Hepatitis C Virus E2 Protein Purified from Mammalian Cells Is Frequently Recognized by E2-specific Antibodies in Patient Sera*

The envelope protein of hepatitis C virus (HCV) is composed of two membrane-associated glycoproteins, E1 and E2. To obtain HCV E2 protein as a secretory form at a high level, we constructed a recombinant chinese hamster ovary (CHO) cell line expressing a C-terminal truncated E2 (E2t) fused to human growth hormone (hGH), CHO/hGHE2t. The hGHE2t fusion protein was purified from the culture supernatant using anti-hGH mAb affinity chromatography at approximately 80% purity. The purified hGHE2t protein appeared to be assembled into oligomers linked by intermolecular disulfide bond(s) when density gradient centrifugation and SDS-polyacrylamide gel electrophoresis were employed. When the purified fusion protein was used for testing its ability to bind to antibodies specific for HCV by enzyme-linked immunosorbent assay, the protein was recognized by antibodies in sera from 90% of HCV-positive patients. Treatment of hGHE2t protein by β-mercaptoethanol, but not by heat and SDS, significantly reduced its reactivity to the antibodies of patient sera, suggesting that intermolecular and/or intramolecular disulfide bonds are important for its ability to recognize its specific antibody and that the E2 protein contains discontinuous antigenic epitope(s).

Hepatitis C virus (HCV) 1 is a major causative agent of posttransfusion and sporadic non-A, non-B hepatitis throughout the world (1,2). In most cases, the virus appears to cause a persistent infection. Previous studies indicate that the development of chronic liver diseases, cirrhosis, and hepatocellular carcinoma is associated with chronic HCV infection (3).
Comparative analyses of the genomes from several HCV strains indicate that HCV is a member of the family Flaviviridae, which includes flaviviruses and pestiviruses (4). The HCV genome is a 9.5-kilobase positive-strand RNA from which a single polypeptide is expressed and processed by cellular and viral proteinases to produce the putative viral structural and nonstructural proteins (4 -6). It was previously shown that structural proteins were composed of the core protein of 18 -22 kDa and two glycosylated envelope proteins, E1 of 31-35 kDa and E2 of 58 -74 kDa (5,(7)(8)(9)(10)(11). Although some lymphocyte cell lines have shown to support the limited replication of HCV, there has not been in vitro cell culture system efficiently enough to be used for viral propagation and for detailed virological studies (12). Expression studies using recombinant cDNA templates are the only means for identifying individual HCV proteins and to study their roles in the pathogenesis of HCV infection.
The hydrophobicity profile of HCV polyprotein suggested that the HCV E2 protein corresponds to the flavivirus NS1 glycoprotein and the major pestivirus envelope protein gp53/ gp55 (E2; gp53 in bovine viral diarrhea virus and gp55 in hog cholera virus), which were reported to induce protective immunity in experimental animals (13,14). HCV envelope proteins are of considerable interest, because experimentally challenged chimpanzees were either protected or shown to ameliorate disease following vaccination with recombinant E1/E2 subunits (15). It was recently reported that HCV E2 protein expressed in Chinese hamster ovary (CHO) cells binds to target cells at a high affinity. In addition, antibody which neutralizes the binding of E2 to target cell appears to correlate with protection from HCV infection (16). These results suggest that HCV E2 protein is a key viral antigen for a hepatitis C vaccine.
The HCV envelope protein expressed in cells infected with recombinant baculovirus and vaccinia virus was used for detection of envelope-specific antibody in patient sera (17)(18)(19). However, the purification of HCV envelope protein at a high yield was thought to be a difficult task. When HCV envelope proteins were purified from HeLa cells infected with recombinant vaccinia virus, approximately 1.5 mg of partially purified E1/E2 protein was obtained from a 120-liter culture of infected HeLa cells (15). Since the yield of purified E1/E2 protein appeared to be extremely low, presumably due to the membrane association of the HCV envelope protein, biochemical and immunological studies with E1 and E2 protein has been hampered so far.
In this study, we established two recombinant CHO cell lines expressing a hGH and a secretory hGHE2t fusion protein, consisting of human growth hormone (hGH), thrombin recognition sequence, and a C-terminal truncated E2 (E2t) region. The fusion protein was shown to be a dimer and higher order oligomer, and purified to greater than 80% purity by using immunoaffinity chromatography from the culture supernatant.
