Overall structure

The structure of recombinant E2s domain from Hepatitis E Virus (HEV) capsid was solved by the Single-wavelength Anomalous Dispersion (SAD) method from a synchrotron data set using Br heavy atom soaked crystals. The model was refined to a final R-factor of 0.198 (R_free = 0.240) at 2.0 Å resolution (Table 1). The E2s model consists of residues from Ser459 to Ala602. Five residues at the N-terminus had no interpretable electron density map and were not modeled. The asymmetric unit consists of one E2s molecule (Figure 1A and 1B). Notably, the symmetry related molecules maintain the tight dimeric architecture of E2s (Figure 1C and 1D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Structure of E2s.

(A) Ribbon diagram of the subunit of the E2s dimer, side view. (B) Top view of the subunit of E2s dimer showing the cavity. β-strands and random coils/turns are depicted in red and green respectively. N- and C-termini are labeled. The dimerization interface and groove region are labeled. (C) The E2s dimer. Subunit A is shown in yellow, subunit B in red. Dimeric interface residues from both subunits are shown in ball-and-stick representation. Notably, the asymmetric unit consists of one subunit of the dimer. This dimer is generated by crystallographic symmetry. These figures were prepared by using Molscript and Raster3D [32],[33]. (D) Close-up view of the dimer interface. Key residues involved in the dimerization are labeled. This figure was prepared using PyMol [34].

https://doi.org/10.1371/journal.ppat.1000537.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Data Collection and Refinement Statistics.

https://doi.org/10.1371/journal.ppat.1000537.t001

E2s mainly consists of a single domain that forms a β-barrel. Residues from β2, β3, β6 and β7, along with loops protrude at one side of the β-barrel structure to form a surface groove of approximately 15 Å in width and 11 Å in depth (Figure 1A). The β-barrel consists of nine anti-parallel β-strands running from one end of the molecule to the other. On one side of the β-barrel, there are three loops which connect adjacent β-strands, whereas on the other side, three loops and a double-strand β-sheet connect adjacent β-strands. The pore inside the β-barrel is highly hydrophobic in nature with side chains consisting of thirteen Leu, seven Val, two Ile, three Tyr, two Phe and one Trp, making up a total of 28 hydrophobic residues lining the inner pore surface (Figure S1). Approximate dimensions of the β-barrel are 30 Å in height and 13 Å in diameter. The top and the bottom side of the cavity of the β-barrel are blocked by loops connecting residues Thr586 and Ala590; Ala467 and Phe462 respectively (Figure 1B). Considering the hydrophobic nature and size of the cavity, we postulate that the cavity could have a role in recognizing hydrophobic ligands. However, the exact role of the cavity as yet remains to be established.

E2s is a dimer

E2s was found to exist exclusively as a 2.55 S particle (Figure 2A), which corresponds to a homodimer in solution, with an apparent molecular mass of 30,651±421Da (Figure 2B), as determined by analytical ultracentrifugation (AUC) experiments. These observations were consistent with a dimeric arrangement observed in the crystal structure (symmetry related subunits), with the dimer having approximate dimensions of 54×30×15 Å. The symmetry related subunits of the dimer are packed in a perpendicular fashion to each other, resulting in a maximum interaction (Figure 1C). The strong hydrophobic cluster at the dimer interface is maintained by side chains of residues Val503, Trp548, Thr552, Ala555, Tyr557, Tyr561, Val598 and Val600 of both subunits of the dimer. In addition, eight hydrogen bonding contacts (<3.5 Å) mainly from Arg542, Lys544, Ser546, Thr552, Thr553, Asn562, Thr564 and Ser566 of both subunits are involved in maintaining the dimer architecture. The surface area buried upon dimer formation was calculated using PISA server [10], and was found to be 1142.7 Ų, or 16.1% of the total surface of each subunit. The dissociation constant (Kd) was estimated as 397±283 nM by the sedimentation equilibrium method in AUC experiments (Figure 2B). The observed tight dimerization of E2s suggests a functionally important role of the dimer structure. Figure 3 shows the omit map of the C- terminal region which is involved in the dimerization.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Dimerization of E2s in solution was investigated by analytical ultracentrifugation (AUC).

