Expression, purification and crystallization of the SARS-CoV macro domain

Macro domains or X domains are found as modules of multidomain proteins, but can also constitute a protein on their own. Recently, biochemical and structural studies of cellular macro domains have been performed, showing that they are active as ADP-ribose-1′′-phosphatases. Macro domains are also present in a number of positive-stranded RNA viruses, but their precise function in viral replication is still unknown. The major human pathogen severe acute respiratory syndrome coronavirus (SARS-CoV) encodes 16 non-structural proteins (nsps), one of which (nsp3) encompasses a macro domain. The SARS-CoV nsp3 gene region corresponding to amino acids 182–355 has been cloned, expressed in Escherichia coli, purified and crystallized. The crystals belong to space group P2₁, with unit-cell parameters a = 37.5, b = 55.6, c = 108.9 Å, β = 91.4°, and the asymmetric unit contains either two or three molecules. Both native and selenomethionine-labelled crystals diffract to 1.8 Å.

1. Introduction

Coronaviruses have the largest genomes of all positive-stranded RNA viruses. Translation of their replicase genes results in two large polyproteins which are subsequently cleaved into a large number of non-structural proteins (16 for the severe acute respiratory syndrome coronavirus; SARS-CoV) whose function is to ensure genomic and subgenomic RNA synthesis (Lai & Holmes, 2001; Siddell, 1995; Siddell et al., 2005). The particularly large size of the genome probably results from the fact that it encodes for unusual enzymatic activities within the viral world (Snijder et al., 2003). The precise roles of some of these proteins have yet to be identified (Ziebuhr, 2005).

SARS-CoV nsp3 is a large (213 kDa) multidomain protein believed to carry several enzymatic activities involved in the replicative complex, which includes a macro domain, also known as an X domain (Gorbalenya et al., 1991, 2004; Snijder et al., 2003). To date, 330 proteins have been reported to possess such a domain (from the SMART database; Letunic et al., 2004). The macro domain is found in a number of multidomain proteins, including nonstructural proteins of several ssRNA viruses (such as rubella virus, alphaviruses and some coronaviruses), some poly-ADP-ribose polymerases and the C-­terminal domain of macro-H2A histone proteins, but it can also constitute a stand-alone protein by itself, as is the case in bacteria, archaebacteria and eukaryotes (F. Forouhar, I. Lee, S. M. Vorobiev, R. Xiao, T. B. Acton, G. T. Montelione, J. F. Hunt & L. Tong, unpublished work; Allen et al., 2003; Shull et al., 2005; Letunic et al., 2004). This wide distribution suggests an essential and ubiquitous role in cellular processes.

Several studies have recently revealed that macro domains act as phosphatases that remove the 1′′-phosphate group from ADP-ribose-1′′-phosphate (Martzen et al., 1999; Karras et al., 2005; Shull et al., 2005; Putics et al., 2005). The role of this activity, particularly in the viral world, is still puzzling. We are interested in identifying the exact role of viral macro domains by a combination of enzymatic characterization studies and structural studies of the enzyme alone or in complex with biologically relevant molecules. Therefore, we have initiated the structural and functional characterization of Semliki Forest virus, human coronavirus 229E and SARS coronavirus macro domains. During the course of this study, the crystal structure of a slightly different construct of another strain of SARS-CoV (strain Tor2) was solved as an apo enzyme and published (Saikatendu et al., 2005). The Tor2 protein is dimeric, both in solution and in the crystal, and its structure reveals that the active site is occluded by the dimeric association, which explains why soaking experiments with various ligands failed for this particular crystal structure.

Here, we report the cloning, expression, purification, crystallization and preliminary X-ray characterization of a monomeric form of the SARS-CoV strain Frankfurt 1 macro domain (an 18.6 kDa protein consisting of amino acids 182–355 of nsp3), which may enable more successful ligand-soaking experiments and hence provide some insight into the mechanism of viral macro domains.

2. Cloning, expression and purification

As soon as the SARS-CoV genome sequence became available, we performed the annotation of its genome independently using the VaZyMolO (`viral enzyme module localization') tool that had been previously developed in our laboratory. The scope of VaZyMolO is to define within large viral polyproteins viral proteins and protein modules that might be expressed in a soluble and functionally active form (Ferron et al., 2005). Our annotation of the proteins (non-structural, structural and accessory) encoded by the SARS-CoV genome was similar to that of Snijder et al. (2003). In the case of nsp3, we defined six independent domains, including a macro domain. Being located between two poorly characterized domains (namely a glutamic acid-rich acidic domain and a SARS-specific unique domain; SUD), the SARS-CoV macro domain did not have clearly defined borders. Our annotation within the VaZyMolO database was based on multiple criteria, including bioinformatics approaches, examination of the available three-dimensional structures of cellular macro domains and existing biochemical data. The latter suggests that in some viruses the N-terminal border definition may be particularly crucial for detection of ADP-ribose-1′′-phosphatase activity (T. Ahola, personal communication). The coding sequence for this domain (consisting of 174 amino acids, amino acids 182–355 of nsp3, isolate Frankfurt 1, DDB/EMBL/Genbank accession No. AY291315) was amplified from the pMal-SARS-CoV-X plasmid kindly provided by Akos Putics and John Ziebuhr (Putics et al., 2005). Amplification was performed by PCR using two primers containing the attB sites of the Gateway recombination system (Invitrogen). A sequence encoding a hexahistidine tag was added at the 5′ end of the gene. The cDNA was then subcloned in the pDest14 plasmid (Invitrogen). The open reading frame of the final construct was checked by sequencing (MilleGen, Toulouse, France).

