|Home | About | Journals | Submit | Contact Us | Français|
The ORF3 protein of hepatitis E virus (HEV) is a multifunctional protein important for virus replication. The ORF3 proteins from human, swine, and avian strains of HEV contain a conserved PXXP amino acid motif, resembling either Src homology 3 (SH3) cell signaling interaction motifs or “late domains” involved in host cell interactions aiding in particle release. Using an avian strain of HEV, we determined the roles of the conserved prolines within the PREPSAPP motif in HEV replication and infectivity in Leghorn male hepatoma (LMH) chicken liver cells and in chickens. Each proline was changed to alanine to produce 8 avian HEV mutants containing single mutations (P64, P67, P70, and P71 to A), double mutations (P64/67A, P64/70A, and P67/70A), and triple mutations (P64/67/70A). The results showed that avian HEV mutants are replication competent in vitro, and none of the prolines in the PXXPXXPP motif are essential for infectivity in vivo; however, the second and third prolines appear to aid in fecal virus shedding, suggesting that the PSAP motif, but not the PREP motif, is involved in virus release. We also showed that the PSAP motif interacts with the host protein tumor suppressor gene 101 (TSG101) and that altering any proline within the PSAP motif disrupts this interaction. However, we showed that the ORF2 protein expressed in LMH cells is efficiently released from the cells in the absence of ORF3 and that coexpression of ORF2 and ORF3 did not act synergistically in this release, suggesting that another factor(s) such as ORF1 or viral genomic RNA may be necessary for proper particle release.
Hepatitis E is a fecally-orally transmitted self-limiting acute disease (1, 49), although cases of chronic hepatitis E have also been reported in patients with HIV infection and organ transplants (28, 31). The causative agent of hepatitis E, hepatitis E virus (HEV), has been genetically identified from humans and several other animal species, including pigs, chickens, rats, mongooses, rabbits, and fish (15, 26, 35, 51). Although the mortality associated with HEV infection is generally <1% in the general population, it can reach as high as 20% in infected pregnant women in some developing countries (32). Recently, HEV has been recognized as a zoonotic virus with pigs and other animal species as reservoirs (40, 42, 50).
HEV belongs to the family Hepeviridae with all mammalian strains of HEV in the genus Hepevirus, and avian hepatitis E virus (avian HEV) is a floating species within the family (38). The avian strain of HEV shares approximately 50% nucleotide sequence identity with mammalian HEV (20, 59) and also shares common antigenic epitopes in the capsid protein with mammalian HEVs (14). Avian HEV infection in chickens is a useful small animal model system for understanding HEV replication and pathogenesis (39). Though there have been significant advances in developing a cell culture system for HEV propagation (41, 48), the current cell culture system is not robust and is of limited use in studying the life cycle of HEV. The genome of HEV contains a short 5′ noncoding region (5′ NCR) followed by three open reading frames (ORFs) and a 3′ NCR: ORF2 overlaps ORF3, but neither ORF2 nor ORF3 overlaps ORF1 (60). ORF1 encodes the nonstructural proteins (33). ORF2 encodes the major viral capsid protein (22, 63). ORF3 codes for a small, multifunctional, cytoskeleton-associated phosphoprotein that is involved in many aspects of viral replication and pathogenesis (6, 7, 9, 10, 29, 30, 43–46, 53, 54, 58, 62).
The ORF3 protein is dispensable for virus infectivity in vitro but is essential for establishing viral infection in vivo as demonstrated in rhesus macaques and pigs (13, 21). The ORF3 protein has been reported to play multiple roles in HEV infection (for a recent review, see reference 2). Overexpression of ORF3 in cultured cells has led to the identification of several interactions with host cellular proteins, including proteins containing the Src homology 3 (SH3) domain (34), microtubule proteins (29), hemopexin (54), alpha-1-microglobulin and bikunin (57). Alpha-1-microglobulin secretion is upregulated via interaction with tumor suppressor gene 101 (TSG101) (56). Most recently, the ORF3 protein interaction with TSG101 is thought to direct virion release through the host proteins forming multivesicular bodies (10, 45, 46, 56, 62). The avian HEV ORF3 protein contains a singular proline-rich amino acid motif PREPSAPP. This motif resembles a conserved PXXP motif which has been noted to serve as a binding site for SH3 domain-containing proteins and as a binding site for host vacuolar sorting machinery proteins (also known as late domains) (11).
