Species-Specific Regions of Occludin Required by Hepatitis C Virus for Cell Entry

Abstract

Hepatitis C virus (HCV) is a leading cause of liver disease worldwide. As HCV infects only human and chimpanzee cells, antiviral therapy and vaccine development have been hampered by the lack of a convenient small-animal model. In this study we further investigate how the species tropism of HCV is modulated at the level of cell entry. It has been previously determined that the tight junction protein occludin (OCLN) is essential for HCV host cell entry and that human OCLN is more efficient than the mouse ortholog at mediating HCV cell entry. To further investigate the relationship between OCLN sequence and HCV species tropism, we compared OCLN proteins from a range of species for their ability to mediate infection of naturally OCLN-deficient 786-O cells with lentiviral pseudoparticles bearing the HCV glycoproteins. While primate sequences function equivalently to human OCLN, canine, hamster, and rat OCLN had intermediate activities, and guinea pig OCLN was completely nonfunctional. Through analysis of chimeras between these OCLN proteins and alanine scanning mutagenesis of the extracellular domains of OCLN, we identified the second half of the second extracellular loop (EC2) and specific amino acids within this domain to be critical for modulating the HCV cell entry factor activity of this protein. Furthermore, this critical region of EC2 is flanked by two conserved cysteine residues that are essential for HCV cell entry, suggesting that a subdomain of EC2 may be defined by a disulfide bond.

Hepatitis C virus (HCV), a member of the family Flaviviridae, is the causative agent of classically defined non-A, non-B hepatitis and is highly prevalent, with approximately 3% of the worldwide population infected (48). HCV infection often results in a chronic, life-long infection that can have severe health consequences, including hepatitis, cirrhosis, hepatocellular carcinoma, and liver failure. There is no HCV vaccine available, and the currently employed interferon-based treatment is inadequate as it has severe side effects and is effective only in half of the major genotype-infected individuals (22, 32). Specific anti-HCV inhibitors targeting the viral proteases and polymerase are currently being developed and will likely improve therapeutic options substantially. Undoubtedly, however, the emergence of viral resistance to such inhibitors will be a problem facing future HCV treatment options. As such, developing a spectrum of inhibitors targeting diverse steps in the virus life cycle, including HCV cell entry, is a priority for HCV research. Such inhibitors may be particularly useful following liver transplantation. Although HCV is the leading cause of liver transplants worldwide (10), the usefulness of such procedures is limited by subsequent universal graft reinfection and often accelerated disease progression (21). Even transiently inhibiting graft reinfection with HCV cell entry inhibitors could greatly improve the effectiveness of this procedure. Therefore, a greater understanding of HCV cell entry is required for the development of therapies targeting this stage of the viral life cycle.

HCV host cell entry is a complex process that culminates in the clathrin-dependent endocytosis of the virion and low-pH-mediated fusion of viral and cellular lipid membranes in an early endosome (9, 12, 26, 27, 36, 51). The entry process requires the two viral envelope glycoproteins, E1 and E2, and many cellular factors, including glycosaminoglycans (GAGs) (3, 27), lipoproteins, the low-density lipoprotein receptor (LDL-R) (1, 38-40), tetraspanin CD81 (43), scavenger receptor class B type I (SR-BI) (47), and two tight junction proteins, claudin-1 (CLDN1) (17) and occludin (OCLN) (31, 44). The polarized nature of hepatocytes and the tight junction roles of OCLN and CLDN1 suggest an entry pathway similar to that of the group B coxsackieviruses, where the virion initially binds readily accessible factors that then provide a mechanism for migration of the virion into the tight junction region, just prior to internalization (14). Indeed, cellular factors are utilized by the incoming HCV virion in a temporal manner. At least GAGs and LDL-R appear to mediate virion binding (1, 3, 27, 38-40). Conflicting evidence has shown that SR-BI acts as either a binding (11) or postbinding entry factor (53), while CD81 (7, 13, 17, 27) and CLDN1 (17, 29) play postbinding roles in the HCV cell entry process. Although the kinetics of OCLN usage have not been clearly defined, this protein does not appear to play a role in virion binding (6). However, recent data showing that CD81 and CLDN1 may form complexes prior to infection (15, 24, 25, 28, 29, 35, 52) and imaging of the cell entry process (12) may contradict such a model.

Human hepatocytes are the major target for HCV infection. While multiple blocks at a number of viral life cycle stages likely exist in other cell types, cell entry is one of the events limiting HCV tropism (45). Although species differences in SR-BI and CLDN1 may exert some influence on this selectivity (11, 23), CD81 and OCLN appear to be largely responsible for the restriction of HCV entry to cells from human and chimpanzee origin (7, 8, 20, 44). In fact, overexpression of the human versions of CD81 and OCLN, along with either mouse or human SR-BI and CLDN1, renders a mouse cell able to support HCV cell entry (44).

We sought to provide greater insight into the species-specific restrictions of HCV cell entry and to elucidate the mechanism by which OCLN acts to mediate HCV cell entry. We examined the ability of OCLN proteins from a range of species to mediate HCV cell entry and how this function correlated with the degree of similarity to the human protein. A six-amino-acid portion of the second extracellular loop (EC2) of human OCLN was found to be responsible for the species-specific differences in entry factor function. OCLN proteins that were less functional than the human protein could be rendered fully functional by adding the human residues at these positions. Conversely, the ability of the human OCLN protein to mediate HCV cell entry was impaired by swapping this region with the corresponding sequence from species with less functional OCLN proteins. Comprehensive alanine scanning of the extracellular loops of human OCLN confirmed that the second half of EC2 was most important for the HCV cell entry process. Two cysteine residues that flank this region were found to be essential for HCV cell entry, suggesting that these residues may define a disulfide-linked subdomain of EC2. None of these amino acid changes influenced OCLN expression or localization, implying that they may serve to modulate an interaction with either another host protein or the incoming HCV virion.

MATERIALS AND METHODS

Cell culture and cell lines.

293T and 786-O cells were grown in Dulbecco's modified Eagle medium (DMEM; Mediatech, Inc., Manassas, VA) with 100 U of penicillin per ml, 100 μg of streptomycin per ml (Mediatech, Inc., Manassas, VA), and 10% fetal bovine serum (FBS; Atlanta Biologicals, Lawrenceville, GA).

Plasmid construction.

