Virology 265, 35–45 (1999) Article ID viro.1999.0005, available online at http://www.idealibrary.com on
Recombinant Hepatitis E Capsid Protein Self-Assembles into a Dual-Domain T 5 1 Particle Presenting Native Virus Epitopes Li Xing,* Kenzo Kato,† Tiancheng Li,† Naokazu Takeda,† Tatsuo Miyamura,† Lena Hammar,* and R. Holland Cheng* ,1 *Department of Biosciences at Novum, Karolinska Institute, 14157 Huddinge, Sweden; and †Department of Virology, National Institute of Infectious Disease, Tokyo 162, Japan Received July 27, 1999; returned to author for revision August 25, 1999; accepted September 13, 1999 The three-dimensional structure of a self-assembled, recombinant hepatitis E virus particle has been solved to 22-Å resolution by cryo-electron microscopy and three-dimensional image reconstruction. The single subunit of 50 kDa is derived from a truncated version of the open reading frame-2 gene of the virus expressed in a baculovirus system. This is the first structure of a T 5 1 particle with protruding dimers at the icosahedral two-fold axes solved by cryo-electron microscopy. The protein shell of these hollow particles extends from a radius of 50 Å outward to a radius of 135 Å. In the reconstruction, the capsid is dominated by dimers that define the 30 morphological units. The outer domain of the homodimer forms a protrusion, which corresponds to the spike-like density seen in the cryo-electron micrograph. This particle retains native virus epitopes, suggesting its potential value as a vaccine. © 1999 Academic Press Key Words: cryo-EM image reconstruction; human hepatitis E virus; quasi-equivalence; recombinant virus-like particle.
INTRODUCTION
(Huang et al., 1992), has revealed that HEV has a 39 poly(A) tail and three open reading frames (ORFs) within the viral genome. ORF1 is the largest reading frame with approximately 5 kb at the 59-terminus of the genome. The ORF1 peptide sequence contains consensus motifs for a methyltransferase, a protease, a helicase, and an RNAdirected RNA polymerase, which strongly suggests its encoding of nonstructural proteins. ORF2 encodes the major capsid protein and is located in the 39-terminal portion of the genome. ORF3 overlaps with the other two ORFs and encodes an immunogenic protein with an unidentified role in the virus life cycle (Tam et al., 1991). The sequence position of the ORF3 is one of the features that distinguish HEV from members of the Caliciviridae (Tam et al., 1991). A difference between HEV and caliciviruses is also found in the length of ORF2 as well as in the arrangement of functional protein domains in the ORF1 products. The ORF2 protein contains a typical signal sequence at its N-terminus, which is followed by an arginine-rich domain. This highly positively charged region is postulated to be involved in RNA encapsidation during virus assembly (Tam et al., 1991). Based on the peptide sequence, the ORF2 protein contains three potential Nglycosylation sites. Two immunodominant regions at residues 394–470 and 546–580 were identified in the ORF2 protein by pepscan analysis (Khudyakov et al., 1994). These two regions display a high degree of sequence similarity even between strains of the most genetically diverged HEV, Burma and Mexican (Yarbough et al., 1991). The C-terminal region of the ORF2 protein is
Human hepatitis E virus (HEV) 2 causes acute hepatitis and is fecal–orally transmitted through contaminated water (Wong et al., 1980; Khuroo, 1980). Young adults are most at risk. The mortality rate among infected pregnant women has been reported to reach 20% (Khuroo et al., 1981; Purcell and Ticehurst, 1988). HEV is similar to calicivirus in morphology and biophysical properties (Bradley et al., 1988). The hepatitis E virion is a spherical particle with a diameter of 270–300 Å, as first shown by immuno-electron microscopy (Balayan et al., 1983). Subsequent studies have revealed that HEV is a nonenveloped icosahedral particle with indentations on the surface. Particles collected by sucrose gradient centrifugation have a diameter of 320–340 Å. The buoyant density of HEV is 1.29 g/cm 3 in a potassium-tartrate/glycerol gradient, and the sedimentation coefficient is 183S (Bradley et al., 1988; Bradley, 1990). The fact that HEV can survive in the intestinal tract suggests that the virus is relatively stable to acid and mild alkaline conditions (Purcell, 1996). HEV is composed of one major structural protein and a single-stranded RNA molecule of approximately 7.5 kb. Nucleotide sequence analysis, based on four geographically distinct isolates, Burma (Tam et al., 1991), Pakistan (Tsarev et al., 1992), China (Bi et al., 1993), and Mexico
1 To whom correspondence and reprint requests should be addressed. Fax: 146-8-774 55 38. E-mail:
[email protected]. 2 Abbreviations used: HEV, hepatitis E virus; rHEV, recombinant HEV; CCMV, cowpea chlorotic mottle virus; ORF, open reading frame.
35
0042-6822/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
36
XING ET AL.
virions from the bile or the stools of experimentally infected monkeys. In this study, the structure of the rHEV particle was determined by cryo-electron microscopy (cryo-EM) and image reconstruction to support a better understanding of its antigenic properties and the viral capsid assembly. RESULTS Intracellular formation of rHEV particles
FIG. 1. Time course study of infected Tn5 cells. (A) The relative concentrations of intracellular ORF2 derived p58 and p50 peptides and p50 recovered from the Tn5 cell supernatant at different time points after infection with recombinant baculovirus of the HEV ORF2 construct. Five microliters of infected cell supernatant and an equal amount of the cell lysate were analyzed by SDS electrophoresis in a 10% polyacrylamide gel with Coomassie brilliant blue staining. The intensity of each protein band was measured and plotted as a function of the postinfection day. (B) The infected Tn5 cells, at day 3 after infection, showed a positive reaction with the HEV patient sera. The immunofluorescence signal is clustered in restricted areas of the cell. (C) Electron micrograph of infected cells demonstrating the time course and compartmentalization of particle formation. The samples were fixed and embedded in Spurr’s resin. Ultrathin sections, 50 nm thick, were subsequently stained with 2% uranyl acetate and Reynolds lead citrate. The infected Tn5 cell contains large dense, exclusion body-like vesicles. No virus-like particles were seen in the electron micrograph of the thin sections at this stage. (D) At day 6 after infection a large number of empty virus-like particles were found in the cytoplasm of Tn5 cells. The bar indicates 2500 and 250 nm in (C) and in (D), respectively.