Expression was performed in Escherichia coli strain C41pROS, a C41 strain into which we have introduced the pLys plasmid from the Rosetta strain (Novagen). Cultures were grown in TB at 310 K and 250g to mid-exponential phase (an OD₆₀₀ of 0.6). Media were supplemented with 100 mg ml⁻¹ ampicillin to select for E. coli recombinants. Protein expression was induced by the addition of a final concentration of 50 µM isopropyl 1-thio-β-D-galactopyranoside (IPTG) and the cells were incubated for a further 20 h at 298 K and 250 rev min⁻¹. Cells were collected by centrifugation at 5000g for 20 min at 277 K and the bacterial pellets were resuspended and then frozen at 253 K in 50 mM Tris–HCl, 150 mM NaCl, 10 mM imidazole pH 8.0 (10 ml per OD₆₀₀ unit and per litre of culture). Cellular suspensions were thawed, supplemented with 1 mM PMSF, 0.2 mg ml⁻¹ lysozyme, 0.1 µg ml⁻¹ DNase, 20 mM MgSO₄ and protease-cocktail inhibitor (Sigma), lysed by sonication and centrifuged (20 000g) for 40 min at 277 K to produce cell-free extract. All purification steps were performed at 277 K. The supernatant was applied onto a 5 ml bed-volume HiTrap nickel immobilized metal-ion affinity chromatography (IMAC) column (Amersham Biosciences) connected to an FPLC system (Amersham Biosciences). The protein was eluted with 50 mM Tris–HCl, 150 mM NaCl, 250 mM imidazole pH 8.0. Fractions containing the protein were determined by SDS–PAGE, pooled and then applied onto a preparative Superdex 200 gel-filtration column pre-equilibrated in 10 mM MOPS, 150 mM NaCl pH 7.0. Protein was eluted with a retention volume corresponding to a monomeric macro domain. The purity of the protein was evaluated by SDS–PAGE. Protein was concentrated to 8 mg ml⁻¹ using a Vivaspin 10 kDa molecular-weight cutoff centrifugal concentrator (Vivascience). The protein concentration was determined from the molar extinction coefficient of the enzyme at 280 nm (9840 M⁻¹ cm⁻¹) as calculated by the Expasy server (https://www.expasy.org/tools/protparam.html). The purified protein was stored at 277 K. Selenomethionine (SeMet) labelled protein was prepared following standard procedures (Doublié, 1997) and the protein was purified using the same protocol employed for the native protein except that 1 mM TCEP was added to the protein immediately after its elution from the IMAC column and to the gel-filtration buffer. Typical yields were 15–­20 mg of pure protein per litre of culture for both native and SeMet-labelled protein.

3. Crystallization

After checking the protein purity, folding and monodispersity using SDS–PAGE, circular dichroism and dynamic light scattering (Zulauf & D'Arcy, 1992), crystallization of the SARS-CoV macro domain was screened at 292 K by the sitting-drop vapour-diffusion method in 96-­well Greiner crystallization plates which had been automatically filled using a Tecan Genesis robot. As the amount of protein was not a limiting parameter, we used a wide-screen approach using the following commercial kits: Wizard Screens I and II (Emerald Biostructures), Structure Screens I and II (Jancarik & Kim, 1991; Hampton Research), Stura Footprint Screen (Stura et al., 1992; Molecular Dimension Limited) and Natrix Screen (Scott et al., 1995; Hampton Research). Using the 8 mg ml⁻¹ purified protein sample, three drops were set up by a Cartesian robot controlled by the AXSYS software above each reservoir; these drops were made up of 300, 200 and 100 nl of protein solution, respectively, and 100 nl reservoir solution (Sulzenbacher et al., 2002). From these 864 conditions, two hits were identified in the Stura Footprint Screen. A primary optimization of these conditions was performed using home-made solutions, which were dispensed in 64 wells of a Greiner crystallization plate in a 8 × 8 format by the Tecan Genesis robot. The best conditions were then transposed to hanging drops in 24-well tissue-culture Limbro plates. As in the Greiner plates, hundreds of plate-like multiple crystals grew from a single nucleation point overnight. Microseeding in undersaturated equilibrated drops led to single crystals that were suitable for X-ray analysis (Fig. 1). The final conditions for crystallization were as follows: 2 µl protein solution (8 mg ml⁻¹) was mixed with 1 µl reservoir solution (total volume of 0.5 ml) containing 0.1 M imidazole pH 7.9, 1.3 M sodium citrate. Multiple crystals grew overnight. They were crushed and used to microseed a pre-equilibrated drop made of 2 µl protein solution (8 mg ml⁻¹) mixed with 1 µl of a reservoir solution containing 0.1 M imidazole pH 7.8, 0.9 M sodium citrate. These conditions differ from those found by Saikatendu and coworkers (see Table 1), which is not surprising considering that the constructs are quite different (considering the N-­terminal tag, the construct of Saikatendu and coworkers is three amino acids longer at the N-terminal and ten amino acids longer at the C-terminal extremities).