OCLN genes were expressed in 786-O cells via lentiviral transduction from the context of pTRIP (49, 54), a self-inactivating lentiviral provirus that expresses no HIV proteins but instead employs an internal cytomegalovirus (CMV) promoter to express cloned genes. TRIP-Venus, TRIP-Venus-hOCLN, and TRIP-Venus-mOCLN, which express the Venus fluorescent protein alone (provided by Atsushi Miyawaki, RIKEN, Saitama, Japan) or fusions of the Venus protein to the amino terminus of human OCLN (hOCLN) and mouse OCLN (mOCLN), respectively, have been previously described (44) (kindly provided by Charles Rice, Rockefeller University, NY). TRIP-Venus-ptOCLN, TRIP-Venus-paOCLN, and TRIP-Venus-mmaOCLN constructs expressing the chimpanzee (Pan troglodytes; GenBank accession number XM_001158243), orangutan (Pongo abelii; GenBank accession number NM_001133668), and rhesus macaque (Macaca mulatta; GenBank accession number XM_001094789) OCLN genes, respectively, were generated by site-directed mutagenesis of the human OCLN sequence to incorporate 1, 6, and 15 species-specific changes, respectively. Standard cloning techniques were used involving two rounds of overlapping PCR with internal oligonucleotides encoding each mutation used in combination with outside oligonucleotides that span unique restriction sites used to clone mutant PCR products back into the TRIP vector (oligonucleotide sequences and additional details are available upon request). All PCR-amplified sequences were fully sequenced. All point mutants and alanine scanning mutants tested were generated by this method as well (oligonucleotide sequences are available upon request).

Similar techniques were used to clone the remaining OCLN orthologs tested. For those that have been previously published, we obtained plasmids encoding the respective open reading frames (ORFs) for PCR amplification to facilitate cloning in the TRIP plasmid. TRIP-Venus-cfOCLN encodes the canine OCLN sequence (Canis lupus familiaris; GenBank accession number NM_001003195) and was created by amplifying this gene from an expression construct provided by Mikio Furuse (Kobe University Graduate School of Medicine, Hyogo, Japan) (2) with the forward oligonucleotide 5′-GAC GAG CTG TAC AGA TCT AGA ATG TCA TCG AGG CCT TTT GAG AGT and reverse oligonucleotide 5′-CGCG CTC GAG CTA TGT TTT CTG TCT ATC ATA GTC TCC AAC. This PCR product was digested with restriction enzymes BsrGI and XhoI and was ligated into similarly digested TRIP-Venus plasmid. The rat OCLN sequence (Rattus norvegicus; GenBank accession number NM_031329) was amplified with forward oligonucleotide 5′-GAC GAG CTG TAC AGA TCT AGA ATG TCT GTG AGG CCT TTT GAG AGT and reverse oligonucleotide 5′-CGCG CTC GAG CTA GGT TTT CCG TCT GTC ATA GTC from an expressed sequence tag (EST) clone (IMAGE clone 5623744; Open Biosystems, Huntsville, AL) in two steps to silently disrupt an internal BsrGI site. This PCR product was digested with restriction enzymes BsrGI and XhoI and was ligated into the similarly digested TRIP-Venus plasmid to generate TRIP-Venus-rnOCLN.

As the complete hamster and guinea pig OCLN sequences have not been reported, we rescued their ORF sequences by reverse transcription-PCR (RT-PCR) with oligonucleotides designed of highly conserved regions of OCLN cDNAs. To ensure that the correct sequences were identified, multiple independent cDNA synthesis and PCRs were used to assemble consensus sequences. Reverse transcription was performed with an AccuScript High Fidelity 1st Strand cDNA Synthesis Kit (Strategene, Santa Clara, CA) according to the manufacturer's instructions. The hamster OCLN sequence was isolated by RT-PCR from total RNA prepared from a hamster (Mesocricetus auratus) liver sample (kindly provided by Alexander Ploss, Rockefeller University, NY). First-strand cDNA synthesis was performed with a reverse oligonucleotide (5′-AAA ATT CTT AAT TGG AGT GTT CAG CCC AGT) that anneals to nucleotide positions 2182 to 2211 of the human OCLN cDNA, and PCR was performed with this reverse oligonucleotide and a forward oligonucleotide (5′-TAA AGA TCA GGT GAC CAG TGA CA) that corresponds to the human OCLN cDNA nucleotide positions 408 to 429. The amplified product includes the entire hamster OCLN ORF, as well as 6 and 193 nucleotides of 5′ and 3′ untranslated sequence, respectively. Reassuringly, our sequence matched a small internal region of hamster OCLN that has been reported previously (GenBank accession number EU856106). For cloning into the TRIP lentivirus, oligonucleotides 5′-TG TAC AGA TCT AGA ATG TCT GTG AGG CCT TTT GAG AGT CCA CCT CCT TA and 5′-CGCG CTC GAG CTA GGT TTT CCG TTT GTC ATA GTC were used to precisely amplify the hamster OCLN ORF. This product was digested with restriction enzymes BsrGI and XhoI and was ligated into like digested TRIP-Venus plasmid to generate TRIP-Venus-maOCLN.

As a template for guinea pig RT-PCR, total RNA was extracted with a Qiagen RNeasy RNA extraction kit (Qiagen, Valencia, CA) from a section of guinea pig (Cavia porcellus) liver (provided by Nicole Bouvier, Mount Sinai School of Medicine, NY). Reverse transcription was performed with a reverse oligonucleotide 5′-TTC AGC CCA GTC AAT TAT CAA AAG C that anneals to nucleotide positions 2168 to 2192 of the human OCLN cDNA, and PCR was performed with this reverse oligonucleotide and the forward oligonucleotide 5′-AGG TGA CCA GTG ACA TCA GCC ATG TC, which corresponds to the human OCLN cDNA nucleotide positions 415 to 441. Although this PCR product includes 161 nucleotides of 3′ untranslated sequence, the forward oligonucleotide anneals to the first five bases of the OCLN ORF, and thus we cannot confirm the sequence of these few nucleotides. This region is absolutely conserved between mammalian OCLN cDNAs and cannot have extensive differences from the oligonucleotide if PCR is to be efficiently performed; thus, it is likely correct as reported. For cloning into the TRIP lentivirus, oligonucleotides 5′-GAC GAG CTG TAC AGA TCT AGA ATG TCA TCG AGG CCA TTT GAA AGT and 5′-CGCG CTC GAG CTA TGT TTT CTG TCT GTC ATA ATC were used to precisely amplify the guinea pig OCLN ORF. This product was digested with restriction enzymes BsrGI and XhoI and was ligated into similarly digested TRIP-Venus plasmid to generate TRIP-Venus-cpOCLN.

