Structural Analysis of a Viral Ovarian Tumor Domain Protease from the Crimean-Congo Hemorrhagic Fever Virus in Complex with Covalently Bonded Ubiquitin

Abstract

Crimean-Congo hemorrhagic fever (CCHF) virus is a tick-borne, negative-sense, single-stranded RNA [ssRNA(−)] nairovirus that produces fever, prostration, and severe hemorrhages in humans. With fatality rates for CCHF ranging up to 70% based on several factors, CCHF is considered a dangerous emerging disease. Originally identified in the former Soviet Union and the Congo, CCHF has rapidly spread across large sections of Europe, Asia, and Africa. Recent reports have identified a viral homologue of the ovarian tumor protease superfamily (vOTU) within its L protein. This protease has subsequently been implicated in downregulation of the type I interferon immune response through cleavage of posttranslational modifying proteins ubiquitin (Ub) and the Ub-like interferon-simulated gene 15 (ISG15). Additionally, homologues of vOTU have been suggested to perform similar roles in the positive-sense, single-stranded RNA [ssRNA(+)] arteriviruses. By utilizing X-ray crystallographic techniques, the structure of vOTU covalently bound to ubiquitin propylamine, a suicide substrate of the enzyme, was elucidated to 1.7 Å, revealing unique structural elements that define this new subclass of the OTU superfamily. In addition, kinetic studies were carried out with aminomethylcoumarin (AMC) conjugates of monomeric Ub, ISG15, and NEDD8 (neural precursor cell expressed, developmentally downregulated 8) substrates in order to provide quantitative insights into vOTU's preference for Ub and Ub-like substrates.

Crimean-Congo hemorrhagic fever (CCHF) is characterized in humans by the sudden onset of fever, myalgia, headache, dizziness, sore eyes, photophobia, and hyperanemia as well as severe hemorrhages (28, 43, 46). The causative agent of CCHF is the CCHF virus, which is a tick-borne, negative-sense, single-stranded RNA [ssRNA(−)] virus of the genus Nairovirus, belonging to the viral family Bunyaviridae. Originally named after outbreaks in the former Soviet Union and in the Congo during the mid-20th century, the affected area of this disease has rapidly spread to large areas of sub-Saharan Africa, the Balkans, Northern Greece, European Russia, Pakistan, the Arabian Peninsula, Iran, Afghanistan, Iraq, Turkey, and recently, the Xinjiang province of China (43, 46). The CCHF viral genome, as well as those of the closely related Dugbe and Nairobi viruses, consists of three negative-sense RNA segments, small (S), medium (M), and large (L). Incubation of CCHF is 5 to 6 days, with fatalities occurring less than 7 days after signs of infection. Fatality rates for patients infected with the CCHF virus ranged from 5% to 70%, depending on phylogenetic variation of the virus, transmission route, treatment facility, and the reporting and confirmation of the case statistics (19, 32, 43, 47).

The innate immune system serves as the human's first line of defense from invading pathogens, including CCHF virus. The type I interferon (IFN) response comprises a key component of this system by upregulating more than 300 IFN-stimulated genes (ISGs) whose products detect viral molecules, promote amplification of the type I IFN response, modulate other signaling pathways, and directly provide antiviral activity (34). Regulation of the type I IFN response has been shown to rely on posttranslational modification by ubiquitin (Ub) and the Ub-like interferon-simulated gene 15 (ISG15) (14, 23). Both Ub and ISG15 are expressed in a proform and cleaved to leave a double-glycine C terminus that forms an isopeptide bond with predominantly the ɛ-NH₂ of lysine residues of a target protein through a three-step enzymatic process. In addition to forming isopeptide bonds with target proteins, Ub, which contains seven lysine residues, has been observed to form poly-Ub chains. The most studied of these moieties are K29-linked, K48-linked, and K63-linked poly-Ub. While K29-linked and K48-linked polyubiquitination of proteins leads to their degradation in the lysosome and proteasome, respectively, conjugation of K63-linked poly-Ub to proteins has an activating effect, resulting in an enhanced type I IFN response (2, 7, 18, 33, 40). Currently, more than 150 proteins have been identified as forming conjugates with ISG15, with the number of proteins forming Ub conjugates far exceeding that number (12, 48). A subset of type I IFN signaling and effector proteins that Ub and ISG15 have been shown to stabilize includes JAK1, STAT1/2, double-stranded RNA-dependent protein kinase (PKR), myxovirus-resistant protein A (MxA), and RIG-I (17). MxA has particularly shown to be important in type I IFN response to CCHF infection. RIG-I and several other proteins have also been shown to be targets for K63-linked poly-Ub (4).