Table 1
Comparison of experimental procedures and results

	Saikatendu et al. (2005)	Present work
Strain	Toronto 2 isolate, gi:34555776	Frankfurt 1 isolate, gi:31581502
Construct	184–365 of nsp3	182–355 of nsp3
No. of amino acids	182	174
N-terminal tag	MGSDKIHHHHHH	MHHHHHH
Protein buffer	25 mM Tris pH 7.8, 150 mM NaCl	10 mM MOPS pH 7.0, 150 mM NaCl
Crystallization conditions	1.85 M sodium malonate pH 7.0	0.1 M imidazole pH 7.9, 1.3 M sodium citrate
Space group	P212121	P21
Unit-cell parameters (Å, °)	a = 76.9, b = 81.2, c = 125.7	a = 37.4, b = 55.4, c = 108.9, β = 91.6
Diffraction limit (Å)	1.4	1.8
Solvent content (%)	51	38.9

Figure 1 Crystals of SARS-CoV macro domain.

4. Data collection and processing

Both native and SeMet crystals were cryoprotected in a solution containing the reservoir solution to which 20% glycerol was added and harvested into a rayon-fibre loop prior to flash-freezing in liquid nitrogen. All data were collected from single crystals at 100 K at beamline ID29 of the European Synchrotron Radiation Facility, Grenoble, France using an ADSC Quantum 210 charge-coupled device detector. Data were processed with MOSFLM (Leslie, 1992) and were reduced and scaled using the SCALA program from the CCP4 suite (Collaborative Computational Project, Number 4, 1994). Crystals belonged to space group P2₁, with unit-cell parameters a = 37.5, b = 55.6, c = 108.9 Å, β = 91.4 °. Data-collection statistics are shown in Table 2. The asymmetric unit may contain either three monomers, giving a V_M value (Matthews, 1968) of 2.0 ų Da⁻¹ and 38.9% solvent, or two monomers, giving a V_M value of 3.0 ų Da⁻¹ and 59.3% solvent. A self-rotation function did not resolve this ambiguity, as no significant peak corresponding to either twofold or threefold non-crystallographic symmetry was observed. The native Patterson map also did not reveal any peaks. Attempts to solve the structure by molecular replacement using various models based on the coordinates of the macro domain from Archaeoglobus fulgidus (Allen et al., 2003; PDB code 1hjz), which shares 29% identity with the SARS-CoV macro domain and using various programs [AMoRe (Navaza, 1994), MOLREP (Vagin & Teplyakov, 1997) and Phaser (McCoy et al., 2005)] proved unsuccessful. Using a SAD data set collected from an SeMet crystal, we have determined the positions of all Se atoms and structure refinement is ongoing.

Table 2
Data-collection statistics

Values in parentheses are for the highest resolution shell.

	Native data	SeMet data
X-ray source	ESRF ID29	ESRF ID29
Wavelength (Å)	0.9792	0.9792
Crystal parameters
Space group	P21	P21
Unit-cell parameters
a (Å)	37.4	37.5
b (Å)	55.4	55.6
c (Å)	108.8	108.9
β (°)	91.6	91.4
Resolution range (Å)	40–1.8 (1.9–1.8)	28–1.8 (1.9–1.8)
Data processing
Observed reflections	238756	432355
Unique reflections	41448	41654
Rmerge (%) †	7.7 (18.2)	10.9 (38.6)
Completeness (%)	99.6 (98.7)	99.9 (99.9)
Multiplicity	4.8 (5.6)	10.4 (8.9)
Mean I/σ(I)	16.2 (7.9)	17.7 (7.1)

†R_merge = .

As previously mentioned, the molecular packing observed by Saikatendu and coworkers did not allow either cocrystallization or ligand-soaking experiments as the active site of both molecules was occluded by the dimer interface. The different crystal form obtained in this study may enable more successful ligand-soaking experiments.

Footnotes

‡These authors made an equal contribution to this work.