For confocal analysis of OCLN localization, wild-type and mutant human, mouse, and guinea pig OCLN sequences were cloned in the TRIP vector as fusions to the amino terminus of enhanced green fluorescent protein (EGFP). For this construction, first a unique PmeI site was inserted upstream of EGFP by amplification of TRIP-EGFP with oligonucleotides 5′-GGA TCC CCA CCG GGT TTA AAC ATG GTG AGC AAG GGC GAG GAG and 5′-TCT CGA GCT AGA GTC GCG GCC GCT TTA. This PCR product was cloned as a BamHI- and XhoI-digested insert into similarly digested TRIP-GFP to produce TRIP-PmeIGFP. The following PCR products were cloned as BglII and PmeI digested into this plasmid digest with BamHI and PmeI. Human OCLN was amplified with oligonucleotides 5′-AGA TCT AGA ATG TCA TCC AGG CCT CTT GAA and 5′-GTT TAA ACC CGG TGG GGA TCT TGT TTT CTG TCT ATC ATA GTC and cloned as a partially BglII- and PmeI-digested insert into TRIP-PmeIGFP digested with BamHI and PmeI. Mouse OCLN with a silently mutated internal BglII site was amplified with oligonucleotides 5′-AGA TCT AGA ATG TCC GTG AGG CCT TTT GAA and 5′-GTT TAA ACC CGG TGG GGA TCT AGG TTT CCG TCT GTC ATA ATC and cloned as a BglII- and PmeI-digested insert into TRIP-PmeIGFP digested with BamHI and PmeI. Guinea pig OCLN was amplified with oligonucleotides 5′-AGA TCT AGA ATG TCA TCG AGG CCA TTT GAA and 5′-GTT TAA ACC CGG TGG GGA TCT TGT TTT CTG TCT GTC ATA ATC and cloned as a partially BglII- and PmeI-digested insert into TRIP-PmeIGFP digested with BamHI and PmeI.

For assays testing the susceptibility of various cell populations to HCV pseudoparticles (HCVpp) and pseudoparticles harboring the vesicular stomatitis virus glycoprotein virus G protein (VSVGpp), a provirus expressing a secreted Gaussia luciferase (GLuc) reporter termed V1-GLuc was utilized. To construct this plasmid, the GLuc gene was amplified with oligonucleotides 5′-G GCC ATT ACG GCC GCC ATG GGA GTC AAA GTT CTG TT and 5′-G GCC GAG GCG GCC TTA GTC ACC ACC GGC CCC CTT. This product was digested with SfiI and cloned into a similarly digested pV1-SfiI0.5link plasmid (17), which is a minimal HIV-1 provirus with most genes deleted but with the Tat, Rev, and Vpu ORFs, as well as all necessary cis-acting sequences intact. In this context, GLuc is expressed in place of the Nef gene.

The no-envelope glycoprotein (no-Env) expression construct is pCDNA3.1 (Invitrogen, Carlsbad, CA) without a transgene, and constructs expressing various genotype HCV envelope proteins have been previously described: H77 (34), Con1 (19, 34), JFH-1 (17), S52 (37), ED43 (37), and SA13 (37).

Pseudoparticle generation and infection.

Pseudoparticle production was performed as previously described (17, 44) by cotransfection of three plasmids encoding (i) a provirus containing the desired reporter (V1-GLuc) or transgene (TRIP provirus based), (ii) HIV Gag-Pol, and (iii) the necessary envelope glycoprotein(s). 293-T cells were seeded at 1.8 × 10⁶ cells/well into a poly-l-lysine (Sigma)-coated six-well plate. Transfection was performed the next day using a total of 1.5 μg of DNA plasmid, with 6 μl of TransIT-LT1 transfection reagent (Mirus, Madison, WI). Supernatants were collected 24, 48, and 72 h posttransfection, filtered (0.45-μm pore size), and mixed with 100 μl of 1 M HEPES buffer. All transductions and infection assays using pseudoparticles were performed in the presence of 4 μg/ml Polybrene (Sigma, St. Louis, MO).

For transgene expression with TRIP provirus-packaged VSVGpp, 786-O cells were seeded at 1.5 × 10⁵ cells per well of a 24-well plate. The following day cells were infected with 1.7 ml of either undiluted pseudoparticle supernatant or a 5-fold dilution of pseudoparticle supernatant in normal medium. To allow the cells to recover from any adverse affects of transduction and to stabilize expression, cells were passed into progressively larger plates for 1 to 2 weeks, until a confluent 10-cm-diameter tissue culture plate was derived.

For infection assays with GLuc reporter HCVpp and VSVGpp, 786-O cells transduced to express various fluorescently tagged OCLN proteins were seeded at a density of 1.2 × 10⁵ cells per well into a poly-l-lysine-coated 24-well plate. In order to challenge cells with approximately equal numbers of infectious units, HCVpp and VSVGpp were diluted in medium plus Polybrene at 1:2 and 1:5,000, respectively. Twenty-four hours postinfection, cells were washed three times with fresh medium to remove GLuc protein that was present in the pseudoparticle inoculum. Luciferase assays were performed at 48 h postinfection. Briefly, 50 μl of supernatant was collected and stabilized by the addition of 50 μl of a 2× dilution of Renilla luciferase assay lysis buffer (Promega, Madison, WI). For the assay, 40 μl of Renilla luciferase substrate (Promega, Madison, WI) was added to 10 μl of diluted supernatant in a 96-well assay plate, and luciferase was quantified with a BioTex Synergy 4 multidetection microplate reader (BioTek, Winooski, VT).

Immunoblot analysis.

For immunoblot analysis of OCLN expression, cells were lysed in a volume of 1× SDS-PAGE sample buffer plus dithiothreitol (DTT) that was proportional to the approximate cell confluence. Cell lysates were passed through a 22-gauge needle several times and heated for 5 min at 95°C. Equivalent volumes of lysate were immunoblotted with mouse anti-β-actin antibodies (AC-15; Sigma, St. Louis, MO) to ensure analysis of equal protein concentrations. Horseradish peroxidase (HRP)-conjugated rabbit anti-mouse secondary antibody (Thermo Scientific, Rockford, IL) was used, and detection was performed with Immobilon chemiluminescent HRP (Millipore, Billerica, MA). For immunoblotting for Venus-OCLN fusion proteins, rabbit anti-GFP (ab290; Abcam, Cambridge, MA) and goat anti-rabbit HRP (Thermo Scientific, Rockford, IL) were used.

Cell staining, fixing, and mounting for confocal microscopy.

786-O cells transduced to stably express various OCLN-GFP constructs were seeded on poly-l-lysine-coated 0.17-mm glass coverslips. At 48 h postseeding cells were stained with 1.56 μg/ml Alexa Fluor 594-conjugated wheat germ agglutinin (WGA) membrane stain and 1.25 μM Hoechst stain for nuclei (Invitrogen, Carlsbad, CA) for 10 min at room temperature according to the manufacturer's directions and then fixed with 4% paraformaldehyde (PFA). Prepared slides were imaged on a Leica TCS-SP5 DMI laser scanning confocal microscope. Images were processed using the ImageJ program (http://rsb.info.nih.gov/ij/). Representative images show Z-sections of approximately 0.5 μm.

Statistical and phylogenetic analysis.

Data were analyzed for statistical significance using Prism software (GraphPad) using an unpaired Student's t test. A P value of ≤0.05 was considered significant. The phylogenetic tree shown in Fig. 1A comparing the OCLN proteins of various species was generated by using the neighbor-joining method (46) with bootstrap testing (1,000 replicates) (18) in the MEGA5 program (50). The evolutionary distances were computed using the Poisson correction method (55).

