Crystallization and Preliminary X-Ray Analysis of Rotavirus Protein VP6

Abstract

As a first step to gain insight into the structure of the rotavirus virion at atomic resolution, we report here the expression, purification, and crystallization of recombinant rotavirus protein VP6. This protein has the property of polymerizing in the form of tubular structures in solution which have hindered crystallization thus far. Using a combination of electron microscopy and small-angle X-ray scattering, we found that addition of Ca²⁺ at concentrations higher than 100 mM results in depolymerization of the tubes, leading to an essentially monodisperse solution of trimeric VP6 even at high protein concentrations (higher than 10 mg/ml), thereby enabling us to search for crystallization conditions. We have thus obtained crystals of VP6 which diffract to better than 2.4 Å resolution and belong to the cubic space group P4₁32 with a cell dimension a of 160 Å. The crystals contain a trimer of VP6 lying along the diagonal of the cubic unit cell, resulting in one VP6 monomer per asymmetric unit and a solvent content of roughly 70%.

Rotaviruses are important human pathogens. They have been identified as the leading cause of severe gastroenteritis in infants throughout the world (13). In addition, many rotavirus strains are pathogenic to farm animals and cause important economic loss in the livestock industry. The viral particles have been very well characterized by structural studies using electron cryomicroscopy (22, 28). These structural studies, along with biochemical and molecular biology approaches, have shown that the virions have a complex structure, composed of three concentric protein layers which enclose the double-stranded RNA viral genome together with the enzymes needed for transcription (8). The external protein layer of mature particles (called triple-layered particles), composed of proteins VP7 and VP4, dissociates from the viral particle upon entry of the virus into the target cell (5, 8). The second layer is formed by protein VP6, which associates in a T=13 levo-icosahedral lattice (4) containing 260 VP6 trimers (24). The third and innermost layer, made up of 120 subunits of protein VP2, is tightly associated with the viral RNA and with the VP6 layer (21). In the cytoplasm of the infected cell, the double-layered particle remains intact and acts as the viral transcription unit. Newly synthesized mRNA molecules have been visualized emerging from the particle through pores at the icosahedral fivefold axes (16). Particles devoid of the VP6 layer are transcriptionally inactive (1), and it has also been shown that certain antibodies directed against VP6 can inhibit transcription by double-layered particles (10, 15). In addition, it has been shown that VP6-specific immunoglobulin A monoclonal antibodies have a protective effect in the mouse model (3), although they lack neutralizing activity.

Despite much progress in the last few years in the study of rotaviruses, the molecular mechanisms underlying many of the events that occur during infection by these pathogens remain unknown. High-resolution structures of the viral proteins, in conjunction with ongoing cryoelectron microscopy studies of intact virions, should provide valuable insight into the roles played by virion components in the viral life cycle. We have thus begun the expression and purification of the viral proteins individually to allow their analysis by X-ray crystallography. We report here our findings on protein VP6: its expression and purification, analysis of the stability of its multimeric states by electron microscopy (EM) and small-angle X-ray scattering (SAXS), and finally its crystallization, along with a preliminary characterization of the crystals, which diffract to better than 2.4Å resolution.

Expression.

Protein VP6 from group A rotavirus is 397 amino acids long and has a molecular mass of 41 kDa. Caterpillars (Spodoptera frugiperda) were infected with the recombinant baculovirus (containing the VP6 gene of bovine strain RF [27]) by injection of 10⁶ PFU of the virus. They were sacrificed 3 to 5 days postinfection, each one was placed in an Eppendorf tube containing 500 μl of buffer A [piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) at 50 mM (pH 6.6), dithiothreitol at 4 mM, aprotinin at 10 μg/ml] containing 20% glycerol. The caterpillars were then ground and cleared by centrifugation at 12,000 × g for 15 min at 4°C. The upper lipid phase was discarded, groups of 10 tubes were pooled (keeping both the pellet and the aqueous supernatant), and the volume was adjusted to 28 ml by adding buffer A. After addition of 14 ml of Freon 113, the mixture was stirred with a polytron and centrifuged for 5 min at 1,500 × g. The supernatant was recovered, and the pellet was resuspended in 28 ml of buffer A and treated twice as described above. A final Freon treatment was done on the pooled supernatants to ensure extraction of most of the VP6 protein from the debris.