Recently, investigators have identified a cysteine viral ovarian tumor domain (vOTU) protease colocated with the RNA-dependent RNA polymerase in the L protein of the CCHF virus (14). Interestingly, as CCHF is an ssRNA(−) virus, no protease is required to cleave a viral polypeptide to facilitate viral replication as in positive-sense ssRNA [ssRNA(+)] viruses. Furthermore, recent reports have observed that vOTU is not required for RNA-dependent RNA polymerase activity and for vOTU protease activity linked to impairment of the type I IFN response through its deubiquitinating and deISGylating activity (6, 14). Additional studies have also tentatively identified the presence of vOTU homologues in the Arterivirus genus of the Arteriviridae family, suggesting that they too may facilitate impairment of the type I IFN response (14). Since the discovery of the first ovarian tumor domain (OTU) protease in Drosophila oogenesis and prior to the identification of vOTU, OTU superfamily members could be divided into three subclasses according to their sequence homology, otubains, A20-like OTUs, and ubiquitin thioesterase ZRANB1 (22). With the addition of the viral OTU subclass, OTU superfamily members in more than 100 eukaryotic, bacterial, and viral proteins have now been identified (6, 27). Predominantly, OTU proteases have been linked to ubiquitin (Ub) removal and/or remodeling of Ub-conjugated proteins, placing them among five protease superfamilies that facilitate signal transduction cascades and play key roles in protein stability (22). However, vOTU is unique in that it is the only OTU to have shown both deubiquitinating and deISGylating activity (14). Instead, Otubain1/2 (OTUB1/2) plays a key role in T cell response and prefers K48-linked poly-Ub or NEDD8 (neural precursor cell expressed, developmentally downregulated 8) as a substrate (12). A20 and A20-like Cezanne OTU proteases are negative regulators of the NF-κB-mediated inflammation response, selectively cleaving K63-linked poly-Ub targets. DUBA also shows preference for K63-linked poly-Ub (20). In attempts to better understand the OTU superfamily, structures of OTUB and A20-like OTU domains have been elucidated (12, 21, 30). An X-ray structure of the yeast ovarian tumor 1 (yOTU1) domain, which interacts with Cdc48 and has a preference for K48-linked poly-Ub, was achieved in complex with mono-Ub (27). However, since yOTU1 has a preference for K48-linked Ub and possesses low sequence identity to vOTU and other OTU domain proteases, only limited information on vOTU could be obtained. In addition to vOTU, several other viral proteases, such as papain-like protease (PLpro) from the severe acute respiratory syndrome (SARS) coronavirus, have also shown deubiquitinating and deISGylating activity to evade the innate immune system (6, 8, 43, 49). However, no viral proteases that are known to possess deISGylating activity have been visualized as being bound to Ub or Ub-like substrates. To address this issue and elucidate the atomic-level structure of a member from the viral OTU superfamily subclass, we have obtained the X-ray crystal structure of vOTU bound with Ub (vOTU-Ub). We also have characterized the vOTU substrate specificity for mono-Ub, ISG15, and NEDD8 and compared the results with those from human OTUB2 (hOTUB2). Additionally, we assessed vOTU's deubiquitinating activity toward K48- and K63-linked poly-Ub.

MATERIALS AND METHODS

Structure and sequence alignments.

The alignment of OTUs in Fig. 1 a was generated using two Protein3Dfit alignments, vOTU (3PRP) to yOTU1 (3BY4) and vOTU (3PRP) to OTUB2 (1TFF) (24). These structural alignments and the amino acid sequence of the NSP2 from porcine reproductive and respiratory syndrome virus (PRRSV) were assembled using CLUSTALWPROF and TEXSHADE (http://workbench.sdsc.edu/) with the following settings: matrix, gonnet; gap-opening penalty, 11; gap extension penalty, 1; and lambda ratio, 0.85. Amino acids are color coded according to their being nonconserved (white background), similar (lime green background), conserved (green background), or completely conserved (dark-green background, orange lettering) across the four sequences. The Ub and Ub-like alignment in Fig. 1b was generated using CLUSTALW and TEXSHADE with the same settings used for alignment of the OTUs.

FIG. 1.

Sequence alignments of various OTU proteases and of Ub and Ub-like proteins. (a) OTUs are from CCHF virus (vOTU; GenBank accession no. AAQ98866.2), Saccharomyces cerevisiae (yOTU1; PDB accession no. 3BY4_A), PRRSV (NSP2PC; GenBank accession no. ACO06904.1), and Homo sapiens Otubain2 (PDB accession no. 1TFF_A). Secondary structure of vOTU according to the defined secondary structure of proteins (DSSP) algorithm is represented by blue cylinders (helical regions), brown arrows (β-sheet regions), and blue lines (loops). Breaks denote regions where vOTU has residues. Residues involved in vOTU's catalytic triad are outlined in black boxes. Residues outlined in orange boxes represent those involved in oxyanion hole formation. Colored bars below the sequence alignment indicate vOTU and yOTU1 positions within 4 Å of Ub as determined by the Contact program of the CCP4 suite (9). Bars under vOTU unique positions are colored light blue with bars under unique yOTU1 positions in red. Equivalent positions in both vOTU and yOTU1 are shown in purple. Positions that form significant salt bridge and H-bond interactions are highlighted by an asterisk. (b) Ub and Ub-like proteins from H. sapiens (PDB accession no. 1UBQ_A), NEDD8 (GenBank accession no. CAG28590), and ISG15 (PDB accession no. 1Z2M_A). Colored bars below the sequence alignment indicate Ub positions within 4 Å of either vOTU and yOTU1 as determined by the Contact program of the CCP4 suite (9). Bars colored light blue and red indicate Ub positions that interact solely with vOTU and yOTU1, respectively. Purple bars indicate Ub residues within 4 Å of both vOTU and yOTU1. Amino acids are color coded according to their being nonconserved (white background), similar (lime green background), conserved (green background), or completely conserved (dark-green background, orange lettering) across the four sequences.

Construction of vOTU and Ub-Br3 expression vectors.

A vOTU expression construct was obtained by use of an Escherichia coli BL21 codon-optimized synthesis of the first 169 amino acids from the L protein in CCHF virus (GenBank accession no. AAQ98866.2) by Biobasic, Inc. Along with the vOTU portion of the L protein, six histidine codons and a stop codon were added to the gene in order to provide for a C terminus histidine tag. The resulting gene was incorporated into a pET11a plasmid using NdeI and BamHI restriction sites. The construct was introduced into E. coli BL21(DE3) competent cells by heat shock transformation. The resulting plasmid was then purified, restriction analyzed, and sequenced to verify the construct. Constructs of expression plasmids for OTUB2 and C-terminally modified Ub with bromopropylamine were designed according to previously published studies (27, 30, 45).