FIG. 1.

Comparison of HCV cell entry activity of OCLN orthologs. (A) Phylogenetic tree comparing the amino acid sequence of the OCLN proteins expressed by the indicated species. The evolutionary history was inferred using the neighbor-joining method (46). The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are shown next to the branches (18). The tree is drawn to scale, with lengths proportional to the evolutionary distances computed using the Poisson correction method (55). The scale bar represents the number of amino acid substitutions per site across this distance. Evolutionary analyses were conducted in MEGA5 (50). (B) Absolute luciferase values resulting from parallel HCVpp (light gray columns) and VSVGpp (dark gray columns) infections of 786-O cells either not expressing a transgene (Naïve) or transduced with lentiviral pseudoparticles to express either the Venus fluorescent protein alone (Venus) or a Venus protein fusion to the OCLN protein from the indicated species. Values are normalized to cells expressing human OCLN. Mean and standard error of greater than four independent experiments, each performed in quadruplicate, are shown. (C) Comparison of the relative HCVpp infectivity from the above 786-O cell populations, where HCVpp values are normalized to cells expressing human OCLN and corrected for variations in VSVGpp infectivity. Mean and standard error of greater than four independent experiments, each performed in quadruplicate, are shown. (D) Immunoblots for either the Venus protein or β-actin of lysates from 786-O cell populations transduced to express the transgenes corresponding to the above graph. Approximate molecular mass (kDa) marker positions are indicated to the left of each blot. α, anti.

Nucleotide sequence accession numbers.

The hamster OCLN sequence was deposited in GenBank as accession number HQ174780. The guinea pig OCLN sequence was deposited in the GenBank under accession number HQ174781.

RESULTS

OCLN species-specific capacity to mediate HCV cell entry.

To investigate the relationship between OCLN sequence and the limited species tropism of HCV cell entry, we tested the OCLN proteins from a variety of species for the capacity to mediate HCV cell entry. In addition to the previously tested human and mouse sequences, seven OCLN orthologs were assayed (Fig. 1A). Although most of these sequences have been previously reported (see Materials and Methods for references), the hamster and guinea pig sequences were determined by RT-PCR analysis of OCLN transcripts in liver tissue samples. While the chimpanzee OCLN sequence differed from the human sequence by only a single amino acid in the intracellular carboxyl terminus, the most divergent sequence was from guinea pig, which has an OCLN sequence 88% identical to that of the human protein. OCLN variants were expressed from lentiviral proviruses fused to Venus yellow fluorescent proteins, which provided a simple means of tracking cell transduction levels and relative protein levels. To test the ability of these OCLN variants to mediate viral entry, human renal carcinoma 786-O cells, which express high levels of CD81, SR-BI, and CLDN1 but are not permissive to HCV infection due to insufficient OCLN levels (44), were transduced with these proviruses packaged within lentiviral pseudoparticles bearing the vesicular stomatitis virus glycoprotein (VSVGpp). Since these cells do not support HCV RNA replication, we assayed their capacity to support the HCV cell entry pathway with lentiviral pseudoparticles that express a reporter gene and bear the HCV glycoproteins (HCVpp) (4, 16, 26). To control for nonspecific effects on lentiviral entry and reporter expression resulting from cell health or density, cells were challenged in parallel with VSVGpp that express that same reporter gene. Expression of the human OCLN protein in these cells enhanced HCVpp infection by approximately 20-fold over naïve or Venus protein-transduced cells (44) (Fig. 1B, light gray columns). The absolute HCVpp infectivity changed dramatically depending on which OCLN ortholog was transduced (Fig. 1B, light gray columns). This effect was specific for the HCV cell entry pathway, as VSVGpp infectivity varied only slightly between these cell populations (Fig. 1B, dark gray columns). To correct for differences in general pseudoparticle permissivity, all subsequent infection data will be reported as relative infectivity, where the HCVpp results are normalized to the VSVGpp values (Fig. 1C).

As shown in Fig. 1B, expression of OCLN proteins of various species in 786-O cells resulted in a wide range of HCVpp susceptibilities. This capacity to mediate HCV cell entry followed the same pattern as sequence similarity to human OCLN (Fig. 1A) and did not correlate with relative expression levels, as gauged by fluorescence-activated cell sorting (FACS) analysis of Venus fluorescence (data not shown) and immunoblotting with antibodies specific for the Venus fluorescent protein, which serves as a common epitope between these proteins, or β-actin as a loading control (Fig. 1D).

While no statistically significant differences were observed between the ability of the primate OCLN sequences to mediate HCVpp infection (P > 0.2 compared to human OCLN), all other OCLN proteins tested were less functional in the HCV cell entry process than the human protein (P < 0.05) (Fig. 1C). Canine OCLN, which is 91.0% identical to the human protein, exhibited 81% of the human protein's activity. The hamster, mouse, and rat OCLN proteins are 90.5%, 89.,5% and 88.6% identical to the human protein, respectively, and mediated HCVpp infection of 786-O cells with relative efficiencies of 33%, 25%, and 23%, respectively. In contrast, guinea pig OCLN (88.0% identical to human OCLN) did not exhibit any detectable capacity to mediate HCV entry. These OCLN proteins served as the foundation of our subsequent analysis to identify specific amino acids within OCLN that are important for the HCV cell entry process.

HCV cell entry determinants that differ between human and mouse OCLN.

OCLN is a four-transmembrane domain-containing protein with intracellular termini and two extracellular loops termed EC1 and EC2 (Fig. 2A). The difference between the capacities of human and mouse OCLNs to mediate HCV cell entry maps to their second extracellular loops (44). As shown previously and confirmed here, replacing mouse OCLN EC2 with that of the human protein renders this chimera as functional as the human protein (44) (Fig. 2C, compare columns 2 and 4). Six amino acids differ between the human and mouse OCLN proteins within this domain (Fig. 2B). To determine which of these variations were responsible for the difference in HCV cell entry functionalities between these species, we tested mutant mouse OCLN proteins bearing the corresponding human EC2 residues.

FIG. 2.

Analysis of human versus mouse OCLN HCV cell entry factor determinants. (A) Illustration of OCLN membrane topology. (B) Alignment of EC2 regions of human and mouse OCLNs. Amino acid positions relative to human OCLN are indicated above. Identical and similar amino acids are indicated by dark and light shading, respectively. (C) HCVpp infectivity (normalized to parallel VSVGpp infections and relative to infections of cells expressing human OCLN) of 786-O cells expressing the indicated mouse or human OCLN mutants. The columns have been numbered to allow a clearer presentation of these data in the results section. Mean and standard error of greater than four independent experiments, each performed in quadruplicate, are shown. (D) Immunoblots for either the Venus protein or β-actin of lysates from 786-O cell populations transduced to express the indicated transgenes. Approximate molecular mass (kDa) marker positions are indicated to the left of each blot. α, anti.