Protein purification.

Recombinant VP6 was found to self-assemble, in a way analogous to that of its virus-derived counterpart (23), to form tubular structures as discussed below. We took advantage of this ability of the protein to polymerize to establish a purification protocol in which VP6 is separated from the other proteins by ultracentrifugation at 100,000 × g (Fig. 1). The average yield of this step was about 1.0 mg of VP6 per caterpillar. Resuspension of the pellet in 100 μl of H₂O (containing 0.02% NaN₃) leads to a VP6 solution containing labile tubes (see the section on EM and SAXS). The purity of the protein resulting from this step is shown in Fig. 1. When this solution was subjected to size exclusion chromatography (SEC), a single peak in the resulting elution profile corresponded to a molecular mass of 150 kDa, consistent with a trimer of VP6. Very little or no protein was eluted in the peak corresponding to the void volume (Fig. 1b, inset). This result indicates that under the elution conditions used, the VP6 tubes dissociate, suggesting that the interactions that hold them together are weak. The concentration of VP6 recovered from the fractions corresponding to the trimer peak was about 0.2 mg/ml. Attempts to concentrate the sample to the higher concentrations required for crystal growth resulted in the reformation of tubes (data not shown). To overcome this limitation, we used EM and SAXS to search for conditions under which the polymerization of VP6 would be inhibited during concentration.

FIG. 1.

(a) Purification of recombinant VP6. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of the preparation was done under denaturing conditions on a 10% polyacrylamide gel stained with Coomassie blue. Lanes: 1, molecular mass markers; 2, supernatant before ultracentrifugation; 3 and 4, supernatant and pellet after ultracentrifugation at 100,000 × g, respectively. (b) Analysis of the oligomerization state of purified recombinant VP6 by SEC on a prepacked Sephacryl S-300 column (Pharmacia) equilibrated with 50 mM Tris buffer (pH 7)–150 mM NaCl and eluted in the same buffer at a flow rate of 0.8 ml/min. The inset shows an analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the two peaks of the chromatogram. Lanes 2 and 3 correspond to the peak at the void volume (concentrated 10-fold) obtained with samples boiled and not boiled in Laemmli buffer, respectively; lanes 4 and 5 correspond to the peak at 150 kDa obtained with boiled and unboiled samples, respectively, showing a trimer in the unboiled sample. This indicates that all of the VP6 protein (detectable by Coomassie blue staining) was found in the second peak, implying that the tubes dissociate into trimers during SEC. The arrows indicate the positions of the molecular size markers thyroglobulin (670 kDa), bovine gamma globulin (158 kDa), ovalbumin (44 kDa), and myoglobin (17 kDa).

EM and SAXS studies of the association states of VP6.

A high Ca²⁺ concentration has been reported to dissociate VP6 from the viral particle (1). This prompted us to investigate the effect of increasing Ca²⁺ concentrations on the solubility of VP6. We found that Ca²⁺ indeed has a major effect on the various equilibria governing the distribution of polymeric species (Fig. 2). When a drop of a VP6 solution is placed on an air glow-discharged carbon-coated grid for a few seconds and then negatively stained with a 2% uranyl acetate solution, the structures observed in an electron microscope (Philips CM12) depend upon the Ca²⁺ concentration (Fig. 2). When the Ca²⁺ concentration is less than 100 mM, mainly tubular structures can be seen. In particular, when we analyze VP6 in water (using the resuspended pellet from the ultracentrifugation step), we see tubes that have numerous defects, as shown in Fig. 2a. The EM observations again suggest that under these conditions, the interactions between trimers are weak since the tubes seem to be easily damaged by the staining procedure, confirming the SEC results. When the Ca²⁺ concentration is higher than 100 mM, no tube can be found on the micrographs, which show mainly isolated trimers, as shown in Fig. 2c (which corresponds to 200 mM Ca²⁺). Indeed, the threefold rotational symmetry is evident on some of the VP6 trimers, depending on their orientation on the grid (data not shown). Because the aggregation state of VP6 may depend upon the method of preparation for EM (like pH shifts during staining, changes in ionic strength, and in particular the protein concentration), we correlated this EM observation with data obtained by SAXS.

FIG. 2.

