map [21] suite, and site refinement was performed in SHARP [22], resulting in phases to 3.7 Å (Table 1). SHARP determined a solvent content of 62.8%, resulting in a Matthews coefficient of 3.4 with four A20 molecules in the asymmetric unit. A native dataset to 3.20 Å resolution was collected, and the obtained phases were extended using DM [23]. The structure was built in Coot [24] and refinement was performed using PHENIX [25], including four-fold NCS, simulated annealing and TLS B-factor refinement. Final statistics can be found in Table 1.

Table 1
Data collection and refinement statistics

Values between brackets are for the highest resolution shell. All measured data were included in structure refinement.

 A20 SeMet   A20 native 
Data collection statistics Peak Inflection Remote  
 Beamline ID29 ID29 ID29 ID14-2 
 Wavelength (Å) 0.97938 0.97948 0.97372 0.9340 
 Space Group P21 P21 P21 P21 
 Unit Cell (Å) a=85.29 a=85.50 a=85.25 a=84.96 
 b=83.34 b=83.46 b=83.32 b=83.02 
 c=165.40 c=165.87 c=165.55 c=164.94 
 β=98.123 β=98.149 β=98.067 β=98.082 
 Resolution (Å) 84.5–3.70 84.5–3.70 71.1–3.70 50–3.20 
 (3.90–3.70) (3.90–3.70) (3.90–3.70) (3.37–3.20) 
 Observed reflections 125481 60929 72464 152339 
 (18297) (8690) (10625) (22338) 
 Unique reflections 24875 23319 23601 36226 
 (3602) (3337) (3459) (5247) 
 Redundancy 5.0 (5.1) 2.5 (2.6) 3.1 (3.1) 4.2 (4.3) 
 Completeness (%) 100 (100) 93.3 (91.8) 95.7 (96.8) 96.0 (95.9) 
Rmerge 0.112 (0.455) 0.110 (0.567) 0.130 (0.497) 0.075 (0.559) 
 <I/σI> 13.3 (3.5) 7.9 (1.6) 11.2 (3.2) 15.9 (2.6) 
Phasing statistics     
 Anomalous completeness 99.8 (99.9) 75.4 (71.4) 89.5 (89.8) − 
 Anomalous multiplicity 2.6 (2.6) 1.5 (1.5) 1.6 (1.6) − 
 Phasing Power (anomalous)  0.914  − 
 <FOM>  0.373  − 
 <FOM>DM to 3.2Å  0.595  − 
Refinement statistics     
 Reflections in test set − − − 1851 
Rcryst − − − 0.204 
Rfree − − − 0.243 
 Number of residues − − − 1304 
 Number of ions − − − 
 Wilson B2− − − 98.1 
 protein (Å2− − − 115.4 
 RMSD from ideal geometry − − − − 
 Bond length (Å) − − − 0.010 
 Bond angles (°) − − − 1.365 
 A20 SeMet   A20 native 
Data collection statistics Peak Inflection Remote  
 Beamline ID29 ID29 ID29 ID14-2 
 Wavelength (Å) 0.97938 0.97948 0.97372 0.9340 
 Space Group P21 P21 P21 P21 
 Unit Cell (Å) a=85.29 a=85.50 a=85.25 a=84.96 
 b=83.34 b=83.46 b=83.32 b=83.02 
 c=165.40 c=165.87 c=165.55 c=164.94 
 β=98.123 β=98.149 β=98.067 β=98.082 
 Resolution (Å) 84.5–3.70 84.5–3.70 71.1–3.70 50–3.20 
 (3.90–3.70) (3.90–3.70) (3.90–3.70) (3.37–3.20) 
 Observed reflections 125481 60929 72464 152339 
 (18297) (8690) (10625) (22338) 
 Unique reflections 24875 23319 23601 36226 
 (3602) (3337) (3459) (5247) 
 Redundancy 5.0 (5.1) 2.5 (2.6) 3.1 (3.1) 4.2 (4.3) 
 Completeness (%) 100 (100) 93.3 (91.8) 95.7 (96.8) 96.0 (95.9) 
Rmerge 0.112 (0.455) 0.110 (0.567) 0.130 (0.497) 0.075 (0.559) 
 <I/σI> 13.3 (3.5) 7.9 (1.6) 11.2 (3.2) 15.9 (2.6) 
Phasing statistics     
 Anomalous completeness 99.8 (99.9) 75.4 (71.4) 89.5 (89.8) − 
 Anomalous multiplicity 2.6 (2.6) 1.5 (1.5) 1.6 (1.6) − 
 Phasing Power (anomalous)  0.914  − 
 <FOM>  0.373  − 
 <FOM>DM to 3.2Å  0.595  − 
Refinement statistics     
 Reflections in test set − − − 1851 
Rcryst − − − 0.204 
Rfree − − − 0.243 
 Number of residues − − − 1304 
 Number of ions − − − 
 Wilson B2− − − 98.1 
 protein (Å2− − − 115.4 
 RMSD from ideal geometry − − − − 
 Bond length (Å) − − − 0.010 
 Bond angles (°) − − − 1.365 