Prevalence of anti-E2 antibody in patients' sera and antigenic character of the purified hGHE2t protein were further identified by enzyme-linked immunosorbent assay (ELISA).

EXPERIMENTAL PROCEDURES
Patients and Plasma-Sera were obtained from blood donors and patients who visited to Korea Cancer Center hospital and Asan Medical Center, located in Seoul, Korea, from 1993 to 1995. Sera were obtained from 24 blood donors who were healthy adults with high alanine aminotransferase levels and negative for anti-HCV antibody assay (HCV ELISA 3.0, Green Cross Corporation, Korea). HCV ELISA 3.0 includes recombinant antigens of core and NS3 purified from bacterial cells and five immunodominant peptides (21-28 amino acids) of NS4 and NS5. When HCV ELISA 3.0 was compared with Ortho HCV 3.0 (Ortho, Neckagemund, Germany) using 990 blood samples, the results of HCV ELISA 3.0 showed 99.6% consistency with that of Ortho HCV 3.0. Four samples showing discrepant results were confirmed by using RIBA 3.0 (Chiron Corporation) and HCV BLOT 3.0 (Genelabs Diagnostics, Singapore). One of four samples was shown to be false positive with HCV ELISA 3.0, while two others were with Ortho HCV 3.0. The last one is undetermined. The diagnosis of chronic hepatitis and liver cirrhosis was made by persistent liver enzyme abnormalities for longer than 6 months duration, physical findings, ultrasonography, and computerized tomography for evidences of portal hypertension, and/or liver biopsy. For the diagnosis of hepatocellular carcinoma, serum ␣-fetoprotein, imaging studies, and histological conformation were performed in all cases. HCV seropositive subjects which showed at least 4-fold elevations in serum transaminases for longer than 6 months were defined as chronic HCV. Sera from 115 patients with chronic renal failure who were on maintenance hemodialysis were used. As a negative control, we used 62 blood samples from healthy adults who had normal alanine aminotransferase levels and were negative for HCV ELISA 3.0.
HCV cDNA Cloning-HCV cDNA covering E1 and E2 regions was obtained by PCR after reverse transcription of RNA extracted from a serum sample of an HCV-positive patient (20). The cDNA was synthesized from an antisense primer 2560A (5Ј-GGC TCT AGA ACA TCA GCA TCC ACA AGC A-3Ј) and then amplified after the addition of a sense primer E1N (5Ј-TAC TCC ATG GTG GGG AAC TGG GCC-3Ј). The complete nucleotide sequence of the clone was determined and classified into type 1b.
Construction of Plasmids-To construct pSK-IRES, internal ribosome entry sequence (IRES) of encephalomyocarditis virus (EMCV) was amplified by PCR from pTM1 (21) using primers 5Ј-GCG GGA TGA ATT CCG CCC CTC T-3Ј and 5Ј-GCC ATG GTA TTA TCG TGT T-3Ј, and the amplified product was inserted into SmaI site of pBluescript SK(ϩ) (Stratagene). pMT3 was constructed by inserting the EcoRI DNA fragment containing IRES of pSK-IRES into EcoRI site of pMT2 (22). The DNA fragments encoding amino acid residues 364 -693 and 364 -740 of HCV polyprotein were amplified by PCR using E1N and 2420A (5Ј-TGT TCT AGA GGA GGT GGA TTA ACC CA-3Ј) or 2560A primers. The PCR products digested with NcoI and XbaI were inserted into downstream of EMCV IRES of pSK-IRES, and the resulting IRES-E2 fusion DNA fragments were digested with EcoRI and XbaI, and then inserted into pMT3 to generate pMT3-E2t and pMT3-E2, respectively. pMT3-hGH was constructed by inserting the cDNA of hGH gene into pMT3 (23). To construct pMT3-hGHT, the DNA fragment encoding the recognition sequence (Leu-Val-Pro-Arg-Gly-Ser) of thrombin, site-specific protease, was amplified by PCR from pGEX-KG (24) using primers, 5Ј-CCG CTG CAG ACT GGT TCC GCG TGG ATC CC-3Ј and 5Ј-GCG CTG CAG TTA AGC TTG AGC TCG AGT C-3Ј. The amplified DNA fragment was in frame fused to amino acid residue 189 of hGH of pMT3-hGH. E2 regions spanning amino acid residues 386 -693 and 386 -740 were obtained by PCR amplification using E2N (5Ј-CCA TAT GCG CGT GAC AGG AGG AAC G-3Ј) and 2420A or 2560A primers, and then were in frame fused to the thrombin recognition sequence of pMT3-hGHT to produce pMT3-hGHE2t and pMT3-hGHE2, respectively. The thrombin recognition sequence allows hGH portion to be cleaved from the HCV envelope protein after purification of fusion protein. Translation of hGHE2 and hGHE2t fusion proteins was stopped by termination codon in primers.