Expression and purification of vOTU, OTUB2, and Ub-Br3.

For vOTU enzymatic studies and native X-ray structural determination of vOTU-Ub, E. coli harboring vOTU was grown at 37°C in 6 liters of LB broth containing 100 μg/ml of ampicillin until the optical density at 600 nm reached 0.6. It was then induced by 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG). The culture was further grown for 4 h at 37°C and then centrifuged at 12,000 × g for 10 min. For incorporation of selenomethionyl (Se-Met) into vOTU for crystal growth, the BL21(DE3) E. coli harboring vOTU was grown in minimal medium supplemented with 19 amino acids (all except methionine). l-Selenomethionine (100 mg per liter; EMD Chemicals) was added immediately before overnight induction at 25°C. OTUB2 was expressed using previously established protocols (30). Cells were collected and stored at −80°C until use. vOTU- and OTUB2-containing cell pellets were lysed by the addition of buffer A (500 mM NaCl, 50 mM Tris-HCl [pH 8.0]) containing 5 mg of lysozyme. The solutions were then sonicated using a Fisher Scientific series 150 sonicator on ice at 30% power with 5-s pulses for 5 min. Insoluble debris was removed by centrifugation at 26,000 × g for 45 min. The clarified extracts were filtered with a 45-μm filter and loaded directly onto Ni-nitrilotriacetic acid (NTA) agarose resin (Qiagen) preequilibrated in buffer A. The column was then washed with two column volumes of buffer A followed by five column volumes of a buffer A supplemented with 10 mM imidazole solution. The protein was eluted using five column volumes of a buffer A supplemented with 250 mM imidazole solution. The elution was applied to a Superdex-200 Hiload 26/60 FPLC column preequilibrated with buffer B (100 mM NaCl, 5 mM HEPES [pH 7.4], 2 mM dithiothreitol [DTT]) and eluted at a flow rate of 1.5 ml/min. Fractions were pooled according to the chromatogram, concentrated to 10 mg/ml in a GE Vivaspin centrifuge concentrator with a 10-kDa molecular mass cutoff and filtered through a 0.65-μm Amicon spin filter. Purity of OTUB2 and vOTU was assessed by use of SDS-PAGE.

Truncated Ub (residues 1 to 75) was expressed according to previously established protocols with the exception that after the IPTG induction, the culture was grown overnight at 25°C (27, 45). E. coli BL21-CodonPlus pellets containing Ub (1 to 75) were lysed with buffer C (25 mM HEPES [pH 6.8], 50 mM sodium acetate, and 75 mM NaCl), which was augmented with 0.16% Triton X-100. As with the vOTU and OTUB2 preparations, the solution was then sonicated on ice at 30% power with 5-s pulses for 5 min. Insoluble debris was removed by centrifugation at 30,000 × g for 45 min. The clarified extract was filtered with a 45-μm filter and then poured over a chitin column preequilibrated with buffer C. The column was washed with three column volumes of buffer C, followed by resuspension in two column volumes of a solution of buffer C supplemented with 100 mM sodium 2-mercaptoethanesulfonate (MESNA). The resuspension was rocked gently overnight at 4°C and eluted by filtering through a XK 26/40 GE column. The resulting Ub thioester was then derivatized with 3-bromopropylamine according to previously described methods (27, 45). Purity of Ub-Br3 was assessed by use of 10 to 20% Ready Gel Tris-Tricine gels (Bio-Rad, CA). All final protein concentrations were determined by absorbance at 280 nm using an experimentally determined extinction coefficient (15).

vOTU and hOTUB2 deubiquitination and deISGylation assay.

All assays were performed with 100 mM NaCl, 50 mM HEPES [pH 7.5], 0.01 mg/ml bovine serum albumin (BSA), and 5 mM DTT, by using a Corning Costar half-volume 96-well plate with a final volume of 50 μl in duplicate. The rates of the reactions were observed using a Synergy HTTR multimode plate reader (Bio-Tek) thermostated at 23°C. Specifically, the increase in fluorescence (excitation λ, 360 nm; emission, 460 nm) of 7-amino-4-methylcourmarin upon cleavage from Ub-aminomethylcoumarin (AMC), ISG15-AMC, and NEDD8-AMC substrates obtained from Boston Biochem, MA, was monitored. The extinction coefficients for all three substrates were determined by adding excess vOTU to various concentrations of each substrate and allowing the reaction to run until completion. The resulting maximum fluorescence values were plotted to determine the slope and subsequently the extinction coefficients. To determine vOTU's K_m values for Ub-AMC, ISG15-AMC, and NEDD8-AMC, the substrate concentration was varied from 0 to 50 μM. The initial rates were fitted to the Michaelis-Menten equation, υ = V_max/{1 + (K_m/[S])}, using the enzyme kinetics (v. 1.3) module of SigmaPlot (v. 10.0; SPSS, Inc.) in order to calculate K_m and V_max. OTUB2 was only assessed using a substrate concentration of 2 μM for each of the three substrates. For K63-linked and K48-linked poly-Ub deubiquitinating assays, tetra- and di-Ub linked by either K48 or K63 isopeptide bonds were purchased from Boston Biochem and Enzo Life Sciences. These poly-Ub substrates (10 μM) were incubated with vOTU (10 nM) in a reaction buffer (100 mM NaCl, 50 mM HEPES [pH 7.5], and 2 mM DTT) at 37°C. Reactions were stopped at various times over 1 h by mixing 9 μl of each reaction with 2× SDS-Tricine sample buffer and boiling for 2 min.