In vitro deubiquitination assays

Prior to assaying the deubiquitinase activity of A20, the protein was activated by incubation with 10 mM DTT at room temperature (22 °C) for 10 min. In vitro deubiquitination assays were performed with 1.5 μg of A20, 2.5 μg of Lys63- or Lys48-linked tetraubiquitin chains as substrate, in 30 μl of DUB buffer [50 mM Tris (pH 7.6) and 5 mM DTT] at 37 °C. At the time points indicated 5 μl aliquots of the reaction mixture were removed and the reaction was stopped by the addition of 5 μl of 2× SDS sample buffer. Samples were subjected to SDS/PAGE analysis with subsequent silver staining using the BioRad Silver Stain Plus Kit according to the manufacturer's protocol.

RESULTS AND DISCUSSION

Structure determination

To investigate structural aspects of A20-mediated deubiquitination, the OTU domain (residues 1–366) was cloned from human cDNA, expressed as a GST-fusion protein in bacteria and purified to homogeneity (see the Material and methods section). The protein crystallized readily in various conditions from commercial screens, but unfortunately, the predominant trigonal crystal form was perfectly twinned, and such crystals were of no use for experimental phasing techniques. A second crystal form could be obtained in a non-twinned monoclinic setting and the structure was determined using the anomalous signal from incorporated SeMet. Phases were obtained to 3.7 Å resolution and extended to 3.2 Å data (obtained from a native crystal, see the Material and methods section). The structure was built and refined with final statistics shown in Table 1.

Overall structure

The structure of the A20 OTU domain is wedge-shaped, with two almost flat large surfaces and a helical stalk projecting from one side of the molecule (Figure 1A). The core domain resembles the deubiquitinase otubain-2 (Figure 1B, [20]), which is the only protein in the PDB with similarities to A20 [Z-score 6.5, RMSD (root mean square deviation) 2.9 Å over 129 aligned residues] as determined using a DALI search [26].

Structure of the A20 OTU domain and comparison with Otubain-2

Figure 1
Structure of the A20 OTU domain and comparison with Otubain-2

(A) Structure of the A20 OTU domain. Colouring of different structural features is described in the main text, and secondary structure elements are labelled. The active-site residues are shown in stick representation. An interactive three-dimensional version of this Figure is available at http://www.BiochemJ.org/bj/409/0077/bj4090077add1.htm. (B) Structure of otubain-2, with labelled secondary structure elements according to PDB accession number 1TFF [20]. (C) Superposition of A20 (coloured as in A) and otubain-2 (in grey).

Figure 1
Structure of the A20 OTU domain and comparison with Otubain-2

(A) Structure of the A20 OTU domain. Colouring of different structural features is described in the main text, and secondary structure elements are labelled. The active-site residues are shown in stick representation. An interactive three-dimensional version of this Figure is available at http://www.BiochemJ.org/bj/409/0077/bj4090077add1.htm. (B) Structure of otubain-2, with labelled secondary structure elements according to PDB accession number 1TFF [20]. (C) Superposition of A20 (coloured as in A) and otubain-2 (in grey).

This catalytic core present in A20 consists of a five-stranded β-sheet (β1–β5), sandwiched between helix α3 on one side and two parallel helices (α4 and α8) packing against the other side. Four more helices (α1, α5, α6 and α7) form a second, perpendicular half-circle around α4 and α8. These regions (coloured blue in Figure 1) are also present in otubain-2, with several differences (Figure 1B, [20]). Otubain-2 has a four-stranded β-sheet (β3 in A20 has no counterpart) and α1 and α2 in otubain-2 are located similarly to α3 of A20 but pack against the β-sheet differently. Helices α3 and α10 in otubain-2 are structurally equivalent to α4 and α8 in A20. An insertion between helices α3 and α7 in otubain-2 (comprising α4, α5 and α6) is not present in A20, but is structurally mimicked by the A20 α1-helix (cyan in Figure 1). The helical insertion in otubain-2 is also present in otubain-1 and otubain-like proteins, but missing in the remaining annotated OTU domains in ProSite (profile PS50802 [27]). Helix α5 of A20 is considerably shorter than its counterpart in otubain-2 (α7), whereas α6 and α7 in A20 (α8 and α9 in otubain-2) are extended, and are connected by a flexible loop of ten residues that is disordered in the crystal structure (indicated by grey dotted lines in Figures 1A and 2A). A second loop between β2 and β3 is disordered in both A20 and otubain-2; these disordered regions are likely to be involved in ubiquitin binding (see below).