Transfection and Cell Lines-COS-7 cells were maintained in Dulbecco's modified Eagle's medium containing penicillin (50 IU/ml), streptomycin (50 mg/ml), and 10% fetal calf serum. CHO cells deficient in dihydrofolate reductase (DHFR) gene, dhfr Ϫ CHO, were maintained in ␣-minimal essential medium containing 10% fetal calf serum, hypoxanthine, and thymidine. COS-7 and dhfr Ϫ CHO cells were transfected with the plasmid DNA (10 g) using calcium phosphate precipitation method (25). dhfr Ϫ CHO cells transfected were maintained as described previously under selective conditions in ␣-minimal essential medium containing dialyzed 5% fetal calf serum (Life Technologies, Inc.) (26). CHO cell lines expressing recombinant HCV E2 protein were initially screened using immunoblotting and ELISA with serum from an HCV positive patient and anti-hGH rabbit antibody and were subjected to five subsequent rounds of methotrexate (Sigma) selection.
Antibodies-Serum of a patient with chronic hepatitis C was assessed as a positive by using HCV ELISA 3.0 (Green Cross Corporation, Korea), anti-hGH rabbit polyclonal antibody was raised against the purified hGH protein expressed in Escherichia coli. Goat anti-rabbit antibody and -human immunoglobulin (Ig) conjugated with horseradish peroxidase were purchased from DAKO diagnostics Ltd (Denmark). Anti-hGH mAb was purified from a hybridoma obtained from the American Type Culture Collection.
Immunoprecipitation and Immunoblot Analysis-COS-7 cells were transfected and metabolically labeled with 75 Ci of 35 S-Express label (NEN Life Science Products) as described previously (27). The labeled cells were lysed with lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.5% SDS) and then clarified by centrifugation at 15,000 ϫ g for 10 min. Portions of each lysate and culture supernatant were incubated either with serum from an HCV-positive patient or anti-hGH rabbit polyclonal antibody. Immune complexes were collected by using Staphylococcus aureus Cowan I (Calbiochem) as described previously (28). Immunoprecipitates were solubilized and analyzed by SDS-polyacrylamide gel electrophoresis.
Immunoblot analysis was done according to the method as described previously (27). After transfer of proteins onto a nitrocellulose membrane, blots were treated with the block solution containing 5% nonfat milk in TBS buffer (50 mM Tris-HCl, pH 7.9, 150 mM NaCl, 0.05% Tween 20) for 1 h at room temperature. Either serum from an HCVpositive patient or anti-hGH rabbit antibody (diluted 1:1000 in block solution) as the primary antibody and either goat anti-human or antirabbit Ig conjugated with horseradish peroxidase (diluted 1:10000 in block solution) as the secondary antibody were used to detect HCV E2 protein and developed by enhanced chemiluminescence (ECL; Amersham Corp.).
Analysis of Disulfide Linkages-The presence of inter-and intramolecular disulfide linkages was analyzed by the method of Allore and Barber (29). Detection of HCV proteins was performed by immunoblotting.
Sedimentation through Sucrose Density Gradients-The purified hGH, the hGHE2t fusion protein (20 g in 200 l), and molecular mass standards (180 g in 200 l; Pharmacia) were layered onto 4.8 ml of 5-40% linear sucrose density gradients containing 100 mM NaCl and 50 mM Tris-HCl, pH 7.5. Centrifugation was performed at 45,000 rpm for 9 h in a Beckman type 55 rotor at 4°C. Gradients were fractionated by puncturing the bottom of the tube and collecting 15 fractions of 330 l each. Gradient fractions were analyzed by immunoblotting.
