Macro Domain from Middle East Respiratory Syndrome Coronavirus (MERS-CoV) Is an Efficient ADP-ribose Binding Module

The newly emerging Middle East respiratory syndrome coronavirus (MERS-CoV) encodes the conserved macro domain within non-structural protein 3. However, the precise biochemical function and structure of the macro domain is unclear. Using differential scanning fluorimetry and isothermal titration calorimetry, we characterized the MERS-CoV macro domain as a more efficient adenosine diphosphate (ADP)-ribose binding module than macro domains from other CoVs. Furthermore, the crystal structure of the MERS-CoV macro domain was determined at 1.43-Å resolution in complex with ADP-ribose. Comparison of macro domains from MERS-CoV and other human CoVs revealed structural differences in the α1 helix alters how the conserved Asp-20 interacts with ADP-ribose and may explain the efficient binding of the MERS-CoV macro domain to ADP-ribose. This study provides structural and biophysical bases to further evaluate the role of the MERS-CoV macro domain in the host response via ADP-ribose binding but also as a potential target for drug design.

Since the severe acute respiratory syndrome (SARS) 2 outbreak in 2003 (1, 2), a newly discovered disease, Middle East respiratory syndrome (MERS), has been spreading from countries in the Middle East to America (3)(4)(5). In the summer of 2015, MERS was reported in North East Asia (6 -8). The causative agent of MERS was identified as an unknown coronavirus (CoV) resembling SARS-CoV and referred to as Middle East respiratory syndrome CoV (MERS-CoV) (9 -12). MERS-CoV belongs to the genus Betacoronavirus and possesses a positivestrand RNA genome that encodes viral proteins essential to the life cycle of the virus (13,14). The mortality of MERS is 4-fold higher than SARS (40% compared with 10%) (15). Since the first case report in Saudi Arabia, MERS has been reported in more than 20 countries and has caused more than 400 deaths worldwide (9).
CoVs utilize the RNA genome to encode structural proteins, including spike glycoprotein (S), membrane protein (M), and nucleocapsid protein (N). They encode a large number of nonstructural proteins (NSPs) for rapid replication. A single large replicase gene encodes all proteins involved in viral replication. The replicase gene contains two open reading frames (ORFs), ORF1a and ORF1b, which encode two polyproteins, pp1a and pp1ab; production of pp1ab requires a ribosomal frameshift to transcribe the portion encoded by ORF1b (16). ORF1a encodes viral proteases, main protease (M pro , also called 3CL pro ), and papain-like protease (PL pro ), which are responsible for cleavage of the ORF1a and ORF1b gene products to produce functional NSPs.
In SARS-CoV, the largest NSP member, NSP3, is a multidomain protein containing the following domains: N-terminal acidic domain, macro domain, SARS-unique domain, PL pro , nucleic acid-binding domain, marker domain (G 2 M), transmembrane domain, and Y-domain (17). The MERS-CoV genome contains 16 NSPs (Fig. 1); except for 3CL pro and PL pro (18,19), most of the functional domains within the NSP3 in MERS-CoV remain structurally uncharacterized.
The macro domain is named after the non-histone motif of the histone variant macroH2A, in which it was originally characterized (20 -22), a protein module ubiquitous in eukaryotes, bacteria, and archaea. This domain is well known for its affinity to adenosine diphosphate (ADP)-ribose (23)(24)(25). Many cellular enzymes bearing macro domains within their structures interact with poly(ADP)-ribose (26 -29). Poly(ADP)-ribosylation is a post-translational modification linked with DNA repair, apoptosis, gene regulation, and protein degradation. Thus, macro domain-containing proteins and enzymes may play important roles in regulating various cellular processes (30). Surprisingly, the CoVs studied so far and a few other viruses such as alphavirus, rubella virus, and hepatitis E virus possess macro domains in their genomes (16). In addition, some viral macro domains were found to have ADP-ribose 1Љ-phosphate phosphatase (ADRP) activity (31)(32)(33), which catalyzes the removal of phosphate from ADP-ribose 1Љ-phosphate (Appr1p) to produce ADP-ribose. ADRP activity has been reported in a yeast protein containing macro domain as well as AF1521 protein in Archaeoglobus fulgidus (23,34,35). The enzymatic activity of viral macro domains in processing Appr1p is low (33, 36 -38) and appears to be dispensable for virus RNA synthesis (31). In addition, the mutant for the CoV mouse hepatitis virus A59 (MHV-A59), encoding a single amino acid substitution of a strictly conserved residue for ADRP activity, replicated to slightly reduced titers in mouse liver but, strikingly, did not induce liver disease (39). The MHV macro domain exacerbates MHV-induced liver pathology, most likely by inducing excessive inflammatory cytokine expression. It was also reported that catalytic residues Asn-809, His-812, Gly-816, and Gly-817 for ADRP activity in hepatitis E virus macro domain are critical for hepatitis E virus replication (40). Accordingly, the development of drugs targeting the viral macro domain may be a strategy for antiviral therapy.
