Origin, antigenicity, and function of a secreted form of ORF2 in hepatitis E virus infection

HEV-Infected Cells Release a Large Amount of Nonvirion-Associated ORF2 both in Vitro and in Vivo.

Previous studies have demonstrated that HEV is released from infected cells as quasi-enveloped particles (eHEV), and the capsid is hidden within a host membrane and cannot be recognized by anticapsid antibodies unless treated with detergent (11). However, using a commercial HEV antigen ELISA, we detected a high level of the ORF2 protein in the supernatant of HEV-infected cell cultures irrespective of detergent treatment (Fig. 1A). Rate-zonal ultracentrifugation revealed that the majority of the ORF2 protein in the culture supernatants was not associated with infectious virions (Fig. 1B). Comparison of the protein abundance of ORF2 and ORF3 in the culture supernatant and those in purified eHEV particles indicated that less than 0.1% (0.009/14.8 × 100%) of the total ORF2 protein in the supernatants was associated with virions (Fig. 1C). Similarly, the majority of ORF2 protein present in the serum of an HEV-infected macaque was not virion-associated (Fig. 1D). Using HEV virus-like particles (VLP) p239 as an ELISA standard, we estimated that the concentration of the ORF2 protein in HEV persistently infected culture supernatants was ∼1.5–3 μg/mL, whereas 10⁹ highly concentrated virions contained ∼0.2 μg ORF2 (Fig. 1E). The detection limit of the commercial HEV antigen ELISA for virions was estimated to be 6.2 × 10⁷ particles [based on HEV genome equivalents (GE)]. Because HEV RNA titers in culture supernatants and patient serum samples rarely exceed 10⁷ GE/mL (11, 19), the secreted form of ORF2 is likely the exclusive target of the commercial antigen ELISA. In the following experiments, we used the term ORF2^S for the secreted, nonvirion-associated ORF2, and ORF2^C for the actual capsid protein.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

HEV-infected cells release a large amount of nonvirion-associated ORF2 protein both in vitro and in vivo. (A) Detection of HEV ORF2 protein in culture supernatants (sup) (1:32 dilution) of mock or HEV (Kernow C1/p6) persistently infected Huh-7 cells, purified quasi-enveloped HEV particles (eHEV) (10⁹ GE), and HEV VLP p239 (12.5 ng) with or without 1% Nonidet P-40 treatment. HEV ORF2 protein was measured by a commercial HEV antigen ELISA (Wantai). Signal to cut-off (S/CO) ratio was determined based on optical density values measured at 450 nm. Data represent mean ± SEM from two independent experiments each in duplicate wells. (B, Upper) Rate-zonal gradient profile of HEV RNA, ORF2 protein, and infectious virions in the culture supernatant of HEV-infected Huh-7 cells. Fractions were collected from the top of the gradient. (Lower) Western blots of HEV ORF2 and ORF3 in the same gradient fractions (each lane contained two consecutive fractions) with a chimpanzee convalescent serum (Ch1313) and a rabbit anti-ORF3 antibody, respectively. (C) Protein abundance of HEV ORF2 and ORF3 in culture supernatants of HEV persistently infected Huh-7 cells and gradient purified eHEV particles (∼10⁹ GE). The relative protein abundance is shown below each blot. (D) Rate-zonal gradient profile of HEV RNA and ORF2 protein in a serum sample from a genotype 3 HEV (Kernow C1)-infected rhesus macaque. (E) ELISA quantitation of HEV ORF2 protein in the culture supernatant with different dilutions and Nonidet P-40-treated purified eHEV particles. Serially diluted VLP (p239) was used for creating a standard curve.

ORF2^S Exists as a Glycosylated Dimer.

The protein size of ORF2^S was larger than ORF2^C (84 kDa vs. 72 kDa) (Fig. 1C), suggesting that ORF2^S is posttranslationally modified. It has been shown that ORF2 becomes glycosylated when overexpressed (20). Consistent with this, treatment of ORF2^S with either PNGase-F, which removes the N-linked glycans from proteins, or O-glycosidase and neuraminidase, which remove O-linked glycans, reduced the size of ORF2^S (Fig. 2A) [ORF2^S (WT)]. The three putative N-linked glycosylation sites described in previous studies likely contributed to the glycosylation of ORF2^S (Fig. 2A) [ORF2^S (N137/310/562Q)]. In contrast, the size of ORF2^C was not reduced by these treatments (Fig. 2A, virions). Moreover, treatment of HEV-infected cells with brefeldin A, an inhibitor of protein transport from the endoplasmic reticulum (ER) to the Golgi apparatus, resulted in intracellular accumulation of slower-migrating ORF2 proteins, along with a dramatic reduction of ORF2^S in the culture supernatant (Fig. 2B). These results demonstrated that ORF2^S is released through the classic secretory pathway and its release is rather efficient.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

