|Home | About | Journals | Submit | Contact Us | Français|
Claudin-1, a component of tight junctions between liver hepatocytes, is a hepatitis C virus (HCV) late-stage entry cofactor. To investigate the structural and functional roles of various claudin-1 domains in HCV entry, we applied a mutagenesis strategy. Putative functional intracellular claudin-1 domains were not important. However, we identified seven novel residues in the first extracellular loop that are critical for entry of HCV isolates drawn from six different subtypes. Most of the critical residues belong to the highly conserved claudin motif W30-GLW51-C54-C64. Alanine substitutions of these residues did not impair claudin-1 cell surface expression or lateral protein interactions within the plasma membrane, including claudin-1-claudin-1 and claudin-1-CD81 interactions. However, these mutants no longer localized to cell-cell contacts. Based on our observations, we propose that cell-cell contacts formed by claudin-1 may generate specialized membrane domains that are amenable to HCV entry.
Hepatitis C virus (HCV) is a major human pathogen that affects approximately 3% of the global population, leading to cirrhosis and hepatocellular carcinoma in chronically infected individuals (5, 23, 42). Hepatocytes are the major target cells of HCV (11), and entry follows a complex cascade of interactions with several cellular factors (6, 8, 12, 17). Infectious viral particles are associated with lipoproteins and initially attach to target cells via glycosaminoglycans and the low-density lipoprotein receptor (1, 7, 31). These interactions are followed by direct binding of the E2 envelope glycoprotein to the scavenger receptor class B type I (SR-B1) and then to the CD81 tetraspanin (14, 15, 33, 36). Early studies showed that CD81 and SR-B1 were necessary but not sufficient for HCV entry, and claudin-1 was discovered to be a requisite HCV entry cofactor that appears to act at a very late stage of the process (18).
Claudin-1 is a member of the claudin protein family that participates in the formation of tight junctions between adjacent cells (25, 30, 37). Tight junctions regulate the paracellular transport of solutes, water, and ions and also generate apical-basal cell polarity (25, 37). In the liver, the apical surfaces of hepatocytes form bile canaliculi, whereas the basolateral surfaces face the underside of the endothelial layer that lines liver sinusoids. Claudin-1 is highly expressed in tight junctions formed by liver hepatocytes as well as on all hepatoma cell lines that are permissive to HCV entry (18, 24, 28). Importantly, nonhepatic cell lines that are engineered to express claudin-1 become permissive to HCV entry (18). Claudin-6 and -9 are two other members of the human claudin family that enable HCV entry into nonpermissive cells (28, 43).
The precise role of claudin-1 in HCV entry remains to be determined. A direct interaction between claudins and HCV particles or soluble E2 envelope glycoprotein has not been demonstrated (18; T. Dragic, unpublished data). It is possible that claudin-1 interacts with HCV entry receptors SR-B1 or CD81, thereby modulating their ability to bind to E2. Alternatively, claudin-1 may ferry the receptor-virus complex to fusion-permissive intracellular compartments. Recent studies show that claudin-1 colocalizes with the CD81 tetraspanin at the cell surface of permissive cell lines (22, 34, 41). With respect to nonpermissive cells, one group observed that claudin-1 was predominantly intracellular (41), whereas another reported associations of claudin-1 and CD81 at the cell surface, similar to what is observed in permissive cells (22).
Claudins comprise four transmembrane domains along with two extracellular loops and two cytoplasmic domains (19, 20, 25, 30, 37). The first extracellular loop (ECL1) participates in pore formation and influences paracellular charge selectivity (25, 37). It has been shown that the ECL1 of claudin-1 is required for HCV entry (18). All human claudins comprise a highly conserved motif, W30-GLW51-C54-C64, in the crown of ECL1 (25, 37). The exact function of this domain is unknown, and we hypothesized that it is important for HCV entry. The second extracellular loop is required for the holding function and oligomerization of the protein (25). Claudin-1 also comprises various signaling domains and a PDZ binding motif in the intracellular C terminus that binds ZO-1, another major component of tight junctions (30, 32, 37). We further hypothesized that some of these domains may play a role in HCV entry.
To understand the role of claudin-1 in HCV infection, we developed a mutagenesis strategy targeting the putative sites for internalization, glycosylation, palmitoylation, and phosphorylation. The functionality of these domains has been described by others (4, 16, 25, 35, 37, 40). We also mutagenized charged and bulky residues in ECL1, including all six residues within the highly conserved motif W30-GLW51-C54-C64. None of the intracellular domains were found to affect HCV entry. However, we identified seven residues in ECL1 that are critical for entry mediated by envelope glycoproteins derived from several HCV subtypes, including all six residues of the conserved motif. These mutants were still expressed at the cell surface and able to form lateral homophilic interactions within the plasma membrane as well as to engage in lateral interactions with CD81. In contrast, they no longer engaged in homophilic trans interactions at cell-cell contacts. We conclude that the highly conserved motif W30-GLW51-C54-C64 of claudin-1 is important for HCV entry into target cells and participates in the formation of cell-cell contacts.