Two structural features of A20 are not present in the structure of otubain-2. First, a helical stalk region extends the OTU core (in shades of yellow in Figure 1A). This stalk is formed by packing of helices α2 and α11. A third region contributing to the stalk is the very N-terminus (residues 1–10), which is partly disordered in the crystal structure (the first residue visible is Pro7). The core between those three regions is of a hydrophobic nature, and termination of the protein at residue 353 (lacking the last two visible turns of the C-terminal α11 helix) results in insoluble protein (results not shown).

The second region only present in A20 is formed by 70 residues (264–337) connecting β-strand β5 with the C-terminal helix α11 (coloured purple in Figure 1A). These residues form two layers of β-hairpin loops connected by helices α9/α10. The top hairpin loop (as in Figure 1A) is shorter, and supported by the lower hairpin loop (β7/β8). In an unusual extended fashion, a gap of 50 Å is bridged from the tip of the lower hairpin loop (residue 320) to the beginning of the C-terminal α11 helix (residue 337), spanning the whole surface of the molecule. Although this region is highly conserved (Figures 2A and 2C) the biological significance of this extension is not clear at present.

Sequence and surface conservation of the A20 OTU domain

Figure 2
Sequence and surface conservation of the A20 OTU domain

(A) Sequence alignment of the A20 OTU domain from different species, indicating sequence conservation in blue. Secondary structure elements are indicated with boxes for α-helices, arrows for β-sheets, and coloured according to Figure 1(A). Residues indicated by arrows are the catalytic site (red) and oxyanion hole (green). Grey dotted lines indicate disordered residues. (B) Surface representation of the A20 OTU domain, coloured according to conservation from green (fully conserved) to red (not conserved). The active site and putative ubiquitin (Ub)-binding sites are indicated. (C) Surface conservation on the A20 surface opposite of the active site. Several highly conserved surfaces are present which are unlikely to contribute to ubiquitin binding.

Figure 2
Sequence and surface conservation of the A20 OTU domain

(A) Sequence alignment of the A20 OTU domain from different species, indicating sequence conservation in blue. Secondary structure elements are indicated with boxes for α-helices, arrows for β-sheets, and coloured according to Figure 1(A). Residues indicated by arrows are the catalytic site (red) and oxyanion hole (green). Grey dotted lines indicate disordered residues. (B) Surface representation of the A20 OTU domain, coloured according to conservation from green (fully conserved) to red (not conserved). The active site and putative ubiquitin (Ub)-binding sites are indicated. (C) Surface conservation on the A20 surface opposite of the active site. Several highly conserved surfaces are present which are unlikely to contribute to ubiquitin binding.

The A20 OTU domain contains conserved surfaces for proximal and distal ubiquitin-binding sites

The structural features of A20 were analysed for conserved surface residues to investigate potential ubiquitin-binding sites. A sequence alignment with the A20 OTU domain from different species was generated (Figure 2A) and mapped on to the surface of the protein (Figures 2B and 2C). A20 shows strongly conserved features around the catalytic site, and in regions that can be assumed to comprise the binding sites for the proximal and distal ubiquitin (Figure 2B). The position of the oxyanion hole relative to the catalytic cysteine residue, and superposition with known deubiquitinase–ubiquitin complexes (see Figure 3 and below) on to A20 allows prediction of the position of the distal ubiquitin. In A20, two loops (residues 152–160 and 212–228) in this distal binding site are disordered despite high sequence conservation, and hence surface analysis in this region is not meaningful. Interestingly, conservation analysis also provides insights into the proximal ubiquitin-binding site. The invariant residues Leu10, Asn98, Asp100, Thr118 and Gln187 are positioned to contact a proximal ubiquitin. However, to understand all features of the A20 OTU domain interactions with ubiquitin, protein complexes will be required.