The secreted form of ORF2 (ORF2^S) is glycosylated and dimerized. (A) Effects of PNGase F, O-glycosidase (O-Gly-ase), and neuraminidase (NA) treatment on ORF2 in virions and in supernatants of Huh-7 cells transfected with a WT HEV RNA or a variant carrying mutations in three putative N-linked glycosylation sites (N137/310/562Q). ORF2 was detected by a chimpanzee convalescent serum (Ch1313). (B) HEV persistently infected Huh-7 cells were treated with brefeldin A (BFA) for 24 h or left untreated (Ctrl). Cells and supernatants were collected and analyzed by immunoblotting with indicated antibodies. (C) Effects of β-mecaptomethanol (β-ME) and heat treatment (95 °C, 10 min) on the size of ORF2^S in a native protein gel. (D) Determination of the size of ORF2^S by size-exclusion chromatography. The majority of ORF2^S was eluted as peak 1 with an estimated size of 180 kDa. The column contaminants were eluted in the second peak (peak 2). The presence of ORF2^S in peak fractions was confirmed by Western blot analysis using Ch1313.

ORF2^S sedimented more slowly in a rate-zonal gradient than HEV VLP p239, which consists of 60 copies of a truncated ORF2 protein comprising amino acids 368–606 (21) (Fig. S1). This suggested that ORF2^S does not exist in a high-order particulate form. In addition to the major 84-kDa species, we routinely detected a minor ∼170-kDa species, suggesting that ORF2^S forms a dimer (Fig. 2C). Unboiled ORF2^S migrated predominantly as an ∼170-kDa species, indicating that the majority of ORF2^S exists in a dimeric form. The ratio of ORF2^S dimers and monomers did not change under a nonreducing condition (Fig. 2C), indicating that ORF2^S dimerization does not require disulfide bonds, a property that is shared with ORF2^C (22). The dimeric nature of ORF2^S was further confirmed by size-exclusion chromatography (Fig. 2D).

ORF2^S and ORF2^C Are Different Translation Products of the ORF2 Gene.

The difference in the size and glycosylation pattern between ORF2^S and ORF2^C suggested that either they are the same proteins subjected to differential posttranslational regulations, or they are different translation products of the ORF2 gene. Sequence alignment revealed a highly conserved internal AUG codon located at 15 codons downstream of the presumed AUG start codon for the capsid (Table S1). We reasoned that both AUGs might be active in infected cells. Because the N terminus of the presumed full-length ORF2 protein contains a putative signal peptide (amino acids 1–23) (www.cbs.dtu.dk/services/SignalP), translation from the first AUG would produce a full-length protein with a signal peptide that could direct the protein into the secretory pathway (i.e., ORF2^S), whereas translation from the internal AUG codon would disrupt the signal sequence, producing a cytoplasmic protein that is responsible for virion assembly (i.e., ORF2^C) (Fig. 3A). To test this hypothesis, we generated mutant HEV RNAs in which the two AUG codons were inactivated, either alone (mut1 and mut2) or in combination (mut1+2). All mutants replicated at comparable levels to the WT HEV, as evidenced by the similar intracellular HEV RNA levels 5 d after RNA transfection (Fig. 3B). We then determined the levels of intracellular and extracellular ORF2 by indirect immunofluorescence assay (IFA) and Western blot analysis (Fig. 3 C and D). Mutation of the first AUG codon resulted in a substantial decrease of intracellular ORF2 expression and a complete loss of ORF2^S in the culture supernatant (Fig. 3 C and D, mut1). In contrast, mutation of the internal AUG codon resulted in no changes in either the intracellular or the extracellular ORF2 levels (Fig. 3 C and D, mut2). Mutation of both AUG codons resulted in a complete loss of both intracellular and extracellular ORF2 expression (Fig. 3 C and D, mut1+2). Importantly, infectious virions were produced in cells transfected with WT or mut1 HEV RNA, but not in cells transfected with mut2 or mut1+2 RNA (Fig. 3 E–G). Similar results were obtained with a genotype 1 HEV strain (Fig. S2). Collectively, these results demonstrated that the first AUG codon, which has long been believed to be the initiation codon for the HEV capsid protein, is indeed responsible for the translation of the secreted form of ORF2 (ORF2^S), whereas the highly conserved internal AUG codon encoding methionine (Met) 16 is responsible for translation of the capsid protein ORF2^C.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