broadly reactive with patient sera, collected from various laboratories (Purdy et al., 1993; Ghabrah et al., 1998). In the absence of an appropriate cell culture for HEV propagation, research has focused on the expression of the ORF2 protein in heterologous systems. Recently, particles of recombinant hepatitis E virus (rHEV) were produced by using a baculovirus system carrying an Nterminally truncated ORF2 gene of the Burma strain (Li et al., 1997b). Thus, rHEV particles were formed in Tn5 cells and could be collected from the supernatant of the culture. These rHEV particles are smaller than the isolated
The infection of Tn5 cells with HEV ORF2 recombinant baculovirus resulted in the formation of two major ORF2derived peptides with apparent molecular masses of 50 (p50) and 58 kDa (p58). The appearance and disappearance of the p58 and the p50 in the cell homogenates and in the cell supernatant were followed by SDS–polyacrylamide electrophoresis during the course of the experiment. Figure 1A shows the relative concentration of the peptides at different times after infection, as the density of Coomassie brilliant blue-stained peptide bands. p58 was the first ORF2 peptide to be detected postinfection. The protein was found only intracellularly and was not recovered in the culture supernatant during the course of the experiment. p50 appeared intracellularly at day 3 postinfection and could be recovered from the cell culture supernatant at day 4. While the amount of the intracellular p58 decreased after day 4 postinfection, p50 was recovered increasingly in the cell supernatant up to day 7. The rHEV particles were regularly harvested at day 6 postinfection and were purified in a yield of about 1 mg per 10 7 cells. At different time points, we further examined histological and ultrathin sections of Tn5 cells infected with the same recombinant baculovirus construct. At day 3 postinfection ORF2 protein-related intracellular immunoreactivity appeared (Fig. 1B), supposedly localized to the transiently occurring, large cytoplasmic bodies, seen by electron microscopy (Fig. 1C). No particle was observed in the ultrathin section of infected cells at this stage. However, at day 6 postinfection, the cells contained numerous particles (Fig. 1D). These particles are empty and of the same diameter as the rHEV particles recovered from the supernatant of the infected cells. The rHEV particles were stable in 10 mM potassiumMES buffer at pH 6.2, but disassociated after overnight incubation at 4°C in 10 mM Tris–HCl buffer at pH 7.4. Analytical molecular sieving chromatography shows that the particles are eluted in the void volume at pH 6.2 (Fig. 2A). To dissociate the particles at pH 6.2, treatment with urea or SDS was needed. However, on treatment at pH 8.3, the same material dissociated into apparent dimers and monomers, as revealed by the front and tail of the saddle peak in the chromatogram (Fig. 2A). To support particle stability, the 10 mM potassium-MES buffer, pH 6.2, was used for the purification of the particles. Highly purified rHEV particles were prepared by CsCl gradient
T 5 1 PARTICLE OF RECOMBINANT HEPATITIS E VIRUS
37
FIG. 6. (A) Radial projections through the structure of rHEV. Sequential radial shells from the twofold oriented density map are displayed. Projections of the density contained within defined spherical shells were calculated by summing only those density points along the viewing directions within specified radial limits. Higher protein density is represented by lighter gray scale. The sections were taken from 125 to 55 Å at 10-Å intervals. (B) Icosahedral surface lattices, showing the packing of 30 dimers at a strictly icosahedral twofold position in a T 5 1 lattice of 270 Å in diameter (left) or the assembly of 90 of the same size dimers in a T 5 3 lattice of 390 Å in diameter (right). The white triangle in the T 5 3 lattice corresponds to one icosahedral face in the T 5 1 lattice.
centrifugation and shown by SDS–polyacrylamide gel electrophoresis to be composed solely of the p50 peptide (Fig. 2B). The molecular mass thus determined was 49–50 kDa. Treatment of the material with 20 mM dithiothreithol showed no major effect on the electrophoretic mobility of the peptide. Isoelectrofocusing of the purified particle, pretreated at pH 8.3, revealed four major peptide forms, focusing in the pH range 5.0–5.8 (Fig. 2C). Electron microscopy and image analysis The excellent productivity of rHEV particles in the baculovirus expression system prompted us to use the material for structure determination. The particles were purified and concentrated by density gradient centrifuga-
tion and molecular sieving chromatography to obtain the homogeneity essential for cryo-EM data collection. Images of unstained, vitrified rHEV particles revealed a profile with prominent surface features (Fig. 3). In addition to their smaller diameter, rHEV particles were less dense in the center than the particles of cowpea chlorotic mottle virus (CCMV, 286 Å in diameter) used for size calibration. The approximately circular feature of projection images indicated that the particles should have a spherical morphology. Projected images of rHEV particles could generally be recognized by the presence of protruding profiles around the edge. Randomly oriented particles could be distinctively observed based on the diversified positioning of their dimeric capsomers. Some
38
XING ET AL.
of diameter deviation, made it possible to achieve an effective Fourier averaging and thus a reliable image reconstruction (Cheng et al., 1995). The resolution determined by referencing a reliability index, R AB, set the final reconstruction truncated at 22 Å (Cheng et al., 1992). This is within the first zero of microscope contrast transfer function according to the defocus level at which the micrograph was recorded. Particles of rHEV display a T 5 1 symmetry The rHEV capsid clearly reveals 60 copies of p50 subunits based on its T 5 1 lattice of icosahedral symmetry (Fig. 4). The maximum diameter of the rHEV particle was 270 Å by using CCMV as an internal size standard. A total of 73 individual particle images of rHEV are included in the final three-dimensional reconstruction. As seen by the depth-cue shaded representation, there are 30 morphological plateaus measured to be 56 Å long and 43 Å wide. These are part of the p50 homodimers projecting along the twofold axes, and they are surrounded by four dimers related by adjacent threefold axes (Fig. 4). Although a lower 522 symmetry is imple-
FIG. 2. Biochemical analyses of the rHEV particles. (A) Molecular sieving chromatography demonstrates the stability of the rHEV particles at pH 6.2 and their dissociation at pH 8.3. The curves showing the run at pH 6.2 and the run at pH 8.3 are labeled. The diagram also shows the molecular mass (left-hand side ordinate) viz. elution volume for standards, run in parallel. (B) Silver-stained electrophoretogram of purified rHEV particles treated with SDS under reducing conditions and run in a PhastSystem 8–25% polyacrylamide gel. Positions of molecular mass standards are indicated (kDa). (C) Isoelectrofocusing of the rHEV particles, dissociated by overnight treatment in 10 mM Tris–HCl, pH 8.3. The sample was run in a PhastSystem IEF gel, pH 3–9, fixed in TCA, and silver-stained. Isoelectric points for simultaneously run markers are indicated.
