The Phosphorylated Form of the ORF3 Protein of Hepatitis E Virus Interacts with Its Non-glycosylated Form of the Major Capsid Protein, ORF2*

Hepatitis E virus (HEV) is a human RNA virus containing three open reading frames. Of these, ORF1 encodes the viral nonstructural polyprotein; ORF2 encodes the major capsid protein, which exists in a glycosylated and non-glycosylated form; and ORF3 codes for a phosphoprotein of undefined function. Using fluorescence-based colocalization, yeast two-hybrid experiments, transiently transfected COS-1 cell co-immunoprecipitation, and cell-free coupled transcription-translation techniques, we have shown that the ORF3 protein interacts with the ORF2 protein. The domains involved in this ORF2-ORF3 association have been identified and mapped. Our deletion analysis showed that a 25-amino acid region (residues 57–81) of the ORF3 protein is required for this interaction. Using a Mexican HEV isolate, site-directed mutagenesis of ORF3 , and a phosphatase digestion assay, we showed that the ORF2-ORF3 interaction is dependent upon the phosphorylation at Ser 80 of ORF3. Finally, using COS-1 cell immunoprecipitation experiments, we found that the phosphorylated ORF3 protein preferentially interacts with the non-glycosyl-ated ORF2 protein. These findings were confirmed using tunicamycin inhibition, point mutants, and as a positive control (27). The Y190 host contains integrated copies of both HIS3 and lacZ reporter genes under the control of GAL4-binding sites. The Y190 yeast strain was transformed with the appropriate plasmids using the lithium acetate procedure and grown on synthetic dextrose (SD) plates in the absence of Trp and Leu (SDTrp (cid:4) and SDLeu (cid:4) , respectively). Protein interaction was tested on SD plates without Leu, Trp, and His (SDLeu (cid:4) Trp (cid:4) His (cid:4) ). After 3 days at 30 °C, individual colonies were streaked out and tested by liquid and filter-lift (cid:1) -galac- tosidase assays, 3-amino-1,2,3-triazole (3-AT) assay (50 m M ), and diploid His assay. The filter (cid:1) -galactosidase assay, a parameter directly reflecting the strength of protein-protein interactions, was performed by streaking doubly transformed yeast colonies onto filter paper and allowing them to grow for 2 days on selection medium. Yeast cells were permeabilized by freezing yeast-impregnated filters in liquid nitrogen and thawing at room temperature. This filter was placed over a second filter that was presoaked in 0.1 M phosphate buffer (pH 7.0) containing 300 mg/ml X-gal and 0.27% (cid:1) -mercaptoethanol. Filters were left for 48 h to develop a blue color, which indicated a positive protein-protein interaction. Liquid (cid:1) -galactosidase activity was determined using the substrate X-gal as described previously (28, 29). Relative enzymatic activity was determined in five independent transformants. Data for quantitative assays were corrected for yeast cell number and are the mean (cid:6) S.E. of triplicate assays. Appropriate positive/negative controls and buffer blanks were used. The Y190 host strain containing pAS2-SNF1 and pACT2-SNF4 was used as a positive control (27). The spec- ificity of the in vivo protein-protein

Hepatitis E is an acute disease endemic in many countries throughout developing parts of the world, in particular on the continents of Africa and Asia, where it causes epidemics and sporadic infections. The causative agent, hepatitis E virus (HEV), 1 is transmitted via the fecal-oral route, predominantly through contaminated water (1). HEV is an RNA virus with a positive-sense genome ϳ7.2 kb in length with three open read-ing frames (ORF1, ORF2, and ORF3) encoding three different proteins (2)(3)(4). ORF1 (5079 bp) is at the 5Ј-end of the genome and is predicted to code for putative nonstructural proteins with sequences homologous to those encoding viral methyltransferases, proteases, helicases, and RNA-dependent RNA polymerases (3)(4)(5)(6). In the absence of a reliable in vitro culture system for HEV, fundamental studies on its replication and expression strategy have not been undertaken. ORF2 and ORF3 have been expressed in Escherichia coli, animal cells, baculovirus, and yeast and in vitro in a coupled transcriptiontranslation system (7)(8)(9)(10)(11). ORF2 encodes the major HEV structural protein, which has been shown to be an 88-kDa glycoprotein that is expressed intracellularly as well as on the cell surface. It is synthesized as a precursor and is processed through signal sequence cleavage into the mature protein, which is capable of self-association (12,13). When expressed through the baculoviral expression system, ORF2 was shown to assemble into virus-like particles (VLPs), which were cell-associated as well as secreted in the culture medium (14,15).