Although no single human amino acid EC2 substitution rendered mouse OCLN as functional as the human protein for mediating HCV cell entry, the fifth and sixth EC2 changes each significantly enhanced this activity (P < 0.001) (Fig. 2C, columns 5 to 10). In this context the G224A mutation (all references to OCLN amino acid positions will be relative to the sequence of the human protein) only slightly enhanced activity of the mouse protein from 25%, for the wild-type version, to 45% as functional as human (Fig. 2C, column 10). In contrast the G223A mutation exerted a larger effect, raising the activity of mouse OCLN to 87% as functional as human OCLN (Fig. 2C, column 9). Although none of the first four EC2 changes on their own rendered mouse OCLN more functional, the combination of the third and fourth changes (M214A and I215L) slightly enhanced the ability of mouse OCLN to mediate HCVpp infection of 786-O cells, making this variant 48% as functional as the human ortholog (Fig. 2C, column 12). However, the combination of the last two changes (G223A and G224A) was sufficient to render mouse OCLN as functional as the human protein (P > 0.2) (Fig. 2C, column 13).

The importance of these two alanine residues at positions 223 and 224 was further demonstrated by testing mouse OCLN mutants with all the EC2 human changes except for either one or both of these last two changes (Fig. 2C, columns 14 to 16). The mouse protein with just the first four human changes (Fig. 2C, column 14) was no more active than the mutant with the third and fourth changes combined (Fig. 2C, column 12). Furthermore, mouse OCLN mutants with five human EC2 substitutions but missing either of the alanine changes at these positions were less functional, although not to a statistically different level (P = 0.365 and 0.878, respectively), than human OCLN (Fig. 2C, column 15 and 16). These results indicate that both of the alanine changes at positions 223 and 224 are essential for generating a mouse OCLN that is as active as the human protein in mediating HCV cell entry.

Human OCLN mutants with the corresponding mouse EC2 changes were also tested for the ability to mediate HCVpp entry into 786-O cells. In this context all but the first mutation caused a statistically different reduction in HCV entry factor activity (P < 0.002) (Fig. 2C, columns 17 to 22). However, only the fifth change (A223G) decreased activity to less than 70% of the wild-type protein. As in the mouse context, the last two changes (A223G and A224G) account for the difference in the ability to mediate HCV cell entry as this human variant was only as active as wild-type mouse OCLN (P < 0.05) (Fig. 2C, column 23).

The expression and transduction efficiencies of each OCLN variant tested were monitored by tracking the fluorescence signature conferred by the Venus yellow fluorescent protein fused to each OCLN. In these experiments, each culture was completely transduced, and no substantial fluorescence differences were detected (data not shown). We further confirmed by immunoblotting lysates from cells expressing the wild-type and mutant versions of each protein that exhibited the greatest impact on HCV cell entry that these differences in HCV cell entry activity were due not simply to differences in protein expression. The wild-type human OCLN and the human A223G A224G mutant were expressed at comparable levels, as were the wild-type and G223A G224A mouse proteins (Fig. 2D). Thus, mutations that completely swapped the abilities of human and mouse OCLNs to mediate HCV entry did not affect the steady-state expression levels of these proteins.

OCLN HCV cell entry determinants between species.

To investigate if the determinants for HCV cell entry in human and mouse OCLNs could be generalized for other OCLN orthologs, the OCLN EC2 sequences of the various species tested in the experiment shown in Fig. 1C were compared (Fig. 3A). Of the nonhuman primate OCLN sequences, all but one encoded identical EC2 sequences. Thus, it is reasonable to hypothesize that primate OCLN orthologs have the same determinants of HCV cell entry as human OCLN. The only sequence difference between human and rhesus macaque OCLN EC2 is a valine at position 224 instead of the human alanine (Fig. 3A). As shown above, the A224G change in human OCLN impairs HCV entry functions somewhat but to a lesser degree than mutations at the previous residue 223 (Fig. 2C). Although such an amino acid change is not generally considered to be dramatic, it is conceivable that this difference is responsible for the slight, but not statistically significant (P > 0.2), reduction in the rhesus macaque OCLN entry function (94% of the human protein) (Fig. 1C).

FIG. 3.

Analysis of HCV cell entry factor determinants between additional OCLN orthologs. (A) Alignment of EC2 regions of OCLN proteins from the indicated species. Amino acid positions relative to human OCLN are indicated above, and identical and similar amino acids are indicated by dark and light shading, respectively. (B) HCVpp infectivity (normalized to parallel VSVGpp infections and relative to infections of cells expressing human OCLN) of 786-O cells expressing the indicated OCLN proteins. Wild-type OCLN proteins are labeled with the species name alone, and mutant versions are labeled with the corresponding amino acid changes (relative to the human sequence). Mean and standard error of greater than four independent experiments, each performed in quadruplicate, are shown. (C) HCVpp infectivity (normalized to parallel VSVGpp infections and relative to infections of cells expressing human OCLN) of 786-O cells expressing the indicated guinea pig and human mutant OCLN proteins. Mean and standard error of greater than four independent experiments, each performed in quadruplicate, are shown. (D) Immunoblots for either the Venus protein or β-actin of lysates from 786-O cell populations transduced to express the indicated transgenes. Approximate molecular mass (kDa) marker positions are indicated to the left of each blot. α, anti.

The remaining OCLN orthologs that were tested are all more evolutionarily divergent from the human sequence, and thus it is not surprising that there is also greater diversity within the EC2 region (Fig. 3A). Even though there are only six residues that differ between the human and canine OCLN proteins in the EC2 region, canine OCLN demonstrated 81% of the human protein's capacity to mediate HCV cell entry. However, changing a single residue at position 223, the critical residue in the murine analysis, from the endogenous threonine to the human alanine enhanced activity to 103% of human OCLN (Fig. 3B). With a glycine at position 223 and an alanine at 224, hamster OCLN appears to have a sequence in this region that is in between human and mouse, which may explain its slightly higher activity than mouse OCLN (Fig. 1B). In the hamster context as well, replacement of an alanine at position 223 raised activity to a level equivalent (97%) to that of the human protein (Fig. 3B). Rat OCLN is very similar to the mouse ortholog, and they encode identical OCLN determinants. In rat OCLN, glycines are present at positions 223 and 224 (Fig. 3A), and, as in mouse OCLN, mutation of both of these to alanines enhances HCV cell entry activity to nearly 94% of human levels (Fig. 3B). None of the changes to the canine, hamster, or rat OCLN proteins influenced the expression levels of these proteins (Fig. 3D). These results indicate that all OCLN proteins that exhibit at least a low-level capacity to mediate HCV cell entry rely on, and can be affected by, changes at amino acid positions 223 and 224.