The macro domain of SARS-CoV NSP3 was previously reported to possess ADP-ribose and poly(ADP)-ribose binding ability, which suggests that the macro domain may regulate cellular proteins involved in an apoptotic pathway via poly-(ADP)-ribosylation to mediate the host response to infection (36). Structural studies of macro domains from CoVs such as human CoV 229E (HCoV-229E) and feline CoV (FCoV) also revealed interactions with ADP-ribose (41)(42)(43) and have offered huge advances in our understanding of viral macro domains. The MERS-CoV genome features a macro domain embedded in NSP3 (Fig. 1). However, we lack structural and functional information regarding the MERS-CoV macro domain.
In the present study, we investigated the MERS-CoV macro domain as an ADP-ribose binding module, with comparison to previously characterized viral macro domains. Furthermore, we determined the crystal structure of the MERS-CoV macro domain in complex with ADP-ribose. Structural comparison of MERS-CoV and other human CoVs revealed divergence in ADP-ribose binding by macro domains. Our study may shed new light on structurally based design of novel antiviral drugs targeting viral macro domains.

Experimental Procedures
Protein Expression and Purification-The DNA sequence containing the MERS-CoV macro domain was synthesized by a local biotechnology company (MDBio, Inc.) and cloned into the pUC57 plasmid. The macro domain fragment was inserted between the NdeI and XhoI sites of the pET28a vector system (Novagen). The forward and reverse PCR primers used for amplification were macro-F (5Ј-AATTCATATGCCACTGA-GCAATTTTGAACA-3Ј) and macro-R (5Ј-AATTCTCGAGT-TAGATGGTCAGGCTCTTATAC-3Ј). The resulting plasmid with the inserted sequence was transformed into Escherichia coli BL21(DE3) cells, which were grown at 37°C up to A 600 1.0 with 50 g/ml of kanamycin. The expression of the recombinant MERS-CoV macro domain with an His tag at the N terminus was induced in cells with 1 mM isopropyl ␤-D-thiogalactoside, followed by growth for 20 h at 16°C. Cells were collected by centrifugation and resuspended in lysis buffer (25 mM phosphate buffer, pH 7.0, 100 mM NaCl). After 20 min of sonication, the cell extract was clarified by centrifugation at 18,900 ϫ g for 30 min at 4°C to remove debris. The clear supernatant was placed in an open column filled with nickel-nitrilotriacetic acid resin. The resin was washed with 10 times volume of lysis buffer containing 50 and 100 mM imidazole, respectively. The Histagged MERS-CoV macro domain was eluted by lysis buffer containing 300 mM imidazole. The purified MERS-CoV macro domain was dialyzed against stabilization buffer (25 mM phosphate buffer, pH 7.0, 100 mM NaCl, 0.5 mM dithiothreitol). The His tag was removed by using thrombin, which resulted in four additional residues (GSHM) at the N terminus. The protein was further purified by gel filtration chromatography with a Super-dex75 XK 16/60 column (GE Healthcare) in 20 mM Tris-HCl buffer (pH 7.0), 100 mM NaCl.
Circular Dichroism (CD) Spectroscopy-Far-UV CD spectra were measured with 10 M protein samples in CD buffer (20 mM phosphate buffer, pH 3.5-8.5) placed into a 1-mm path length cuvette and recorded on a JASCO J-810 spectropolarimeter equipped with a Peltier temperature control system (JASCO International Co.). Thermal transition of protein samples with or without preincubation of 1 mM ADP-ribose were monitored at 220 nm from 25 to 95°C at a scan rate of 1°C/min. Baseline subtraction, smoothing, and data normalization involved the use of SigmaPlot. The melting temperature (T m ) was calculated with the maximum of the first derivative of the CD signal.