of the rHEV particles deviated in size but with preserved circular symmetry. Uniform rHEV particles tended to be found in the thicker regions of the vitrified ice. The presence of protease inhibitors did not affect the particle size variation during the purification. A polar Fourier transform (PFT) was used to refine the origin and orientation with intermediate 3D reconstructed models (Baker and Cheng, 1996). With Fourier bandpass filtering, we ejected the lowest spatial-frequency data of circularly symmetric features and the highest spatial-frequency noise to sharpen the icosahedral symmetric components of the images during the orientation assignment. Additional radial filtering of the data used in the phase alignment was found to be essential to sensitize the refinement of viewing orientations. Such size exclusion, implemented with a 3% cut-off
FIG. 3. Micrograph of the vitrified rHEV and CCMV sample suspended over holes in a carbon support film. It was recorded at 1.4-mm underfocus and with an electron dose of 7 e 2/Å 2. The rHEV particles show an indentation profile as indicated by arrowheads. The smooth, round feature of CCMV can be easily characterized as a calibration reference in the micrograph. Bar, 500 Å.
T 5 1 PARTICLE OF RECOMBINANT HEPATITIS E VIRUS
39
is connected to form a shell, which covers the surface area of the icosahedral faces. This shell contributes to the main part of the depressed flat revealed around fivefold axes. The subunits meet again at the radius of 75 Å and revive the dimer connection underneath the tunnel at the twofold axes. On top of the tunnel, the protrusion domain not only stretches out radially, but also skews counterclockwise toward the threefold symmetry position. The hole around the fivefold axis at this radial section is part of the central cavity. The tunnel under the dimeric protrusion couples two of these holes from each side of the protein (Fig. 4). As a result, the capsid shells of rHEV particles are not completely sealed under the confined T 5 1 lattice. The tunnels pervade the interstices of the basal protein layer, conjoined at the central cavity underneath the fivefold flats, where they open to the surroundings. Therefore, the interior void volume is freely accessible to the surrounding solvent. Configuration of morphological dimers FIG. 4. Surface-shaded representations of the rHEV particle. The reconstruction of rHEV, computed from 73 independent particle images, is viewed along the two-, (A), three-, (B), and fivefold axes (C). The subunits are arranged as dimeric plateaus at the interface of two icosahedral faces and form a T 5 1 surface lattice. The depth-cued representation (D), viewed along a threefold axis, is overlaid with a T 5 1 icosahedral lattice net in which the icosahedral elements, two-, three-, and fivefold, are identified by ovals, triangles, and pentagons, respectively. Bar, 100 Å.
mented in the averaging of the assigned symmetry, the final reconstruction shows strong dimer connections at threefold positions. There is a flat depression located in the regions of the fivefold axis, where five tunnel passageways lead radially toward the nearby twofold axes. A radial density map of the reconstruction shows this plane at a radius of 90 Å, which is about 45 Å below the maximum radius of the protruding platform. The interior view from the bottom half shows a central cavity with a maximum and a minimum radius of 80 and 50 Å at icosahedral five- and twofold positions, respectively (Fig. 5). The protein shell consists of two radially distinct domains, namely the distal protrusion and the basal layer, which can easily be observed in the surface rendering and the density distribution of the equatorial section (Fig. 5A). The thickness of the protein shell varies from 85 Å at twofold positions to 10 Å at fivefold positions. At the radial shell of 75–95 Å, the subunits of the dimeric capsomer are separated by a cavity of ;20 Å in diameter. In forming an arch-like tunnel, these cavities eventually connect through the dimers with a tilting angle of ;30° away from the equator between two nearby fivefold axes (Fig. 5B). The tunnel reaches its maximal length of 62 Å at a radius of 85 Å and closes up at a radius of 95 Å. At this radial distance, the protein density
In Fig. 5, we examined the radial density distribution along the direction parallel to the axes of the dimeric capsomer. These density maps reveal that the capsid protein starts at a radius of 50 Å at the twofold axes and forms a partly sealed shell at a radius under 90 Å. The protein extends further to form the protrusion domain of dimeric
FIG. 5. Equatorial cross section of the rHEV reconstruction along a twofold axis, using the gray scale to indicate the protein density. (A) The left panel is a surface representation, shaded at a contour level of 1 standard deviation above the average density, whereas the right panel is a continuous, gray-level representation of the section. (B) Shaded density distribution at specific radii of the rHEV reconstruction. Note the arch-like tunnels passing all the twofold axes at the shell domain regions. The contour level is chosen based on an assumption that the averaged protein density is 1.30 g/cm 3. With a conversion factor of 1.66 g z Å 3/Da/cm 3, the 60 copies of p50 give the proximate volume of 3.8 3 10 6. This is 11.6% of the total volume of our EM density map, which corresponds to a contour level slightly higher than the value of 1 standard deviation above the solvent density. The appearance of the capsid tunnel starts, and stays stable, at the contour level above 0.5 standard deviations.