ORF2^S and ORF2^C are different translation products of the HEV ORF2 gene. (A) The N terminus sequence of the HEV subgenomic RNA (Kernow C1/p6) and strategies for site-directed mutagenesis. The AUG start codons for ORF3, ORF2^S, and ORF2^C are underlined, and the corresponding protein sequences are shown below. The first AUG and the internal AUG codons of the ORF2 gene are mutated either singly (mut1 and mut2) or in combination (mut1+2) without altering the amino acid sequence of ORF3. The putative signal peptide cleavage site in ORF2^S is indicated (arrowhead). (B) qRT-PCR analysis of intracellular RNA levels in Huh-7 cells transfected with in vitro transcripts of WT or mutant HEV RNA (5-d posttransfection). Data are shown as mean ± SEM of two independent experiments each performed in duplicate wells. LOD, limit of detection. (C) Huh-7 cells transfected with WT or mutant HEV RNA for 5 d were stained with a chimpanzee convalescent serum Ch1313 (green). Nuclei were stained with DAPI (blue). (Scale bars, 100 μm.) (D) Immunoblots of intracellular and extracellular ORF2 and ORF3 proteins at 5 d after HEV RNA transfection into Huh-7 cells. β-Actin was used as a loading control. ORF2 was detected by Ch1313. ORF3 was detected with a rabbit anti-ORF3 antibody. (E) Isopycnic gradient profile of intracellular HEV RNA (as a surrogate marker of virions) in Huh-7 cells transfected with different HEV RNA (15-d posttransfection). Cells were freeze-thawed three times and treated with RNase before gradient centrifugation. Fractions were collected from the top of the gradient. (F and G) HepG2/shMAVS cells were inoculated with equal amounts of WT or mutant HEV (10,000 GE per cell) isolated from HEV RNA-transfected Huh-7 cells. Cells were stained with Ch1313 (green) and DAPI (blue) at 5-d postinoculation (F). (Scale bars, 100 μm.) Supernatant HEV RNA levels (5 d after inoculation) were measured by qRT-PCR (G). Data are shown as mean ± SEM of two independent experiments each performed in duplicate wells.

To corroborate the above results, we adopted a transcomplementation system developed by Ding et al. (23). In this system, ORF2 and ORF3 are provided in trans by lentivirus vectors, and transfection of a replication-competent HEV subgenomic RNA into cells expressing ORF2 and ORF3 results in infectious virus particle assembly and release. We created cell lines that stably express ORF3 along with different ORF2 variants (WT, mut1, and mut2), followed by transfection with a GFP-expressing HEV replicon RNA that lacks both ORF2 and ORF3. We found that cells expressing WT or mut1 ORF2 produced comparable amounts of infectious virions, but cells expressing mut2 produced none (Fig. S3). These results again demonstrated that the internal AUG codon is responsible for initiating the translation of the capsid protein.

Montpellier et al. (18) recently reported that the N-terminal sequence of the ORF2 in purified HEV virions begins at leucine (Leu) 14, instead of Met16. However, cells expressing ORF2 that contains amino acids 14–660 barely produced any infectious particles (Fig. S3). The low level of infectious virions produced from these cells was likely due to leaky translation of ORF2^C from Met16, as inactivation of this internal AUG codon led to a further reduction in infectious virus production. Mass spectrometry (MS) analyses of highly purified HEV virions identified Leu17 as the N-terminal residue of ORF2^C (Fig. S4), providing further evidence that ORF2^C is translated from the internal AUG start codon.

Interestingly, the ORF2 protein level in mut1 RNA-transfected cells was much less than in cells transfected with WT or mut2 RNA. This result indicates that most intracellular ORF2 is derived from ORF2^S. In agreement with this, a Gaussia luciferase placed behind the AUG codon for ORF2^S was more efficiently translated (∼fivefold) than that placed behind the AUG codon for ORF2^C (Fig. S5). Confocal microscopy indicated that neither the intracellular ORF2^S nor ORF2^C was associated with the ER or the Golgi apparatus (Fig. S6), suggesting that the intracellular ORF2^S predominantly resides in the cytoplasm. To better understand how ORF2^S accumulated in the cytoplasm, we determined the N-terminal and C-terminal sequences of both extracellular and intracellular ORF2^S by MS. The result indicated that the extracellular ORF2^S included amino acids 34–660, consistent with a recent report (18), while the intracellular ORF2^S included amino acids 12–660 (Fig. S4). The absence of the first 11 amino acids would disrupt the transmembrane domain of the signal sequence, providing a plausible explanation for the observation that some ORF2^S was retained in infected cells.

ORF2^S Is Antigenically Similar but Not Identical to ORF2^C in Virions.