Comparison of catalytic architectures between A20 and other deubiquitinases

Figure 3
Comparison of catalytic architectures between A20 and other deubiquitinases

(A) Structure of A20, and close-up stereoview of the active site. Catalytic residues are shown as ball-and-stick models, and hydrogen bonds are indicated as grey dotted lines. Catalytic residues are labelled in black, whereas residues forming the oxyanion hole are labelled in red. The side chains of the oxyanion hole generating residues have been omitted for clarity in (A) and (B). (B) Otubain-2 (PDB accession number 1TFF, [20]), superposed on to A20, shown as in (A). (C) Structure of UCH-L1 in complex with ubiquitin vinyl sulfone (PDB accession number 1XD3, [31]). The ubiquitin molecule is drawn in yellow, with stick representation for the C-terminal glycine residues. The position of the C-terminal glycine within the active site is shown; extra atoms from the vinyl sulfone group have been omitted for clarity. (D) Structure of HAUSP in complex with ubiquitin aldehyde (PDB accession number 1NBF, [30]), shown as in (C).

Figure 3
Comparison of catalytic architectures between A20 and other deubiquitinases

(A) Structure of A20, and close-up stereoview of the active site. Catalytic residues are shown as ball-and-stick models, and hydrogen bonds are indicated as grey dotted lines. Catalytic residues are labelled in black, whereas residues forming the oxyanion hole are labelled in red. The side chains of the oxyanion hole generating residues have been omitted for clarity in (A) and (B). (B) Otubain-2 (PDB accession number 1TFF, [20]), superposed on to A20, shown as in (A). (C) Structure of UCH-L1 in complex with ubiquitin vinyl sulfone (PDB accession number 1XD3, [31]). The ubiquitin molecule is drawn in yellow, with stick representation for the C-terminal glycine residues. The position of the C-terminal glycine within the active site is shown; extra atoms from the vinyl sulfone group have been omitted for clarity. (D) Structure of HAUSP in complex with ubiquitin aldehyde (PDB accession number 1NBF, [30]), shown as in (C).

Further examination of conserved residues surprisingly showed that the surface opposite of the active site is also highly conserved (Figure 2C). This surface is mainly formed by the linker region between β8 and α11 (residues 330–340) and loop 263–271, which contribute numerous invariant residues (Glu332, Asn334, Arg271, Lys264 and Ser266) to this surface. This surface region of the molecule is unlikely to contribute to ubiquitin binding, but might be involved in other protein–protein interactions, or in interactions with the C-terminal ZnF domains. Overall, several conserved surface patches in A20 can be derived from this analysis, and it will be interesting to see how and whether these contribute to A20 function.

Catalytic site environment

Cysteine proteases share a mechanistically common active site, which contains a catalytic triad consisting of the catalytic cysteine residue (Cys103 in A20), a histidine residue to abstract the cysteine proton and stabilize the nucleophile (His256 in A20), and a residue to stabilize the histidine residue (often negatively charged, potentially Asp70 in A20) (Figure 3A). A second feature is an oxyanion hole close to the catalytic cysteine residue to stabilize the reaction intermediate. These features are conserved in cysteine-dependent deubiquitinases of the USP, UCH and OTU families [28,29]. Although these three protein families have entirely different folds, structure superposition aligns the active site remarkably well (Figure 3). For the USP and UCH families, complex structures with ubiquitin suicide substrates are available {HAUSP (herpes-associated USP)–ubiquitin complex, PDB accession number 1NBF, [30], Figure 3C; UCH-L1–ubiquitin complex, PDB accession number 1XD3, [31], Figure 3D}, illuminating the position of the C-terminal residues of a distal ubiquitin at the catalytic site, and indicating the overall position of the ubiquitin molecule. In USP and UCH families, a glutamine or asparagine residue N-terminal of the catalytic cysteine residue forms the oxyanion hole.

As mentioned above, A20 and otubain-2 display several differences in structure, but share a conserved core domain harbouring the catalytic residues, which varies from the UCH/USP architecture (Figure 3). Although the catalytic cysteine and histidine residues are structurally conserved, the third residue in the catalytic triad is different. In otubain-2, two residues (Asn226 and Thr45) stabilize the His224 residue (Figure 3B). The loop preceding the catalytic Cys51 in otubain-2 is structurally in a different position from any other deubiquitinase, allowing Thr45 to maintain hydrogen bonds to His224 via side chain and backbone atoms. Strikingly, this different loop also changes the oxyanion hole architecture, which is formed entirely by backbone amide groups from this loop [20].