ELISA-Microtiter plates (Immulon 2, Dynatech) were coated with either the culture supernatant from transfected cells or the purified hGHE2t protein (400 ng/well). A portion of the purified protein pretreated with heat, 0.2% SDS, 100 mM ␤-mercaptoethanol (␤-ME) with or without heat, and N-glycosidase F at 37°C for 2 h, and then washed with PBS containing 0.05% Tween 20. The wells were incubated with 1% bovine serum albumin and followed by incubation with sera from HCV-positive patients at 37°C for 2 h. After aspiration of the unbound material and washing of the wells, wells were incubated with goat anti-human Ig coupled with horseradish peroxidase. The reaction was developed by the addition of tetramethylbenzidine.
Purification of hGHE2t and the C-terminal Truncated E2 Protein-Anti-hGH mAb (25 mg), purified from ascites fluid by saturated (NH 4 ) 2 SO 4 precipitation, was coupled to 3 g of CNBr-activated Sepharose-4B (Pharmacia) according to the manufacturer's instructions. When CHO cells expressing hGHE2t protein were grown to subconfluent monolayers, culture medium was changed with serum-free medium (Life Technologies, Inc.). After additional 72-h incubation, the medium was harvested and applied to the hGH mAb affinity column equilibrated with PBS. The column was washed extensively with PBS and 0.5 M NaCl in PBS, and the bound hGHE2t protein was eluted with 3 M NaSCN in 10 mM sodium phosphate, pH 7.2. The eluate was immediately dialyzed against PBS. For the purification of C-terminal truncated E2 protein, the purified hGHE2t fusion protein was dialyzed against thrombin digestion buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2.5 mM CaCl 2 , 0.1% ␤-ME) and digested with thrombin (Boehringer Mannheim) for 2 h, and then applied to an hGH mAb affinity column. The C-terminal truncated E2 protein was purified by collecting the unbound protein.
Deglycosylation of Purified Protein-The purified hGHE2t protein was adjusted to 0.2% SDS and 100 mM ␤-ME, and boiled for 2 min. Treatment of N-glycosidase F was performed in the buffer containing 20 mM sodium phosphate, pH 8.0, 10 mM EDTA, 0.1% SDS, and 100 mM ␤-ME for 1 h at 37°C.
Reverse Transcription PCR-HCV RNA was extracted from serum by proteinase K digestion-phenol/chloroform extraction method as described previously (30). Reverse transcription and amplification using an antisense primer 300A (5Ј-CAC TCG CAA GCA CCC TAT CAG GCA-3Ј) and a sense primer 80S (5Ј-ATC ACT CCC CTG TGA GGA ACT AC-3Ј) were performed as described previously (20), and amplified products were analyzed by electrophoresis on 1.5% agarose gels.

Construction and Identification of Various Recombinant Plasmids Expressing HCV E2 as a Secretory Form-Biochem-
ical and immunological studies of HCV envelope proteins in virion have been limited due to a lack of an in vitro cell culture system allowing virus propagation. In addition, the purification of a native HCV envelope protein without denaturation appears to be difficult because HCV envelope proteins are membrane-associated. To express and purify HCV E2 protein at a high level, we designed several expression vectors for establishing a recombinant CHO cell line. pMT3-E2 and pMT3-E2t plasmids were designed to express the HCV E2 gene with and without a C-terminal hydrophobic region, respectively. Also, the signal sequence of E2 was replaced with the coding region of hGH, because it was found that the signal peptide of E2 was not appropriate for the efficient expression and that hGH was produced at a high level and efficiently secreted into culture medium in CHO cells. 2 Therefore, we designed two fusion constructs, pMT3-hGHE2t and pMT3-hGHE2, consisting of the E2 gene connected downstream of hGH in which the thrombin recognition sequence was inserted upstream of E2 to remove the hGH portion from the fusion protein (Fig. 1).
To determine if the resulting constructs were capable of expressing immunologically relevant protein, the recombinant constructs were tested by the transient transfection assay in COS-7 cells. As shown in Fig. 2, E2t, E2, hGHE2t, and hGHE2 proteins were expressed as molecular masses of 45-54, 52-62, 64 -74, and 72-82 kDa, respectively, when cell lysates were specifically precipitated with sera from an HCV-positive patient (Fig. 2, lanes 2, 3, 5, and 6). In contrast, any specific protein bands were not detected in cell lysates transfected with either pMT3 or pMT3-hGH as a control (Fig. 2, lanes 1 and 4).