ORF3 encodes a small 13.5-kDa phosphoprotein that is expressed intracellularly and shows no major processing. It associates with the cytoskeletal and membrane fractions of cells (16,17). Recently, the ORF3 protein has been shown to dimerize in a yeast cellular environment, and its dimerization domain has been mapped to a 43-amino acid region overlapping the SH3 binding and phosphorylation domains (13). Furthermore, ORF3 has been recently shown to interact with SH3 domains and to activate MAPK (18).
Heterotypic interactions of the HEV proteins have not been studied as yet. Although ORF3 is located in the 3Ј-third of the genome and has been termed a structural protein, there is no evidence to date for its involvement in HEV structural assembly. Because ORF2 is the major capsid protein, we undertook studies to examine colocalization of ORF2 and ORF3 in transfected cells and to test for heterotypic interactions between these two viral proteins to evaluate a structural role for ORF3.
In the few years since its introduction, the yeast two-hybrid system has proven invaluable for studying physical interactions between genetically defined partners, for identifying contacts among the subunits of multiprotein complexes (19 -21), and for mapping specific domains involved in protein-protein interactions (22)(23)(24). In this system, two plasmid-borne gene fusions are cotransformed into yeast cells, and the interaction between these two fusion proteins is measured by the reconstitution of a functional transcriptional activator that triggers the expression of reporter genes lacZ and HIS3.
We have used the yeast two-hybrid system along with fluorescence-based colocalization experiments, transiently transfected COS-1 cell immunoprecipitation, and coupled transcription-translation techniques to show the interaction of ORF3 with ORF2. We have further mapped the interaction domain of ORF3 to a 25-amino acid region. Within this region, we have shown that a single amino acid (Ser 80 ) is responsible for this heterotypic protein-protein interaction. Ser 80 has been shown to be the site for phosphorylation of ORF3. Our yeast twohybrid analysis using a Mexican HEV isolate, site-directed mutagenesis, and a phosphatase digestion assay revealed that the ORF2-ORF3 interaction is phosphorylation-dependent. Finally, using immunoprecipitation experiments, followed by tunicamycin inhibition, point mutations, and deletion mutations expressing only the non-glycosylated form of ORF2, we have shown that phosphorylated ORF3 preferentially interacts with non-glycosylated ORF2. A possible role of ORF3 in a posttranslational modification-dependent interaction with ORF2 is discussed in light of our results.

EXPERIMENTAL PROCEDURES
Strains, Media, and Plasmid Constructs-All strains, plasmids, and plasmid constructs used in this study are described in Table I. The full-length ORF2 and ORF3 genes of HEV were excised from the pMT-ORF2 and pSG-ORF3 vectors (9,25), respectively, and cloned into the yeast two-hybrid vectors, resulting in an N-terminal in-frame fusion with either the GAL4 DNA-binding or activation domain. DNA manipulations were carried out as described by Sambrook et al. (26). All deletion constructs were generated by subcloning the full-length ORF2 and ORF3 genes of HEV and are described in Table I. Plasmid constructs not containing fully compatible ends were screened for the correct reading frame by sequencing, whereas all other constructs with fully compatible ends were verified by restriction digestion and sequencing.
Immunofluorescence Analysis-COS-1 cells were plated at a confluency of ϳ50% on coverslips 1 day before transfection and grown for 18 h. 40 h post-transfection, phosphate-buffered saline (PBS)-washed cells were fixed with 2% paraformaldehyde in PBS at room temperature for 10 min, permeabilized with 100% methanol at Ϫ20°C for 3 min, and then rehydrated with PBS for 20 min at room temperature. The cells were blocked with 5% normal goat serum for 2 h at room temperature and then incubated with appropriately diluted primary antibodies in PBS and 0.5% Tween 20 (PBST) containing 1% normal goat serum for 2 h at room temperature. The primary antibodies used were mouse monoclonal anti-ORF3 antibody (1:200 to 1:500 dilution) and rabbit polyclonal anti-ORF2 antibody (1:100 to 1:200 dilution). Cells were washed three times with PBST for 5 min each and then incubated for 1 h at room temperature with a 1:1000 dilution of conjugated secondary antibodies. For colocalization experiments, the secondary antibodies used were goat anti-mouse IgG coupled to Alexa 488 dye and goat anti-rabbit IgG coupled to Alexa 594 dye (Molecular Probes, Inc., Eugene, OR). These were chosen to label the ORF3 protein with Alexa 488 (green) and the ORF2 protein with Alexa 594 (red). Cells were washed as described above and mounted in 90% glycerol in PBS. Fluorescence images were collected using a 60ϫ or 100ϫ planapo objective in a Bio-Rad 1024 LSM attached to a Nikon inverted microscope. To prevent cross-talk in dual labeling experiments, only one dye was excited at a time, keeping the other channel completely closed. The images were processed using Confocal Assistant, followed by Adobe Photoshop Version 5.0.