A20 displays the same catalytic loop organization as otubain-2, and the catalytic residues Cys103 and His256 are structurally conserved, as is the oxyanion loop (Figures 3A and 3B). However, stabilization of His256 for catalysis seems to be different in A20 compared with otubain-2. In A20, similar backbone interactions with Thr97 are formed, however this residue is not conserved throughout species (Figure 2A) and hence the side-chain interaction might not be required for stabilization. Furthermore, there is no Asn226 equivalent residue in A20; the residue in this position (Val258) is unable to stabilize the catalytic His256. Instead, a negatively charged residue, Asp70, is located in close proximity to His256. This residue does not form other electrostatic contacts, and although its distance from His256 is too far for a hydrogen bond (4.4 Å), minor rearrangement would suffice to allow this interaction. Furthermore, this residue is fully conserved (Figure 2A). Asp70 is located on helix α3 that has no structural equivalent in otubain-2.

In the annotated OTU domains in Prosite (profile PS50802), 43 out of 55 proteins contain an aspartate/asparagine/glutamate residue two positions C-terminal to the catalytic histidine residue, mimicking the His224–Asn226 stabilization observed in otubain-2. The remaining 12 proteins contain residues incapable of histidine stabilization such as Val258 of A20. Comparison of sequences N-terminal to the catalytic cysteine residue in these proteins reveals that 10 out of these 12 proteins contain a negatively charged glutamate or aspartate residue at a position equivalent to Asp70 (the remaining two proteins are bacterial and viral OTU domains containing proteins with poor sequence similarity). These proteins are Cezanne-1 and 2, TRABID (TRAF-binding domain)/ZRANB1 and VCIP135 [VCP(p97)/p47 complex-interacting protein of 135 kDa]. Cezanne and TRABID are known interactors of TRAF proteins and may function similarly to A20 [32], whereas VCIP135 interacts with the ubiquitin-dependent chaperone VCP/p97/cdc48 and plays roles in Golgi formation after mitosis [33]. It is hence a possibility that the catalytic triad found in A20 is also present in Cezanne, TRABID and VCIP135.

In vitro deubiquitination activity

The A20 OTU domain possesses catalytic activity as a deubiquitinating enzyme, as it specifically cleaves isopeptide bonds formed between the C-terminal Gly76 carboxy group of ubiquitin and the Nε amino group of a substrate lysine residue. Isopeptide bonds link ubiquitin monomers, and as the seven different lysine residues on ubiquitin are located spatially in different positions, ubiquitin chain structures differ radically depending on the ubiquitin lysine employed for chain formation. We tested the ability of the isolated A20 OTU domain to hydrolyse Lys63- and Lys48-linked tetraubiquitin chains in vitro. Interestingly, we found that the purified A20 OTU domain was able to hydrolyse Lys48-linked ubiquitin tetramers with higher activity compared with Lys63-linked tetraubiquitin (Figure 4A).

In vitro deubiquitination activity of A20

Figure 4
In vitro deubiquitination activity of A20

(A) In vitro activity of the isolated A20 OTU domain against Lys48- and Lys63-linked ubiquitin tetramers in a time course experiment, resolved on a silver-stained denaturing gradient gel. The various ubiquitin chain products are labelled, and the A20 OTU domain is indicated. M, molecular mass marker. A20 does contain activity against Lys63-linked ubiquitin chains at later time points in the experiment. (B) In vitro activity of catalytic site point mutants against Lys48-linked tetraubiquitin compared with wild-type A20 as in (A). Cys103, His256 and Asp70 are required for full catalytic activity.

Figure 4
In vitro deubiquitination activity of A20

(A) In vitro activity of the isolated A20 OTU domain against Lys48- and Lys63-linked ubiquitin tetramers in a time course experiment, resolved on a silver-stained denaturing gradient gel. The various ubiquitin chain products are labelled, and the A20 OTU domain is indicated. M, molecular mass marker. A20 does contain activity against Lys63-linked ubiquitin chains at later time points in the experiment. (B) In vitro activity of catalytic site point mutants against Lys48-linked tetraubiquitin compared with wild-type A20 as in (A). Cys103, His256 and Asp70 are required for full catalytic activity.

The A20 OTU domain structure revealed the identity of the catalytic triad residues, indicating that Asp70 might serve in positioning and charging of the histidine residue in the catalytic triad. Using site-directed mutagenesis, the catalytic triad residues were mutated individually to alanine in residues 1–366 of A20 (A201–366), and the ability of the mutants to cleave Lys48-linked tetraubiquitin was tested, and compared with wild-type A201–366. While A20C103A and A20H256A were devoid of catalytic activity against Lys48-linked tetraubiquitin, reduced activity was observed for A20D70A (Figure 4B). Residual activity of A20D70A is not surprising as the backbone interaction of His256 with Thr97 may also contribute to stabilizing the imidazolium ion.