(A) Sedimentation velocity experiment shows that E2s behaves as a single globular species having the sedimentation coefficient of 2.55S and a hydrated friction ratio of 1.30. (B) Sedimentation equilibrium experiment indicates that E2s mainly exists as a dimer with M.W. 30,651±421 Da. The dissociation constant of the E2s dimer, Kd was fitted as 397±283 nM using the self-association model.

https://doi.org/10.1371/journal.ppat.1000537.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Stereo view of the electron density map.

Simulated annealing Fo-Fc omit map of the C-terminal region of E2s, which is crucial for the dimerization. Residues Val600, Leu601 and all atoms within 2.0 Å were omitted prior to refinement. The map contoured at a level of 3σ. This figure was prepared using PyMol [34].

https://doi.org/10.1371/journal.ppat.1000537.g003

Structural comparison

All the positive-stranded RNA eukaryotic viruses have been shown to have a capsid protein folded as β-barrels with jelly roll topology [11],[12]. These capsid proteins mostly consist of a shell and a projection domain [13],[14]. The shell domain is a tightly closed domain protecting the viral RNA with its jelly roll β-barrel oriented such that the β-strands run tangentially to the particle surface [12]. However, the projection domain, if present, appears to be structurally more variable. The N-terminal region of the HEV capsid protein pORF2 is most likely to represent the shell domain, whereas E2s, the C-terminal region of pORF2, is considered as the projection domain of the HEV capsid protein. A search for topologically similar proteins within the PDB database performed with the program DALI [15] revealed no significant structural homology for E2s. Notably, E2s does not share any significant sequence or structural homology with any known viral proteins. It appears to represent the structurally more variable features of the projection domains.

It is worth mentioning here that Hepatitis E Virus was initially grouped with Caliciviruses, which comprise Norwalk Virus (NV) and San Miguel Sea lion Virus (SMSV). P2 domains of rNV and SMSV have been shown to be recognized by neutralizing monoclonal antibodies, which would suggest that they play a role in virus host interactions [13],[14]. Similar to the E2s domain of HEV, these P2 subdomains of rNV and SMSV are the most exposed regions, and contain determinants of strain specificity for Norwalk Virus (NV) and SMSV, respectively [13],[14]. Therefore, independent structural comparisons of the E2s dimers with the P2 domains of rNV and SMSV were performed. These three domains adopt the β-barrel architecture as shown in Figure 4A and 4B. Because the projection domain of the virus is highly structurally variable, the strand connectivity is not the same between E2s and the other two P2 domain structures (Figure 4B). In addition, the number of β-strands of these three domains is different (Figure 4B). The core part of this domain comprises of nine β-strands in E2s, whereas, there are seven β-strands in both rNV and SMSV P2 subdomain, respectively. However, cores of β-barrels of these three domains can be superposed up to certain extent (RMSD of 3.8 Å for 64 Cα atoms; 7% sequence identity with rNV P2 domain; RMSD of 3.4 Å for 87 Cα atoms; 8% sequence identity with SMSV P2 domain).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Structural comparison of E2s with P2 domains.