SH3 binding domain epitopes are often distinguished via a conserved amino acid motif consisting of X-P-p-X-P where X is an aliphatic amino acid, P is always a proline, and p is sometimes a proline (37). Late domains are conserved amino acid motifs first identified in the structural Gag protein of retroviruses (12). Late-domain motifs fall into three predominant types, PS/TAP, PPXY, and YPXL (27). These conserved motifs interact with members of the endosomal sorting complex required for transport (ESCRT) pathway (4). The ESCRT pathway is involved in multivesicular body transport within cells and, when usurped by viral proteins, plays a role in enveloped particles pinching off from the cellular membrane (36).
The objective of this study was to determine the roles of the prolines within this PXXPXXPP motif in HEV infectivity and release.
The pGEM-7zf(+) vector containing the avian HEV infectious cDNA clone pT7-aHEV (aHEV stands for avian HEV) has been previously described (19). Fluorescent vectors used for ORF2 and ORF3 expression in this study were enhanced cyan fluorescent protein (eCFP), enhanced green fluorescent protein (eGFP), and eYFP-N1 (eYFP stands for enhanced yellow fluorescent protein) vectors (Clontech, Mountain View, CA). Leghorn male hepatoma (LMH) cells (ATCC CRL-2117) passages 8 to 60 were used to assess viral replication competence and protein release.
The ORF2 and ORF3 expression constructs were generated by PCR amplification from the avian HEV infectious clone pT7-aHEV. Primers spk104 and spk8 (Table 1) were used for amplification of the ORF2 fusion constructs, and primers spk5 and spk6 were used for amplification of ORF3 fusion constructs.
Using overlap extension PCR, we introduced mutations in ORF3 of the full-length avian HEV infectious cDNA clone pT7-aHEV by changing the proline (P) to alanine (A) singly or in combination (Fig. 1, ORF3 sequences). Eight avian HEV mutants containing single mutations (P64, P67, P70, and P71 to A), double mutations (P64/67A, P64/70A, and P67/70A), and triple mutations (P64/67/70A) were generated. Mutagenic primers were as follows: primers spk163 and spk164 for mutant P64A, primers spk54 and spk55 mutant for P67A, primers spk165 and 166 for mutant P70A, and primers spk167 and spk168 for mutant P71A. The mutagenic primers were used in conjunction with primer spk171 and primer spk172 (Table 1). PCR fragments were inserted using EcoRV and SacII in pT7-aHEV. Double mutants were created by using the P64A mutagenic primers with P67A or P70A viral cDNA as a template and using P70A cDNA as a template with P67A primers. The triple mutant was created using P67/70A cDNA with the P64A mutagenic primers. Since the ORF3 coding sequence overlaps the ORF2 coding sequence, we were unable to silently mutate the ORF3 proline residues, resulting in concurrent mutation of the ORF2 protein. Consequently, mutation of proline 64 in ORF3 resulted in a proline-to-arginine mutation at amino acid 46 in ORF2. Proline 67 to alanine results in a glycine 48-to-alanine mutation in ORF2, and the ORF3 proline 70-to-alanine mutation alters proline 52 to arginine in ORF2 (Fig. 1, ORF2 sequences).
The ORF3 mutant and wild-type avian HEV infectious cDNA clones were linearized with the restriction enzyme XhoI. Capped RNA transcripts were synthesized using either the Promega T7 Ribomax kit or mMessage machine T7 kit (Ambion) as previously described (28).
In vitro-generated viral RNA transcripts were transfected into LMH cells using Lipofectamine LTX reagent (Invitrogen). Two days posttransfection, the cells were split and left until day 7 when cells were replated onto gelatin-coated LabTek chamber slides (Nunc). The cells were incubated overnight, fixed in cold 80% acetone and 20% methanol, followed by incubation in blocking buffer (5% goat serum and 5% nonfat dried milk in phosphate-buffered saline [PBS]). The cells were probed with anti-avian HEV chicken antiserum (19), followed by incubation with a fluorescein isothiocyanate (FITC)-labeled goat anti-chicken secondary antibody. The cells were mounted in antifade gold (Invitrogen) and imaged using a Nikon TE2000 SFC confocal microscope.