Characterization of the association states of VP6 by negative-staining EM (a and c) and SAXS (b and d). Panels a and b correspond to VP6 in water, and panels c and d correspond to VP6 in 200 mM CaCl₂. (a) Electron micrograph of a VP6 pellet dissolved in water and negatively stained with a few drops of a 2% uranyl acetate aqueous solution. Tubular structures with a fairly constant diameter are observed. Note the defects in the structures, which suggest that under these conditions the interactions between VP6 trimers that hold the tubes together are rather weak. (b) SAXS pattern of VP6 in water. Eight successive frames of 100 s each were recorded for the VP6 solution and water. The average pattern was computed after visual inspection of each frame for radiation damage; none was found. Finally, the scattering intensity [I(s)] of the buffer was subtracted from that of the VP6 solution after scaling to the transmitted intensity. The curve shows oscillations which correspond to the intensity maxima expected from the presence of tubes of regular diameter in solution. The shallowness of the intervening minima is accounted for by the presence of defects in the tubes, which are visible in the electron micrograph in panel a. (c) Electron micrograph of a VP6 sample 200 mM CaCl₂. No tubular structures, only isolated VP6 trimers, are visible. (d) SAXS pattern of VP6 in 200 mM Ca²⁺ (recorded under the same conditions as panel b). Note the absence of oscillations compared to panel b, indicating that the tubes have dissociated and thus the intensity maxima have disappeared. The sharp rise at very small angles is due to the presence of some aggregates, which dominate this part of the spectrum. Inset: Guinier plots of VP6 in 200 mM Ca²⁺ at a protein concentration of 10 mg/ml (circles) and of VP6 eluted from a SEC column (Fig. 1b) at a protein concentration of 0.3 mg/ml (squares). In this case, since the solution of trimeric VP6 was very dilute, the weak scattering intensity could only be obtained by using a 60° sector-shaped detector. The top and bottom pairs of arrows delimit the s range used for the calculation of the radius of gyration (Rg) from the slope of the linear regression fit on the top and bottom curves, respectively. The Rg values obtained (around 32.5 Å in both cases) are consistent with the presence of VP6 trimers in solution.

The SAXS pattern of VP6 in solution was recorded, at different CaCl₂ and protein concentrations, on the small-angle scattering instrument D24 at the Laboratoire pour l’Utilisation du Rayonnement Electromagnétique (LURE) (Orsay, France) by using synchrotron radiation with a wavelength of 1.488 Å. The instrument (6), the data acquisition system (2), and the evacuated cell (7) have been described previously. The scattering curve obtained from a VP6 pellet dissolved in water, at a concentration of 10 mg/ml, is shown in Fig. 2b. It displays regular but shallow oscillations due to the cylindrical symmetry of the tubes. The shallowness of the minima is consistent with our EM data, which show tubes that are fragile and easily damaged (Fig. 2a). Upon addition of Ca²⁺ at concentrations higher than 100 mM, there is a dramatic change in the scattering pattern. The curve shown in Fig. 2d (corresponding to 200 mM Ca²⁺) displays a monotonic decrease, showing that the tubes have completely dissociated, in agreement with the EM data. This was the case at protein concentrations in the range of 10 to 20 mg/ml, indicating that the dilution step necessary for the EM experiments was not responsible for the disappearance of the tubes. A sharp rise in intensity, arising from the presence of large scattering objects, is visible in the innermost part of the curve. The presence of a few aggregates could account for the observed rise in intensity, since the scattering of large objects largely dominates in this region of the scattering curve. In addition, the magnitude of the intensity rise, and therefore the proportion of aggregates, varied from one sample to another (data not shown). This is in agreement with the EM data showing occasional contaminants. We believe that the purification protocol of VP6 did not get rid of large-size but minor contaminants, which sedimented with the VP6 tubes. At larger angles, the scattering originates mainly from smaller particles. According to Guinier and Fournet, the scattering curve of an object can be approximated at small angles by a Gaussian curve, the width of which yields the radius of gyration of the object (12), which in our case (ignoring the innermost part of the curve [Fig. 2d]) was found to be 33.5 ± 1 Å. This value is in agreement with the expected radius of gyration of a VP6 trimer and compares well with the value of 32 ± 1 Å obtained with very dilute (0.2 to 0.5 mg/ml) solutions of trimeric VP6 eluted from a SEC column (in 50 mM Tris-HCl buffer [pH 7.6] containing 150 mM NaCl) a few minutes before the SAXS measurement (inset in Fig. 2d). These data indicate, in agreement with the EM observations, that in the presence of high concentrations of Ca²⁺ (>100 mM), the vast majority of protein VP6 is present in solution as trimers, even at the high protein concentrations required for crystal growth.