The corresponding bands of hGHE2t and hGHE2 fusion proteins identified by sera from an HCV-positive patient were also detected when the same lysate was precipitated with anti-hGH rabbit antibody (data not shown). These results indicate that E2t, E2, hGHE2t, and hGHE2 proteins expressed in COS-7 cells were immunoreactive with circulating antibodies in HCVinfected individuals.
To determine whether the expressed E2 protein can be secreted, the culture supernatant of transfected COS-7 cells was also immunoprecipitated with sera from an HCV-positive patient. Neither E2t nor E2 protein was detected in culture supernatants of cells transfected with pMT3-E2t and pMT3-E2 (Fig. 2, lanes 8 and 9). In contrast, a significant amount of hGHE2t and hGHE2 proteins with molecular masses of 74 -84 and 82-94 kDa, respectively, were found in the culture supernatant (Fig. 2, lanes 11 and 12). It is likely that the hGHE2t protein is more efficiently secreted than the hGHE2 protein.
These results indicate that E2 proteins, with and without a C-terminal hydrophobic domain, E2 and E2t, are capable of being secreted into culture medium when an efficient secretory protein, such as hGH, was fused to the E2 protein. The secreted form appeared to be higher molecular weight than its intracellular form, presumably due to the amount of added sugar residue through the secretory pathway. On the basis of these results, we selected pMT3-hGHE2t vector for the establishment of a recombinant CHO cell line expressing E2 as a secretory form.  pMT3-E2 and pMT3-E2t were designed to express the full-length and the C-terminal-truncated E2 gene, which encompass the indicated regions, respectively. The DNA fragment encoding the recognition sequence of thrombin (Leu-Val-Pro-Arg-Gly-Ser) was inserted between the hGH region and HCV E2 to generate pMT3-hGHE2 and pMT3-hGHE2t, respectively. EMCV IRES, the hGH gene, and the E2 region of HCV are designated as thick black bars, open boxes, and hatched boxes, respectively. The gene cloned was expressed under the control of adenovirus major late promoter. Numbers in parentheses indicate the number of amino acids encoded by the DNA fragment.

Establishment of Recombinant CHO Cell Lines
Expressing hGHE2t and hGH-pMT3-hGHE2t and pMT3-hGH as a control were transfected into dhfr Ϫ CHO cells, and then positive cell lines were screened using ELISA and immunoblotting with anti-hGH rabbit antibody. After subsequent rounds of methotrexate selection, the recombinant CHO/hGHE2t and CHO/ hGH cell lines were adapted at a medium containing up to 20 and 1 M methotrexate, respectively. At this concentration of methotrexate, the expression level of hGHE2t and hGH protein was shown to be 7 and 2 g/ml by an anti-hGH ELISA kit (Boehringer Mannheim), respectively. The hGHE2t protein was, as expected, detected as the molecular mass of 74 -84 kDa by sera from an HCV-positive patient in immunoblot analysis (data not shown). These results suggest that the hGHE2t protein produced in CHO cells as well as COS-7 cells is a soluble secretory protein which is immunologically relevant.