We employed a procedure that can bypass the inefficient in vitro cell culture system to determine the infectivity of avian HEV and mutant viruses in chickens. Capped RNA transcripts from pT7-aHEV clones were intrahepatically inoculated into the livers of specific-pathogen-free (SPF) chickens (19). Thirty 4-week-old SPF chickens were divided into 10 groups with 3 chickens in each group. The chickens in each group were intrahepatically inoculated with the capped RNA transcripts from each mutant, the wild-type avian HEV infectious clone as a positive control, and PBS buffer as a negative control (Table 2). Birds under full anesthesia (isoflurane) had a 2-cm parasternal incision made to visualize the right lobe of the liver. RNA transcripts were injected into two sites of the liver, with approximately 250 μl (63 μg) per injection site. The birds were housed in environmentally controlled biosafety level 2 (BSL-2) isolation cages for 12 weeks. Fecal swabs and sera were collected from each individual chicken at weekly intervals and tested by reverse transcription-PCR (RT-PCR) for avian HEV RNA (17). Additionally, serum samples were tested by an enzyme-linked immunosorbent assay (ELISA) for seroconversion to anti-avian HEV antibodies and for viremia by RT-PCR as described previously (18, 20). Pooled feces from each isolator floor from the same group were also collected three times per week and tested for avian HEV RNA by RT-PCR. The chickens were necropsied at week 12 postinoculation or at the time of death.
Total RNA extracted using Tri-Reagent (Molecular Research Centers Inc.) was tested for the presence of avian HEV RNA via a verso one-step RT-PCR kit (Thermo Scientific). Primers spk217 and spk218 (Table 1) were used to amplify the ORF3 gene, including the region encoding the PXXPXXPP motif. First-round PCR products were used as the template for the second round of PCR amplification using nested primers spk219 and spk220. PCR products were sequenced to confirm mutations within the PXXPXXPP coding region.
Serum samples diluted 1:100 were tested for reactivity to a recombinant truncated avian HEV capsid antigen (1 μg/ml) (14) bound to the wells on a 96-well plate. The plates were then incubated with goat anti-chicken horseradish peroxidase (HRP)-conjugated secondary antibody for 30 min at 37°C and washed again before color development using 2,2′-azinobis(3-ethylbenthiazolinesulfonic acid) (ABTS) as previously described (14).
LMH cells grown on coverslips were transfected using Lipofectamine LTX. At 18 to 24 h posttransfection, the cells were fixed in 4% paraformaldehyde and mounted using aquapolymount (Polysciences Inc.). The cells were imaged on a Zeiss LSM510 confocal microscope using a plan-apo 63× oil immersion objective. CFP fluorophores were excited using 458-nm light, and YFP was excited using 514-nm light. Colocalization was determined by using the RG2B colocalization plugin with autothresholding (Christopher Mauer, Northwestern University) in Image J software (NIH). For FRET, the cells were initially imaged and bleached for 11 to 16 iterations of 100% 514-nm laser intensity. The cells were bleached to 30% or less of initial YFP intensity. CFP fluorescence intensity was determined using fluorescence intensity (FI) NIH Image J software. FRET efficiency was calculated as follows: (FI for the donor after FRET corrected for nonspecific bleaching − donor pre)/FI for the donor before FRET) × 100. A minimum of 16 total cells were analyzed for each construct from at least two separate transfections.
LMH cells were transfected using Lipofectamine LTX. At 6 h posttransfection, the medium was changed to the primary medium for an additional 14 to 16 h before harvest. Cell viability was determined by the addition of Alamar blue cell viability dye (Invitrogen) 30 min prior to the harvest of the medium. The culture medium was removed from the cells and centrifuged at 1,500 rpm to pellet floating debris. Radioimmunoprecipitation assay (RIPA) buffer with HALT protease inhibitors (Fisher) was added to the medium. The cells were lysed in 2 ml of RIPA buffer plus protease inhibitors. Viral proteins were immunoprecipitated using polyclonal antibodies generated in guinea pigs against bacterially expressed 6×His recombinant avian HEV ORF2.1 (14) or ORF3 protein. Antibody complexes were precipitated using magnetic protein A beads (Millipore) that had been washed one time with RIPA buffer and PBS. Proteins boiled in Laemmli sample buffer were separated via polyacrylamide gel electrophoresis. Western blotting using rabbit polyclonal antibody to GFP was performed. Proteins were visualized via an anti-rabbit IRdye680-conjugated secondary antibody and a Licor Odyssey imaging device (Licor Biosciences). Densitometry was performed on blots using NIH Image J software.