40
XING ET AL. TABLE 1 Biological and Physicochemical Properties of HEV and rHEV Particles Property
Morphology Symmetry Size in diameter Physicochemistry Antigenic sites Genome
HEV native virions
rHEV particles
Nonenveloped, spherical particles with indentations on the surfaces Icosahedral a 320–340 Å (immuno-EM) b 385–410 Å (estimated, see Discussion) Buoyant density 1.290 g/cm 3 (K-tartrate/glycerol gradient); stable to acid and mild alkaline Two immunodominant epitopes c 7.5KB (1)ssRNA
Nonenveloped, spherical particles with indentations on the surfaces Icosahedral (see Fig. 6). 220–230 Å (conventional EM) 270 Å (cryo-EM) Buoyant density 1.28 g/cm 3 (sucrose gradient); stable to acid, labile at and above neutral pH Immunogenicity similar to that of native d Empty particles
a
Krawczynski (1993). Bradley et al. (1988). c Khudyakov et al. (1994). d Li et al. (1997b). b
capsomers above this basal layer. These dimeric capsomers appear to make stronger contacts with neighboring capsomers at threefold axes than at fivefold axes. Proceeding with the radial density from inside out, the sections at 55 and 65 Å show that two protein subunits are connected at the twofold symmetry axes (Fig. 6A). The next three sections, from radius 75 to 95 Å, are in the region of the arch-like tunnel, where the protein gradually moves to a position between the threefold and the fivefold axes to form an almost closed capsid shell. In this layer, the density of the capsid protein is much more condensed than that at other radial regions. The density around fivefold axes locates mostly at a radius of 85 Å, in which the flat surface forms. At this radius, the subunits establish a steady clockwise cluster around the threefold axes. The dimeric contacts around the tunnels basically serve as connectors to couple neighboring threefold clusters together. Further upward in the shell, the protein density takes a sharp turn and orients 90° to adjacent twofold positions at the radius of 105 Å. This type of dimeric assembly brings the density of icosahedral faces together at twofold axes. In Fig. 6A, the density of subunits turns from being parallel to the tunnel at the shell domain (radii 85–95 Å) toward the vertically oriented dimeric contacts at the protruding region (sections 95– 105 Å). The 90° twist at this region constructs a hinge separator between the basal shell and the protrusion domains. A detailed tracing of the protein shows that such a twist happens during the merging of the monomers into the dimers between 90 and 100 Å in radius. The tight cluster at the threefold axes provides one of the essential contacts to place the capsid proteins into a T 5 1 shell domain (Caspar and Klug, 1962; Johnson and Speir, 1997). DISCUSSION The truncated ORF2 peptide is able to self-assemble into virus-like particles in Tn5 cells. HEV ORF1 and ORF3
are not included in the rHEV construct, and therefore, they are not essential for particle assembly. The recombinant particles share morphological and antigenic properties with the native virion although they are empty and smaller than the native virions (Table 1). In total, 60 copies of the single subunit cluster as 30 dimers into a T 5 1 lattice. The self-assembly of p50 into a dualdomain capsid apparently preserves the proper topology of important HEV epitopes. While the Escherichia coli expressed ORF2 protein does not adequately present essential epitopes (Li et al., 1997a), the baculovirus expressed p50 closely mimics the native capsid configuration in particles of T 5 1 surface lattice. The high productivity, in combination with retained native antigenicity, makes rHEV particles an ideal candidate for a hepatitis E vaccine. Sequence analysis reveals that the N-terminus of the full-length ORF2 peptide contains basic residues and, hence, may be involved in RNA binding. As a result of N-terminal deletion, the rHEV capsid particles are probably devoid of nucleic acid. The presence of a large cavity inside the rHEV particles further confirms that these capsid particles do not contain substantial amounts of RNA. The N-terminus of the ORF2 peptide also contains a functional signal peptide, which was recently shown to be responsible for the peptide translocation in monkey kidney COS-1 cells (Zafrullah et al., 1999). Although this domain is truncated in our ORF2 construct, the expressed p50 seems to utilize a cellspecific compartmentalization route to form particles in the Tn5 cells. Cell-dependent proteolytic modification and particle formation The expression of foreign proteins in baculovirus systems opens the prospect of studying HEV capsid assembly with a single structural protein. A difference in the formation of rHEV particles is observed with the ORF2
T 5 1 PARTICLE OF RECOMBINANT HEPATITIS E VIRUS
baculovirus construct expressed in two insect cell lines. Though two peptides with measured molecular masses of 58 and 50 kDa are retrieved from the infected Tn5 cell lysate, the particle harvested from the culture supernatant comprises mainly the p50. The transient formation of HEV immunoreactive intracellular vesicles corresponds mainly to the appearance of p58 (Figs. 1B and 1C). The observed amounts of p50 and rHEV particles appear to correlate with the disappearance of p58 (Figs. 1A and 1D). This indicates that a posttranslational proteolytic cleavage event has taken place and is required for particle formation in the Tn5 cells. Furthermore, protein characterization based on terminal sequencing suggests that p58 is the proteolytic precursor of p50, where a cleavage occurs at residue 608 of the ORF2 peptide (in preparation). No rHEV particle was found in Sf9 culture supernatant when the cells were infected with a similar recombinant baculovirus of ORF2 constructs from the Burma and Pakistan strains (Li et al., 1997b; Robinson et al., 1998). In addition to 58- and 50-kDa peptides of the Burma strain, the presence of an additional peptide of intermediate size suggests an alternative strategy of posttranslational modification in the Sf9 cell. Thus, the deficiency of an appropriate proteolytic cleavage in the C-terminus may link to the failure of particle assembly in Sf9 cells. Constructs of ORF2 with sufficient C-terminal truncations further indicate that, in the Burma strain, only the expressed peptides of ;50 kDa assemble into a capsid particle in both Tn5 and Sf9 cells (unpublished data). Therefore, a valid cleavage of the ORF2 peptide appears to be the key step in forming the empty rHEV particles. The ORF2 peptide may be modified by various means, such as glycosylation, phosphorylation, sulfatation, or fatty acid conjugation when expressed in the two different insect cell lines (McAtee et al., 1996; Robinson et al., 1998). The single peptide, found in highly purified particles, shows a close fit with the predicted mass of the sequence, i.e., very close to 50 kDa (Fig. 2B). Nevertheless, the peptide presents several forms on isoelectric focusing (Fig. 2C). This may reflect variations due to a conjugated ligand, like any of these mentioned, or be a phenomenon occurring as a result of pH-dependent assembly. In relation to the observed dissociation (Fig. 2A), pH conditions seem to play a role in the particle stability. As a hierarchical model of protein folding seems to prevail (Baldwin and Rose, 1999), correct configuration of the subunit peptide, as well as of capsid subclusters, should be essential for the assembly of the capsid particle. This topic needs to be explored further, together with information on the function of the C-terminal domain, to reveal the critical elements in the ORF2 derived protein for rHEV particle formation. The fecal–oral dissemination of HEV demands that the native virion survive under extreme environmental conditions including those in the gastric and bile compart-