Purification of the hGHE2t Protein Expressed in CHO Cells-To purify hGHE2t fusion protein, the hGH mAb affinity column was prepared by coupling the hGH mAb to the activated Sepharose-4B. Culture supernatants of recombinant CHO/hGHE2t and CHO/hGH cells were applied to the hGH mAb affinity column and washed serially with PBS and 0.5 M NaCl in PBS, and then hGHE2t protein was eluted with 3 M NaSCN. By repeating this simple immunoaffinity chromatography twice, approximately 4.0 mg of hGHE2t protein with about 80% purity were obtained from 1 liter of culture of the recombinant CHO/hGHE2t cells. The hGHE2t protein purified from recombinant CHO cell line showed a broad band of 70 -86 kDa on SDS gel stained with Coomassie Brilliant Blue and was identified with sera from an HCV-positive patient (Fig. 3, A and B, lane 1). The E2t protein was obtained after thrombin digestion, followed by hGH mAb affinity chromatography. The purified E2t protein showed a smaller molecular mass (45-65 kDa) and a broader band than expected in SDS-polyacrylamide gel electrophoresis analysis (Fig. 3, A and B, lane 2). Sequencing of the amino-terminal residues resulted in Val 387 and Cys 607 at the amino terminus. The predicted amino acid sequences around the cleavage sites of thrombin are Leu-Val-Pro-Arg-Ń-Gly-Ser-Pro-His-Met-Arg 386 -Ň-Val 387 and Leu 603 -Thr 604 -Pro 605 -Arg 606 -Ň-Cys 607 , in which closed and open arrows indicate the predicted and unexpected cleavage site of thrombin, respectively. Therefore, it is likely that Val 387 and Cys 607 at the amino terminus are derived from the cleavage at Arg 386 -Ň-Val 387 and Arg 606 -Ň-Cys 607 of the purified hGHE2t fusion protein, respectively. It is likely that the smaller and broader E2t bands are caused by partial thrombin cleavage of E2t at other unexpected cleavage sites. The hGHE2t fusion protein was used to further characterize the HCV E2 protein for the following reasons. First, the fusion protein appears to be cleaved by thrombin at unexpected sites, resulting in a smaller E2 protein. Second, the binding activity of the E2 protein to an E2-specific antibody was shown to be highly sensitive to ␤-ME, which should be included in the buffer of thrombin cleavage (see below).
The Secreted E2 Protein Forms Oligomeric Complexes-Although it was reported that E1 and E2 protein form a heterodimer, it is still controversial that this interaction is mediated by either intermolecular disulfide bonds and/or noncovalent association (10,27,31). In addition, a homo-oligomeric complex formation of E1 or E2 protein remained to be elucidated. Therefore, we characterized the intermolecular association of purified hGHE2t fusion protein by the method of Allore and Barber (29), which detects disulfide linkages by band shifting caused by diffusion of ␤-ME from reduced samples into adjacent lanes containing nonreduced samples. Under nonreducing conditions, additional high molecular mass bands were observed (Fig. 4, A, lanes 5-7). These high molecular mass bands disappeared in nonreducing samples adjacent to reduced lanes (lanes 4 and 8) as well as lanes containing reducing agents (lanes 1-3 and 9 -11), indicating that they were gener- ated by the formation of disulfide bond(s) among monomers. To further confirm oligomeric complexes of the hGHE2t protein, the hGHE2t fusion and the hGH proteins were analyzed on 5 to 40% sucrose gradients, followed by immunoblotting. Sedimentation analysis revealed that the hGHE2t fusion protein, but not the purified hGH protein (data not shown), exists as monomer, dimer, trimer, and tetramer forms (Fig. 4B). Taken together, our results suggest that the HCV E2 protein as a secreted form appears to be a mixture of monomer and oligomers such as dimer, trimer, and tetramers.
Prevalence of Anti-E2 Antibodies in Sera of Patients-To investigate the prevalence of anti-E2 antibody in chronic non-A, non-B hepatitis and hemodialysis patients in Korea, patients' sera were analyzed by an ELISA using the hGHE2t protein (Table I). The diluted patients' sera was incubated with the protein coated in microtiter plates and detected with antihuman Ig coupled with horseradish peroxidase. It was found that anti-hGH antibody was not detected in any patient's serum, when the hGH protein purified from CHO/hGH cell line, as a control, was coated in microtiter plates (data not shown). The hGHE2t protein was shown to react with antibodies contained in 87, 86, and 96% of sera from anti-HCV-positive patients with chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma, respectively ( Table I). None of sera from both 62 normal adults and 24 anti-HCV-negative patients with a high alanine aminotransferase level appeared to react with the hGHE2t protein. These results suggest that anti-E2 antibody is detected in about 90% of sera from anti-HCV-positive patients.
The prevalence of anti-E2 antibody in immunosuppressed patients with chronic renal failure was also examined by ELISA using the hGHE2t protein. Sera of 6 patients from 115 hemodialysis patients are shown to be positive by both an HCV ELISA 3.0 and an ELISA using the hGHE2t protein. Interestingly, an additional four patients' sera, which were anti-HCVnegative, reacted with the hGHE2t protein (Table I). To determine if these four patients were HCV viremic, HCV RNA was extracted and amplified by reverse transcription PCR with primers 80S and 300A. Three of them showed HCV viremia by HCV reverse transcription PCR analysis (Table I). These results indicate that E2 protein is recommended to be included in the diagnostic assay kit of HCV infection for patients with chronic renal failure.
