4th European Conference on Tetraspanin

Structural characterization of CD81–Claudin-1 hepatitis C virus receptor complexes

Nicklas Bonander, Mohammed Jamshad, Ke Hu, Michelle J. Farquhar, Zania Stamataki, Peter Balfe, Jane A. McKeating, Roslyn M. Bill


Tetraspanins are thought to exert their biological function(s) by co-ordinating the lateral movement and trafficking of associated molecules into tetraspanin-enriched microdomains. A second four-TM (transmembrane) domain protein family, the Claudin superfamily, is the major structural component of cellular TJs (tight junctions). Although the Claudin family displays low sequence homology and appears to be evolutionarily distinct from the tetraspanins, CD81 and Claudin-1 are critical molecules defining HCV (hepatitis C virus) entry; we recently demonstrated that CD81–Claudin-1 complexes have an essential role in this process. To understand the molecular basis of CD81–Claudin-1 complex formation, we produced and purified milligram quantities of full-length CD81 and Claudin-1, alone and in complex, in both detergent and lipid contexts. Structural characterization of these purified proteins will allow us to define the mechanism(s) underlying virus–cell interactions and aid the design of therapeutic agents targeting early steps in the viral life cycle.

  • CD81
  • Claudin-1
  • hepatitis C virus (HCV)
  • oligomerization
  • tetraspanin


Future advances in understanding the biology of the cell will rely on our increasing knowledge of protein behaviour and the complex interplay of proteins with other biomolecules. The tetraspanin family of four-TM (transmembrane) domain proteins is involved in multiple biological functions, including cell proliferation and cell–cell adhesion, as well as being key players in cancer, the immune system, fertilization and several infectious diseases [1]. However, the biochemical function of tetraspanins is not well understood. They are reported to co-ordinate the lateral movement and trafficking of molecules into TEMs (tetraspanin-enriched microdomains), which are rich in cholesterol. Emerging evidence also reveals tetraspanin association with signalling lipids, cytoplasmic proteins [2] and adhesion receptors of the integrin family in TEMs. In the specific case of the tetraspanin CD81, it has been reported to have a role in B- and T- cell activation [2,3] and interacts with Ig superfamily members, CD4, CD8, CD19 and EWI proteins [4]. It also has a role in HCV (hepatitis C virus) and Plasmodium falciparum infection of the liver [5,6].

The Claudin superfamily of four-TM domain proteins comprises the junctional strands that form cellular TJs (tight junctions). Although the family displays low sequence homology and appears to be evolutionarily distinct from the tetraspanins, the tetraspanin CD81 and the TJ protein Claudin-1 (with 15.2% identity and 38.3% similarity) are critical molecules defining HCV entry [79]. We recently reported CD81–CD81 homodimers and CD81–Claudin-1 heterodimers at the plasma membrane of hepatoma cells [10]. Perturbation of CD81–Claudin-1 complexes inhibited HCV entry, supporting their role in the viral entry process and raising questions about protein–protein interactions in TEM formation and receptor-mediated infectivity [11].

The structural characteristics of CD81 and Claudin-1

A total of 33 members of the tetraspanin family have been identified in humans. Common to all of them are the four TM regions containing several conserved polar residues and, in most cases, relatively short cytoplasmic N- and C-terminal tails [12]. These TM domains are linked via a short EC1 (extracellular loop 1) and a larger EC2 (extracellular loop 2) containing the highly conserved CCG (Cys-Cys-Gly) motif unique to tetraspanins. Between four and eight cysteine residues can be found in the EC2 loop. These residues form several disulfide bridges [13], which are critical for correct folding of the EC2 domain and subsequent interaction with ligands and partner proteins [2,12]. In human CD81, there are four such residues. The characteristic structural features of tetraspanins are shown in Figure 1.

Figure 1 Cartoon of Claudin-1 and CD81 illustrating their basic topologies

The first and second extracellular loops (EC1 and EC2) are shown.

To date, because of the lack of appropriate tools, the oligomeric status of tetraspanins within TEMs is largely unknown and the relationship between protein conformation and biological activity is poorly defined: the only crystal structure of any tetraspanin is that of the soluble EC2 domain of human CD81 [13], which shows a mushroom-shaped loop confirming the presence of the highly conserved CCG motif and two intact disulfide bridges as well as a potential hydrophobic interface for dimerization. The 6 Å (1 Å=0.1 nm) cryo-electron microscopy structure of a naturally occurring uroplakin tetraspanin 2D (two-dimensional) array (or ‘urothelial plaque’) revealed a rod-shaped structural morphology consisting of four TM helical bundles bound to a single TM helix partner [14]. However, the uroplakin tetraspanins are atypical members of the superfamily on account of their natural assembly into hexagonally packed 2D crystals and their highly specific interactions with a limited number of partner ligands. Moreover, differences between the proposed uroplakin dimer and EC1–EC2 contacts with predicted models of CD81 [15], CD82 [16] and CD9 [17] exist, and structural information is lacking about the critical intracellular C-terminal domain that plays key roles in the interaction with a variety of effector proteins [12,16]. Elucidation of the full-length CD81 structure is therefore necessary to increase our understanding of the role of the TM domains in specific interactions with biologically relevant ligands.