41
ments. Accordingly, the isolated native virus is resistant to acid and mildly alkaline conditions (Purcell, 1996). In contrast, above neutral pH, the rHEV particles fall apart into subassemblies containing oligomers of the p50 peptide (Fig. 2A). The rHEV particles are, however, stable under acidic conditions and retain their characteristic features, as seen by negative stain EM, on extensive purification at a pH of around 6.2. Despite the limited stability at and above neutral pH, the use of a slightly acidic buffer of low ionic strength allowed us to produce rHEV preparations suitable for cryo-EM imaging. The N-terminal 111 residues of the native ORF2 protein may be crucial for the incorporation of the viral genomic RNA into virus particles. Packaging of the genome may stabilize native HEV particles by multiple interactions between the capsid protein and the viral RNA. Specific interactions between the viral RNA and the capsid protein may serve as a checkpoint in the assembly of virions (Caspar and Klug, 1962). Since the rHEV capsid does not contain this N-terminal domain, the assembled particles should not contain a substantial amount of nucleic acid. Consequently, the rHEV particles may lack such a stabilizing scaffold and thus become fragile and easily damaged during the purification. The uneven composition of the truncated ORF2 peptides may result in a slight variation in the rHEV particle configuration. This may be one of the reasons for the heterogeneity of particle size observed during the image processing. This also explains why the iterative refinements, by the exclusion of the size-deviated particles, have been essential for a successful reconstruction (Cheng et al., 1995). Nevertheless, stability of the rHEV particles may be improved with constructs of baculovirus recombinants, which express the ORF2 peptide of appropriate length to form the capsid particles without going through the proteolytic modification in the insect cells (unpublished data). The conformation of T 5 1 and T 5 3 lattice The native HEV virion is a nonenveloped, spherical particle enclosing a 7.5-kb molecule of RNA. The virus particle was originally described to be between 270 and 300 Å in diameter. However, its diameter was found to range between 320 and 340 Å measured after sucrose gradient centrifugation (Bradley et al., 1988). Negatively stained rHEV particles in our study were between 220 and 230 Å in diameter, which is smaller than the averaged 270 Å obtained from ice-embedded specimen. Without a staining artifact, cryo-EM analysis often reveals better accuracy and a larger size estimate than the conventional procedure. Therefore, the actual diameter of the native HEV virions is likely to be in the range 385–410 Å (Table 1). The rHEV particle is assembled with 60 identical copies of 50-kDa subunits. In our 3D reconstruction, the outer radius of the particle is 135 Å, and the inner radius
42
XING ET AL.
of the central cavity is 50 Å, measured along the icosahedral twofold axis. Such a capsid shell provides a volume of only 5.3 3 10 5 Å 3 for potential RNA packing in the central cavity. Taking the averaged volume of a single hydrated ribonucleotide as 600–700 Å 3 (Johnson and Wikoff, 1998), the central cavity of the rHEV particle is only large enough to carry an RNA molecule smaller than 1 kb. To accommodate the entire 7.5 kb of the HEV RNA would be possible only if the capsid shell provides an internal volume with at least a diameter of 205–215 Å. With a retained capsid topology of 85-Å thickness, the capsid protein needs to build a much more expanded shell in order to hold the volume of the entire genome. Together with the size assessment of the capsid and the calculated indispensable volume of the viral RNA, the native virion may assemble into a capsid lattice with its triangulation number equal to, or larger than, 3. A similar polymorphism has been previously reported in some plant viruses where the capsid protein assembled into various triangulation numbers (Golden and Harrison, 1982). While a T 5 3 lattice is found in the native particles of turnip crinkle virus and tomato bushy stunt virus, trypsin cleavage of the N-terminal charged peptide causes the capsid protein to assemble into a T 5 1 particle (Golden and Harrison, 1982; Crowther and Amos, 1971). The truncated ORF2 protein with N-terminal deletion of 111 residues forms the rHEV particles. Although the N-terminal region may be essential for genome packaging, the omission of this domain does not seem to affect the particle assembly of the recombinant capsid. The estimated size of the native HEV virion is similar to the measured size of Norwalk virus, a T 5 3 calicivirus. On a schematic sphere of 395 Å in diameter, the dimension of the T 5 3 facet fits well to accommodate the rHEV dimers that are arranged equivalently in the T 5 1 icosahedral face (Fig. 6B). Individual faces of the T 5 1 icosahedron can serve to fill in the larger T 5 3 lattice as equivalent triangles, in which sets of three dimers are arranged around quasi-threefold axes (Caspar and Klug, 1962; Johnson, 1996; Johnson and Speir, 1997). As a result, a T 5 3 shell needs only 60 equivalent dimers added to the locations of quasi-twofold symmetry. In addition to the 30 sets of the X-shaped twists at icosahedral twofold axes, another 60 sets of similar linkage would have to allocate their central joints at the quasitwofold axis of an expanded shell (Fig. 6B). Thus, HEV virions most likely possess a T 5 3 lattice with an equivalent topological configuration of the capsomers as the T 5 1 rHEV particles. This explains why the native and the rHEV particles share important epitopes. Conclusion and perspectives HEV resembles calicivirus in genome organization and strategy of gene expression though it has not been definitively classified (Tam et al., 1991). Our structure