Crystallization.

As a consequence of the studies described above, we did all of our crystallization trials at a protein concentration of 10 to 20 mg/ml in the presence of Ca²⁺ concentrations higher than 100 mM to avoid polymerization in the form of tubes. Under these conditions, we found that the precipitant polyethylene glycol monomethyl ether (molecular weight, 550) at concentrations ranging between 14 and 20%, pH 7.5, in the presence of 200 mM CaCl₂, leads to the formation of small, cubic crystals of VP6 which grow slowly, the biggest reaching a size of 0.2 mm per side after several months (Fig. 3a). The minimal Ca²⁺ concentration at which crystals were found to grow was 150 mM, but these crystals remained very small. CaCl₂ concentrations higher than 200 mM (250 to 300 mM) led to crystals that grew rapidly and deteriorated in a few days.

FIG. 3.

(a) Crystals of recombinant VP6. The crystals were grown by vapor diffusion using the hanging-drop technique (17). Typically, 1 μl of a VP6 solution (in H₂O) at a concentration of 10 mg/ml was mixed with an equal volume of a solution containing 17% polyethylene glycol 550 monomethyl ether and 200 mM CaCl₂ in HEPES buffer (pH 7.5). Cube-shaped crystals appear after about 3 weeks and grow steadily for several months, to a maximum size of about 200 μm per side. (b) Diffraction pattern of VP6 crystals. Shown is the diffraction pattern (0.5° oscillation image) of a VP6 crystal collected on a charge-coupled device (CCD) detector using synchrotron radiation at ESRF beam line BM2. The diffraction pattern extends to the edge of the detector, which corresponds to a resolution of 2.4 Å (note that the image is not symmetric, going to 2.4 Å resolution on only one side, because the detector was swung by an angle of 2.5°. The images were processed by using the HKL package (20). The symmetry of the diffraction pattern corresponds to either space group P4₁32 or its enantiomorph P4₃32. The cubic unit cell edge is 160 Å long.

Characterization of VP6 crystals.

The cubic crystals were characterized by using the synchrotron radiation of LURE beam lines DW32 and D41A and of the European Synchrotron Radiation Facility (ESRF) (Grenoble, France) beam line BM2 and found to belong to space group P4₁32 (or P4₃32) with a cell edge a of 160 Å. These crystals diffract X-rays to at least 2.4 Å resolution, as shown in Fig. 3b. The dimensions of the unit cell dictate that the trimer must lie with its threefold axis coincident with the crystallographic threefold axis, along the diagonal of the cubic unit cell (the presence of a trimer in the asymmetric unit would lead to unacceptably tight packing). This gives rise to a solvent content in the crystals of roughly 70% of the volume at an average specific volume for protein molecules of about 0.73 cm³/g. We have collected a native diffraction data set to 3.2 Å at LURE. With these data, we have attempted to determine the structure by molecular replacement (18, 25) by using the atomic model of bluetongue virus (BTV) protein VP7 (11) as the search object. BTV is an orbivirus in which VP7 forms the middle layer of the triple-layered mature particles and thus has functional homology to rotavirus VP6, although the sequence identity is only 19% (orbiviruses and rotaviruses belong to the same virus family, Reoviridae [9, 26]). Using the trimer of BTV VP7, we searched for molecular replacement solutions by rotating the model about the diagonal of the cubic unit cell and translating along it in a unidimensional search with the program AMoRe (19). This procedure did not provide any clear solution (16a). We conclude from this negative result that the atomic model of BTV VP7 is different enough from rotavirus VP6 that it cannot be used to obtain starting phases. The determination of the crystal structure by the isomorphous replacement method (see reference 14 and references therein for a review of this method) is well under way, and a description of the structure of the VP6 molecule and its refinement will soon be published elsewhere.