The human kidney endothelial (HEK) cell line was obtained through the American Type Culture Collection. Huh-7 and Huh-7.5 cells were provided by C. Rice (Rockefeller University, NY) and H1H cells by R. Chowdhurry (Albert Einstein College of Medicine, Bronx, NY). Mutagenesis of the claudin-1-coding sequence was performed using QuikChange (Stratagene) according to the manufacturer's instructions, and the sequence was subcloned into the pQCXIN retroviral expression vector (Clontech) using NotI and BamHI restrictions enzyme sites. Vesicular stomatitis virus pseudoparticles (VSVpp) for the delivery of pQCXIN-claudin-1 were generated in HEK cells and used to transduce HEK or H1H cells, which were selected with G418 (1 mg/ml; Invitrogen) for 1 to 2 weeks, as previously described by us (28).
Pseudoparticles were made by cotransfecting the NLenv-luc+ vector with a vector encoding HCV, VSV, or envelope glycoproteins, as previously described by us (10, 14, 28, 29). HCV envelope glycoprotein sequences derived from subtypes 1a, 1b, 2b, 3a, 4, and 5 were also described by us (28). Viral titers were determined by infecting Huh-7 cells with the viral stocks and measuring luciferase activity in cell lysates at 48 h postinfection. Similar luciferase titers were then used to infect HEK or H1H cells stably expressing wild-type or mutant claudin-1. The percentage of entry of HCV pseudoparticles (HCVpp) was calculated as a percentage of entry into cells expressing the wild-type claudin-1 protein.
Proteins in the lysates of HEK or H1H cells (3 × 106) were separated by 14% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Invitrogen) and transferred to pure nitrocellulose membranes (Bio-Rad). These were probed with a JAY.8 rabbit polyclonal antibody against claudin-1 (1:500; Invitrogen) or a rabbit polyclonal antibody against β-tubulin (1:500; Santa Cruz Biotechnology). Staining was revealed with a horseradish peroxidase (HRP)-conjugated donkey anti-rabbit antibody (1:10,000; Amersham Biosciences) and developed using Western Lightning Chemiluminescence Reagent Plus (PerkinElmer).
HEK cells (8 × 106) expressing mutant or wild-type claudin-1 were biotinylated at 4°C for 30 min with EZ-Link Sulfo-NHS-LC-biotin (0.5 mg/ml; Pierce) according to the manufacturer's instructions. The reaction was quenched by washing with 10 mM glycine (Sigma) in phosphate-buffered saline. Radioimmunoprecipitation assay buffer (10 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.05% SDS, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 2 mM Na3VO4 and Pierce protease inhibitor cocktail) was used to lyse the cells, and biotinylated proteins were pulled down with streptavidin agarose resin (Pierce). Proteins were stripped off the resin by heating for 5 min at 95°C in radioimmunoprecipitation assay buffer, and Western blotting of proteins was performed as described in the previous section. Golgin-97 was detected with a rabbit polyclonal antibody (1:50; Abcam), followed by HRP-conjugated donkey anti-rabbit antibody (1:10,000; Amersham Biosciences).
The biomolecular fluorescence complementation (BiFC) assay was adapted from that described by Yang et al. (41). Briefly, sequences encoding either the C- or N-terminal moiety of the Venus fluorescent protein (termed vc and vn, respectively) were appended to the 3′ ends of wild-type or mutant claudin-1 or to CD81-coding sequences. Plasmids expressing complementary chimeric proteins were transiently transfected in a 1:1 ratio into HEK cells (3 × 106) using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. After 48 h, cells were harvested and analyzed by flow cytometry (fluorescein isothiocyanate, FACSCalibur). Background fluorescence was determined by cotransfecting wild-type claudin-1 appended with vn or CD81 appended with vc and pQCXIN vectors encoding only vc or vn, respectively.