Although Claudin-1 has a similar proposed topology to CD81, it is not a member of the tetraspanin family, since it lacks sequence homology and key structural features, including the characteristic disulfide bonds and a CCG motif in EC2 (Figure 1). Although it is well established that Claudin polymerization is critical for establishing membranous strands and TJs [18], the molecular structure of TJ subunits or their organization principle is unknown. Claudins are thought to associate in the plasma membrane of a single cell and between opposing cells via interactions between the extracellular loops (reviewed in [19]). We recently reported Claudin-1 dimerization based on FRET (Förster resonance energy transfer) between tagged molecules, suggesting that dimers are the primary building block(s) of TJ strands.

CD81 and Claudin-1 as co-receptors for HCV entry

HCV is an enveloped positive-stranded RNA virus of the Flaviviridae family that primarily infects the liver. Advances in the development of infectious retroviral pseudo-particles bearing HCV glycoproteins and an infectious system generating native virus particles in cell culture have allowed studies on HCV entry and replication [20]. HCV encodes two glycoproteins, E1 and E2, and both are required to initiate infection via a pH- and clathrin-dependent pathway. Current evidence suggests that four host cell molecules are important for HCV entry: scavenger receptor class B member I, CD81, Claudin-1 and Occludin, an additional member of the TJ protein family [21]. Thus the entry pathway of HCV is complex and likely to involve a cascade of events culminating in particle internalization.

A proportion of Claudin-1 in hepatocytes [10] and polarized hepatoma cells localizes outside TJs [22]. Non-junctional pools of Claudin-1 associate with CD81 in TEMs, supporting a function for Claudin-1 beyond its role in TJs [23]. Further studies demonstrate that anti-CD81 antibodies co-precipitate Claudin-1 [24] and FRET occurs between tagged CD81 and Claudin-1, demonstrating homotypic (CD81–CD81, Claudin-1–Claudin-1) and heterotypic (Claudin-1–CD81) interactions [10]. The tagged proteins maintain viral receptor activity, and treatment of hepatoma cells with agents that inhibit HCV entry reduce the frequency of CD81–Claudin-1 FRET complexes, supporting their essential role in the viral life cycle [10].

Our results show that HCV E2 interacts with soluble dimeric forms of recombinant CD81 EC2 and fails to associate with monomeric versions of the protein, suggesting that CD81 dimers are the minimal oligomeric form of the receptor. Soluble CD81 EC2 and anti-CD81 antibodies neutralize infectivity after viral attachment to the target cell, consistent with CD81 acting as a co-factor or internalization receptor. Although Claudin-1 has also been identified to have an essential role in late stage(s) of the HCV cell entry process [25], cell-based experiments have failed to detect an interaction between Claudin-1 and HCV glycoproteins, which may reflect a requirement for the virus to bind its receptors in a defined sequence or the relatively low sensitivity of cell-based methods.

Studies with full-length recombinant CD81 and Claudin-1

In order to address oligomerization of CD81–Claudin-1 complexes, we have focused on purifying the full-length proteins from recombinant sources. Both have been difficult to overproduce in a purified form for detailed biophysical analyses, which is typical for the majority of human membrane proteins. We have reported the recombinant production of milligram quantities of human CD81 in the methylotrophic yeast Pichia pastoris [26]. This yeast species allows ease of cultivation, optimization of production and access to the full complement of higher eukaryote-like post-translational modifications. This is in contrast with the widely used prokaryotic microbe Escherichia coli, in which accumulation of the recombinant protein in inclusion bodies requires refolding and can lead to much lower levels of a fully functional recombinant protein.

Monomers, dimers and higher oligomers of CD81 were observed in recombinant P. pastoris membranes comparable with the forms seen in mammalian membranes. Immuno-fluorescent and flow cytometric staining of P. pastoris protoplasts with mAbs (monoclonal antibodies) specific for CD81 EC2 confirmed the antigenicity of the recombinant molecule. Full-length CD81 was solubilized with an array of detergents and subsequently characterized using CD and analytical ultracentrifugation. These biophysical techniques confirmed that the protein can be isolated in monomeric and oligomeric forms in a detergent-dependent manner. The monomeric form isolated using n-octyl-β-D-glucopyranoside interacted with HCV E2 glycoprotein and had a highly defined α-helical secondary structure (77.1%). This prediction from CD data fits remarkably well with what would be expected (75.2%) from knowledge of the protein sequence together with the data from the crystal structure of the second extracellular loop. This study represents the first biophysical characterization of a full-length recombinant tetraspanin, and opens the way for structure–activity analyses of this ubiquitous family of four-TM domain proteins [26].