(A) Side-by-side ribbon diagram of HEV-E2s, the SMSV-P2 domain (pdb code 2gh8) and the rNV-P2 domains (Pdb code 1ihm). β-strands and α-helices are numbered. N- and C-termini are also labeled. These figures were prepared using Molscript and Raster3D [32],[33]. (B) Topology diagrams of HEV-E2s, the SMSV-P2 domain and the rNV-P2 domain. β-strands, α-helices and connecting loops are represented by red arrows, blue cylinders and green lines respectively.

https://doi.org/10.1371/journal.ppat.1000537.g004

Dimerization of E2 and HEV neutralization

Analysis of the dimer interface of E2s provided new insight into its tight dimeric architecture. In the case of NV and SMSV the P2 subdomains are dimers and are shown to interact with the host [13],[14],[16],[17],[18]. To verify that the dimerization of E2s is crucial for the host interaction of HEV, several mutations on dimer interface regions of E2/E2s were carried out (Figures 5 and 6), and their roles in destabilizing the dimer formation were studied. A total of eleven point mutants of E2/E2s were constructed. Non-reducing SDS-PAGE was used to verify the dimerization of these mutants in comparison with the standard MW markers (Figure 5). Independently, the dimeric nature of most of E2/E2s constructs was verified by analytical ultracentrifugation and gel filtration experiments (Figures S2 and S3). In addition to C-terminal residues, Arg542 and Tyr557 were revealed by the structure of E2s to be possibly involved in the dimerization. Arg542 is involved in two inter-subunit hydrogen bonding contacts. However, the mutant of Arg542 to Ala was still observed as a dimer in solution and reacted with mAb 8C11. Tyr557 is a part of the hydrophobic cluster, which consists of Tyr559, Tyr561 and Tyr584 from both subunits of the E2s dimer. The mutant of Tyr557 to Ala still showed up as a dimer in solution (Figure S2). However, it migrated as a monomer on a non-reducing SDS-PAGE (Figure 5). It was observed that this mutant retained mAb 8C11 reactivity only as a dimer, but lost its ability to recognize antibodies when it became a monomer in the presence of 0.1% SDS. Furthermore, we observed that five more mutants, namely, E2-T564A, E2-V598E, E2-A599E, E2-L601E and E2-A602E, existed exclusively as monomers in phosphate-buffered saline at pH 7.4. These mutants were found to lack the ability to interact with HEV-neutralizing monoclonal antibody, mAb 8C11 (Figure 5). These results confirm that HEV neutralization sites are associated with the dimeric nature of E2/E2s. In order to further verify the integrity of secondary structures in these mutants, circular dichroism (CD) spectra were recorded for wild-type E2, as well as, for all other mutants. In all cases, CD spectra showed the existence of similar secondary structures (Figure S4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Mutational studies on the dimer interface.

(A) The schematic representation of wild-type E2 and eleven point mutations targeting the dimer interface region. Secondary structural elements are shown for the E2s region. The mutated residues located on the β-strands and coils are shown in red and black, respectively. (B) These mutants and wild-type E2 were subjected to non-reducing SDS-PAGE and Western Blotting with the neutralizing mAb 8C11 and 8H3 to study the effects of these mutations on dimerization and neutralization, respectively. [+] denotes dimerization or reactivity with 8C11 or 8H3, [−] denotes loss of the respective property. Note that both the capacity to form dimers and the reactivity with mAb 8C11 and 8H3 were abolished simultaneously in six of these mutants: Y557A, T564A, V598E, A599E, L601E and A602E.

https://doi.org/10.1371/journal.ppat.1000537.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Mapping of site-directed mutation on E2s.

A transparent surface representation of the subunit of the E2s dimer is shown at two different orientations. (A) Shows the dimerization mutants and (B) shows the groove region mutants. Figure 6(B) is 180° rotated with respected to Figure 6(A). Further the view of Figure 6(A) is related to the view of Figure 1(A) with an anticlockwise rotation of 90°. All mutated residues are shown in ball and stick model. Residues playing roles in the E2s dimerization are shown in blue. Residues in the groove region that were mutated to study the reactivity of mAb are shown in magenta. This figure was prepared using PyMol [34].

https://doi.org/10.1371/journal.ppat.1000537.g006

Moreover, our studies confirmed that various constructs of E2, such as E2a, p239 and E2s, contain the major dimerization site in the C-terminal region between aa597 and aa602 (Table S1) [6]. This is supported by the structure of E2s, which reveals the participation of hydrophobic residues of the β11 strand (Gly589 to Ala602) in forming the dimeric interface.