Yeast Two-hybrid Techniques-The GAL4-based two-hybrid system (kindly provided by Dr. Stephen Elledge) contained pAS2 (DNA-binding domain vector) and pACT2 (activation domain vector), together with yeast reporter Saccharomyces cerevisiae strain Y190 (see Table I). The host strain containing pAS2-SNF1 and pACT2-SNF4 was used as a positive control (27). The Y190 host contains integrated copies of both HIS3 and lacZ reporter genes under the control of GAL4-binding sites. The Y190 yeast strain was transformed with the appropriate plasmids using the lithium acetate procedure and grown on synthetic dextrose (SD) plates in the absence of Trp and Leu (SDTrp Ϫ and SDLeu Ϫ , respectively). Protein interaction was tested on SD plates without Leu, Trp, and His (SDLeu Ϫ Trp Ϫ His Ϫ ). After 3 days at 30°C, individual colonies were streaked out and tested by liquid and filter-lift ␤-galactosidase assays, 3-amino-1,2,3-triazole (3-AT) assay (50 mM), and diploid His assay. The filter ␤-galactosidase assay, a parameter directly reflecting the strength of protein-protein interactions, was performed by streaking doubly transformed yeast colonies onto filter paper and allowing them to grow for 2 days on selection medium. Yeast cells were permeabilized by freezing yeast-impregnated filters in liquid nitrogen and thawing at room temperature. This filter was placed over a second filter that was presoaked in 0.1 M phosphate buffer (pH 7.0) containing 300 mg/ml X-gal and 0.27% ␤-mercaptoethanol. Filters were left for 48 h to develop a blue color, which indicated a positive protein-protein interaction. Liquid ␤-galactosidase activity was determined using the substrate X-gal as described previously (28,29). Relative enzymatic activity was determined in five independent transformants. Data for quantitative assays were corrected for yeast cell number and are the mean Ϯ S.E. of triplicate assays. Appropriate positive/negative controls and buffer blanks were used. The Y190 host strain containing pAS2-SNF1 and pACT2-SNF4 was used as a positive control (27). The specificity of the in vivo protein-protein interaction was confirmed using a yeast genetic assay for reconfirming positive two-hybrid interactions (30,31). Plasmid constructs were extracted from the positive Y190 cotransformants (BD-ORF3/AD-ORF2, clones 1 and 2). The plasmids isolated from these clones were separated and verified using E. coli HB101 cells on M9 synthetic medium lacking Leu. Subsequently, these plasmids were singly transformed into the PJ69 -4a and PJ69 -4␣ haploid yeast strains (32). After genetic crossing, the His3 prototrophy of the diploid strains was tested by plating for growth on SD plates in the absence of His (SDHis Ϫ ). All possible control transformations were conducted and were verified to be negative for His3 prototrophy.
In Vitro Transcription-Translation Assay-The full-length ORF2 protein (pRSET-ORF2, encoding 660 amino acids of ORF2 with an N-terminal His 6 tag) and [ 35 S]methionine-radiolabeled full-length ORF3 protein (123 amino acids), along with its deletion mutations (see Table I), were expressed in separate reactions using an in vitro coupled transcription-translation system (TNT coupled reticulocyte lysate system, Promega) following the manufacturer's instructions. The unlabeled ORF2 protein was then bound to Ni 2ϩ -NTA beads (Amersham Biosciences) and washed three times with PBS (pH 7.4). [ 35 S]Methionine-labeled ORF3 protein (full-length or deletion mutation) was then added to the same tube and incubated for 4 h at 4°C with gentle shaking. The beads were washed three times with PBS, resuspended in 10 ml of SDS-PAGE loading buffer (50 mM Tris-HCl (pH 6.8), 5% 2-mercaptoethanol, 2% SDS, 0.1% bromphenol blue, and 10% glycerol), and boiled for 4 min to dissociate the bound proteins. Aliquots (10 l) of the supernatants were subjected to SDS-PAGE, and the [ 35 S]methionine-labeled proteins were detected by autoradiography.
Transfection and Labeling of Cultured Cells-COS-1 cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 20 g/ml gentamycin. Cells were transfected at a confluency of ϳ50% with plasmid DNA using Lipofectin (Invitrogen) according to the manufacturer's guidelines. For each 60-mm diameter culture dish, 2.5 g of DNA and 10 l of Lipofectin were used in 1.2 ml of Dulbecco's modified Eagle's medium without serum or antibiotics, and DNA uptake was allowed to proceed for 6 h at 37°C in a CO 2 incubator. Forty hours post-transfection, cells were washed with 3 ml of methionine-deficient Dulbecco's modified Eagle's medium (Invitrogen) and metabolically labeled with [ 35 S]methionine (Amersham Biosciences), with each 60-mm diameter plate receiving 100 Ci of label in 1 ml of methionine-deficient Dulbecco's modified Eagle's medium. After 4 h of labeling, cells were washed with ice-cold PBS and harvested for further analysis. In addition to HEV ORF-containing expression plasmids, each experiment also included a control (or mock) transfection, in which the same amount of the parent vector (pSGI) was used. For phosphate labeling, at 40 -44 h post-transfection, cells on 60-mm plates were washed once with phosphate-deficient Dulbecco's minimal essential medium (Invitrogen) and incubated in 3 ml of deficient medium for 1 h. Following this step, each plate was labeled for 4 h in a CO 2 incubator at 37°C with 250 Ci of [ 32 P]orthophosphate (Amersham Biosciences or PerkinElmer Life Sciences) in 1 ml of deficient medium.