Guinea pig OCLN, which does not exhibit detectable HCV cell entry factor activity (Fig. 1C), has an alanine at position 223. This presents a conflict with our above model, according to which an alanine at this position should confer at least some capacity to mediate HCVpp infection. However, there are numerous other differences between the guinea pig and human proteins clustered around this residue (Fig. 3A). Residue 224 exerts some influence on the capacity of guinea pig OCLN to mediate HCV entry as mutation of the endogenous glycine residue to alanine (G224A) resulted in a statistically significant increase (P < 0.05) in activity to the detectable level of 13% of the human protein (Fig. 3C). Testing guinea pig OCLN mutants with additional surrounding residues changed to the corresponding human sequence reveals that the serines at positions 225 and 226 are antagonistic to HCV cell entry functions. Although the S226G mutation did not exert a detectable effect on its own (P = 0.436), the S225T mutation alone enhanced entry activity to 40% of human OCLN and to 58% in combination with S226G (Fig. 3C). The triple G224A S225T S226G guinea pig OCLN mutant was 82% as active as the human protein. This appears to be the maximal activity that mutations in this region can bestow upon guinea pig OCLN as altering the amino acids at positions 222 and 227 to those of the human sequence actually slightly reduced the activity to 71% of human OCLN (Fig. 3C).

Testing the guinea pig substitutions in the human OCLN protein showed similar contributions of each residue toward the HCV cell entry activity of this protein. The A224G, T225S, and G226S single mutations slightly impaired the capacity of human OCLN to mediate HCVpp infection (Fig. 3C), and combining these mutations revealed progressively larger impairments. While no mutant was completely inactive, the human OCLN mutants with three (A224G T225S G226S) and five (P222A A224G T225S G226S L227I) changes to the guinea pig sequence had less activity than any of the human variants tested thus far, with just 11% and 8.1% of wild-type human OCLN activity, respectively. Although the latter mutant has very low activity, it is still significantly higher than the background level of 4.9% (P = 0.0004) that was observed with Venus-transduced cells. Again, such determinants did not influence OCLN expression levels as human OCLN with these five mutations was detected at a similar steady-state level by immunoblotting (Fig. 3D). In addition, although guinea pig OCLN is expressed at slightly lower levels than human OCLN, the three mutations (G224A S225T S226G) that greatly enhanced its entry factor activity did not enhance its expression (Fig. 3D).

Determinants of HCV cell entry factor activity in OCLN are viral genotype independent and do not affect localization.

The above analyses examined the ability of OCLN variants to mediate infection of 786-O cells with pseudoparticles bearing the envelope glycoproteins from a single HCV genotype 1a isolate. To determine if other HCV genotype glycoproteins utilize similar OCLN entry determinants, we challenged 786-O cells expressing either the Venus protein alone or Venus fused to wild-type human OCLN, the moderately impaired human OCLN mutant bearing two mutations from the mouse and rat orthologs (A223G and A224G), or the highly impaired human OCLN mutant bearing five guinea pig-specific changes (P222A A224G T225S G226S L227I) with lentiviral pseudoparticles bearing various glycoproteins. Because each HCV glycoprotein pair produced pseudoparticles with a wide range of infectivities, we also infected in parallel the naturally HCV-infectible Huh-7.5 cell line to gauge the relative infectivity of each virus (Fig. 4). As previously determined, VSVGpp infected all cells with similar efficiencies independent of which OCLN was expressed (Fig. 4). In Fig. 4, we also show that the infectivity of pseudoparticles that do not contain a viral glycoprotein (no Env) is not enhanced by the expression of any of the forms of human OCLN. Thus, these results indicate that the effects of OCLN on viral entry are specific for HCVpp. Although the general titers of each HCVpp isolate vary greatly depending on which isolate's envelope proteins are incorporated, the general trend of the effects of OCLN mutations is consistent between the viruses. All HCVpp display little to no infectivity in 786-O cells not expressing OCLN, and each isolate is enhanced by the expression of wild-type human OCLN (Fig. 4, compare white and light gray columns). All HCVpp infections are moderately impaired by the mouse and rat changes, and this effect was even greater in cells expressing the human OCLN protein bearing the guinea pig-specific mutations (Fig. 4, dark gray and black columns, respectively). One exception was observed with HCVpp bearing the genotype 4a strain ED43 glycoproteins, which were so inefficient that neither human OCLN mutant supported infection of this challenge virus above background levels. Thus, it could not be determined if the guinea pig-specific mutants had a greater impact than the mouse mutations on infection with the envelope protein of this particular genotype. Taken together, these results indicate that the effects of OCLN determinants identified above that differ between species are conserved in a wide range of HCV genotypes.

FIG. 4.

HCV cell entry factor determinants in OCLN function similarly for HCV envelope proteins from a variety of genotypes. Pseudoparticles bearing either no envelope glycoprotein (No Env), as a negative control, or glycoproteins E1 and E2 from the indicated HCV genotype and strain were used to infect 786-O cells expressing the Venus fluorescent protein alone (Näive) or the Venus protein fused to either the wild-type, A223G A224G mutant, or P222A A224G T225S G226S L227I mutant human OCLN protein, as indicated on the figure. As a comparison of general pseudoparticle infectivity, Huh-7.5 cells were infected in parallel (dotted). Shown are the mean relative light units (RLU) and standard errors measured from quadruplicate infections.

If OCLN cell surface localization is required to mediate HCV cell entry, then it is possible that the mutations we examined above may be enhancing or impairing the ability of OCLN to participate in this process by altering its trafficking to that location. To evaluate this possibility, we visualized the location of various OCLN proteins in 786-O cells by confocal fluorescence microscopy. For this analysis we chose to focus on a subset of the above OCLN variants. These included the wild-type human protein and mutants bearing moderately and highly impaired substitutions from the mouse and guinea pig sequences (Fig. 5B). In addition, we also examined localization of wild-type mouse OCLN and the G223A G224A mouse mutant version (Fig. 5C) as well as guinea pig OCLN proteins that lack detectable function and are greatly enhanced (wild type and G224A S225T S226G, respectively) (Fig. 5D). Each OCLN variant was cloned as a carboxyl terminus fusion with EGFP. Although cell membrane fluorescence was observed with the above Venus fusion proteins, this context also produced a large amount of intracellular, punctate fluorescence. We found the GFP fusion protein to produce similar levels of protein at the cell membrane but less intracellular fluorescence, which may have simply been a result of less expression of the GFP fusion protein (data not shown). The capacity of the different OCLN forms to mediate HCVpp entry of 786-O cells in the GFP context was identical to that observed in the Venus-OCLN context (data not shown). Cells were counterstained with red fluorescent Alexa Fluor 594 wheat germ agglutinin (WGA) and Hoechst 33342 dye to label the plasma membrane and nuclei, respectively. While a substantial portion of all OCLN-GFP fusion proteins localized intracellularly, some GFP signal is present on the cell membrane, as determined by colocalization with red WGA (Fig. 5, yellow in merged). Importantly, the degree of membrane localization was not altered by the mutations that influenced HCV cell entry activity. Thus, the mutations we identified that either impair the ability of human OCLN or enhance the ability of mouse and guinea pig OCLNs to mediate HCV cell entry do not dramatically change the presence of this protein on the plasma membrane.