A20 contains seven A20 ZnF domains within its C-terminus, and recent work indicates that these domains have the ability to bind ubiquitin [13,14], and might therefore serve as ubiquitin chain receptors and contribute to A20 specificity. We therefore tested whether longer versions of A20, including the first (A201–413), the first two (A201–505), three (A201–565) and four (A201–635) ZnF domains have different abilities to cleave Lys63-linked tetraubiquitin. All proteins were similarly active towards Lys48-linked tetraubiquitin, and none cleaved Lys63-linked tetraubiquitin with increased activity compared with A201–366 (results not shown). In contrast with the CYLD USP domain, the second deubiquitinase in the NF-κB pathway, which is intrinsically Lys63-specific (D. Komander, C. J. Lord, H. Scheel, S. Swift, K. Hofmann, A. Ashworth and D. Barford, unpublished work), the A20 OTU domain lacks Lys63 specificity. This suggests the requirement for additional specificity determinants. One possible mechanism could be that A20 utilizes interacting proteins as ‘specificity factors’. ABIN family proteins contain a NEMO-like ubiquitin-binding domain and bind to Lys63-linked ubiquitin chains (results not shown). ABINs interact with the C-terminal ZnF domains of A20 and might serve to co-localize A20 with Lys63-modified substrates. Further structural studies of full length A20 and complexes with interacting proteins as well as ubiquitin chains will be required to shed light on the specificity determinants of A20.

Conclusions

In the present study we provide insights into the structure and catalytic mechanism of the deubiquitinase A20. This second structure of an OTU domain highlights the variability of the cysteine-protease fold. A20 lacks the three helices only present in otubain family members, and features a characteristic C-terminal extension with two β-hairpin loops and a C-terminal α-helix. Further OTU domain structures will be required to understand the contribution of such regions to the OTU domain fold. A second interesting insight is the identification of Asp70 as an imidazolium ion stabilizing residue in A20, which might define a subset of OTU domains which have similar catalytic architecture. It will be important to obtain an (A20) OTU domain structure in complex with ubiquitin. Active site rearrangements, common in deubiquitinases upon substrate binding, are also to be expected for A20; especially movement of helix α3, which would be necessary to bring Asp70 closer to His256, and several disordered loops might become structured in the presence of ubiquitin. Finally, our biochemical data reveal an in vitro preference of the A20 OTU domain for Lys48-linked polyubiquitin. This activity would directly antagonize a ligase activity conferred by the A20 C-terminus; hence modulation of the deubiquitinase activity towards Lys63-modified substrates in vivo seems necessary. Further characterization of A20 interactions and regulation is likely to provide important insights into the cellular role of A20 and the signalling pathways leading to NF-κB activation.

D.K. is supported by a Beit Memorial Fellowship for Medical Research. This work was funded by Cancer Research UK.

