We applied a bio-informatic modelling-based mutagenesis approach to characterize the CD81/claudin-1 extracellular loop interactions essential for HCV infection. Our in silico
model identified interacting residues between CD81 and claudin-1 and in vitro
mutagenesis confirmed their role in HCV entry. We predict that the first extracellular loop of claudin-1 is relatively unstructured, with a high beta turn composition and two motifs, incorporating amino acids 33–35 and 63–66, that interact with CD81. Previous reports have shown that mutation of claudin-1 ECL1 residues 32 and 48 ablated CD81 association and virus entry (Harris et al
), suggesting a direct role in receptor complex formation. However, in silico
modelling of these variants showed the formation of an α-helical region, leading to a marked change in loop orientation, providing an alternative explanation for their inability to associate with CD81 and failure to support HCV infection.
To validate our in silico
model we studied a panel of claudin-1 mutants, previously reported to modulate HCV entry (Cukierman et al
). We demonstrate a significant association between claudin-1 viral receptor activity and interaction with CD81 in silico
and in vitro
. Three claudin-1 mutants (L50A, C54A and C64A – group III) were predicted to interact with alternative CD81 residues L162, I182, N184 and F186. All three variants displayed significant association with CD81 in vitro
by FRET analysis and yet failed to support HCV entry, demonstrating that this altered orientation of the two proteins is receptor inactive. Of note, all of these residues were previously reported to be involved in binding HCV E2, suggesting that the second extracellular loop of CD81 needs to interact with both HCV E2 and claudin-1 to confer virus entry (Drummer et al
In silico modelling of claudin-7 ECL1 demonstrated minimal contact with CD81 (E), with the loop projecting outwards in a similar fashion to the group I claudin-1 mutants. This projection fills the space that would be occupied by the α-helix containing CD81 residues T149, E152 and T153. Introduction of M32I/K48E changes into claudin-7 (F) re-orientates the loop and allows claudin-7 C64 to interact with CD81 E152 in a similar way to that seen for WT claudin-1.
Mutagenesis studies have shown the importance of the highly conserved claudin motif, W30–GLW51–C54–C64 in HCV entry (Cukierman et al
). Our data provide a structural rationale for these observations. Mutation of W30, G49 or W51 induced substantial changes in the local conformation and orientation of the ECL1 motifs predicted to interface with CD81, such that group I claudin-1 mutants disrupt the interaction of region 63–66 with CD81 (). In contrast, mutation of claudin-1 residues L50, C54 or C64 resulted in more modest changes to ECL1 orientation, leading to an interaction with alternative CD81 residues.
In forming tight junctions, claudins form homodimers between identical claudin molecules and heterodimers with different claudin family members. The intermolecular interface of claudin-5 has been shown to include both aromatic (F147, Y148, Y158) and hydrophilic (Q156, E159) residues in ECL2 (Piontek et al
). Given the conserved nature of these residues across the claudin family, our model is consistent with distinct roles for the extracellular loops in dimerizing with other claudins and associating with tetraspanins (Kovalenko et al
). Our model predicts that a claudin-1 protein is capable of simultaneous association with other claudin proteins and CD81.
We confirmed that mutation of CD81 residues T149, E152 and T153 to alanine reduced claudin-1 association and HCVpp entry. In contrast, mutation of residue K148A had no effect on CD81 interaction with claudin-1 in vitro
and supported HCVpp infection. Importantly, a panel of conformation-dependent anti-CD81 antibodies that neutralize HCV infection (Farquhar et al
) showed comparable binding to all mutants. These data, together with the observation that HCV E2 protein bound mutant and parental CD81 with comparable values, suggest that the mutations have minimal effect on CD81 conformation. However, a CD81 variant bearing the double mutation K148A/T149A was receptor active, suggesting that K148A suppresses the receptor inactive T149A mutant. Our model predicts that CD81 K148 interacts with claudin-1 Y33, whereas T149 interacts with claudin-1 motif 63–66. Importantly, the K148A/T149A double mutation reinstates the interaction of CD81 residue E152 with claudin-1 Q63. Although murine CD81 differs from human CD81 at 17 of the 87 residues of the ECL2 domain, the region identified here is conserved, leading us to speculate that the functions defined by this motif are similar in both organisms (Flint et al
). These experimental results validate our structural predictions and highlight claudin-1 region 63–66 interaction with T149 and E152 of CD81 as the site of primary molecular association. In contrast, the 33–35 region of claudin-1 appears to determine the orientation and packing of the claudin-1 and CD81 extracellular loops.
Importantly, all of the CD81 mutants studied (K148, T149, E152 and T153) bound HCV E2, consistent with a report by Drummer and colleagues that residues outside this region form the HCV E2 binding site (Drummer et al
). It is interesting to note that these residues, which were identified by in silico
modelling and verified by site-directed mutagenesis, have previously been shown to have a role in the association between the ECL1 and ECL2 domains of CD81, suggesting that this region may have a role in both homotypic and heterotypic interactions of CD81 (Yalaoui et al
). Our data suggest that distinct regions of CD81 ECL2 engage HCV glycoproteins and claudin-1, highlighting the potential to form a ternary HCV/CD81/claudin-1 complex. Although claudin-1 does not appear to modulate the ability of CD81 to bind HCV E2 (data not shown), claudin-1 limits CD81 lateral movement at the plasma membrane and promotes receptor endocytosis (Farquhar et al
; Harris et al
), suggesting a role for the CD81–claudin-1 complex in virus internalization and fusion with early endosomes. Our data uncover a novel role for CD81 residues T149, E152 and T153 in HCV entry independent of viral glycoprotein-receptor interaction, and substantiate the key role of CD81–claudin-1 complexes in HCV internalization. Our model of the CD81/claudin-1 interface provides a new conserved target for anti-viral drug design and will allow the rational design of small molecule and peptide mimetics targeting the receptor complex.