HCV entry into host cells is most likely to occur through a multistep process. Current evidence suggests that scavenger receptor class B member I may define the attachment of HCV virolipoparticles to the cell surface, and this may prime particle interaction(s) with CD81 and Claudin co-receptors that are essential for subsequent particle internalization events (49
). In the present study we demonstrate a relationship between receptor-active CLDNs and their association and organization with CD81 at the plasma membrane. Given the reported role of Occludin in HCV entry, we investigated the stoichiometry and FRET between CLDN proteins and Occludin. The majority of CLDN proteins interacted with Occludin, and there was no discernable relationship between CLDN receptor activity and ability to associate with Occludin (). Mutation of residues 32 and 48 in CLDN1 EC1 ablated CD81 and Occludin association and HCV receptor activity (), which may reflect an altered EC1 conformation. Importantly, mutation of the same residues in CLDN7 enabled CD81 complex formation and virus entry without any detectable Occludin association, demonstrating an essential role of CLDN-CD81 receptor complexes in HCV infection.
Tagging CLDN proteins with AcGFP or DsRED enabled us to assess protein localization in the absence of CLDN-specific mAbs. The majority of AcGFP-tagged CLDNs localized to the plasma membrane, with only CLDN15 showing a dominant intracellular staining pattern. As viral receptor activity is dependent on cell surface expression (51
), the inability of CLDN4, -7, -11, -12, and -17 to mediate HCVpp entry is unlikely to be ascribed to low expression levels. The observation that only receptor-active CLDNs associate with CD81 () provides supporting evidence for a role for these protein complexes in the HCV entry process.
CLDNs oligomerize to form tight junction strands, and several reports demonstrate homotypic and heterotypic interactions between different members of the CLDN family (52
). FIR and FRET analysis demonstrated interactions between CLDN1 and CLDNs 4, 6, and 9 in 293T cells (). In contrast there was no significant association between CLDN1 and CLDN7 (r2
= 0.11) (). These data are supported by surface plasmon resonance data showing CLDN1 EC1-EC1 interaction in the absence of any significant association between CLDN1 and CLDN7 EC1 regions (). It was interesting to note that overexpression of non-receptor CLDNs with g.CLDN1 in 293T cells had minimal effects on CLDN1-CLDN1 association or HCV entry (data not shown). Mutation of residues 32 and 48 in EC1 of CLDN1 and CLDN7 had no significant effects on homotypic cis-interactions (), consistent with a recent report that residues in the EC2 of CLDN5 define homotypic cis-interactions (28
). We hypothesize that CLDN1 may associate with both CD81 and CLDN1 via interactions through its EC1 and EC2 loops, respectively, suggesting the presence of receptor heterodimers.
Anti-CD81 mAbs 2s20 and 2s66 reduced CLDN1-CD81 FIR and FRET values and yet had minimal impact on CD81-CD81 association (), reflecting potential differences in protein conformation or temporal association of CD81-CD81 and CLDN1-CD81 complexes. 2s66 FAb had a minimal effect on CLDN1-CD81 association, despite showing a comparable Kd for recombinant CD81 LEL to the complete IgG molecule (2s66 IgG Kd 1.7E−07; 2s66 FAb Kd 8.6E−06), suggesting that bivalent cross-linking of CD81 may be necessary to perturb CD81 association with CLDN1. It is interesting to note that 2s66 FAb demonstrated a 5-fold reduced capacity to neutralize HCV infection compared with 2s66 IgG, suggesting that antibody-induced disruption of CLDN1-CD81 complexes may contribute to the neutralizing activity of anti-CD81 antibodies.
To investigate whether CLDN association with CD81 is critical for viral receptor activity we investigated the effect of known receptor inactivating mutations in CLDN1 EC1 on protein association with CD81. We selected two previously identified amino acids (residues 32 and 48) in EC1, where the reciprocal interchange between CLDN1 and CLDN7 sequences abrogated receptor activity of CLDN1 and conferred receptor activity to CLDN7 by unknown mechanism(s) (14
). Neither mutation had an effect on CLDN1/CLDN7 expression or transport to the cell surface, consistent with a recent report by Cuikerman and colleagues demonstrating that alanine substitution at residue 32 ablated receptor activity with minimal perturbation of protein localization (32
). In contrast Liu and colleagues reported that CLDN1 I32M was not expressed at the cell surface (19
). These differences may reflect the use of different vectors to express CLDN1 in the various studies, because we and others have noted that high level expression of CLDNs can lead to their intracellular accumulation (30
). Importantly, we found that introduction of both mutations in CLDN1 destroyed lateral associations with CD81, whereas substitution of both residues in CLDN7 promoted heterotypic association with CD81 (A
). In contrast, mutation of both residues in CLDN1 or CLDN7 had a modest to negligible effect on homotypic cis-associations (C
), respectively, demonstrating a critical role for CLDN-CD81 complexes in HCV entry.
The major reservoir supporting HCV replication in vivo
is thought to be hepatocytes in the liver. Hepatocytes polarize with at least two basal surfaces facing the circulation and a branched network of grooves between adjacent cells that constitute the apical or bile canalicular surface (55
). Tight junction strands encircle the apical region and comprise multiple transmembrane, scaffolding, and signaling proteins (reviewed in Ref. 56
). We recently reported CLDN1 expression at basolateral and apical hepatocyte membranes in normal liver tissue (57
) and in polarized HepG2 cells, with an enrichment at tight junction-associated apical sites (33
). To extend our studies on CLDN1-CD81 association in 293T cells and to investigate the presence and location of receptor complexes in polarized HepG2 cells, we transduced HepG2 cells to stably express AcGFP- and DsRED-tagged versions of CLDN1 and CD81. Overexpression of either molecule had no effect on the polarization of HepG2 cells (data not shown). CLDN1 was found to associate with basolateral pools of CLDN1 and CD81 with reduced FRET values (36 and 25%, respectively) compared with 293T cells, most likely representing competition from endogenous CLDNs, as previously reported (30
). In contrast, CD81 was largely excluded from the tight junction and exhibited a minimal association with CLDN1 ().
HCV enters the liver via the sinusoidal blood, and therefore the virus will encounter receptors expressed on the sinusoidal or basal surface of the hepatocyte. Our current data, showing that CLDN1 receptor activity correlates with the formation of CLDN1-CD81 complexes that localize at the basolateral surface of polarized hepatoma cells, support a model where virus engagement of CD81-CLDN1 at the basal membrane may initiate the particle internalization process (33
). The role of Occludin in HCV entry into polarized cells is poorly defined. Our current data fail to demonstrate a role for CLDN-Occludin complexes in HCV infection. HCV internalizes via a clathrin-dependent process and fusion is believed to occur within the early endosomes (49
). At present it is unknown whether any of the viral receptors, including CD81 and CLDN1, are endocytosed with HCV, and further research on the trafficking and endocytic routing of receptor complexes and virus particles in polarized hepatocytes is required to fully appreciate the complex entry process of HCV in the liver.