Differential Scanning Fluorimetry (DSF)-Thermal shift assay with DSF involved use of a CFX48 Real-time PCR Detection System (Bio-Rad). In total, a 25-l mixture containing 2 l of SYPRO Orange (Sigma), 1.25 l of dialysis buffer (20 mM Tris-HCl, and 100 mM NaCl, pH 7.0), 10 l of 1 M protein sample, and various concentrations of ADP-ribose were mixed on ice in an 8-well PCR tube. Fluorescent signals were measured from 25 to 95°C in 0.1°C/30-s steps (excitation, 450 -490 nm; detection, 560 -580 nm). The main measurements were carried out in triplicate. Data evaluation and T m determination involved use of the Bio-Rad CFX Manager, and data fitting and dissociation constant (K d ) calculations involved the use of SigmaPlot.
Isothermal Titration Calorimetry (ITC)-Binding of ADP-ribose to the MERS-CoV macro domain was measured by ITC with the Nano Isothermal Titration Calorimeter (TA Instruments). Aliquots of 3 l of 1.14 mM ADP-ribose were titrated by injection into protein (0.057 mM in 0.98 ml) in 20 mM Tris-HCl (pH 7.0) and 100 mM NaCl. Experiments were carried out at 25°C with 250 rpm stirring. Background heat from ligand to buffer titrations was subtracted, and the corrected heat from the binding reaction was used to derive values for the stoichiometry of the binding (n), K d , apparent enthalpy of binding (⌬H), and entropy change (⌬S). Data were fitted by use of an independent binding model with Launch NanoAnalyze version 2.3.6.
Crystallization and Data Collection-The MERS-CoV macro domain and ADP-ribose were mixed in a molar ratio of 1:15. Initial protein crystallization trials were performed at 283 K by the sitting-drop vapor-diffusion method with commercial crystallization screen kits, 96-well Intelli-plates (Art Robbins Instruments), and a HoneyBee 963 robot (Genomic Solutions). Each crystallization drop was prepared by mixing 0.3 l of macro domain/ADP-ribose at 10 mg/ml with an equal volume of mother liquor, and the mixture was equilibrated against 100 l of reservoir solution. The crystals for data collection were grown in 1 week at 283 K with the optimal condition of 100 mM phosphate/citrate (pH 4.2), 2.0 M ammonium sulfate, and 10 mM nicotinamide adenine dinucleotide as the additive. For subsequent anomalous phasing, the crystal was soaked for 8 h in 3 mM mercuric(II) chloride, cryoprotected in mother liquor sup-plemented with 20% glycerol, and flash-frozen in liquid nitrogen at 100 K. The diffraction images were recorded in a 100-K nitrogen gas stream with use of BL13B1 or BL13C1 beamlines (National Synchrotron Radiation Research Center, Taiwan) and processed by using HKL2000 software (44).
Structure Determination and Refinement-The crystal structure of the MERS-CoV macro domain in complex with ADPribose was solved by the mercury(II) derivative single-wavelength anomalous dispersion method by using SHELXD/ SHELXE software (45). The initial model was refined by the maximum likelihood method implemented in REFMAC5 (46) as part of the CCP4 suite (47) and rebuilt interactively by inspecting the -weighted electron density maps with coefficients 2mF o Ϫ DF c and mF o Ϫ DF c in COOT (48). During the later stages, restrained positional and B-factor refinement involved the program phenix.refine (49). Water molecules were manually added at the final stages. The models were evaluated with use of PROCHECK (50) and MOLPROBITY (51). The data collection and structure refinement statistics are in Table 1.