HEK cells stably expressing wild-type or mutant claudin-1 or Huh-7 cells were incubated on Lab-Tek chambers (Fisher) at 70% confluence, fixed in 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, treated with NaBH4 (1 mg/ml), and blocked with 0.1% Tween-2% bovine serum albumin-2% normal donkey serum. Cells were costained for detection of claudin-1 or mutants using a JAY.8 rabbit polyclonal antibody (1:100; Invitrogen) and for ZO-1 using a ZO-1-1A12 monoclonal mouse antibody (1:100; Invitrogen). As secondary antibodies we used Cy3-conjugated donkey anti-rabbit immunoglobulin G (IgG) or Cy2-conjugated donkey anti-mouse IgG antibody (1:500; Jackson Immunoresearch). Nuclei were stained with Hoechst 33258 (Invitrogen). Images were generated using a Leica AOBS laser scanning confocal microscope (63× objective). Claudin-1 expression in Huh-7 cells was analyzed using the same protocol.
Recently identified as a cofactor for HCV replication, claudin-1 is known to act at a late stage of entry, but its exact function remains unclear. To better understand the role of claudin-1 in HCV entry, we performed alanine scanning mutagenesis of single residues predicted to be involved in various putative functions (4, 16, 25, 35, 37, 40) (Fig. (Fig.1).1). Alanines were introduced in place of C-terminal domain residues belonging to an internalization motif (Y193A, P194A, Y199A, or P200A) or a phosphorylation site (T195A). Cysteines were alanine substituted to study the role of palmitoylation sites (C104, C107, C183A, C184A, or C186A). We then generated stable HEK cell lines expressing the various claudin mutants and tested their ability to be infected by HCVpp bearing the envelope glycoproteins of the H77 (1a) isolate, as previously described by us (10, 14, 28). None of the mutations described in this section significantly affected HCVpp entry into HEK cells (Fig. (Fig.2A).2A). We confirmed these results using the human hepatoma cell line H1H, which expresses very low levels of endogenous claudin-1 and is therefore not permissive to HCV entry (Fig. (Fig.2B).2B). Based on our observations, we concluded that intracellular functional domains of claudin-1 are not important for HCV entry.
Next we explored the role of residues in ECL1 of claudin-1, with special emphasis on its N-terminal half, which was previously shown to be important for HCV entry (Fig. (Fig.1)1) (18). We introduced single alanine substitutions at positions occupied by bulky, charged, or polar residues, including R31A, I32A, Y33A, Y35A, D38A, N39A, T42A, Q44A, A45G, M52A, S53A, K65A, and N72A, a potential glycosylation site (Fig. (Fig.1).1). We also mutated each of the residues in the highly conserved motif W30-GLW51-C54-C64 (Fig. (Fig.1).1). HEK cells stably expressing wild-type or mutant claudin-1 were infected with HCVpp bearing the envelope glycoproteins of isolate H77 (1a). We confirmed that the previously described residue I32 is important for entry (18) and also showed that the proximal D38 residue was equally important (Fig. (Fig.3A).3A). The following ECL1 mutants, however, had no significant effect on HCVpp entry: R31A, Y33A, Y35A, N39A, T42A, Q44A, A45G, M52A, S53A, K65A, and N72A (Fig. (Fig.3).3). In contrast, all six residues within the conserved motif, including W30-GLW51-C54-C64, were critical, because their alanine substitutions enabled only ≤15% of the HCVpp entry observed for the wild-type protein (Fig. (Fig.3B).3B). We also tested the ability of human T-cell leukemia virus type 1 pseudoparticles and VSVpp to enter the various HEK-claudin-1 derivatives and found no effect of the claudin-1 mutants, indicating that the observed effects were specific for HCVpp (data not shown).
In order to determine whether ECL1 residues that were found to impair HCVpp 1a entry were also important for entry mediated by the envelope glycoproteins of other HCV subtypes, we extended our studies to include HCVpp generated with the envelope glycoproteins from patient isolates previously described by us (28). We found that the residues that we identified as being important for entry of the H77 (1a) isolate were also important for entry of isolates belonging to subtypes 1b, 2b, 3a, 4, and 5 (Fig. (Fig.4).4). There were some differences in the degree to which entry mediated by the various envelope glycoproteins was affected, such that each isolate represented a unique signature pattern of residue usage. For example, HCVpp 1b entry was most severely suppressed by substitutions at positions 30, 32, 50, 54, and 64; HCVpp 2b entry was most sensitive to substitutions at positions 30, 49, 50, and 64; and HCVpp 3a entry was most sensitive to the substitution at position 64 (Fig. (Fig.4).4). Together, our results confirm the importance of seven ECL1 residues in entry of six HCV subtypes.