vOTU and vOTU-Ub static light scattering.

vOTU and vOTU-Ub were loaded onto a Wyatt Technology Corp. size exclusion column with a pore size of 300 Å preequilibrated in buffer D [100 mM NaCl, 5 mM HEPES (pH 7.4), 1 mM Tris(2-carboxyethyl)phosphine hydrochloride [TCEP HCl]) and attached to a miniDawn Treos (Wyatt, CA). Molecular weights (MW) were calculated using ASTRA (v. 5.3.4.16) software and the experimentally determined extinction coefficients for vOTU and vOTU-Ub (15).

vOTU-Ub formation and crystallization.

vOTU was combined with Ub-Br3 in equal molar ratios, incubated at 37°C for 2 h and left overnight at 4°C. Complex formation was monitored using 10 to 20% Ready Gel Tris-Tricine gels (Bio-Rad). vOTU-Ub was purified using a GE mono-Q column with buffer E (50 mM Tris-HCl, pH 8.0) and buffer F (50 mM Tris-HCl [pH 8.0], 1 M NaCl). Fractions were pooled according to the chromatogram and loaded onto a AP-1 (Waters) column packed with Superdex-75 preequilibrated with buffer D and eluted at a flow rate of 1.0 ml/min. The resulting vOTU-Ub was concentrated in a GE Vivaspin 6 concentrator with a 10-kDa molecular mass cutoff and filtered in a 0.65-μm Amicon spin filter.

Initial crystal conditions for vOTU-Ub were determined from high-throughput screening of Qiagen Classics and polyethylene glycol I (PEG I) screens in a 96-well sitting drop format using an Art Robbins Phoenix robot. Drops contained 0.4 μl of protein solution and 0.4 μl of precipitate with a 100-μl reservoir volume. Initial screening resulted in several hits; however, a solution containing 100 mM Sodium cacodylate (pH 6.5), 200 mM magnesium acetate, and 24% PEG 8000 produced the most viable crystals. These crystals were optimized using Additive HT Screen from Hampton Research. Final native and Se-Met vOTU-Ub crystals were obtained through vapor diffusion using a 500-μl reservoir with 4-μl hanging drops mixed 1:1 with protein solution and using a precipitant gradient of 22 to 28% PEG 8000, 100 mM Sodium cacodylate (pH 6.5), 100 to 250 mM magnesium acetate, and 2% n-octyl-β-d-glucoside.

Data collection and X-ray structural determination of vOTU-Ub.

All X-ray data sets were collected using crystals mounted on nylon loops and submerged in a 5 μl cryosolution of 28% PEG 8000, 100 mM Sodium cacodylate (pH 6.5), and 200 mM magnesium acetate. Crystals were subsequently flash frozen in liquid nitrogen. Frozen crystals were mounted under a stream of dry N₂ at 100 K. A single-wavelength anomalous dispersion (SAD) data set with a resolution to 2.3 Å and a native data set with a resolution to 1.7 Å were collected at the 21-ID-D and 21-ID-F Life Science-Collaborative Access Team beamlines, respectively, at the Advanced Photon Source Synchrotron, Argonne National Laboratory. For the SAD data collected, a single crystal of the Se-MET containing vOTU-Ub protein was collected at 0.9789 Å with a MarMosaic300 charge-coupled-device (CCD) detector (Rayonix) with the native data collected at 0.97872 Å with a MarMosaic225 CCD detector (Rayonix). All data were subsequently processed and scaled using Denzo and Scalepack (31). The heavy atom sites from the SAD data were determined using the program HYSS from Phenix (16), subsequent phases were determined and refined with Phaser (25), Resolve was used for density modification (41), and an initial model was autobuilt with Resolve (41). The entire path was administered via the AutoSol procedure within Phenix (1). Once an initial model was constructed utilizing winCoot with the initial phases, the vOTU-Ub model was used to create a molecular replacement solution for the native data using Phaser (25) and further refined using iterative cycles of model building and structure refinement using Coot (13) and REFMAC (29), respectively. Water molecules were added to 2F_o-F_c density peaks that were greater than 1σ using the “Find Water” winCoot program function. The final model was checked for structural quality utilizing the CCP4 suite programs Procheck and Sfcheck. Structure factors and coordinates have been deposited in the Protein Data Bank (PDB), and data processing and refinement statistics are shown in Table 1 .

TABLE 1.

Data collection and refinement statistics^a

Statistic	Se-Met	Native
Data collection
Space group	P212121	I222
Wavelength (Å)	0.978	0.978
Unit cell dimensions
a, b, c (Å)	60.5, 65.7, 133.13	79.3, 105.8, 113.0
α = β = γ (degrees)	90.0	90.0
Resolution (Å)	30.0-2.3	50.0-1.7
No. of reflections observed	175,494	319,461
No. of unique reflections	23,928	50,853
Rmerge (%)	7.2 (32.2)	5.5 (18.6)
I/σI	39.5 (7.3)	22.7 (2.8)
% Completeness	98.7 (100)	96.5 (74.2)
Phasing statistics
No. of Se-Met sites found	8
Phasing FOM	36.4
Resolve FOM	65
Refinement
Resolution range	30.0-2.3	50.0-1.7
No. of reflections in working set	22,660	48,272
No. of reflections in test set	1,212	2,581
Rwork (%)	20.4	17.1
Rfree (%)	26.7	21.2
Avg B factor (Å2)	34.8	25.5
RMS deviation
Bond length (Å)	0.02	0.01
Bond angle (degrees)	1.64	1.35
No. of protein atoms/no. of water molecules	3,948/220	4,013/492

^aRmerge, Σ_hΣ_i|i_i(h)−<I(h)>|/Σ_hΣ_iI_i(h), where I_i(h) is the ith, measurement and <I(h)> is the weighted mean of all measurements of I(h). I/σI, signal intensity-to-noise ratio; FOM, figure of merit; Rwork, and Rfree, h(|F(h)obs|−|F(h)calc|)/h|F(h)obs| for reflections in the working test sets, respectively. RMS, root mean square. Data for the last resolution shell are shown in parentheses.