determination of rHEV indicates a similarity in the capsid morphology between HEV and caliciviruses. Like HEV, native calicivirus possesses a single capsid protein, which is encoded by the ORF2. The known structure of caliciviruses, such as Norwalk virus, shows that the particle is composed of 180 copies of the capsid protein in a T 5 3 surface lattice (Prasad et al., 1994a). Expressed in the baculovirus system, the dual-domain configuration of the recombinant calicivirus capsid appears similar to the rHEV particles in our study (Thouvenin et al., 1997; Prasad et al., 1994b). In contrast to the rHEV particle formation, the fulllength expression of Norwalk ORF2 produces two sizes of particles (White et al., 1997). While the large particle may have assembled a T 5 3 surface lattice, the smallsize particle of ;230 Å in diameter has been hypothesized to be of T 5 1 symmetry. In the absence of protease inhibitors, the capsid protein was found to be less stable in the small particle than it is in the normal particle during storage. An additional cleavage of the ORF2 peptide is required for the empty rHEV particle to form. Despite many shared features in morphology, the HEV capsid protein is about 10 kDa larger than that of the caliciviruses. The structural protein of HEV may contain additional functional domains at its termini, which are not used in T 5 1 particle formation. In calicivirus, the expressed capsid protein does not need to remove a large portion of N-terminus before the assembly of virus-like particles takes place. Our structural analysis shows that the rHEV T 5 1 particle probably represents the assembly principle of the native virus. The capsid protein configuration essentially provides not only the lateral interactions of dimers at the basal domains, but also the subunit contacts of the monomers in the protrusion region. It makes the rHEV T 5 1 particle a valuable system with which to explore how icosahedral particles may assemble with subunits in different quasi-equivalent environments with a small modification of their building blocks. The presentation of native antigenic epitopes suggests that such a virus-like particle is a putative vaccine against HEV. The dualdomain configuration may allow foreign epitopes to be presented at its external protruding domain without disturbance of the assembly principles. This renders the rHEV particle a widely useful prototype vaccine. MATERIALS AND METHODS Expression of self-assembled rHEV particles The baculovirus recombinant Ac5480/7126, encoding amino acids 112–660 of the HEV-ORF2 protein, was used as previously described (Li et al., 1997b). The virus stock was propagated in Sf9 cells (Riken Cell Bank, Tsukuba, Japan), which were maintained in the TC-100 medium (Gibco), supplemented with 8% fetal bovine serum and 0.26% Bacto tryptose phosphate broth (Difco Laborato-
T 5 1 PARTICLE OF RECOMBINANT HEPATITIS E VIRUS
ries, Detroit, MI), and cultured for 5 days at 26.5°C (Stewart and Possee, 1993). For the large-scale expression of rHEV particles, the cell line Tn5 (BTL-Tn5B1-4, Invitrogen) was infected with the seed virus at an m.o.i. of 10. Cells were incubated in Ex-cell 405 medium (JRH Biosciences, Lenexa, KS) for 6 days at 26.5°C. Purification of virus-like particles The rHEV particles were harvested on day 6 after infection. The supernatant of 100-ml cultures was centrifuged for 1 h at 10,000 rpm (20,000 g), room temperature, to remove debris. All subsequent steps were then performed at 0–4°C. The supernatant was spun at 25,000 rpm (113,000g) for 3 h in a Beckman SW28 rotor. The resulting pellet was resuspended in Ex-cell 405 medium, transferred to a 1.5-ml eppendorf tube, and cleared at 15,000 rpm for 30 min to remove remaining debris. The supernatant was then centrifuged in a Beckman TLA45 rotor at 45,000 rpm (125,000 g) for 2 h to pellet the rHEV particles. These particles were resuspended in 100–500 ml of Ex-cell 405 medium and temporarily stored at 4°C. The material was then diluted with 10 mM potassiumMES buffer at pH 6.2. Material used for cryo-EM was purified by centrifugation through a cushion of 10% sucrose, in potassium-MES buffer, pH 6.2, using a Beckman SW50.1 rotor at 35,000 rpm (147,000 g) for 2 h. The particles were then resuspended in the same buffer and loaded on a Sephacryl S-1000 (Pharmacia-Amersham Biotech, Uppsala, Sweden) chromatography column (10 mm 3 130 mm), equilibrated against 10 mM potassium– MES buffer, pH 6.2. The column was eluted at 0.1 ml/min and the rHEV particle-containing fractions were pooled. The particles were collected by centrifugation in a Beckman SW50.1 rotor at 30,000 rpm (108,000 g) for 1 h and resuspended in the same buffer. The quality of the rHEV particles was checked by negative-staining electron microscopy with uranyl acetate before freezing. Biochemical analyses Recombinant HEV particles used for biochemical analyses were purified by repeated pelleting and centrifugation in a CsCl gradient at 147,000 g for 2 h in a Beckman SW50.1 rotor. The rHEV particles were suspended to a concentration of 1 mg/ml in the buffers 10 mM potassium MES, pH 6.2, or 100 mM Tris–HCl, 150 mM NaCl, pH 8.3, and left overnight at 4°C. Analytical molecular sieving chromatography was performed at 4°C and run in the BioLogic System (Bio-Rad Laboratories, Hercules, CA) with a BioSilect SEC 250-5 column (5 mm i.d., height 300 mm). Samples of 25 ml were applied to the column and preequilibrated against the same buffer, and the chromatogram was developed at a flow of 0.4 ml/min. Polyacrylamide gel electrophoreses were run in the PhastSystem (Amersham Pharmacia Biotech) or in the MiniProtean electrophoresis system (Bio-Rad Laboratories).
43
Cryo-electron microscopy Cryo-EM was performed by established procedures with a Philips CM120 (Philips Electronics Instruments) operated at 120 kV (Cheng et al., 1992). Low-dose images of rHEV were recorded at about 1.4-mm underfocus to facilitate image processing and three-dimensional reconstruction from noisy data. The images were recorded at a nominal magnification of 45,000 with Kodak SO163 film. The accumulated exposure dose corresponds to ;7e 2/Å 2 at the specimen. Image analysis and three-dimensional reconstruction Micrographs exhibiting minimal astigmatism and specimen drift were digitized at 14-mm intervals, corresponding to 3.1-Å sampling of the specimen, on a SCAI microdensitometer (Zeiss, Jena, Germany). The magnification and microscope contrast transfer function were calibrated using CCMV particles as an internal calibration standard with cross-correlation procedures and radial density fitting (Cheng et al., 1994b, 1995). The structure of CCMV, with a diameter of 286 Å, was determined by X-ray crystallography (Speir et al., 1995). Images of the frozen-hydrated rHEV particle samples were analyzed with icosahedral symmetry procedures to reconstruct the three-dimensional structures of the viruses (Baker and Cheng, 1996; Crowther, 1971; Fuller et al., 1996) on DEC Alphastations (Digital Equipment Co., Maryland, MA). The origin and orientation of the individual images were first estimated based on cross-correlation with circularly averaged image and the Fourier phase residues with symmetry common lines. PFT was used to refine the origin and orientation with intermediate 3D reconstructed models (Baker and Cheng, 1996). In addition to the PFT correlation coefficient, Fourier phase residues of cross common-lines and the Fourier-ring correlation between projected reconstruction and raw-image data were used to aid in the particle selection. In the final three-dimensional reconstruction, all transform data of rHEV images below spatial frequency of (1/22) Å 21 were used in computing. On the basis of correlation coefficients, the best 73 of 347 images from a single micrograph were combined to compute reconstruction to 22-Å resolution. A representative Fourier transform of the rHEV particle images and the corresponding plot of the electron microscope contrast transfer function were used to constrain the resolution in the refinements. This resolution is within the limit imposed by the first zero of the contrast function of the electron microscope (Erickson and Klug, 1971; Cheng, 1992). The resolution of the reconstruction was assessed quantitatively by calculating the reliability index, based on the consistency between two divided data sets, as a function of spatial frequency (Cheng et al., 1992). We also assessed the correctness of the recon-
44
XING ET AL.