FIG. 5.

Subcellular localization of OCLN variants. The localization of OCLN-GFP fusion proteins was visualized by confocal fluorescence microscopy as described in the Materials and Methods section. In each set blue, red, and green represent Hoechst 33342 nuclear DNA staining, Alexa Fluor 594 WGA staining of the cell membrane, and GFP fluorescence, respectively. Yellow in the third panel for each set is indicative of an overlap of the red and green signals. Representative images include naïve 786-O cells (A) or cells expressing the indicated wild-type and mutant versions of human (B), mouse (C), and guinea pig (D) OCLNs.

Species-independent OCLN HCV cell entry determinants.

The above comparison of the HCV cell entry activities of OCLN proteins from different species would reveal only determinants that differed between these various OCLN proteins. Some OCLN regions that are evolutionarily highly conserved might also be important for HCV cell entry (Fig. 6B). Because these would not be identified in the above species survey, we also conducted a more directed alanine scanning mutagenesis survey of the extracellular loops of human OCLN, where five residues of each domain were progressively replaced with alanine residues. For simplicity, naturally occurring alanines were not mutagenized, which had the added benefit of leaving intact the alanine residues at positions 223 and 224 found above to be important for HCV cell entry activity. Although all mutants were similarly expressed, as determined by fluorescence associated with the Venus fluorescent protein fusion (data not shown) and immunoblotting (Fig. 6C), many of these mutants were highly impaired for HCV cell entry factor activity (Fig. 6A). Broadly speaking, alanine mutagenesis was better tolerated within EC1 than within EC2. Within EC1, although most alanine substitution mutants retained their ability to mediate HCVpp infection of 786-O cells, the first two alanine sets (amino acids [aa] 86 to 90 and 91 to 95) and the last set (aa 136 to 139) were markedly handicapped at mediating this function (Fig. 6A). The first 20 amino acids of human OCLN EC2 also appear to be flexible in their sequence requirements as the four alanine substitution mutants that span this region retained most, if not all, of their abilities to mediate HCV cell entry (Fig. 6A). In contrast, alanine substitutions within the rest of this region either completely abrogated function or reduced function to just 23% of the wild-type human protein.

FIG. 6.

Analysis of HCV cell entry determinants in OCLN extracellular domains by alanine scanning mutagenesis. (A) HCVpp infectivity (normalized to parallel VSVGpp infections and relative to infections of cells expressing human OCLN) of 786-O cells expressing either the Venus fluorescent protein alone or Venus-human OCLN fusion proteins encoding the wild-type OCLN sequence or alanine mutants. Each mutant is labeled with the amino acid numbering for the region replaced with alanines. Mean and standard error of greater than three independent experiments, each performed in quadruplicate, are shown. (B) An alignment of the extracellular loop sequences EC1 and EC2 of the indicated OCLN orthologs, with the regions of alanine replacement marked by vertical lines, to illustrate the amino acid conservation of these regions. Numbering below the sequence is relative to the human OCLN protein. (C) Immunoblots for either the Venus protein or β-actin of lysates from 786-O cell populations transduced to express either the Venus fluorescent protein, Venus-human OCLN fusion proteins encoding wild-type OCLN (Human), or mutants with the indicated regions of EC1 and EC2 replaced with alanines. Approximate molecular mass (kDa) marker positions are indicated to the left of each blot. (D) HCVpp infectivity (normalized to parallel VSVGpp infections and relative to infections of cells expressing human OCLN) of 786-O cells expressing either the Venus fluorescent protein alone or Venus-human OCLN fusion proteins encoding the wild-type OCLN sequence or serine replacements of cysteine residues at positions 216 (C216S) and 237 (C237S). Mean and standard error of two independent experiments, each performed in quadruplicate, are shown. (E) Immunoblots for either the Venus protein or β-actin of lysates from 786-O cell populations transduced to express either the Venus fluorescent protein, Venus wild-type human OCLN, C216S, or C237S. Approximate molecular mass (kDa) marker positions are indicated to the left of each blot. α, anti.

The human OCLN with alanine mutations at positions 214 to 218 was particularly interesting to us. This mutant exhibited no HCV cell entry activity, yet the residues that were mutated to alanines are not well conserved (Fig. 6A). In fact, changes to three of these residues were well tolerated in the species analysis above. One of the remaining two residues is a completely conserved cysteine residue at position 216. We hypothesized that a disulfide bond involving this cysteine may be required for HCV cell entry activity. In support of this possibility, we found that human OCLN bearing a serine in place of this cysteine (C216S) was completely inactive for mediating HCV cell entry (Fig. 6D). The only other cysteine residue within the extracellular domains of OCLN is at position 237 in EC2. This residue is also completely conserved among the orthologs analyzed and is in a region found important for HCV cell entry by alanine scanning mutagenesis (Fig. 6A, mutant at position 234 to 238). Similar to C216S, mutagenesis of this residue to a serine (C237S) disrupted the ability of OCLN to mediate HCVpp infection of 786-O cells (Fig. 6D). Both the C216S and C237S OCLN mutants were expressed at similar levels as wild-type OCLN (Fig. 6E). These results suggest that a disulfide bond between these cysteine residues is important for the ability of OCLN to mediate HCV cell entry.

DISCUSSION

The HCV cell entry process requires numerous cellular factors, which are likely engaged by the incoming virion in a sequential manner to mediate a number of steps involved in this event. Although many cellular factors are known to be required for HCV cell entry, much work remains to be done to determine the mechanism by which they mediate this process. To gain further insight into the role of OCLN in HCV cell entry, we conducted an analysis of the regions of this protein required for mediating this process.

This analysis has particular value in terms of understanding the species tropism of HCV replication. Only human and chimpanzee cells are permissive to HCV infection, and this tropism is confined at least partially at the level of cell entry. CD81 and OCLN influence the species specificity of HCV cell entry as the human versions of these proteins are more active than those from more divergent species (7, 8, 20, 44). Although the mouse versions of SR-BI and CLDN1 may not function as well as the human versions, the differences in their activities are considerably less drastic than the differences between mouse and human CD81 and OCLN, and they are not seen in all assays or cell types (11, 23).

To further define how OCLN modulates the species specificity of HCV cell entry, we tested the OCLN proteins from a range of species for their ability to support this process. All nonhuman primate OCLN proteins were as efficient as the human protein at mediating HCV cell entry. Similarly, CD81 proteins from primates are also fully functional for mediating HCV cell entry (20). These findings along with the fact that neither gorillas nor rhesus macaques are infectible with HCV indicate that entry is likely not the only HCV life cycle restriction in animals other than humans and chimpanzees. To various degrees, all other OCLN proteins tested were less functional than the primate proteins. Canine OCLN was actually only moderately impaired compared to human OCLN. This result raises the possibility that HCV entry may efficiently occur in canine cells. In support of this possibility, although canine CD81 has not been formally tested for HCV entry factor function, it does not appear to contain any sequence previously shown to be detrimental for this activity (7, 20).