Protein structure accession numbers.

Structure factors and coordinates have been assigned PDB codes 3PRM (Se-Met) and 3PRP (native).

RESULTS

Kinetic characterization of vOTU using mono-Ub-AMC and Ub-like-AMC substrates.

Initial reports outlining the domain boundaries of vOTU to the first 169 residues of the CCHF viral L protein qualitatively illustrated its ability to cleave both Ub and ISG15 conjugates (14). To gain a quantitative understanding of the vOTU's order of preference for Ub and some Ub-like substrates, the cleavage of C-terminally linked 7-amino-4-methylcourmarin conjugates of Ub, ISG15, and NEDD8 by vOTU was monitored. Using initial velocities, the K_m and V_max were determined for each substrate (Fig. 2 a and b). Interestingly, vOTU possesses 6-fold higher activity for Ub than for ISG15. Furthermore, its affinities for NEDD8 are 200-fold and 30-fold less than those for Ub and ISG15, respectively. To gain a snapshot of how vOTU compares to other known OTU family members, human OTUB2 (hOTUB2) was also assessed using Ub-AMC, ISG15-AMC, and NEDD8-AMC at a fixed substrate concentration of 2 μM. Beyond the overall higher deubiquitinating/deISGylating activity of vOTU over hOTUB2 at this concentration and the mutual preference of the proteases for Ub-AMC, the proteases differ in their activity toward ISG15-AMC and NEDD8-AMC. Similar to results of previous qualitative studies utilizing band shifts on a sodium dodecyl sulfate-polyacrylamide gel, OTUB2 shows minimal deISGylating activity (Fig. 2c) (14). However, it does possess the ability to cleave NEDD8-AMC, reversing the order of vOTU's preference for ISG15-AMC and NEDD8-AMC.

FIG. 2.

vOTU mono-Ub and Ub-like cleavage activity. (a) vOTU cleavage activity for mono-Ub-AMC (▵), ISG15-AMC (•), and NEDD8-AMC (○). The initial substrate concentration was 2 μM. Error bars represent the average errors. (b) vOTU K_m and V_max constants for substrates in panel a. (c) hOTUB2 cleavage activity for mono-Ub-AMC (▵), ISG15-AMC (•), and NEDD8-AMC (○). Error bars represent the average errors.

X-ray structure elucidation of vOTU-Ub.

To gain a fuller understanding of vOTU's mechanism for specificity of Ub and Ub-like substrates, Ub with its 76th residue replaced by bromopropylamine was used as a vOTU substrate in order to form the covalent complex vOTU-Ub. By screening vOTU-Ub against commercial crystal precipitant screens and using an additive screen for optimization, single diffracting crystals were obtained in a mixture of PEG 8000, sodium cacodylate (pH 6.5), magnesium acetate, and n-octyl-β-d-glucoside. A native data set was subsequently obtained to 1.7 Å in an I222 space group, but molecular replacement employing composite models generated from yOTU1-Ub, OTUB1, and OTUB2 failed to elucidate a solution (12, 27, 30). However, using selenomethionyl (Se-Met)-substituted vOTU-Ub, phases were obtained using SAD to 2.3 Å in the space group of P2₁2₁2₁ (Table 1). The resulting solution was two vOTU-Ub molecules in the asymmetric unit. Once an initial model was constructed, a molecular replacement solution also containing two vOTU-Ub molecules in the asymmetric unit was readily obtained from the native data set (Fig. 3 a; Table 1). In the final model, all of the Ub residues and vOTU residues 1 to 165 per vOTU-Ub were visualized. Although Se-Met and native crystals were obtained under similar conditions and resulted in solutions containing two vOTU-Ub molecules in the asymmetric unit, their differing space group symmetries underline the fact that the molecular arrangement of the dimer and the crystal lattice differ radically. To further investigate the oligomeric state of vOTU-Ub and vOTU, static light scattering was performed for vOTU and vOTU-Ub. For vOTU, 95.7% of the sample's molecular mass was calculated at 18.7 kDa with a 2% error, confirming that vOTU is a monomer in solution. As for vOTU-Ub, 72.5% of the sample possessed a monomeric complex molecular mass of 27.5 kDa with a 1% error. The remaining 25% of the sample was found to possess a molecular weight of 54.2 kDa with a 1% error. Overall, the results coupled with the differences in lattice arrangement suggest that the most probable biologic assembly is that of one vOTU interacting with one Ub.

FIG. 3.

I222 asymmetric unit of vOTU-Ub crystal and vOTU monomer alone. (a) Cartoon representation of the I222 asymmetric unit of vOTU-Ub. Ub is rendered in light green, with vOTU rendered according to its secondary structures: helices in light blue, β strands in orange, and loops in blue. (b) vOTU monomer labeled according to secondary structure.