We performed experiments to ascertain expression, transport, and folding of the various claudin-1 mutants no longer capable of enabling HCVpp entry. Initially, lysates of HEK cells transformed and selected for expression of wild-type or mutant claudin-1 were probed by Western blotting using a rabbit anti-claudin-1 antibody specific for the C-terminal domain of claudin-1 (Fig. (Fig.5A,5A, top). The various HEK cell populations were found to express similar levels of the 22-kDa band corresponding to wild-type or mutant claudin-1 protein. Equal amounts of lysates were deposited in the wells as shown by the consistency of the beta-tubulin control (Fig. (Fig.5A,5A, bottom). Note that expression of claudin-1 in transduced HEK cells is similar to its endogenous expression levels in Huh-7 and Huh-7.5 hepatoma cells as well as primary human hepatocytes (Fig. (Fig.5B5B).
We then performed experiments to establish whether mutant claudin-1 proteins were transported to the cell surface, like the wild-type protein. Because the only commercially available anti-claudin-1 antibody is directed against an epitope in the intracellular C terminus, detection of claudin-1 could not be realized by flow cytometry. Consequently, we used a cell surface biotinylation approach to purify and detect plasma membrane-associated claudin-1. Briefly, whole cells were treated with NHS-biotin and biotinylated proteins pulled out with streptavidin-conjugated agarose. After elution and Western blotting with the anti-claudin-1 antibody, we showed that all of the mutants that were impaired for HCV entry were transported to the cell surface, similar to the case for the wild-type protein (Fig. (Fig.5C).5C). To ascertain that only plasma membrane-associated proteins were biotinylated and detected, we also tested for the presence of an intracellular protein, golgin-97, which is an integral membrane protein in the Golgi apparatus. Only trace levels of golgin-97 were detected in the biotinylated fractions, indicating that the majority of biotinylated claudin-1 is from the cell surface (Fig. (Fig.5D).5D). Overall, our data show robust expression of entry-impaired claudin-1 mutants, similar to that of the wild-type protein. The lack of function in HCVpp entry of our ECL1 mutants is not due to changes in their expression levels or their ability to reach the cell surface.
We used a BiFC assay to determine whether the claudin-1 mutations that impair HCVpp entry affect claudin-1 lateral interactions with itself or with CD81. The C termini of wild-type or mutant claudin-1 were appended with Venus fluorescent protein moieties, either vn or vc. Similarly tagged versions of CD81 were also generated. In this assay, vn and vc are not able to fluoresce separately. However, if two proteins each bearing one of the moieties interact and bring the C and N termini within ≤15 nm of each other, a detectable fluorescence signal is generated by complementation. Using flow cytometry to detect Venus fluorescence, we showed that wild-type claudin-1 proteins interact with each other, as do the mutants that were shown to be impaired for HCV entry (Fig. (Fig.6).6). Though there is a slight decrease in BiFC generated by the claudin-1 mutants compared to the wild type, it is not statistically significant (P = 0.6124 for I32 and P = 0.5981 for W-GaW-C-C in an unpaired, two-tailed t test). Moreover, we observed that mutations in the conserved domain do not affect lateral interactions between claudin-1 and CD81 as there was no significant difference between BiFC generated by coexpression of CD81 with wild-type or mutant claudin-1 proteins (P = 0.4437 for I32 and P = 0.5981 for W-GaW-C-C in an unpaired, two-tailed t test) (Fig. (Fig.66).
Six of the seven residues that we identified as being critical for HCVpp entry belong to the highly conserved motif W30-GLW51-C54-C64. Because this motif is conserved in all claudins, we hypothesized that it is involved in the formation of cell-cell contacts. We explored claudin-1 trans interactions by studying the ability of wild-type and mutant proteins to colocalize with another component of tight junctions, ZO-1, at cell-cell contacts. Based on immunofluorescence analyses, we showed that wild-type claudin-1 as well as a mutant that does not affect HCVpp entry localized at cell-cell contacts and colocalized with ZO-1 (Fig. 7A and B). In contrast, mutants of the conserved motif that impaired HCVpp entry were not detected at cell-cell contacts, nor did they colocalize with ZO-1 (Fig. 7C and D and data not shown). Note that claudin-1 tends to accumulate intracellularly in transduced HEK cells as well as Huh-7 hepatoma cells (data not shown). We conclude that the highly conserved motif W30-GLW51-C54-C64 is required for the formation of cell-cell contacts as well as for HCV entry.