To test the effect of the proline mutation within the PXXPXXPP motif on virus replication in vitro, equal amounts of full-length capped RNA transcripts were transfected into LMH cells using Lipofectamine LTX. Eight days posttransfection, an IFA was performed. Bright fluorescence was observed in the cytoplasm of cells transfected with capped RNA derived from wild-type pT7-aHEV and from all cells transfected with HEV mutants containing ORF3 PXXPXXPP single, double, and triple mutations but not in mock-transfected cells (Fig. 2), indicating that avian HEV mutants containing mutations in the ORF3 PXXPXXPP motif are replication competent in LMH cells.
Viremia was detected in infected birds sporadically until week 11, while the chickens inoculated with the PBS buffer remained negative for the duration of the experiment (Table 2). Viremia was detected mostly within the first 4 weeks postinfection (wpi) with eight out of nine mutant groups testing positive for serum viral RNA. The P64A single mutant virus had delayed viremia first detected in week 7 (bird 12) and lasting through week 10 (bird 11). We failed to detect viremia or fecal virus shedding for birds 17 (P64/70A), 18 (P64/70A), 21 (P67A), 25 (P64/67/70A), and 27 (P64/67/70A) throughout the 12-week study. When monitoring both fecal swabs from individual birds and pooled isolator floor feces from each group at different time points, fecal shedding of virus was detected in all groups (except PBS-inoculated control) at some point during the 12-week period. Wild-type, P64A, and P71A virus-inoculated chickens had the most consistent and longest virus shedding (Tables 2 and and3).3). Wild-type virus shedding in feces was detected in weeks 1 to 6 and week 8, P64A fecal virus shedding was more variable and longer, detected in weeks 1 and 2, 5, and 7 to 9, whereas the P71A virus fecal shedding was detected more consistently in weeks 2 to 9. Viruses containing P67A, P67/70A, and P64/67/70A mutations were detected in only 1 out of the 12 weeks (weeks 2, 1, and 2, respectively). The P70A virus was detected in weeks 1 and 5, the P64/70A virus was detected in weeks 1, 5, and 6, and the P64/67A virus was detected in weeks 1, 3, 4, and 8 to 10 (Table 3). PCR products amplified from fecal and serum samples of inoculated chickens were sequenced for ORF3, and sequence analyses confirmed the origin of each mutant virus (data not shown).
Birds injected with viral RNA from wild-type avian HEV seroconverted between 1 and 5 wpi (Fig. 3A) and remained seropositive for the duration of the study. Chickens inoculated with P64A and P71A mutants resulted in a delayed and transient seroconversion from 7 to 8 wpi for mutant P64A and 4 to 8 wpi for mutant P71A (Fig. 3C and E). All other mutant virus-inoculated groups had no detectable seroconversion exemplified by P70A (Fig. 3D and data not shown). At least one bird in each of the mutant virus-inoculated groups failed to seroconvert except the P71A group (Fig. 3E). The bird that did not seroconvert in the P64A group died at week 9, approximately when the other birds were seroconverting (Fig. 3C, bird 10).
To examine possible reasons for the observed differences in viral infectivity in vivo, we constructed recombinant vectors to express the avian HEV ORF2 and ORF3 proteins as separate gene products fused C terminally with fluorescent protein tags. Transient expression of the fluorescently tagged ORF3 protein in LMH cells resulted in a diffuse localization mostly within the cytoplasm. Punctate fluorescence signals were also observed throughout the cytoplasm and at the plasma membrane (Fig. 4B, F, and J). Coexpression of fluorescently tagged proteins localizing to different endosomal compartments (Rab7, TGN38, and Rab11) revealed that the wild-type or mutant ORF3 proteins could localize to the late endosomes, trans-Golgi network, and recycling endosomal compartments, respectively (Fig. 4C, G, and K and data not shown). Mutation of the prolines within the PREPSAPP motif did not alter subcellular localization (data not shown).