Comparison of vOTU to other OTU superfamily members.

The overall structure of the Ub in vOTU-Ub is consistent with previous Ub structures; however, despite vOTU's structure sharing many tertiary elements with other structurally elucidated OTUs, several divergent regions are readily apparent (10, 27). In keeping with the OTU superfamily, vOTU's core is composed of a β sheet surrounded by helices. In vOTU's case, five α helices and a 3₁₀ helix flank the core β sheet on one side, with two 3₁₀ helices on the other (Fig. 3b). Although vOTU possesses a β-sheet core with surrounding helices, similarities with other OTU superfamily members end there. Unlike previous yOTU1 and OTUB1/2 structures that possessed a core formed from only five to six β strands, vOTU's β sheet is comprised of seven β strands arranged as follows: ↓ β2, ↑ β6, ↑ β5, ↑ β3, ↓ β4, ↑ β1a, and ↑ β1 (Fig. 3b) (12, 27, 30). An additional β strand, β2a, forming an anti-parallel β sheet with a β strand formed from Ub residues 73 to 75, was also observed (Fig. 3b).

A comparison of vOTU directly to two other OTUs, yOTU1, and OTUB2, highlights the structural uniqueness of vOTU. By employing secondary structure matching to align the structures of yOTU1, OTUB2, and vOTU, four significant structurally divergent regions can be identified (Fig. 4 a and b). Region 1 concerns the arrangements of β strands that form the core of each structure. Within that region, vOTU has a two-β-strand extension of its β sheet formed from the presence of β1 and β1a. There is no equivalent to vOTU's β4, β1, and β1a in OTUB2, whereas yOTU1 does have a β1, but this β1 forms against β2 on the other end of the β-sheet core. The existence of these extra β strands in vOTU requires that vOTU's polypeptide pass behind the β sheet, forming a 3₁₀ helix prior to forming β2. This loop fills in region 2, which is normally occupied by a conserved α6 helix in all known OTU superfamily members. The presence of β1 and β1a also sterically forbids vOTU from binding Ub in the same orientation as yOTU1 does (Fig. 4c and d). The result is a 30° twist of vOTU's bound Ub in comparison to that of the yOTU-Ub structure. These differences in orientation of Ub or its absence in relationship to the core β sheet of vOTU, yOTU1, and OTUB2 result in differences in the orientation of α3, which comprises region 3 and is a component of the Ub binding interface. Last, region 4 highlights the absence of α helices present in OTUB1/2 but absent in yOTU1 and vOTU.

FIG. 4.

vOTU and vOTU-Ub comparison to OTUB2, yOTU1, and yOTU1-Ub. (a) Divergent-eyed stereo view of vOTU (orange and blue) aligned with yOTU1 (gray) using secondary structure matching though Coot. (b) Divergent-eyed stereo view of vOTU colored as in panel a aligned with hOTUB2 (magenta). (c) Divergent-eyed stereo view of vOTU-Ub with Ub sourced from a secondary structure alignment of vOTU-Ub and yOTU1-Ub. vOTU is rendered as a cartoon and colored according to secondary structure, with helices in light blue, β strands in orange, and loops in blue. Ub of the vOTU-Ub complex is also rendered as a cartoon and colored light green. Ub from a yOTU1-Ub that was aligned with vOTU-Ub using the secondary structure matching of vOTU and yOTU1 is rendered as a cartoon in red. (d) Shown is a 180° y-axis rotation of panel c.

vOTU binding interface with Ub.

The 1,065 Ų of buried surface area between vOTU and Ub can be optimally described by its classification into three areas (Fig. 5 a). One of the two most prominent areas of interaction, area I, is comprised of the interface between Ub residues 72 to 75, including the propylamine adduct of Ub and numerous residues of vOTU (Fig. 5b). Four major factors drive the formation of this interface. The first of these is the formation of hydrogen (H) bonds between the main chain atoms of Ub residues 72 to 75, including the amine in the propylamine adduct, and the main chain atoms of vOTU's W99, G100, S101, and G150. The overall result of these H bonds is the creation of a β sheet joining vOTU and Ub. In addition to H-bond formation, a network of polar bonds and salt bridges is formed between Ub's R72 and R74 and vOTU's D98, E78, and S101. Beyond electrostatic interactions and despite the distinctly negative electropotential of area I, a row of vOTU residues, I118, I131, V18, and V12, forms a hydrophobic trough deep in vOTU's binding surface. Part of this trough resides in area I and facilitates a favorable interaction with Ub's L73. The hydrophobic trough continues into another prominent area of interaction, area II, with the accommodation of Ub's V70 and I44. Similar to area 1, a network of polar bonds and salt bridges is formed between Ub's E51, Q49, and R42 and vOTU's R80, E78, P77, and Q16 (Fig. 5c) in area II. Just as the vOTU hydrophobic trough creates favorable interactions in areas I and II, area III also benefits from Ub's L8 being buried in this trough between vOTU's V18 and I131. Also, a polar interaction is observed in area III by Ub's K6 interacting with vOTU's N20 through a water molecule. Overall, 25 residues from vOTU and 18 residues from Ub are involved in the formation of the interface. Not surprisingly, the 18 residues of Ub involved are isolated to one side of the Ub. However, a comparison to Ub bound to yOTU1 illustrates that the Ub surface that vOTU interacts with is not the same as that of yOTU1.