Immunoprecipitation-PBS-washed transfected COS-1 cells were harvested directly in 0.5 ml of GST binding buffer (20 mM Tris (pH 7.9), 180 mM KCl, 0.2 mM EDTA, 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 0.01% Nonidet P-40, and 1 mM dithiothreitol containing 1 g/ml bovine serum albumin) after incubation on ice for 15 min. Lysates were clarified at 10,000 ϫ g for 10 min, and the supernatant was incubated on ice for 1 h with 5 l of rabbit antiserum. To this were added 100 l of a 10% suspension of GST buffer-washed protein A-Sepharose beads (Amersham Biosciences), and the mixture was incubated with constant shaking at 4°C for 1 h. The beads were washed five times, each time with 0.5 ml of GST buffer, after being centrifuged at 10,000 rpm for 10 s. Washed beads were resuspended in 50 l of SDS-PAGE loading buffer, heated at 100°C for 4 min, and centrifuged, and the supernatant was subjected to SDS-PAGE and autoradiography.
Phosphatase Treatment-Lysates from cells transfected with the ORF3 expression vectors and labeled with [ 32 P] were subjected to im-munoprecipitation with rabbit polyclonal anti-ORF3 antibody as described above. The immunoprecipitates were washed once with 250 l of -protein phosphatase reaction buffer (50 mM Tris-HCl (pH 7.8), 5 mM dithiothreitol, 2 mM MnCl 2 , and 100 mg/ml bovine serum albumin). The washed immunoprecipitates were resuspended in 50 l of -protein phosphatase reaction buffer with or without 1 l of -protein phosphatase (400,000 units/ml; New England Biolabs Inc.) and incubated for 1 h at 30°C. A control reaction with the same amount of 35 S-labeled ORF3 was also conducted to test for protein stability, against -protein phosphatase treatment. After washing the beads with GST binding buffer once, clarified cell lysate containing 35 S-labeled ORF2 was applied to the tube with phosphatase-treated ORF3 and incubated at 4°C for 1 h with gentle shaking. The beads were washed five times with GST binding buffer. A control reaction with the same amount of immunoprecipitated 32 P-labeled ORF3 was subjected to the same treatment to show that ORF2 was capable of binding under the above experimental conditions. The beads were centrifuged, resuspended in SDS-PAGE loading buffer, heated at 100°C for 4 min, and centrifuged, and the supernatant was subjected to SDS-PAGE, followed by autoradiography.

Colocalization of HEV Structural Proteins ORF2 and
ORF3-Dual labeling immunofluorescence microscopy revealed colocalization of ORF3 and ORF2 in COS-1 cells transiently transfected with the expression vectors pMT-ORF3 and pMT-ORF2 (Fig. 1). The distribution of ORF3 in these cells was cytoplasmic and displayed punctate green staining (Fig. 1,  ORF3-488). Distribution of ORF2 was observed in the cytoplasm, too, and was denser around the nucleus, possibly in the endoplasmic reticulum, and stained red (Fig. 1, ORF2-594). Both proteins colocalized in the cytoplasm and did not aggregate in the nucleus or other organelles as shown in yellow (Fig.  1, MERGE). With these initial results showing colocalization of both ORF2 and ORF3, we decided to test the heterotypic interactions of these two HEV structural proteins in vivo and in vitro.