Because the commercial HEV antigen ELISA detects ORF2^S, VLP, and virions (Fig. 1A), we sought to compare their antigenicity using a panel of conformation-dependent monoclonal antibodies (mAbs) that recognize six different antigenic clusters (C1–C6) on the HEV capsid (24) (Fig. 4A). As some of these mAbs were generated using genotype 1 ORF2, we included different genotypes for comparison. A recombinant ORF2 dimer (rDimer) comprising amino acids 394–606 was also included (21). mAbs 12F12 and #4, which are used in the commercial HEV antigen ELISA (25), recognized all four forms of ORF2 (rVLP, rDimer, ORF2^S, and virions) of all genotypes tested (g1, g3, and g4), therefore were used to adjust the input amounts of different ORF2 forms.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

ORF2^S antigenically overlaps with virions and interferes with antibody-mediated neutralization but not entry of HEV. (A) Location of six distinct antigenic clusters (C1–C6) on the protrusion domain of the HEV capsid (genotype 1). (B) Comparison of antigenicity of HEV recombinant VLP (rVLP, genotypes 1 and 4), recombinant ORF2 dimer (rDimer, genotypes 1, 3, and 4), ORF2^S (O^S), and virions (V) [Xinjiang-1 (genotype 1), Kernow C1/p6 (genotype 3)] in a sandwich ELISA with a rabbit anti-ORF2 antibody as a capture antibody and indicated anti-ORF2 monoclonal antibodies as the detection antibody. (C) Nonenveloped HEV or eHEV particles (1,000 GE per cell) were incubated with indicated concentrations of purified ORF2^S (37 °C, 1 h) before inoculation of HepG2/shMAVS cells. Relative infectivity was calculated based on the numbers of foci determined by indirect IFAs with Ch1313 5 d later. Data represent mean ± SEM of two independent experiments each performed in duplicate wells. (D) Serially diluted monoclonal (6H8 and 9F7) or polyclonal (Ch1313) anti-HEV antibodies were mixed with indicated concentrations of purified ORF2^S at 37 °C for 2 h before incubation with HEV (1 × 10⁷ GE) for another 2 h. The mixture was then inoculated to HepG2 cells (4 × 10⁴ cells per well). Percent neutralization was calculated based on the number of HEV⁺ foci 5 d after inoculation. Shown are representative results from two independent experiments.

This analysis revealed substantial antigenic overlap between ORF2^S and other forms of ORF2-including virions (Fig. 4B). The most noticeable difference was the loss of the C3 and C4 epitopes on ORF2^S for genotype 3 HEV. These epitopes are located at the top region of the protrusion (P) domain of the capsid and have been implicated for receptor binding (26).

It is worth noting that mAbs specific to the C5 and C6 epitopes, which are located in the “groove zone” of ORF2 in the crystal structure, possess potent neutralizing activity (24). This suggests that major neutralizing epitopes are preserved in ORF2^S. Unexpectedly, the C2 epitope, which is recognized by a pan-genotypic mAb 8G12 (27), appeared to be lost in the Kernow strain that we used in this study. The loss of this epitope is likely due to an amino acid change at position 554 (K554 → R), because K554 constitutes a key residue for the C2 epitope and is highly conserved in human HEV strains (27). Indeed, changing R554 to the conserved residue K reinstated the C2 epitope on both ORF2^S and virions (Fig. 4B, p6 554K). Interestingly, this mutation also reinstated the C3 epitope on virions, possibly due to a change in the local conformation as an adjacent residue T553 is important for the C3 epitope (24). However, the C3 epitope remained lost on ORF2^S in the 554K mutant.

ORF2^S Does Not Block HEV Entry, but Impairs Antibody-Mediated Neutralization.

Because ORF2^S antigenically overlapped with virions and was released in great excess to virions, we investigated if ORF2^S inhibits HEV entry by competition for the cell receptors. This appeared not to be the case, as preincubation of either nonenveloped HEV or eHEV particles with high concentrations of purified ORF2^S had no impact on virus infectivity (Fig. 4C). Consistent with this, cell binding of ORF2^S was much less efficient than VLPs (Fig. S7A).

We next determined the effect of ORF2^S on antibody-mediated neutralization of HEV. Purified ORF2^S dose-dependently inhibited the neutralizing activity of mAb 9F7, a potent neutralizing antibody recognizing the C6 epitope (24) (Fig. 4D). However, under the same condition, no inhibition was observed for mAb 6H8, which recognizes the C4 epitope. These results are consistent with the presence of the C6 epitope and absence of the C4 epitope on ORF2^S. To investigate if ORF2^S also affects neutralization by antibodies generated by infection, we used a hyperimmune plasma from a chimpanzee infected with HEV (Ch1313) (28). A dose-dependent inhibition by ORF2^S was also observed. The inhibitory effect of ORF2^S on antibody-mediated neutralization of HEV was also confirmed with a VLP cell-binding assay (Fig. S7B).