struction by visually comparing individual particle images with corresponding back-projected views of the reconstruction (Cheng et al., 1994a). A small magnification deviation was adjusted from projected images with a variety of real-space processing procedures (Cheng et al., 1994b, 1995). There is no handedness in a T 5 1 lattice of icosahedral symmetry, and the detailed reconstructed protein density was computed with discretionary enantiomorphism. ACKNOWLEDGMENTS We thank Drs. Mary Ng and Twan deVries for helpful discussions and Dr. Kenji Suzuki, Laurent Joffre, Sintau Kan, and Anna Bjorkman for their analyses and assistance in preparing rHEV particles suitable for structural study. The study is sponsored by the Medical Research Council (MFR-12175), the Natural Science Research Council (NFR11691), and the Structural Biology Network in Sweden. We also thank the Human Science Foundation and especially the Wallenberg Foundation for making the structural study possible at the Karolinska Institute.
REFERENCES Baker, T. S., and Cheng, R. H. (1996). A model-based approach for determining orientations of biological macromolecules imaged by cryoelectron microscopy. J. Struct. Biol. 116, 120–130. Balayan, M., Andjaparidze, A., Savinskaya, S., Ketiladze, E., Btaginsky, D., Savinov, A., and Poleschuk, V. (1983). Evidence for a virus in non-A non-B hepatitis transmitted via the fecal–oral route. Intervirology 20, 23–31. Baldwin, R. L., and Rose, G. D. (1999). Is protein folding hierarchic? II. Folding intermediates and transition states. Trends Biochem. Sci. 24, 77–83. Bi, S., Purdy, M., McCaustland, K., Margolis, H., and Bradley, D. (1993). The sequence of hepatitis E virus isolated directly from a single source during an outbreak in China. Virus Res. 28, 233–247. [Published erratum appears in Virus Res. 1994 Jul; 33(1), 98] Bradley, D., Andjaparidze, A., Cook, E., McCaustland, K., Balayan, M., and Stetler, H. (1988). Aetilogical agent of enterically transmitted non-A non-B hepatitis. J. Gen. Virol. 69, 731–738. Bradley, D. W. (1990). Enterically-transmitted non-A, non-B hepatitis. Br. Med. Bull. 46, 442–461. Caspar, D. L. D., and Klug, A. (1962). Physical principles in the construction of regular viruses. Cold Spring Harbor Symp. Quant. Biol. 27, 1–24. Cheng, R., Caston, J., Wang, G., Gu, F., Smith, T., Baker, T., Bozarth, R., Trus, B., Cheng, N., Wickner, R., et al. (1994a). Fungal virus capsids, cytoplasmic compartments for the replication of double-stranded RNA, formed as icosahedral shells of asymmetric Gag dimers. J. Mol. Biol. 244, 255–258. Cheng, R., Olson, N., and Baker, T. (1992). Cauliflower mosaic virus: A 420 subunit (T 5 7), multilayer structure. Virology 186, 655–668. Cheng, R., Reddy, V., Olson, N., Fisher, A., Baker, T., and Johnson, J. (1994b). Functional implications of quasi-equivalence in a T 5 3 icosahedral animal virus established by cryo-electron microscopy and X-ray crystallography. Structure 2, 271–282. Cheng, R. H. (1992). Correlation of cryo-electron microscopic and X-ray data and compensation of the contrast transfer function. Proc. Electr. Microsc. Soc. Am. 50, 996–997. Cheng, R. H., Kuhn, R. J., Olson, N. H., Rossmann, M. G., Choi, H. K., Smith, T. J., and Baker, T. S. (1995). Nucleocapsid and glycoprotein organization in an enveloped virus. Cell 80, 621–630. Crowther, R. A. (1971). Procedures for three-dimensional reconstruction
of spherical viruses by Fourier synthesis from electron micrographs. Philos. Trans. R. Soc. London Ser. B 261, 221–230. Crowther, R. A., and Amos, L. (1971). Three-dimensional image reconstructions of some small spherical viruses. Cold Spring Harbor Symp. Quant. Biol. 36, 489–494. Erickson, H. P., and Klug, A. (1971). Measurement and compensation of defocusing and aberrations by Fourier processing of electron micrographs. Phil. Trans. R. Soc. London B 261, 105–118. Fuller, S. D., Butcher, S. J., Cheng, R. H., and Baker, T. S. (1996). Three-dimensional reconstruction of icosahedral particles—the uncommon line. J. Struct. Biol. 116, 48–55. Ghabrah, T., Tsarev, S., Yarbough, P., Emerson, S., Strickland, G., and Purcell, R. (1998). Comparison of tests for antibody to hepatitis E virus. J. Med. Virol. 55, 134–137. Golden, J. S., and Harrison, S. D. (1982). Proteolytic dissection of turnip crinkle virus subunit in solution. Biochemistry 21, 3862–3866. Huang, C., Nguyen, D., Fernandez, J., Yun, K., Fry, K., Bradley, D., Tam, A., and Reyes, G. (1992). Molecular cloning and sequencing of the Mexico isolate of hepatitis E virus (HEV). Virology 191, 550–558. Jameel, S., Zafrullah, M., Ozdener, M., and Panda, S. (1996). Expression in animal cells and characterization of the hepatitis E virus structural proteins. J. Virol. 