The Two Proteins ORF2 and ORF3 Interact with Each Other-The full-length ORF2 gene was cloned into the yeast twohybrid vector containing the GAL4 DNA-binding domain, resulting in the expression of a fusion protein with ORF2 fused to the C terminus of the GAL4 DNA-binding domain. Similarly, the full-length ORF3 gene was cloned in-frame into the twohybrid vector containing the GAL4 activation domain, resulting in the expression of a fusion protein with ORF3 fused to the C terminus of the GAL4 activation domain (Table I). Cotransformation of S. cerevisiae with plasmids encoding BD-ORF3 and AD-ORF3 induced strong GAL4-dependent HIS3 and lacZ expression as determined by growth on SDTrp Ϫ Leu Ϫ His Ϫ dropout medium and the blue color from filter-lift ␤-galactosid-ase assays, respectively (Fig. 2). The yeast extract/peptone/ dextrose (YPD) plate (Fig. 2B) showed unrestricted growth of all transformants shown in the template ( Fig. 2A). Neither plasmid alone induced HIS3 or lacZ expression in yeast. Single transformants, the host strain, and the BD-SNF1/AD-SNF4positive control (27) were plated on all the restrictive media plates (Fig. 2, C-E). Only transformants that possessed the BD plasmid or constructs containing it grew on SDTrp Ϫ plates (Fig. 2C), whereas only transformants containing the AD plasmid or constructs derived from it grew on SDLeu Ϫ plates (Fig.  2D). Only the positive control (BD-SNF1/AD-SNF4) and the transformants containing both BD-ORF2 and AD-ORF3 were able to grow on SDTrp Ϫ Leu Ϫ His Ϫ plates (Fig. 2E). The second reporter gene (lacZ) was also tested for expression by a filterlift assay, resulting in a blue color for the positive cotransformants and the positive control (Fig. 2F).
Liquid ␤-galactosidase activity was determined for the positive clones along with all appropriate negative and positive controls using the substrate chlorophenol red ␤-D-galactopyranoside. The mean relative ␤-galactosidase activities are shown in Fig. 3A. The host strain (Y190) along with transformants with single plasmids (AD-ORF2 and BD-ORF3) showed negligible ␤-galactosidase activity. Cotransformants containing none or one of the two ORFs (BD/AD, BD/AD-ORF2, and BD-ORF3/AD) also showed negligible ␤-galactosidase activity; however, cotransformants containing AD-ORF2 and BD-ORF3 together showed a high ␤-galactosidase response.
We further investigated the level of activation of the HIS3 reporter genes for the full-length ORF2-ORF3 interaction in the presence of 50 mM 3-AT. Hundredfold serial dilutions of log-phase cultures of Y190 strains expressing BD-SNF1 and AD-SNF4 (BD-SNF1/AD-SNF4) and AD-ORF2 and BD-ORF3 (BD-ORF3/AD-ORF2) along with appropriate controls were plated on YPD, SDHis Ϫ , and SDHis Ϫ plus 50 mM 3-AT (Fig.  3B). These results indicate the strength of the protein-protein interactions as a function of His prototrophy. The BD-ORF3/ AD-ORF2 cotransformants showed growth up to a 10 Ϫ4 -fold dilution on the SDHis Ϫ plus 50 mM 3-AT plate. This experiment showed that the ORF2-ORF3 interaction is strong and true.
The specificity of the ORF2-ORF3 interaction was also confirmed using a yeast genetic approach (30). After genetic crossing of the single transformants (haploids), the His3 prototrophy of the diploid strains (a/␣) was tested. Only the diploids containing both BD-ORF3 and AD-ORF2 (BD-ORF3/AD-ORF2) showed a positive phenotype, similar to the positive diploid control (BD-SNF1/AD-SNF4) (Fig. 4). From all the above ex- FIG. 1. Colocalization of ORF2 and ORF3 proteins. Two sets of COS-1 cells transiently transfected with pMT-ORF3 and pMT-ORF2 were doubly labeled with mouse monoclonal anti-ORF3 and rabbit polyclonal anti-ORF2 antibodies, followed by Alexa 488-conjugated anti-mouse antibodies (ORF3-488) and Alexa 594-conjugated anti-rabbit antibodies (ORF3-594), respectively. Separate images were acquired showing ORF3 distribution (green) and ORF2 distribution (red) and were merged (yellow) using Adobe Photoshop Version 5.0 software. periments, we conclude that the two HEV proteins ORF2 and ORF3 interact with each other in a yeast two-hybrid system.
A 25-Amino Acid Region of the ORF3 Protein Binds to the Full-length ORF2 Protein-To characterize the domains involved in the ORF2-ORF3 interaction, an array of deletion mutations were constructed for both ORF2 and ORF3 and were cloned into the yeast two-hybrid AD and BD vectors as described in Table I. Combinations of full-length fusion constructs and deletion mutants of each fusion construct were tested for in vivo protein-protein interactions as shown in Fig. 5.