The Binding Activity of the Purified E2 Protein to Patients' Sera in Various Conditions-The major antigenic epitope(s) of HCV envelope proteins is thought to be conformation-dependent, because the E2 protein is highly glycosylated and recognized more frequently in a native form than in a denatured form by sera from HCV-positive patients (8). To examine the character of antigenic epitopes, the hGHE2t fusion protein was treated with denaturing and/or reducing agents and analyzed by ELISA using 19 HCV-positive patients' sera. Heat, SDS, and ␤-ME were used at maximal concentrations that do not inhibit ELISA. The protein denatured with either 0.2% SDS or boiling for 5 min appeared to have the binding activity to patients' sera as efficiently as did nondenatured protein (Table  II). In contrast, treatment with either ␤-ME or ␤-ME/heat showed decreased reactivity in about 80% of patients' sera. These results suggest that intermolecular and/or intramolecular disulfide linkage(s) of hGHE2t protein molecules are important for the preservation of the antigenic determinant.
When treated with glycosidase, the secreted hGHE2t protein was shown to be an N-linked glycoprotein (data not shown), which is consistent with previous reports (27). To examine the role of carbohydrate in the binding activity of E2 protein, purified protein was pretreated with ␤-ME/heat, followed by digestion with N-glycosidase F. Deglycosylation did not further decrease the binding ability of the denatured hGHE2t protein to anti-E2-specific antibodies, suggesting that the carbohydrate moiety of the E2 protein is not critical for the binding activity of the denatured E2 protein to its specific antibodies. DISCUSSION In this report, we established a recombinant CHO cell line expressing HCV E2 as a secretory form. The hGHE2t fusion protein was produced at a high level (7 mg/liter) and purified to greater than 80% purity using simple immunoaffinity chromatography (4 mg/liter). The purified hGHE2t protein was recognized by sera from 90% of HCV-positive patients and 9% of hemodialysis patients. Reactivity of the purified protein to anti-E2 antibody from patients' sera appeared to be reduced by FIG. 4. Detection of oligomeric complexes of the purified hGHE2t protein. A, the presence of inter-and intramolecular disulfide linkages in the purified hGHE2t protein was analyzed by the method of Allore and Barber (29). The center five lanes (lanes 4 -8) were loaded with a sample mixture prepared without ␤-ME as a reducing agent, whereas the outer three lanes on either side of the center (lanes 1-3 and 9 -11) were loaded with a sample mixture containing ␤-ME. The hGHE2t protein was visualized by immunoblot with anti-hGH rabbit antibody. Molecular mass markers are indicated at the left on the gel. B, the purified hGHE2t protein was sedimented through 5 to 40% sucrose density gradients. The gradient fractions was separated on 6% SDS-polyacrylamide gel electrophoresis and detected by immunoblotting using anti-hGH rabbit antibody. Arrows indicate monomer and higher order oligomeric forms of hGHE2t protein from the bottom, respectively. the treatment with reducing agent, but not with denaturing agents, such as 0.2% SDS and boiling.
The hydrophobicity profile of HCV polyprotein showed that the C-terminal region of the E2 protein is hydrophobic (4). It was previously reported that the full-length E2 protein remained membrane-associated, presumably due to the putative transmembrane domain, and that deletion of the C-terminal hydrophobic region appeared to facilitate the secretion of the truncated E2 protein (10,27,31,32). The signal peptide of other proteins such as tissue plasminogen activator and rabies virus glycoprotein was shown to further facilitate the secretion of the C-terminal-truncated E2 protein, but not the full-length E2 protein (11,32). In contrast, our data showed that the fusion of hGH to the E2 protein enables full-length E2 protein as well as the C-terminal-truncated form to efficiently secrete into culture medium, suggesting that HCV E2 protein could be secreted, depending upon the signal peptide and/or secretory protein fused to the E2 protein.