Abbreviations

     
  • ABIN

    A20 binding inhibitor of NF-κB

  •  
  • DTT

    dithiothreitol

  •  
  • GST

    glutathione transferase

  •  
  • HAUSP

    herpes-associated USP

  •  
  • IκB

    inhibitor of κB

  •  
  • IKK

    IκB kinase

  •  
  • NEMO

    NF-κB essential modifier

  •  
  • NF-κB

    nuclear factor κB

  •  
  • OTU

    ovarian tumour

  •  
  • RIP

    receptor-interacting protein

  •  
  • SeMet

    seleno-methionine

  •  
  • TAK

    TGFβ (transforming growth factor β)-activated kinase-1

  •  
  • TNFα

    tumour necrosis factor α

  •  
  • TRABID

    TRAF binding domain

  •  
  • TRAF

    tumour-necrosis-factor-receptor-associated factor

  •  
  • UCH

    ubiquitin C-terminal hydrolase

  •  
  • USP

    ubiquitin specific protease

  •  
  • ZnF

    zinc finger

References

References
1
Hayden
M. S.
Ghosh
S.
Signaling to NF-κB
Genes Dev.
2004
, vol. 
18
 (pg. 
2195
-
2224
)
2
Perkins
N. D.
Integrating cell-signalling pathways with NF-κB and IKK function
Nat. Rev.
2007
, vol. 
8
 (pg. 
49
-
62
)
3
Opipari
A.
Boguski
M.
Dixit
V.
The A20 cDNA induced by tumor necrosis factor α encodes a novel type of zinc finger protein
J. Biol. Chem.
1990
, vol. 
265
 (pg. 
14705
-
14708
)
4
Beyaert
R.
Heyninck
K.
Van Huffel
S.
A20 and A20-binding proteins as cellular inhibitors of nuclear factor-κB-dependent gene expression and apoptosis
Biochem. Pharmacol.
2000
, vol. 
60
 (pg. 
1143
-
1151
)
5
Heyninck
K.
Beyaert
R.
A20 inhibits NF-κB activation by dual ubiquitin-editing functions
Trends Biochem. Sci.
2005
, vol. 
30
 (pg. 
1
-
4
)
6
Lee
E.
Boone
D.
Chai
S.
Libby
S.
Chien
M.
Lodolce
J.
Ma
A.
Failure to regulate TNF-induced NF-κB and cell death responses in A20-deficient mice
Science
2000
, vol. 
289
 (pg. 
2350
-
2354
)
7
Boone
D.
Turer
E.
Lee
E.
Ahmad
R.
Wheeler
M.
Tsui
C.
Hurley
P.
Chien
M.
Chai
S.
Hitotsumatsu
O.
, et al. 
The ubiquitin-modifying enzyme A20 is required for termination of Toll-like receptor responses
Nat. Immunol.
2004
, vol. 
5
 (pg. 
1052
-
1060
)
8
Wertz
I.
O'Rourke
K.
Zhou
H.
Eby
M.
Aravind
L.
Seshagiri
S.
Wu
P.
Wiesmann
C.
Baker
R.
Boone
D.
, et al. 
De-ubiquitination and ubiquitin ligase domains of A20 downregulate NF-κB signalling
Nature
2004
, vol. 
430
 (pg. 
694
-
699
)
9
Adhikari
A.
Xu
M.
Chen
Z. J.
Ubiquitin-mediated activation of TAK1 and IKK
Oncogene
2007
, vol. 
26
 (pg. 
3214
-
3226
)
10
Chen
Z. J.
Ubiquitin signalling in the NF-κB pathway
Nat. Cell Biol.
2005
, vol. 
7
 (pg. 
758
-
765
)
11
Evans
P.
Regulation of pro-inflammatory signalling networks by ubiquitin: identification of novel targets for anti-inflammatory drugs
Expert Rev. Mol. Med.
2005
, vol. 
7
 (pg. 
1
-
19
)
12
Kovalenko
A.
Wallach
D.
If the prophet does not come to the mountain: dynamics of signaling complexes in NF-κB activation
Mol. Cell
2006
, vol. 
22
 (pg. 
433
-
436
)
13
Lee
S.
Tsai
Y.
Mattera
R.
Smith
W.
Kostelansky
M.
Weissman
A.
Bonifacino
J.
Hurley
J.
Structural basis for ubiquitin recognition and autoubiquitination by Rabex-5
Nat. Struct. Mol. Biol.
2006
, vol. 
13
 (pg. 
264
-
271
)
14
Penengo
L.
Mapelli
M.
Murachelli
A.
Confalonieri
S.
Magri
L.
Musacchio
A.
Di Fiore
P.
Polo
S.
Schneider
T.
Crystal structure of the ubiquitin binding domains of rabex-5 reveals two modes of interaction with ubiquitin
Cell
2006
, vol. 
124
 (pg. 
1183
-
1195
)
15
De Valck
D.
Heyninck
K.
Van Criekinge
W.
Contreras
R.
Beyaert
R.
Fiers
W.
A20, an inhibitor of cell death, self-associates by its zinc finger domain
FEBS Lett.
1996
, vol. 
384
 (pg. 
61
-
64
)
16
Van Huffel
S.
Delaei
F.
Heyninck
K.
De Valck
D.
Beyaert