A phenomenon very clearly observed with the ORF2 protein was that none of its deletion mutants could interact with full-length ORF3 when tested in the yeast two-hybrid system. The ORF2 deletion mutants used in this study represented different parts of the protein and were from a variety of regions and lengths as shown in Fig. 5. On the other hand, ORF3 deletion mutants showed both positive and negative results with various different deletion mutants when tested with full-length ORF2 for two-hybrid interactions. The ORF3-(1-81) deletion fragment showed a positive interaction with full-length ORF2 upon the yeast two-hybrid assay. Consequently, when the ORF3-(83-123) deletion fragment was tested, it showed negative. These initial experiments indicated that amino acids 1-81 of the ORF3 protein contain the interaction domain. Furthermore, when the ORF3-(1-57) deletion fragment was tested with ORF2, the result was again negative, whereas ORF3-(57-123) showed a positive interaction with ORF2. All the above results indicate that the region between amino acids 57 and 81 of ORF3 contains the interaction domain. Our mapping results were confirmed when ORF3-(57-81) was constructed and tested for its in vivo interaction with ORF2. This interaction showed positive upon all yeast two-hybrid assays. From these experiments, we were able to clearly map a 25-amino acid region of the ORF3 protein that is responsible for its interaction with ORF2.
Ser 80 Plays a Key Role in ORF2-ORF3 Interactions-The 25-amino acid interaction domain of ORF3 contains Ser 80 , which has earlier been shown by us to be the site for phosphorylation of this protein (17). An S80A point mutant of fulllength ORF3 was cloned into the yeast two-hybrid BD vector. When full-length AD-ORF2 was tested with this BD-ORF3(S80A) point mutation, the results showed negative in the yeast two-hybrid analysis (Fig. 6). This result indicates that the ORF2-ORF3 interaction depends on the presence of Ser 80 .
The above experiments have clearly been able to pinpoint a single amino acid residue responsible for this protein-protein interaction. We designed a co-immunoprecipitation procedure to study the ORF2-ORF3 interaction and to validate our twohybrid findings. Heterotypic interactions of the two HEV proteins ORF2 and ORF3 were studied by transiently transfecting COS-1 cells with either pMT-ORF2 alone (as a control) or in combination with one of the following: pMT-ORF3 (full-length, containing Ser 80 ), pSG-BD-ORF3-(57-81) (containing the binding domain fused to the 25-amino acid interaction domain from ORF3), or pSG-ORF3(S80A) (full-length, containing an S80A point mutation) (Fig. 7). [ 35 S]Methionine-labeled cell lysates were then immunoprecipitated with polyclonal antibodies. In a control experiment, the expression of both ORF2 and ORF3 was detected by their respective antibodies (Fig. 7, lanes 1-3). All binding studies were conducted with anti-ORF2 antibodies (Fig. 7, lanes 4 -10).
Subsequently, we tested for ORF2 interaction with fulllength ORF3, the 25-amino acid interaction domain of ORF3, and the S80A point mutation of ORF3 by an in vitro coupled transcription-translation immobilization assay. The full-length ORF2 protein was cloned into the pRSET cloning vector, thus expressing a fusion protein with a His 6 tag fused to its Nterminal end. The ORF2 protein was immobilized to Ni 2ϩ -NTA-charged beads for these experiments. The full-length ORF3 protein and the BD-ORF3-(57-81) (containing the 25amino acid interaction domain) and ORF3(S80A) point mutation constructs were individually transcribed and translated using [ 35 S]methionine. Fig. 8 (A and B) shows the results from these experiments. Lanes 1 and 7 show the full-length ORF2-ORF3 interaction in vitro after radiolabeled ORF3 was allowed to bind to immobilized ORF2, Ser 80 of ORF3 is conserved in all known isolates of HEV, except the Mexican isolate (33). We thus cloned the ORF3 coding region of the Mexican isolate of HEV into the yeast two-hybrid BD vector and tested it for an interaction against the AD-ORF2 (full-length) protein. The results for this experiment are shown in Fig. 8C. As positive and negative controls in this experiment, we used the ORF3-(57-81) protein (containing the 25-amino acid interaction domain) and the ORF3(S80A) point mutant, respectively. The Mexican ORF3 protein clearly showed negative upon yeast two-hybrid assays. Hence, we have proved through various in vivo and in vitro methods that Ser 80 is essential for the ORF2-ORF3 interaction.
Phosphorylation at Ser 80 of the ORF3 Protein Is Essential for the ORF2-ORF3 Interaction-Ser 80 of ORF3 has been shown to be the major site for phosphorylation (17). Our experiments described above showed that Ser 80 is essential for the ORF2-ORF3 interaction. We thus designed assays using -protein The open boxes represent regions deleted from the full-length gene. The numbers above the boxes represent the first and last amino acids of the regions included. His Ϫ represents growth on SDTrp Ϫ Leu Ϫ His Ϫ plates. ␤-gal represents the filter-lift ␤-galactosidase assay results and the liquid ␤-galactosidase assay results (in brackets) from the yeast two-hybrid analysis. AT represents the 3-AT assay for growth on SDTrp Ϫ Leu Ϫ His Ϫ plates with 50 mM 3-AT. Dip His Ϫ represents growth of diploids tested by the genetic two-hybrid analysis. phosphatase to investigate the requirement of phosphorylation of Ser 80 for the ORF2-ORF3 interaction. In these experiments, two aliquots of COS-1 cells were starved for phosphate and sulfate separately. These cultures were radiolabeled with [ 32 P] and [ 35 S], respectively. The ORF3 protein was immunoprecipitated using anti-ORF3 antibodies from the 32 P-labeled cell lysate.