The HEV ORF3 protein contains a PSAP motif known in other viruses to promote interaction with the ESCRT pathway in cells to promote enveloped-particle release from the plasma membrane, and previous studies suggested a role for the ORF3 protein in release of HEV particles (10, 45, 62). In this study, we demonstrated that coexpression of the ORF3 protein with tumor suppressor gene 101 (TSG101) in LMH cells resulted in colocalization in membranous ring-like structures within the cytosol of the cells (Fig. 5A). Mutation of the PXXPXXPP motif did not abolish this colocalization (Fig. 5A, mutant P67A).
To further determine whether the avian HEV ORF3 protein interacts with TSG101, we performed acceptor photobleaching FRET (Fig. 5B). FRET is a biophysical phenomenon in which proteins tagged with two fluorophores whose emission and excitation overlap (CFP and YFP) allow the acceptor YFP to quench the donor CFP if the two proteins are within 10 to 100 Å (a biologically relevant distance for protein-protein interactions to occur) (23, 55). Acceptor photobleaching cells expressing just CFP and YFP resulted in a FRET efficiency of 9.08. Photobleaching cells expressing wild-type avian HEV ORF3 and P64A and P71A mutants exhibited FRET efficiencies of 18.75, 18.32, and 13.42, respectively, which are significantly higher than that of CFP/YFP alone (P = 0.0004, <0.0001, and 0.02, respectively). Cells expressing P67A and P70A mutants produced FRET values of 9.7 and 3.55 with no significant difference with those CFP- and YFP-expressing cells (Fig. 5B).
Since the ORF3 protein interacts with TSG101, we decided to determine whether this interaction was required for release of ORF3 and whether ORF2 was required to be released from the cell. To determine the ability of ORF2 and ORF3 to be released from the cell, we transiently expressed ORF3 or ORF2 alone or in conjunction and analyzed the amount of protein released into the medium. The results showed that approximately 44.9% of the ORF2 protein is released from the cells without any other viral protein (Fig. 6B). Approximately 13.7% of the ORF3 protein is released from the cells in the absence of other viral proteins, an unexpected result considering that it possesses the putative late-domain motif. Coexpression of ORF2 and ORF3 does not alter the amount of ORF2 protein released from the cells (39.9%; P = 0.575). The amount of ORF3 protein released from cells in the presence of ORF2 became more variable and increased to 33.64%; however, this increase was not statistically significant (P = 0.1326). There was no significant difference between ORF2 expressed alone (44.94%) compared to ORF2 coexpressed with ORF3 (39.90%) (P = 0.575). Also, ORF2 containing a P52R (P70A ORF3) mutation, which was the least viable virus in terms of viremia and fecal virus shedding, was released at 41.96%, which is not significantly different from wild-type ORF2 (44.94%) (P = 0.8803) (data not shown), suggesting that mutations within ORF2 did not alter its ability to be released. Finally, the release of ORF2 protein did not appear to be dependent on the multivesicular body (MVB) pathway, since coexpressing ORF2 with a dominant-negative VPS4dsRed protein led to an ORF2 release efficiency (53.15%) that is similar to that of ORF2 alone (44.94%) (P = 0.415) (data not shown).
The steps involved in the assembly of infectious HEV particles remain largely unknown. In this study, we utilized an avian strain of HEV along with a chicken HEV model and LMH chicken liver cells to determine the role of the conserved PXXPXXPP motif within the avian HEV ORF3 protein in virus infectivity. In addition, for the first time, we examined the release of ORF2 from eukaryotic cells in the presence and absence of ORF3 at the protein level.
Experimental infections with genotype 1 human HEV mutants containing mutations in the ORF3 PXXP motif in rhesus macaques (10) produce undetectable virus infection. In our chicken model with avian HEV mutants, we show that, although inefficient, viruses with mutations of the prolines in the PSAP motif were capable of establishing an infection in chickens. The observed discrepancy may be a result of differences between avian HEV and genotype 1 human HEV in the ORF2 and ORF3 proteins or that there are additional ways in which the avian HEV ORF3 protein interacts with the host MVB machinery.