The major glycoprotein of pestiviruses was shown to form disulfide-linked heterodimers (E1 and E2) and homodimers (E0 and E2) (13,33). The association mode of HCV E1 and E2 is still a matter of controversy. It was reported that a fraction of E1 and E2 present in lysates of cells infected with recombinant vaccinia virus was associated via disulfide linkage (27). In contrast, others demonstrated that E1 and E2 proteins expressed in insect and HeLa cells are noncovalently associated (5,10). Our results showed that the secreted hGHE2t protein appeared to form a homodimer and higher order oligomer, which are linked by intermolecular disulfide bond(s). A homooligomerization is likely to be innate property of E2 protein, because it was not observed in hGH protein. Oligomerization of envelope protein in other enveloped RNA viruses, such as vesicular stomatitis virus G protein and simian and human immunodeficiency virus envelope protein, has also been observed (34 -36). It was reported that oligomerization was shown to be required for intracellular transport and cell surface expression TABLE I Detection of anti-E2 antibody present in patients' sera using the purified hGHE2t protein Anti-E2 antibody in various patients' sera was detected by an enzyme-linked immunosorbent assay with the recombinant purified hGHE2t protein (400 ng/well). Anti-hGH antibody was not detected in patients' sera when purified hGH protein was used as a control protein. We determined the cutoff value of the assay based on the S.D. of 62 patients' sera found to be negative for anti-HCV assay.   19 patients with various levels of anti-E2 antibody were tested for the reactivity to hGHE2t protein at various conditions. The purified hGHE2t protein was treated with 0.2% SDS (SDS-treated), heat by boiling (heat-treated), 100 mM ␤-ME (␤-ME-treated), boiling in the presence of 100 mM ␤-ME (␤-ME/heat-treated), and N-glycosidase F (Degly) pretreated with ␤-ME/heat. The treated hGHE2t protein was coated into microtiter plates, and ELISA was performed as described under "Experimental Procedures." The numbers represent the value of optical density measured at 450 nM. of many viral glycoproteins (37) and that the preM and the E protein of West Nile virus are present as heterodimers in cell-associated virus, whereas the E protein of extracellular virus has a tendency to oligomerize into a trimer (38). Although the actual arrangement of E1 and E2 proteins in the virion is not known, our data support that the processing and maturation of envelope proteins of HCV would be similar to those of pestiviruses, in which the homo-oligomeric complex of envelope proteins, E0 and E2, is formed by an intermolecular disulfide bond. When anti-E2 antibody levels of patient sera were determined by ELISA using the purified hGHE2t protein, a wide range of anti-E2 antibody levels were observed in chronically infected patients (range, 420 to 44320; mean 4352; n ϭ 28) (data not shown). This result is well consistent with the previous results (15). However, the correlation between the titer of anti-E2 antibody and specific liver disease was not observed. It is very interesting to note that several cases of hemodialysis patients' sera were determined as positive with both anti-E2 antibody detection and reverse transcription PCR analysis, even though assay using HCV ELISA 3.0 was shown to be negative. Since immune responses might be down-regulated in immunosuppressed patients such as patients with chronic renal failure, it is possible that the antibody response to intracellular protein is weakly generated, not enough to be detected by the current anti-HCV diagnostic test. In contrast, the antibody response to envelope protein would be elicited because the E2 glycoprotein on the surface of the infected cell and the virion might be taken up by antigen-presenting cells and/or can act as a T cell-independent antigen owing to its heavy glycosylation. In this regard, the E2 protein of HCV has the potential to be used for the detection of HCV infection in immunocompromised patients.
It was previously reported that E2-specific antibodies from patients with hepatitis C were able to bind to the native E2 protein much better than to the denatured E2 protein (10). In addition, monoclonal antibodies against the envelope protein of hog cholera virus (E0; gp44/48) appeared to react only with its native form, suggesting that these antibodies appear to recognize discontinuous epitope(s) of E0 which may be generated by the formation of higher structures (38). In contrast, our purified hGHE2t protein denatured with SDS or boiling did not show any significant differences in the reactivity to anti-E2 antibody of patients' sera. The discrepancy might be due to the denaturation condition and/or the E2 protein used. The reactivity of the purified hGHE2t protein to anti-E2 antibody was, however, decreased by treatment of a reducing agent, suggesting that the purified hGHE2t protein has a discontinuous antigenic epitope(s) generated by disulfide linkage(s).