Upon -protein phosphatase treatment of 32 P-labeled ORF3, the protein was allowed to interact with the 35 S-labeled ORF2expressing cell lysate (Fig. 9, lane 6). 35 S-labeled ORF3 was subjected to -protein phosphatase and shown to be unaffected (data not shown). On the other hand, phosphatase treatment of 32 P-labeled ORF3 resulted in no visible ORF3 band on the autoradiogram (Fig. 9, lane 5), showing complete removal of the phosphate group from ORF3. Lanes 1-4 show all required expression and immunoprecipitation controls for this experiment. Thus, phosphorylation at Ser 80 of ORF3 is required for the ORF2-ORF3 interaction.
The Phosphorylated ORF3 Protein Preferentially Interacts with the Non-glycosylated Form of ORF2-ORF2 is a glycoprotein with three glycosylation sites (Asn 137 , Asn 310 , and Asn 562 ). The glycosylation site at Asn 310 is the major site for ORF2 glycosylation. We designed experiments to investigate whether phosphorylated ORF3 binds primarily to the glycosylated or non-glycosylated fraction of ORF2. Fig. 10A shows the results of our initial experiments. Lane 1 shows the glycosylated (gORF2) and non-glycosylated (ORF2) forms of the ORF2 protein expressed in COS-1 cells and immunoprecipitated by anti-ORF2 antibodies. Upon cotransfection with both ORF2-and ORF3-expressing vectors and immunoprecipitation with anti-ORF3 antibodies, we found primarily the non-glycosylated form of ORF2 binding to ORF3 (Fig. 10A, lane 2). When -protein phosphatase was added to the lysate coexpressing ORF2 and ORF3 and immunoprecipitated with antibodies against ORF3, none of the forms of the ORF2 protein were detected (Fig. 10A, lane 3). These results gave us preliminary evidence that ORF3 preferentially binds to the non-glycosylated fraction of ORF2.
We subsequently used the ORF2 mutants ORF2(N137A, N310A), ORF2(N310A,N562A), and ORF2(⌬2-34) described in Table I , respectively. All samples were co-immunoprecipitated using antibodies against the ORF2 protein. When these coexpression lysates were analyzed for the ORF3 protein, each one of the corresponding lysates containing the non-glycosylated form of ORF2 pulled out the ORF3 protein from the lysates (Fig.  10C). This clearly proves that the non-glycosylated form of ORF2 is capable of binding to ORF3. The combined results in this report prove that the phosphorylated ORF3 protein interacts preferentially with the non-glycosylated form of the ORF2 protein of HEV. DISCUSSION HEV cannot be cultured routinely, although it has recently been propagated in primary macaque hepatocytes (34,35), and a virus resembling HEV has been cultured in A549 cells (33). As a result, studies of HEV protein synthesis, processing, and assembly have been limited to heterologous expression systems. We chose the yeast two-hybrid system to study the heterotypic interactions of the two proteins encoded by ORF2 and ORF3 located in the structural part of the HEV genome. Our interests in this interaction increased significantly when we found these proteins to colocalize upon immunofluorescence microscopy of cotransfected cells. Using the yeast two-hybrid approach, we showed interactions between these two proteins, mapped the interaction domains, and showed that Ser 80 of ORF3 protein is responsible for this interaction. Results thus obtained were verified using other in vitro binding and immunoprecipitation techniques. Furthermore, we showed that the ORF2-ORF3 interaction is dependent upon phosphorylation at Ser 80 of the ORF3 protein. Because the ORF2 protein exists in both glycosylated and non-glycosylated forms (25), we designed experiments to observe any preference shown by ORF3 for the glycosylated and non-glycosylated forms of ORF2. Analysis using ORF2 mutants and tunicamycin inhibition revealed that ORF3 preferentially interacts with the non-glycosylated form of ORF2.
There are three sites for ORF2 glycosylation (Asn 137 , Asn 310 , and Asn 562 ), with Asn 310 being the primary one (25). Torresi et al. (36) have shown that the glycosylated ORF2 species is much less stable than non-glycosylated ORF2, which is present in the cytosol and represents the major product accumulated in the cell. It is postulated from this work that the non-glycosylated form may be involved in capsid assembly. Our results show that ORF3 preferentially binds with the non-glycosylated form of ORF2.