FIG. 5.

vOTU-Ub binding interface. (a) Electropotential surface rendering of vOTU generated with the adaptive Poisson-Boltzmann solver (APBS) plug-in of Pymol software (5, 36-38). APBS settings were −4 for the negative maximum and 4 for the positive maximum. Ub is illustrated as a light-green transparent cartoon. Ub residues that reside within 4 Å of vOTU as determined by the use of the Contact program part of the CCP4 suite are colored tan (9). Three colored ovals highlight the three regions of interactions between vOTU and Ub: area I (cyan), area II (yellow), and area III (green). (b) Close-up of area I with vOTU rendered in cartoon and stick format with color according to the secondary structure: helices in light blue, β strands in orange, and loops in blue. Ub is also rendered in cartoon and stick format but is colored in green. A cartoon view of the secondary structure of residues 70 to 75 in Ub has been hidden for clarity. White labels indicate Ub residues, black labels indicate vOTU, and pink labels and dashes indicate distances. All distance numbers are in angstroms. (c and d) Close-up of area II and area III, respectively, with vOTU and Ub colored as in panel d. (e) Surface rendering of a 180° x/y-axis rotation of the Ub bound in panel a. In general, Ub is colored light green, with the regions colored tan in panel a colored the same in panel b. (f) Surface rendering of Ub from yOTU1-Ub in a similar orientation as Ub in panel b. Ub is colored in red, with residues within 4 Å of yOTU1, as determined by the use of the Contact program from the CCP4 suite (9), colored yellow.

vOTU's catalytic triad.

As vOTU was predicted to be a cysteine protease and only possesses one cysteine (C40), research groups were quick to identify it as critical to vOTU's protease activity. However, identification of the remaining two residues involved in vOTU's catalytic triad has remained elusive. Recent studies have proposed H151 of vOTU as a member of the vOTU catalytic triad but have only reported on activity of the C40/H151 double mutant (14). Additionally, the presence of a histidine at the equivalent position of vOTU's E98 in the proposed viral OTU subclass arterivirus member's NSP2 proteases has also spurred speculation on the location of the catalytic histidine within the viral OTU subclass (14, 39). As for the aspartic acid component of the catalytic triad, D37 and D153 have been suggested as candidates (11). To clarify the identity of the complete catalytic triad, the active site of vOTU was inspected. Although several vOTU histidine and aspartic acid residues, including H43 and D37, were within the general area of that catalytic cysteine, only H151a and D153 were in close enough proximity and in the correct orientation to assist in the deprotonation of vOTU's C40 (Fig. 6 a). To confirm the essentiality of these amino acids, they were mutated to alanine, and the activity of the resulting protein containing vOTU was assessed. vOTU mutants C40A and W99A exhibited an ∼1,000-fold loss of V_max compared to that for wild-type (wt) vOTU. The vOTU mutant H151A possessed only a slightly higher V_max than that of vOTU mutants C40A and W99A, with ∼350-fold loss of activity compared to that of the wt. The most active vOTU mutant, D153A, was still ∼40-fold less active than wt vOTU. All of the vOTU mutants exhibited a K_m of ∼4-fold more than that for wt vOTU (Fig. 6b).

FIG. 6.

Active site of vOTU. (a) Cartoon and stick rendering of vOTU's active site. vOTU is colored according to its secondary structures, with helices in light blue, β strands in orange, and loops in blue. (b) Mono-Ub K_m and V_max constants for catalytic triad vOTU mutants.

vOTU's K63- and K48-linked deubiquitinating activity.

To investigate vOTU's activity toward both K63- and K48-linked di-Ub species, both K63- and K48-linked species of di-Ub were acquired. Cleavage of 10 μM di-Ub species by 10 nM vOTU was monitored over the course of 1 h by their products being resolved on a 10 to 20% Ready Gel Tris-Tricine gel and visualized by staining with Coomassie blue (Fig. 7 a). Interestingly, there is no difference in vOTU's activity toward the two di-Ub substrates. To assess whether larger poly-Ub species could also be cleaved by vOTU, K48- and K63-linked tetra-Ub was obtained. These tetra-Ub species were cleaved in the same manner as their di-Ub counterparts. Unlike the comparable activity between K48- and K63-linked di-Ub species, vOTU appears to have a slight preference for the two tetra-Ub substrates. For K63-linked tetra-Ub, almost no poly-Ub species are visible after 10 min, whereas several poly-Ub species appear at this time frame, and faint bands can also be seen at the 60-min time frame in the K48-linked tetra-Ub reactions (Fig. 7a).

FIG. 7.

Polydeubiquitination as well as Poly-Ub and vOTU-ISG15 models. (a) vOTU's cleavage of K48-linked di-Ub and K63-linked di-Ub (top panel) and K48-linked tetra-Ub and K63-linked tetra-Ub (bottom panel). (b) Location of lysine residues on vOTU-bound Ub. The surface of vOTU is colored blue, with Ub shown as a cartoon in light green. Lysine residues located in Ub are colored magenta and labeled accordingly. (c) ISG15 (2JF5) aligned using a secondary structure-matching tool in Coot to the Ub bound to vOTU(3). ISG15 is rendered as a cartoon in orange. (d) K63-linked di-Ub (3H7P) and K48-linked di-Ub (2BGF) aligned using a secondary structure-matching tool in Coot to the Ub bound to vOTU(3). K63-linked di-Ub rendered as a cartoon in dark blue, with K48-linked di-Ub in pink (42, 44).

DISCUSSION

vOTU and other viral OTU superfamily members.