Previously we had shown that mAb 8C11 and 8H3 neutralize HEV infection in monkeys [8] and block virus-host interaction [7]. Consequently, all E2 constructs were verified for the ability to form dimers and were tested for their reactivity with the neutralizing monoclonal antibodies (Figure 5). The structure of E2s reported here is the representative structure for all E2 constructs. E2s is the shortest among all constructs of E2 (Figure 5) that can dimerize and recognize HEV antibodies, in a way similar to other E2 constructs, as well as the native HEV [8]. Furthermore, E2s contains most of the conserved residues. More importantly, it contains the crucial dimerization region of E2. The structure of E2s shows that C-terminal residues, such as Ala597, Val598, Ala599, Leu601 and Ala602, are directly engaged in dimer formation, with additional interactions from Tyr557 further stabilizing dimers.

HEV antibody recognition and E2s

We have investigated the functional relevance of E2 constructs through a panel of 33 previously reported mAb that are reactive against the E2 fragment (aa 394–606) and p239 (aa 368–606) (Table S1). These antibodies target the unique structural features of E2/E2s. Thirteen of them are linear epitope - reactive antibodies and 20 of them are conformational determinants. Moreover, two of 13 linear epitope-reactive antibodies and 15 of 20 conformational determinant -reactive antibodies can bind genotype I and/or genotype IV HEV. We have further identified that two of them (mAb 8C11 and 8H3) can neutralize the infectivity of HEV, thus preventing the virus from infecting primates [8]. Notably, one of the recombinant E2 constructs, p239 (ORF2 aa368–606), forms a virus like particle (VLP) with a diameter of 23 nm [9]. The p239 VLP was found to specifically adsorb and penetrate susceptible host cells similarly to live HEV viruses. This interaction could be blocked by the neutralizing mAb 8C11 and 8H3 [7]. This showed that the recombinant E2 constructs shared many common features of the native virus. Thus, it can be suggested that the host and antibody interaction sites/regions of HEV are similar if not identical.

Based on our studies on various E2 constructs, we hypothesize that the surface exposed groove region is the most likely antibody recognition site of HEV (Figure 1A). To verify this hypothesis we have carried out several mutations targeting the groove region, and studied their interactions with mAb 8C11 and 8H3 (Figures 6, 7 and S5). Interestingly, the structure of the groove region of E2s is unique, and no such groove region was observed in the P2 domain in either rNV or SMSV [13],[14]. Our analysis showed that (1) all these mutants remained as dimers, (2) only D496A mutant did not recognize mAb 8C11 and 8H3, and (3) mutants E479A, Y485A, I529A and K534A abrogated the reactivity of mAb 8H3, while retaining the 8C11 reactivity (Figure 7B). These mutagenesis studies suggest that the groove region may contain a neutralization site. It is possible that the required positioning of the groove region might be preserved only in the dimeric form of E2, and it might collapse when it becomes a monomer (Figure 7B, lane E2-N, H). Hence we propose that the dimeric form of E2 is essential to position these groove regions for the neutralization of HEV. We see that antibody interactions are associated with the E2 region of HEV, a region which may mediate the first contact with the host cell to initiate viral infection [7]. Thus, the capsid protein domain E2 is a functionally important region of HEV, and it reacts against different mAb that are capable of HEV immune capture and virus neutralization.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Mutational studies on the groove region.