ADP-ribose Binding Ability of MERS-CoV Macro Domain-
The MERS-CoV macro domain (pp1a residues 1110 to 1273) was expressed and purified from E. coli. The final purified pro- tions of symmetry related reflections of h. c CC 1/2 is a percentage of correlation between intensities from random half-datasets (56). d CC ano is a percentage of correlation between random half-datasets of anomalous intensity differences. e R work /R free ϭ ⌺͉F obs Ϫ F calc ͉/⌺F obs , where F calc is the calculated protein structure factor from the atomic model (R free was calculated with 5% of the reflections selected).
tein was a 167-amino acid protein (20 kDa), with four additional residues at the N terminus resulting from removal of the hexahistidine tag after thrombin cleavage. CD spectra revealed that the macro domain exhibited a stable ␣/␤-type folding pattern under various pH conditions (Fig. 2). The T m of the macro domain from thermal transition monitored by CD was 43°C.
However, the addition of ADP-ribose significantly increased the T m to 51°C (Fig. 3A). The significant increase in T m suggests the interaction between the MERS-CoV macro domain and ADP-ribose.
To understand the affinity of ADP-ribose binding to the MERS-CoV macro domain, both DSF and ITC measurements were used to examine the equilibrium dissociation constant (K d ) of ADP-ribose. After fitting DSF data, the K d was determined to be 3.12 Ϯ 0.42 M (r 2 ϭ 0.9628) (Fig. 3B), which is similar to the calculated K d of 2.95 M based on ITC data (Fig.  3C). In addition, ITC data indicated that ADP-ribose bound to the MERS-CoV macro domain with favorable enthalpy change (exothermic, ⌬H ϭ Ϫ91.04 KJ/mol). The binding reaction was spontaneous at 25°C with exergonic Gibbs energy of binding (⌬G ϭ Ϫ31.56 KJ/mol). The thermodynamic profile (⌬G Ͻ 0, ⌬H Ͻ 0, and ϪT⌬S Ͼ 0) of ADP-ribose binding to the MERS-CoV macro domain suggests that ADP-ribose is likely stabilized by hydrogen bond formations (52).
We reviewed the results of previously reported binding assays of ADP-ribose binding to CoV macro domains ( Table 2). Overall Structure of MERS-CoV Macro Domain in Complex with ADP-ribose-We determined the crystal structure of ADP-ribose-bound MERS-CoV macro domain for further molecular elucidation. The orthorhombic crystals gave good quality x-ray diffraction and belonged to the space group C222 1 with the following unit cell dimensions: a ϭ 41.798 Å, b ϭ 120.807 Å, c ϭ 67.659 Å, and ␣ ϭ ␤ ϭ ␥ ϭ 90°. The structure of the MERS-CoV macro domain was solved by mercury singlewavelength anomalous dispersion (see "Experimental Procedures"). The final protein structure (Fig. 4A) was refined to 1.43-Å resolution with R-factor and R-free values of 0.1273 and 0.1619, respectively ( Table 1). The core of the structure of MERS-CoV macro domain is a seven-stranded ␤-sheet in the order of ␤1-␤2-␤7-␤6-␤3-␤5-␤4 (Fig. 4B). The central ␤-sheet is sandwiched between six ␣-helices, with ␣1, ␣2, and ␣3 packing onto one face and ␣4, ␣5, and ␣6 onto the other. In the initial refinement cycle, a strong bent electron density (continuous at 1 cutoff) (Fig. 4A) located at the central pocket, was unambiguously identified as an ADP-ribose molecule. This molecule is tightly bound in an uncharged crevice located at the C-terminal end of strands ␤3 and ␤6 in the loop regions between ␤3-␣2 and ␤6-␣5 (Fig. 4A).
A search of the DALI database (53) with the structure of the MERS-CoV macro domain in complex with ADP-ribose used as a model revealed several structural homologs. Top-ranked structures were macro domains of CoVs in complex with ADPribose such as those for SARS-CoV (PDB code 2FAV; Z score 27.9; r.m.s. deviation 1.3; sequence identity 45%; sequence similarity 65%) (36), HCoV-229E (PDB code 3EWR; Z score 22.8; r.m.s. deviation 1.8; sequence identity 33%; sequence similarity 56%) (43), FCoV (PDB code 3JZT, Z score 22.6; r.m.s. deviation 1.8; sequence identity 30%; sequence similarity 53%) (42), and infectious bronchitis virus (PDB code 3EWP; Z score 19.5; r.m.s. deviation 1.9; sequence identity 28%; sequence similarity 47%) (43). This finding reflects that the viral macro domains are structurally well conserved. However, variability between all these structures arises from the loops connecting the core secondary structure elements, which display great diversity in sequence, length, and conformation and may correspond to different ADP-ribose binding ability.