To complement our survey of species-specific OCLN HCV cell entry determinants, we used alanine scanning mutagenesis to further define the extracellular domains that are important for this process. Within the EC1 of OCLN, only the first two and the last alanine sets were dramatically impaired for mediating HCV cell entry. As residues directly proximal to transmembrane regions often act to brace these domains, these deficiencies may be a result of disruption of membrane topology rather than a requirement for specific amino acid residues at these positions. The observation that the sequence of EC1 can tolerate numerous sets of alanine substitutions and retain HCV cell entry activity does not mean that this domain is not required for this process. Indeed, the EC1 sequence is extremely repetitive, with an abundance of glycine and tyrosine residues (Fig. 6B). It is possible that HCV cell entry requires EC1; however, merely replacing five residues at a time in this region with alanines is not enough to visualize a reduction in activity.

Alanine scanning of EC2 showed the first 20 amino acids to be flexible in sequence; however, substitutions within the remainder of this domain either completely ablated or highly impaired entry factor activity. As these mutant sets overlap with the species-specific HCV cell entry determinants identified above, these results further support the importance of the latter half of EC2 in mediating this function. These defined mutants also extend the critical domain in a manner the species analysis could not as the last three regions mutagenized were almost completely evolutionarily conserved (Fig. 6B). Furthermore, we found two conserved cysteine residues that flank this region to be essential for entry factor activity. This is reminiscent of CD81, whose large extracellular loop, which is the region of this protein critical for HCV cell entry activity, has four cysteines that cross-link to form a subdomain of this protein (42). It remains to be determined if such cross-links occur within OCLN and if they are homo- or heterotypic within a single OCLN protein or between adjacent proteins, as previously hypothesized (33).

OCLN mutant sets derived in this study, which encode a small number of amino acid changes and yet exhibit dramatic differences in HCV cell entry factor activity, should serve as valuable tools for probing the mechanism by which OCLN mediates HCV cell entry. Several possibilities can be envisioned. On the most basic level, such mutations may influence OCLN expression levels or stability. However, probing the steady-state levels of the above OCLN variants by immunoblotting did not reveal any such correlation. Another possible mode of action is that mutations that impair OCLN function may prevent the proper localization of this protein to a site within the host cell. We analyzed the localization of the OCLN proteins from human, mouse, and guinea pig, including their wild-type sequence and mutants that dramatically influence their entry factor activity, by confocal fluorescence microscopy. Although significant amounts of the overexpressed OCLN proteins appear to be sequestered intracellularly, all OCLN variants exhibited at least some detectable surface expression. While it is possible that localization to discrete compartments on the cell surface is required for OCLN to mediate HCV cell entry, our analysis at least rules out gross disruption of localization, such as intracellular sequestration, as the reason why the nonfunctional HCV variants in this study do not mediate this event efficiently.

A third possible mechanism by which OCLN HCV cell entry factor determinants may exert an effect is by influencing interactions between OCLN and other factors required for the entry process. The simplest model for how a virion utilizes an entry factor is that there is a direct interaction with the host cell factor. Only SR-BI and CD81 have been conclusively shown to bind HCV E2 (43, 47). Interactions between both CLDN1 and OCLN and the HCV glycoproteins have been reported (5, 31, 52) although it is not clear if these are direct or if they actually occur in the context of an incoming virion. To determine the specificity and functionality of such interactions, it will be particularly important to show that versions of these proteins that do not support HCV cell entry also do not make such associations. For instance, although wild-type CLDN1 can coimmunoprecipitate with either E1 or E2, this interaction may not reflect entry factor activity because CLDN7, which does not function as an entry factor, and CLDN1 containing mutations that impair its entry factor activity still copurify with these glycoproteins with similar affinities (data not shown).

It is also possible that OCLN does not function as a classical viral receptor that makes direct contact with incoming virions but, instead, indirectly mediates a process required for cell entry through interactions with other cellular factors. This scenario could actually be the case for CLDN1 as it has been shown that this protein is able to interact with CD81 (15, 24, 25, 28, 35, 52), and the complex of these proteins could be required for HCV cell entry (24, 25, 29). In fact, CLDN1 mutations that disrupt HCV cell entry activity also impair this interaction (24), and anti-CLDN1 antibodies that bind extracellular regions of CLDN1 appear to inhibit HCV infection by disrupting this interaction (29). Interestingly, this interaction does not occur within the tight junction (12, 35) and may take place in some polarized cells (24) but not others (28). Other groups have identified CLDN1 mutations that impair HCV cell entry functions in a manner that does not disrupt CD81 associations and instead may act by impairing associations between claudins within cell-cell contacts (15). While these findings do not rule out that both CD81 and CLDN1 interact with incoming HCV particles, they do substantiate the possibility that an interaction between cellular proteins may be required during this process.

Although this study presents the first detailed analysis of extracellular OCLN regions required for HCV cell entry, several additional lines of study may be useful for further defining determinants for this activity within this protein. In this study none of the species chimeras we created revealed species-specific differences outside the extracellular domains that were important for HCV cell entry activity. However, the intracellular carboxyl terminus of OCLN is responsible for mediating interactions with many intracellular proteins that influence the stability and functions of this and other tight junction proteins (41). If this domain is required for HCV cell entry, then at least its functions are conserved between the species analyzed above. We also intend to study how these OCLN cell entry determinants influence not only HCVpp entry but also infection with authentic HCV produced in cell culture (HCVcc) (30). Unfortunately, 786-O cells are not able to support intracellular HCV RNA replication, which is required to detect HCVcc infection (data not shown), and thus we could not use this system to address this question. That being said, no differences and many similarities have been documented in the cell entry pathways of HCVpp and HCVcc; thus, the former is a valid surrogate system to analyze HCV cell entry requirements. As the 786-O cells used in this study likely do not form tight junctions, it will also be interesting to test how the OCLN variants function in HCV entry in polarized cells that correctly form tight junction networks.

In summary, we have conducted a high-resolution analysis of the extracellular regions of OCLN required for HCV host cell entry. We found that EC2 and perhaps a subdomain of this region defined by a disulfide bond are critical for this process. Such information is critical to future studies aimed at defining how OCLN mediates HCV cell entry. OCLN is a major player in determining the species tropism of HCV infection (44). Our comparison of the sequences and activities of OCLN proteins from a range of species precisely defines the species-specific molecular requirements within OCLN for HCV cell entry. Such information will help elucidate HCV cell entry mechanisms and has the potential to contribute to the development of small-animal model systems capable of supporting HCV infection, as well as anti-HCV therapies targeting this stage of the viral life cycle.