Our data also show that ORF3 interacts only with full-length ORF2 and not with any of its deletion mutants. Using baculovirus constructs expressed in insect Tn5 cells, Xing et al. (37) have shown that post-translational proteolytic cleavage is required for particle formation. Similarly, Li et al. (15) have shown that an N-terminally 111-amino acid truncated ORF2 protein shows empty particle formation. Both these studies suggest that there is proteolytic processing of ORF2 prior to capsid assembly. Our data show that full-length ORF3 inter-acts only with full-length ORF2 and not with any of its deletions mutants, including ORF2-(112-660). This result indicates that, during the course of ORF2 processing in the viral replication cycle, the ORF2-ORF3 interaction occurs prior to the processing of the ORF2 protein into its ϳ50-kDa processed form, which later forms VLPs.
To date, there is no evidence for an RNA-binding activity for either ORF2 or ORF3. Sequence analysis reveals, however, that the N-terminal 111 residues of native ORF2 contains basic residues and hence may be involved in RNA binding. Alternatively, ORF3 may be the RNA-binding protein and thus may interact with ORF2 during capsid assembly. The ORF3 dimerization (13) is independent of the ORF2-ORF3 interaction domain and the phosphorylation domain. So, it is possible that the ORF3 protein forms a dimer prior to interacting with fulllength ORF2. After dimerization, ORF3 probably gets phosphorylated, which makes it capable of binding to ORF2. If the ORF3 protein were to bind RNA and get phosphorylated, the ORF2-ORF3 interaction may have a major role to play in RNA packaging.
The stability of VLPs after proteolytic modification of the ORF2 protein has been shown to decrease (37). It has been thus postulated from previous work that recombinant HEV particles lack a stabilizing scaffold and thus become fragile and easily damaged during purification. This suggests the possibility of another protein interacting with ORF2 that may provide it more stability.
The ORF2 protein has been shown to self-associate in the absence of ORF3 to form dimers (12,13) and VLPs (15,37). Although these VLPs mimic the HEV virion, there are detectable differences in size, and the internal cavity thus formed is too small to accommodate the ϳ7.2-kb HEV genomic RNA (37). Virions of this size have not been found in the bile or stool of patients suffering from hepatitis E or of experimentally infected monkeys (1). Also, the calculated capacity of these VLPs for packaging RNA is only ϳ1 kb in size, whereas the size of the HEV genome is 7.5 kb. Together with the size assessment of the capsid and the calculated indispensable volume of the viral RNA, the possibility of the involvement of another viral protein, ORF3 in this case, may be a possibility for correct HEV capsid assembly. With ORF3 selectively binding to the nonglycosylated form of ORF2, which is the one involved in capsid formation, our study shows ORF3 to be an important candidate for participation in capsid assembly.
The ORF3 phosphoprotein has been shown to self-associate (13), to bind proteins containing SH3 domains, and to activate cellular MAPK (18). Although ORF3 maps in the structural region of the HEV genome, it has, in these recent reports, shown indications of regulatory functions. Also, with this report showing ORF2 interacting with ORF3, the possibility of a nonstructural function for the ORF2 protein also exists.
In this report, we have observed that the ORF3 protein from the Mexican isolate, which contains leucine instead of serine at position 80, fails to associate with ORF2. Upon close examination of the sequence of the Mexican ORF3 amino acid sequence in that region, we observed that it was considerably different, with an upstream serine at position 76 also replaced with leucine. Either the ORF2-ORF3 interaction is not critical for virion assembly or infection, as suggested by the Mexican HEV isolate, and has a nonstructural role, or the Mexican isolate has a weak serine-free ORF2-ORF3 interaction that we were unable to detect as fusion proteins in a yeast environment. In the absence of comparative infectivity data on the prototypic and Mexican isolates of HEV, it is difficult to assess the effects of lack of Ser 80 phosphorylation on viral pathogenesis. Finally, because the ORF2-ORF3 interaction is phosphorylation- A, immunoprecipitation of ORF3 using antibodies against the ORF3 protein was examined for the ORF2 protein. Lane 1 shows the glycosylated (gORF2) and non-glycosylated (ORF2) forms of the ORF2 protein as detected by antibodies against ORF2. Lane 2 shows the non-glycosylated form of the ORF2 protein preferentially interacting with the ORF3 protein, which was immunoprecipitated by its antibody. Upon phosphatase treatment of this lysate, non-glycosylated ORF2 did not bind to non-phosphorylated ORF3 (lane 3). B and C, shown is the preferential interaction of the ORF3 protein with the non-glycosylated form of ORF2. B is an expression control for C. Plus and minus signs indicate the presence and absence of tunicamycin in the reactions, respectively. 137,310, 310 -562, and ⌬2-34 represent the two double point mutants (ORF2(N137A,N310A) and ORF2(N310A,N562A)) and the deletion mutant (ORF2(⌬2-34)) of ORF2. dependent, this seems like a good post-translational control mechanism to check the level of ORF2-ORF3 interaction within the hepatocyte.