In general, viral genomes undergo a higher rate of mutation and recombination than their eukaryotic and prokaryotic counterparts. The CCHF virus is no different. Interestingly, the L protein tends to be the most conserved among those in the three-segmented genome, suggesting an evolutionary need to minimize substantive changes in the L protein (3). For the L protein itself, the RNA-dependent RNA polymerase's variable region (amino acids 752 to 851) tends to be the source of most L-protein diversity (26). Intriguingly, this leaves the vOTU-containing region of the L protein only a few mutations between GenBank entries from strains from Kosovo (ABW17160), Turkey (ACT88368), and China (ADD64466) (see entry for vOTU, AAQ98866.2). Of the 16 polymorphic sites in vOTUs from different CCHF strains, only three are located at the vOTU-Ub interface: A14D, R80K, and E128G. As vOTU's R80 is involved in a salt bridge with Ub's D98, the change of arginine to a shorter lysine residue may impact binding (Fig. 5b). Unlike the case for the R80K polymorphism, any impact of A14D and E128G on protease activity and its effects on virulence are hard to predict, as these residues have no readily apparent side chain interactions in vOTU-Ub (Fig. 5c and d). Nairovirus homologues of Dugbe and Nairobi viruses are also very similar, 72 to 76%, respectively, to the CCHF virus vOTU. Most of the differences between the nairoviruses lay outside the binding interfaces. However, as with other strains of vOTU, differences can be found at R80 and E128 might suggest that even closely related nairovirus homologues might not have the exact same protease specificity or catalytic capacity.

Although vOTU and its nairovirus homologues are highly conserved, other vOTU homologues from the arterivirus genus, including PRRSV, show only less than 40% similarity with vOTU. Despite the low sequence similarity in general, the elucidation of the vOTU structure coupled with the analysis of an arterivirus PRRSV amino acid sequence allows some insight into possible tertiary features of arterivirus vOTUs. One striking example is that regions of higher conservation between vOTU and the PRRSV vOTU homologues are at the same amino acid positions as those involved in substrate binding. This suggests that arterivirus vOTUs may possess a substrate binding interface similar to that of PRRSV (Fig. 1a). In addition, arterivirus vOTUs have a high degree of sequence similarity with the region containing the catalytically relevant cysteine as well as a shared amino acid motif among OTU superfamily members HXD/N, where X is a 5- to 6-member side chain ring containing residues. Since this motif now contains the catalytically relevant histidine and aspartic acids in a viral OTU superfamily member, there is a strong likelihood that the arterivirus vOTUs, catalytic triad residues also reside in this motif.

Potential vOTU primary and tertiary structural determinates for substrate recognition.

yOTU1 shares only 40% of the amino acids positions that comprise the vOTU-Ub binding interface. This, along with the significant difference in the orientation of Ub between the vOTU-Ub and yOTU1-Ub structures, underlines the potential diversity of substrate recognition within the OTU superfamily. Although only through an intensive mutagenesis approach will the complete understanding of vOTU substrate recognition be ultimately known, some likely candidates can be envisioned. vOTU's R80, E51, and N20 are some of these candidates. As a methionine or isoleucine in OTUB2 and yOTU1, respectively, R80 forms a salt bridge with E51 in Ub. Interestingly, E51 is also a glutamate in ISG15 but is a glutamine in NEDD8. In the case of vOTU's E78, this residue creates two salt bridges with R72 and R74 of Ub. As R74 of Ub is a leucine in NEDD8 and E78 of vOTU is a histidine in OTUB2, the presence of both of these salt bridges is not likely in vOTU-NEDD8 or in OTUB2-Ub. Last, N20 is located on vOTU's unique β1-β1a tertiary element and forms a set of polar interactions with Ub's K9 through a water molecule. This interaction should be absent completely in OTUB1/2 and yOTU1.

In addition to vOTU's N20 proposed role in mono-Ub and Ub-like recognition, its interaction with K6 of Ub suggests that vOTU may not be able to cleave K6 linked poly-Ub conjugates. Analysis of other lysine side chains in the vOTU-Ub structures also suggests that through steric hindrance, vOTU may not prefer K27- or K11-linked poly-Ub (Fig. 7b). These side chains are buried within the vOTU-Ub interface. When vOTU-Ub is used as a scaffold for modeling vOTU-ISG15, vOTU-K63-linked di-Ub, and vOTU-K48-linked di-Ub, the results would suggest that vOTU might also have difficulty also cleaving K48-linked poly-Ub (Fig. 7b, c, and d). Within the vOTU-Ub structure, K48 is located near vOTU's surface, and when the K48 di-Ub NMR structure (2BGF) is overlaid, significant steric clashes are observed. Interestingly, vOTU has only a slight preference for K63-linked tetra-Ub over its K48-linked counterpart. Of course, this is in contrast to the slight preference that Otu1 shows for K48-linked poly-Ub over K63-linked poly-Ub (27). Additionally, several di-Ub products are observed when vOTU cleaves K48-linked tetra-Ub, suggesting that K48-linked poly-Ub can adopt a conformation that allows vOTU access to its isopeptide bonds. vOTU's robust activity toward K63-linked poly-Ub is also intriguing, as a cysteine protease from murine cytomegalovirus shows a more distinct preference for K48-linked di-Ub over K63-linked di-Ub (35). vOTUs cleavage of K63-linked poly-Ub protein conjugates with rates slightly greater than K48-linked tetra-Ub might be reflective of the K63-linked poly-Ub quaternary structure being similar to that of ISG15 (Fig. 7c and d). With K63-linked poly-Ub and ISG15 activating and/or stabilizing key antiviral proteins, and K48-linked poly-Ub's proteasome-vectoring attributes, vOTU activity likely has a significant impact on these cellular activities.