70, 207–216. Jiang, X., Wang, M., Wang, K., and Estes, M. K. (1993). Sequence and genomic organization of Norwalk virus. Virology 195, 51–61. Johnson, J. E. (1996). Functional implications of protein–protein interactions in icosahedral viruses. Proc. Natl. Acad. Sci. USA 93, 27–33. Johnson, J. E., and Speir, J. A. (1997). Quasi-equivalent viruses: A paradigm for protein assemblies. J. Mol. Biol. 269, 665–675. Johnson, J. E., and Wikoff, W. R. (1998). Macromolecular assembly: Chemical stabilization of a viral capsid. Curr. Biol. 8, 914–917. Khudyakov, Y., Favorov, M., Jue, D., Hine, T., and Fields, H. (1994). Immunodominant antigenic regions in a structural protein of the hepatitis E virus. Virology 198, 390–393. Khuroo, M. (1980). Study of an epidemic of non-A non-B hepatitis: Possibility of another human hepatitis virus distinct from post-transfusion non-A non-B type. Am. J. Med. 68, 818–824. Khuroo, M. S., Teli, M. R., Skidmore, S., Sofi, M. A., and Khuroo, M. I. (1981). Incidence and severity of viral hepatitis in pregnancy. Am. J. Med. 70, 252–255. Krawczynski, K. (1993). Hepatitis E. Hepatology 17, 932–941. Li, F., Torresi, J., Locarnini, S., Zhuang, H., Zhu, W., Guo, X., and Anderson, D. (1997a). Amino-terminal epitopes are exposed when fulllength open reading frame 2 of hepatitis E virus is expressed in Escherichia coli, but carboxy-terminal epitopes are masked. J. Med. Virol. 52, 289–300. Li, T., Yamakawa, Y., Suzuki, K., Tatsumi, M., Razak, M., Uchida, T., Takeda, N., and Miyamura, T. (1997b). Expression and self-assembly of empty virus-like particles of hepatitis E virus. J. Virol. 71, 7207– 7213. McAtee, C. P., Zhang, Y., Yarbough, P. O., Bird, T., and Fuerst, T. R. (1996). Purification of a soluble hepatitis E open reading frame 2-derived protein with unique antigenic properties. Protein Expression Purif. 8, 262–270. Muramatsu, T. (1993). Carbohydrate signals in metastasis and prognosis of human carcinomas. Glycobiology 3, 291–296. Prasad, B., Matson, D., and Smith, A. (1994a). Three-dimensional structure of calicivirus. J. Mol. Biol. 240, 256–264. Prasad, B., Rothnagel, R., Jiang, X., and Estes, M. (1994b). Threedimensional structure of baculovirus-expressed Norwalk virus capsids. J. Virol. 68, 5117–5125. Purcell, R. (1996). Hepatitis E virus. In “Fields Virology” (B. Fields, D. M. Knipe, and P. M. Howley, Eds.), Vol. 2, pp. 2831–2843. Lippincott– Raven, Philadelphia. Purcell, R., and Ticehurst, J. (1988). Enterically-transmitted non-A non-B hepatitis: Epidemiology and clinical characteristics. In “Viral Hepatitis and Liver Disease” (A. Zuckerman, Ed.), pp. 131–137. A. R. Liss, New York.
T 5 1 PARTICLE OF RECOMBINANT HEPATITIS E VIRUS Purdy, M., McCaustland, K., Krawczynski, K., Spelbring, J., Reyes, G., and Bradley, D. (1993). Preliminary evidence that a trpE-HEV fusion protein protects cynomolgus macaques against challenge with wildtype hepatitis E virus (HEV). J. Med. Virol. 41, 90–94. Robinson, R., Burgess, W., Emerson, S., Leibowitz, R., Sosnovtseva, S., Tsarev, S., and Purcell, R. (1998). Structural characterization of recombinant hepatitis E virus ORF2 proteins in baculovirus-infected insect cells. Protein Expression Purif. 12, 75–84. Speir, J. A., Munshi, S., Wang, G., Baker, T. S., and Johnson, J. E. (1995). Structures of the native and swollen forms of cowpea chlorotic mottle virus determined by X-ray crystallography and cryo-electron microscopy. Structure 3, 63–78. Stewart, L. M. D., and Possee, R. D. (1993). Baculovirus expression vectors. In “Molecular Virology: A Practical Approach” (A. J. Davison and R. M. Elliots, Eds.), pp. 227–256. IRL Press, Oxford, England. Tam, A., Smith, M., Guerra, M., Huang, C., Bradley, D., Fry, K., and Reyes, G. (1991). Hepatitis E virus (HEV): Molecular cloning and sequencing of the full-length viral genome. Virology 185, 120–131. Thouvenin, E., Laurent, S., Madelaine, M., Rasschaert, D., Vautherot, J., and Hewat, E. (1997). Bivalent binding of a neutralising antibody to a
45
calicivirus involves the torsional flexibility of the antibody hinge. J. Mol. Biol. 270, 238–246. Tsarev, S., Emerson, S., and Reyes, G. (1992). Characterization of a prototy strain of hepatitis E virus. Proc. Natl. Acad. Sci. USA 89, 559–563. White, L., Hardy, M., and Estes, M. (1997). Biochemical characterization of a smaller form of recombinant Norwalk virus capsids assembled in insect cells. J. Virol. 71, 8066–8072. Wilson, I. B. H., Gavel, Y., and von Heine, G. (1991). Amino acid distribution around O-linked glycosylation sites. Biochem. J. 275, 529–534. Wong, D., Purcell, R., Sreenivasan, M., Prasad, S., and Pavri, K. (1980). Epidemic and endemic hepatitis in India: Evidence for a non-A non-B hepatitis virus aetiology. Lancet 2, 876–879. Yarbough, P., Tam, A., Fry, K., Krawczynski, K., McCaustland, K., Bradley, D., and Reyes, G. (1991). Hepatitis E virus: Identification of typecommon epitopes. J. Virol. 65, 5790–5797. Zafrullah, M., Ozdener, M., Kumar, R., Kumar Panda, S., and Jameel, S. (1999). Mutational analysis of glycosylation, membrane translocation, and cell surface expression of the hepatitis E virus orf2 protein. J. Virol. 73, 4074–4082.