(A) The schematic representation of wild-type E2 and nine point mutations targeting the solvent-accessible residues near the groove region. (B) The wild type E2 and its mutants were subjected to non-reducing SDS-PAGE and Western Blotting with the HEV-neutralizing antibody 8C11 or 8H3. In this figure the lanes with H indicate samples in the reduced condition (i.e. these samples were heated up to 100°C for 3 minutes). These samples were mainly resolved as monomers. The lanes with N indicate samples in the non-reducing condition (i.e. these samples with 0.1% SDS, no BME and were not heated). These samples were resolved mainly as dimers. All nine mutants remained as dimers. Western Blotting showed that the dimeric E2 wild type and eight mutants were reactive with mAb 8C11. Of these, only E479A, Y485A, I529A, K534A and D496A abolished the 8H3 reactivity. Interestingly, mutant D496A abolished the HEV neutralizing antibodies 8C11 and 8H3 reactivity while maintaining the dimeric arrangement.

https://doi.org/10.1371/journal.ppat.1000537.g007

Plasmid and strain construction

The E2s gene (encoding HEV ORF2 aa 455–602) was PCR amplified from the E2 gene (ORF2 aa 394–606) [6], then subcloned into a vector pMD 18-T with a TA cloning kit (Takara, Dalian, China), and ligated to a non-fusion pTO-T7 expression plasmid [20] in NdeI/EcoRI restriction sites. The alanine scanning mutageneses on E2s or E2 were carried out with site-directed PCR reactions. All resultant DNA fragments were cloned into a pTO-T7 plasmid. The E.coli ER2566 strain was transformed to express these non-fusion proteins with an additional start Met at N-terminus.

Purification and crystallization

All recombinant proteins formed inclusion bodies in bacterial cells. Inclusion bodies were separated from cellular debris by extensive washing with buffer containing 2% Triton X-100, and then dissolved by homogenization with 4 M (for E2 and its mutants) or 8 M urea (for E2s and its mutants). Proteins were renatured by dialysis against phosphate-buffered saline at pH 7.4 (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄ and 2 mM KH₂PO₄) at room temperature and further purified by gel filtration HPLC using a TSK gel SW3000 25 mm×60 cm column (TOSOH, Japan). E2s was dialyzed with pure H₂O and concentrated to 15 mg/ml for crystallization. Crystallization drops containing 1 µl E2s protein solution (15 mg/ml) and 1 µl reservoir solution were equilibrated by hanging drop vapor diffusion at 21°C. The best crystals were grown from a reservoir solution consisting of 0.1 M HEPES pH 7.5, 12% PEG3350 and 5 mM of cobalt chloride hexahydrate, nickel (II) chloride hexahydrate, cadmium chloride dehydrate and magnesium chloride hexahydrate. Crystals measuring ∼0.2 mm in length grew over the course of 3 days, belonged to the space group R32, and contained one molecule in the asymmetric unit. Obtaining the diffraction quality single crystals and the phasing were the most challenging task in this project. The present data set is the best of over one hundred data sets collected. The X-ray data collection and refinement statistics are given in Table 1.

Data collection, structure solution and refinement

Crystals were cryo-protected in the reservoir solution supplemented with 25–30% glycerol, and flash cooled at 100 K. The structure was determined using Br heavy atom soaked crystals of recombinant E2s protein by the single-wavelength anomalous dispersion (SAD) method. X-ray diffraction data were collected at the beamline X12C, Brookhaven National Laboratory using a Quantum-210 CCD detector (ADSC). A single data-set was collected at the wavelength corresponding to the peak. All data-sets were processed with HKL2000 [21]. Three Br sites of an asymmetric unit were located by using the program BnP [22]. Phases were further improved by density modification using RESOLVE [23],[24], which gave a final overall Figure of merit 0.70. Over 67% of the backbone atoms of the model were built by the RESOLVE iteration method [25]. Remaining residues of the molecules were added after several cycles of manual model building by using O [26], followed by refinement using CNS [27]. Finally, 244 well-defined water molecules were added, and refinement was continued until the R-value converged to 0.198 (R_free = 0.240) for reflections I>σ (I) to 2.0 Å resolution. The model had good stereochemistry, with all residues within allowed regions of the Ramachandran plot (Table 1) analyzed by PROCHECK [28].