We have further demonstrated in vitro that ORF3 PXXPXXPP mutant viruses are replication competent in LMH cells. Detection of viral RNA in serum and fecal samples from chickens inoculated with avian HEV mutants along with a lack of seroconversion for all PSAP mutant viruses suggests that release of these PXXPXXPP mutant viruses was not sufficient to generate a robust humoral immune response from the host. It is unlikely that the lack of seroconversion in birds inoculated with some of the mutant viruses is due to mutations introduced in ORF2 reducing antibody avidity to the wild-type ORF2.1 protein used in the ELISA, since the P64A and P71A mutant viruses did not appear to have a loss of antigenicity and convalescent-phase sera recognizes all infected cells in vitro.
Numerous RNA viruses, including retroviruses, rhabdoviruses, filoviruses, flaviviruses, orbiviruses, orthomyxoviruses, and arenaviruses (8, 12, 16, 25, 47, 52, 61) use the ESCRT pathway to release virus particles from the cells. Recent work suggests that a PXXP motif within the HEV ORF3 protein may play a similar role during HEV particle formation (10, 45, 46, 62). In the case of enveloped viruses, expression of the structural proteins is sufficient to produce virus-like particles (VLPs) which can be released from the cell. Coexpression of additional viral protein(s) often creates a synergistic effect in which particle morphology closely resembles the authentic virions. In hepatitis B virus assembly, surface protein drives VLP formation, whereas addition of the L glycoprotein and capsid protein results in authentic particle formation (5). In filoviruses, the glycoprotein (GP) and matrix protein (VP40), when singly expressed in cells, drive amorphous VLP assembly at an inefficient rate, but coexpression of both proteins leads to increased release of VLPs more closely resembling the authentic virus particles (24). Therefore, we reasoned that expression of the avian HEV ORF2 or ORF3 protein may be sufficient to produce VLPs and that release of these proteins should be augmented by coexpression with the complementary protein. We were surprised to find that, in the absence of other viral proteins, the ORF2 protein is released from LMH cells at a high efficiency (45%). The ORF3 protein containing the PXXP motif which has been proposed to be a late domain was released much less efficiently than ORF2 (13%). Combining ORF2 and ORF3 expression resulted in no significant enhancement of release of either ORF2 or ORF3 from the cells.
Our inability to detect a synergistic effect on protein release when the ORF2 and ORF3 proteins are coexpressed in LMH cells differs from published results in which particle release is abolished when mutations of the PSAP motif within the ORF3 protein are analyzed (10, 45, 46, 62). It is important to mention that these published studies were in the context of the full-length viral genome, whereas our study looked at ORF2 and ORF3 proteins independently of other viral proteins and RNA. Additionally, previous studies used RT-PCR to quantify viral RNA as a surrogate for virion release without looking specifically at ORF2 or ORF3 protein release (10, 45, 46, 62). It is possible that, by overexpressing these proteins, we are driving protein release through an unnatural pathway so that when the ORF2 and ORF3 proteins are expressed via the viral promoter, they are in much lower quantities and follow a different mechanism of release. It is also possible that binding of the ORF2 protein to the viral RNA, which was not provided in our system, may alter the pathway used for viral assembly. In addition to viral RNA, the presence of the ORF1 protein may also contribute to how the forming virions are released from the cell. Another possibility is that interaction of the ORF3 protein with TSG101 may alter viral RNA packaging or protection of the viral RNA within the assembling virions as seen with tomato bushy stunt virus (3), an explanation that has not been addressed by the previous studies.
In summary, in this study we used a system in which viral protein release can be studied in isolation from other factors. The ability of the ORF2 protein to be released from the cell in the absence of the ORF3 protein was unexpected, since previous studies suggest the need of the ORF3 protein for virion release (10, 45). Results from this study suggest that the ORF2 protein can be released from the cell in the absence of the ORF3 protein. It is possible that other viral factors such as the ORF1 protein or full-length viral genomic RNA may also play a role in proper assembly and release of virus particles. Therefore, further studies are warranted to delineate the specific mechanisms required for particle assembly and release of HEV.
This work was supported by grants from the National Institutes of Health (AI074667 and AI050611).
We thank the animal care staff at Virginia Tech for their assistance in the animal study. We also thank Nathan Beach, Dianjun Cao, Caitlin Cossaboom, Barbara Dryman, Alicia Feagins, and Sara Smith for their assistance in sample collections of chickens. We thank Walther Mothes (Yale University) for providing the YFP MVB expression plasmids and Wes Sundquist (University of Utah) for providing the dominant-negative VPS4 expression construct.
Published ahead of print 21 March 2012