CD81 and SR-B1 initially entered the HCV field as envelope glycoprotein E2-binding proteins (14, 15, 33, 36). With the development of the HCVpp and HCV cell culture assays, it was demonstrated that these cell surface molecules are HCV entry receptors (9, 14, 21, 26, 27, 38). At that time it also became clear that SR-B1 and CD81 were necessary but insufficient to mediate HCV entry into target cells. Using a functional cloning approach, Evans et al. identified the tight junction component claudin-1 as yet another, indispensable HCV entry cofactor (18). Claudin-1 is expressed by all HCV-permissive cells, and its expression in CD81+ and SR-B1+ cells renders them permissive to the virus. Follow-up studies by us and others demonstrated that claudin-6 and -9, but not -2, -3, -4, -7, -11, -12, -15, -17, and -23, could also fulfill the role of HCV entry cofactor (28, 43).
In their study, Evans et al. (18) analyzed chimeras of claudin-1 and the closely related claudin-7, observing that the intracellular domains of claudin-1 and claudin-7 were interchangeable vis-a-vis HCV entry. However, this did not address whether functions associated with these domains were in fact required. Using an alanine scanning mutagenesis strategy, we investigated the necessity for various putative functional domains of claudin-1 in HCV entry, including motifs for internalization, glycosylation, palmitoylation, and phosphorylation. Here, we report that none of these putative functions of claudin-1 play a role in HCV entry, a surprising finding in light of the observation that claudin-1 intervenes late in the process of viral entry. Furthermore, our observations suggested that the cofactor function of claudin-1 was strictly associated with cellular functions fulfilled by its extracellular domain.
The requirement for specific residues in ECL1 was therefore explored. In addition to previously identified residues I32 and E48, we found seven amino acids that are critical for entry of six HCV isolates belonging to subtype 1a, 1b, 2b, 3a, 4, or 5. Six of the residues belong to the highly conserved motif (W30-GLW51-C54-C64) (25, 37), and one (D38) is external to this motif. We confirmed that the lack of function of our mutants was not due to a defect in expression or protein transport to the cell surface. Residues I32, D38, and E48 may be involved in the passage of cations across tight junctions (2, 3, 13), but this function of claudin-1 has not yet been directly linked to HCV entry. The role of I32, D38, and E48 therefore remains to be explored, and these residues may determine HCV specificity for claudin-1.
In this study, we focused on exploring the function of the residues in the highly conserved, hydrophobic motif W30-GLW51-C54-C64, which is found in all 24 members of the human claudin family. The two cysteines included in this motif, in positions 54 and 64 in claudin-1, probably do not form a disulfide bond, since we did not observe a size difference between reduced and nonreduced protein by Western blotting or between alanine mutants and wild-type claudin-1 (T. Dragic, unpublished data). However, the two cysteines have been described to be important for tightness function as measured by transepithelial electrical resistance (25, 39). This led us to speculate that the conserved motif is involved in the formation of cell-cell contacts.
The ability of W30-GLW51-C54-C64 mutants to polymerize via lateral interactions (cis interaction) within the same membrane and to participate in head-to-head interactions (trans interaction) for adhesion of adjacent membranes was explored. Using a BiFC assay, we showed that motif mutations do not affect claudin-1-claudin-1 or claudin-1-CD81 lateral interactions within the membrane. On the other hand, we observed by confocal microscopy that mutants within the conserved motif were not able to form homophilic trans interactions since they did not localize at cell-cell contacts along with ZO-1. These data suggested that residues important for HCV entry do not affect cis interactions but do affect trans interactions required for cell-cell adhesion. Thus, cell-cell contacts formed by claudin-1 may represent specialized membrane domains required for HCV entry. Formation of these domains involves residues in the highly conserved motif W30-GLW51-C54-C64.
Evans et al. (18) showed that claudin-1 intervenes at a very late stage of HCV entry. We hypothesize that claudin-1 interacts either with the HCV envelope glycoproteins or with one of the other HCV entry cofactors. To date, there are no reports of claudin-1-HCV envelope glycoprotein interactions, either because claudin-1 interacts with E1, for which there is currently no binding assay, or because it interacts with a conformation of E2 that is not reproduced by available models, including soluble E2, HCVpp, or HCV cell culture. Based on our data, we propose that the formation of cell-cell contacts by residues in the highly conserved motif W30-GLW51-C54-C64 is necessary for the recruitment of the HCV-receptor complex to membrane domains that are amenable to its internalization and trafficking into fusion-permissive intracellular compartments.
We thank Tianyi Wang for generously providing us with constructs encoding the N- and C-terminal moieties of Venus fluorescence protein.
This work was supported by NIH grant AI060390 and the Burroughs Wellcome Fund, Investigators in Pathogenesis of Infectious Diseases. This work was also supported in part by NIAID Centers for AIDS Research grant AI051519 to Albert Einstein College of Medicine.
Published ahead of print on 18 March 2009.