Analytical Ultra-Centrifugation (AUC)

The AUC velocity experiment was to independently establish the homogeneity of the molecules in solution, and subsequently determine their molecular mass by equilibration experiment. Sedimentation velocity (SV) and sedimentation equilibrium (SE) experiments were conducted at 20°C on a Beckman XL-A analytical ultracentrifuge, equipped with absorbance optics and an An60-Ti rotor. The molecular mass and partial specific volume of E2s, the solvent density and viscosity of the solvent were calculated from the amino acid or buffer composition using the program SEDNTERP (John Philo, Amgen, Thousand Oaks, CA, and RASMB). For SV experiments E2s was diluted to 1 mg/ml (∼1.2 OD_{280 nm}) in phosphate-buffered saline at pH 7.4. The rotor speed was set at 60,000 rpm for the highest resolution. The sedimentation coefficient and f/f₀ were obtained with c(s) method [29] using the Sedfit software (kindly provided by Dr. P. Schuck, National Institutes of Health, http://www.analyticalultracentrifugation.com). Similarly, for SE experiments E2s was diluted to 0.69OD, 0.36OD and 0.14OD in phosphate-buffered saline at pH 7.4 (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄ and 2 mM KH₂PO₄). Samples were centrifuged, first at 18,000 rpm, and, subsequently, at 21000, 24000 rpm, respectively, and finally at 42000 rpm for solute depletion. Data sets were processed using the program Origin (Beckman) for detecting multiple equilibria and were fitted to a single ideal species model and a self-association monomer-dimer model using a nonlinear least squares fit [30]. In subsequent models the monomer molecular mass was fixed at the value calculated from the E2s amino acid sequence (15,788 Da).

SDS-PAGE and Western Blotting (WB)

Analysis of proteins by SDS-PAGE was performed according to the method of Laemmli with minor modifications [31]. Polyacrylamide gels with 12% or 15% acrylamide in the separating gel and 5% in the stacking gel were used. Protein samples were mixed with equal volumes of 2× loading buffer (100 mmol/L Tris-HCl pH 6.8, 200 mmol/L BME, 4% SDS, 0.2% Bromophenol blue and 20% Glycerol). Sample mixtures were heated at 100°C for 3 minutes and subsequently loaded onto the separating gel. For the non-reducing SDS gel, the buffer contained only 0.1% SDS, no BME, and the sample was not boiled.

For Western Blotting experiments, separated proteins were transferred from an SDS gel onto a nitrocellulose membrane. Membranes were soaked in 1∶2,000 diluted HEV-neutralizing monoclonal antibody (8C11 or 8H3), incubated at room temperature for 1 h, and subsequently washed with 0.2% Tween 20 in phosphate-buffered saline (at pH 7.4). The bound antibody was detected with alkaline phosphatase conjugated secondary antibody (DAKO), and developed with a mixture of nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate.

Gel filtration chromatography

Purified proteins were passed through a Superdex 75 10/300GL column equilibrated in phosphate-buffered saline (at pH 7.4) using an AKTA Explorer 100 (GE, USA) at a flow rate of 0.5 ml/min. Molecular weights of eluted proteins were determined by using the following molecular weight standards: Conalbumin (75 kDa), Ovalbumin (43 kDa), Carbonic Anhydrase (29 kDa), Ribonuclease A (13.7 kDa), and Aprotinin (6.5 kDa) (GE, USA).

Circular dichroism spectrometry

Far UV spectra (260–190 nm) of the E2 wild-type and mutants were measured in a phosphate-buffered saline at pH 7.4 at room temperature using a J-810 spectropolarimeter (JASCO, Tokyo), with 0.1 cm path length and stoppered cuvettes. A total of 5 scans were recorded and averaged for each spectrum, and the baseline was subtracted.

Coordinates

Coordinates have been deposited in the Protein Data Bank (accession code 3GGQ).