Molecular Basis of ADP-ribose Binding in MERS-CoV Macro
Domain-To gain insights into the molecular mechanism of ADP-ribose binding, we further investigated the binding pocket for ADP-ribose in the MERS-CoV macro domain. The adenine moiety resides in the hydrophobic cavity containing Gly-19, Ala-21, Ile-47, Pro-123, Leu-124, and Val-152 (Fig. 5A). Coordination of ADP-ribose involves serial hydrogen bond formations and hydrophobic interactions provided by surrounding amino acid residues (Fig. 5B). The side chain of Asp-20 contacts the N-6 atom of the pyrimidine ring in adenine moiety via direct hydrogen bonding. This residue is critical for binding specificity of the macro domain AF1521 in A. fulgidus (54). Structure-based multiple sequence alignment showed that this aspartic acid is conserved among CoV macro domains (Fig. 6A). Oxygen atoms of the pyrophosphate in ADP-ribose contact surrounding residues via hydrogen bonding with nitrogen atoms in backbone amides of Ile-47, Ser-126, Gly-128, Ile-129, and Phe-130. The second ribose is stabilized by complex hydrogen bonding with surrounding residues and water molecules (Fig. 5A). The ribose-3Љ oxygen atom forms a hydrogen bond with a nitrogen atom in the side chain of Asn-38. The ribose-2Љ oxygen atom forms hydrogen bonds with the oxygen and nitrogen atoms in the backbone amides of Lys-42 and Gly-44, respectively. The ribose-1Љ oxygen atom forms a hydrogen bond with the nitrogen atom in the backbone amide of Gly-46. A water molecule serves as a bridge between the ribose-1Љ oxygen atom, Asn-38, and His-43. This organization of the terminal ribose and surrounding molecules was also observed in the yeast ADRP enzyme (34), which suggests that Asn-38 and His-43 may be critical for the hydrolysis reaction of ADP-ribose 1Љ-phosphate to ADP-ribose. In addition, equivalent residues critical for ADRP activity in the SARS-CoV macro domain (36)  HCoV-229E macro domains shows that the major structural divergence lies in the ␣1 helices, which participates in stabili-  Pull-down based binding assay    tures. As compared with Asp-20 in MERS-CoV and Asp-23 in SARS-CoV, the equivalent residue in HCoV-229E is Asp-19, which does not contact ADP-ribose. Instead of forming a hydrogen bond directly with ADP-ribose, the side chain of Asp-19 in HCoV-229E contacts Thr-22 in the ␣1 helix via hydrogen bonding with oxygen and nitrogen atoms in the side chain and backbone of Thr-22, respectively. Hydrogen bonding with Thr-22 drags the side chain of Asp-19 in the HCoV-229E macro domain away from the adenine cavity as compared with the position of Asp-20 in the MERS-CoV macro domain (Fig.  6D). Consistent with the previous study, the thermodynamic profile (⌬G Ͻ 0, ⌬H Ͻ 0, and ϪT⌬S Ͻ 0) of ADP-ribose binding to the HCoV-229E macro domain suggests less contribution of the hydrogen bond to stabilization of ADP-ribose (41) ( Table  2). Variations in strength of the hydrogen bond and orientation of the side chain in Asp residues may result in differential binding affinities of ADP-ribose observed in macro domains of MERS-CoV (K d 2.95 M), SARS-CoV (K d 24 M) (36), and HCoV-229E (K d 28.9 M) (41). The relationship between binding affinities of ADP-ribose in macro domains and differential pathogenicity of human CoVs needs further investigation. Conclusion-Taken together, our biochemical study shows higher binding affinity for ADP-ribose in the MERS-CoV macro domain than macro domains of CoVs characterized to date. Structural analysis revealed that differences in the context of hydrogen bonds formed by the conserved Asp with ADPribose and residues in ␣1 helices in macro domains of MERS-CoV, SARS-CoV, and HCoV-229E may result in differential binding affinities for ADP-ribose. Our studies provide a biochemical basis for further investigating the role of macro domain in MERS-CoV infection and also the precise structural information for the design of novel antiviral drugs.