R.
Identification of a novel A20-binding inhibitor of nuclear factor-κB activation termed ABIN-2
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
30216
-
30223
)
17
De Valck
D.
Jin
D. Y.
Heyninck
K.
Van de Craen
M.
Contreras
R.
Fiers
W.
Jeang
K. T.
Beyaert
R.
The zinc finger protein A20 interacts with a novel anti-apoptotic protein which is cleaved by specific caspases
Oncogene
1999
, vol. 
18
 (pg. 
4182
-
4190
)
18
Evans
P.
Ovaa
H.
Hamon
M.
Kilshaw
P.
Hamm
S.
Bauer
S.
Ploegh
H.
Smith
T.
Zinc-finger protein A20, a regulator of inflammation and cell survival, has de-ubiquitinating activity
Biochem. J.
2004
, vol. 
378
 (pg. 
727
-
734
)
19
Evans
P.
Smith
T.
Lai
M.
Williams
M.
Burke
D.
Heyninck
K.
Kreike
M.
Beyaert
R.
Blundell
T.
Kilshaw
P.
A novel type of deubiquitinating enzyme
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
23180
-
23186
)
20
Nanao
M.
Tcherniuk
S.
Chroboczek
J.
Dideberg
O.
Dessen
A.
Balakirev
M.
Crystal structure of human otubain 2
EMBO Rep.
2004
, vol. 
5
 (pg. 
783
-
788
)
21
Pape
T.
Schneider
T. R.
Hkl2map: a graphical user interface for macromolecular phasing with shelx programs
J. Appl. Cryst.
2004
, vol. 
37
 (pg. 
843
-
844
)
22
Bricogne
G.
Vonrhein
C.
Flensburg
C.
Schiltz
M.
Paciorek
W.
Generation, representation and flow of phase information in structure determination: recent developments in and around SHARP 2.0
Acta Crystallogr. Sect. D Biol. Crystallogr.
2003
, vol. 
59
 (pg. 
2023
-
2030
)
23
Cowtan
K.
Main
P.
Miscellaneous algorithms for density modification
Acta Crystallogr. Sect. D Biol. Crystallogr.
1998
, vol. 
54
 (pg. 
487
-
493
)
24
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. Sect. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2126
-
2132
)
25
Adams
P. D.
Grosse-Kunstleve
R. W.
Hung
L. W.
Ioerger
T. R.
McCoy
A. J.
Moriarty
N. W.
Read
R. J.
Sacchettini
J. C.
Sauter
N. K.
Terwilliger
T. C.
PHENIX: building new software for automated crystallographic structure determination
Acta Crystallogr. Sect. D Biol. Crystallogr.
2002
, vol. 
58
 (pg. 
1948
-
1954
)
26
Holm
L.
Sander
C.
Protein structure comparison by alignment of distance matrices
J. Mol. Biol.
1993
, vol. 
233
 (pg. 
123
-
138
)
27
Hulo
N.
Bairoch
A.
Bulliard
V.
Cerutti
L.
De Castro
E.
Langendijk-Genevaux
P. S.
Pagni
M.
Sigrist
C. J.
The PROSITE database
Nucleic Acids Res.
2006
, vol. 
34
 (pg. 
D227
-
D230
)
28
Amerik
A. Y.
Hochstrasser
M.
Mechanism and function of deubiquitinating enzymes
Biochim. Biophys. Acta
2004
, vol. 
1695
 (pg. 
189
-
207
)
29
Nijman
S. M.
Luna-Vargas
M. P.
Velds
A.
Brummelkamp
T. R.
Dirac
A. M.
Sixma
T. K.
Bernards
R.
A genomic and functional inventory of deubiquitinating enzymes
Cell
2005
, vol. 
123
 (pg. 
773
-
786
)
30
Hu
M.
Li
P.
Li
M.
Li
W.
Yao
T.
Wu
J. W.
Gu
W.
Cohen
R. E.
Shi
Y.
Crystal structure of a UBP-family deubiquitinating enzyme in isolation and in complex with ubiquitin aldehyde
Cell
2002
, vol. 
111
 (pg. 
1041
-
1054
)
31
Johnston
S. C.
Riddle
S. M.
Cohen
R. E.
Hill
C. P.
Structural basis for the specificity of ubiquitin C-terminal hydrolases
EMBO J.
1999
, vol. 
18
 (pg. 
3877
-
3887
)
32
Evans
P. C.
Taylor
E. R.
Coadwell
J.
Heyninck
K.
Beyaert
R.
Kilshaw
P. J.
Isolation and characterization of two novel A20-like proteins
Biochem. J.
2001
, vol. 
357
 (pg. 
617
-
623
)
33
Wang
Y.
Satoh
A.
Warren
G.
Meyer
H. H.
VCIP135 acts as a deubiquitinating enzyme during p97-p47-mediated reassembly of mitotic Golgi fragments
J. Cell. Biol.
2004
, vol. 
164
 (pg. 
973
-
978
)

Author notes

Model co-ordinates and structure factors have been deposited with the PDB under accession number 2VFJ.