No crystallographic analysis of full-length CD81 has yet been performed due to the difficulties in obtaining sufficient amounts of stable, active monodispersed protein; a problem that besets the membrane protein field as a whole. There is also significant evolutionary divergence in the protein sequences of tetraspanin soluble domains, making extrapolation to other family members difficult. Recent studies suggest further important differences in the interaction of HCV with full-length cell-expressed forms of CD81 and recombinant forms of CD81 EC2: some amino acid mutations exert a phenotype in the context of EC2, but not in the full-length molecule [27]. Notably, Drummer et al. [28] identified mutations that ablate CD81 dimerization within the context of soluble recombinant EC2. These were suggested to be important for inter-monomer contacts (F150S and V146E), salt bridge formation (K124T) and intra-monomer disulfide bonding (T166I, C157S and C190R). Two monomeric mutants retained the ability to bind HCV E2, K124T and V146E, whereas F150S, T166I, C157S and C190R did not. However, K124T, V146E and F150S mutants in full-length cell-expressed CD81 had minimal effects on CD81 oligomerization and HCV E2 binding. These results suggest that the EC2 has a more robust structure in the full-length tetraspanin, with regions in EC1 and the TM regions playing a role in CD81 dimerization [29]. Hence, analysis of interactions between full-length CD81 and HCV glycoproteins are required to identify the critical amino acids defining this interaction [25].

The production of high yields of Claudin-1 has been substantially more challenging than that of CD81, but we are now also able to purify milligram quantities of this protein. Recombinant Claudin-1 can also be isolated in monomeric and oligomeric forms, and we have recently examined its complexation with recombinant CD81. Earlier studies identified the importance of residues Ile32 and Glu48 in Claudin-1 EC1 for HCV receptor activity [25] and recent results highlight their role in associating with CD81 [23]. In contrast, Ile182, Asn184 and Phe186 in CD81 EC2 were identified as critical for viral receptor activity; however, their effect(s) on CD81 structure or association with Claudin-1 is unknown. Macromolecular docking simulations of the predicted interactions between our homology model of Claudin-1 and the crystallized CD81 EC2 structure have allowed us to model the protein interface. This will provide a solid framework for our structure–activity studies on the full-length proteins.

Membrane protein conformation and activity is often regulated by lipid composition, especially cholesterol, as seen for the serotonin transporter [30], the GABA (γ-aminobutyric acid) transporter [31] and Na+/K+-ATPase [32]. We have reconstituted CD81 into proteoliposomes alone and with Claudin-1. These experiments are allowing us to identify the precise structural determinants of receptor oligomerization and activity. Fusion of P. pastoris protoplasts expressing an irrelevant protein or CD81 and Claudin-1 enabled us to analyse the effect(s) of co-expression on virus association. Protoplasts bound currently available antibodies; however, cells expressing either CD81 or Claudin-1 alone did not bind HCV at greater than background levels. The efficiency of polyethylene-mediated fusion was modest and yet protoplasts co-expressing CD81 and Claudin-1 (9% were CD81+/Claudin-1+) bound virus particles, suggesting a role for CD81–Claudin-1 complexes in associating with HCV particles.

All currently available mAbs to CD81 recognize the same epitope in the hypervariable region of EC2. To investigate the role of EC1 and epitopes outside the hypervariable EC2 region in HCV entry, we studied the immunogenicity of full-length CD81 in detergent micelles. We successfully cloned 32 mAbs mapping to a series of novel epitopes in EC2 and non-EC1/EC2 regions, many of which neutralize HCV infectivity. Several mAbs discriminate between monomeric and oligomeric forms of CD81 EC2 and provide a panel of reagents to study tetraspanin conformation. These data highlight the immunogenicity of full-length CD81 and its ability to elicit antibodies that recognize novel epitopes expressed on the native molecule. We are now in a unique position to capitalize on these advances to investigate the biophysical and biological properties of these important proteins and make significant contributions towards understanding their interactions.


The structural and functional characterization of two-membrane proteins each with four TM domains, CD81 and Claudin-1, will enable the mechanism underlying their protein–protein interactions to be elucidated. The oligomerization status of these two co-receptor components is critical to HCV infectivity. We have shown that CD81 oligomerizes at the plasma membrane, with the identification of CD81–CD81 homodimers and CD81–Claudin-1 heterodimers suggesting a role for non-TJ pools of Claudin-1 in regulating tetraspanin CD81 function. Perturbation of CD81–Claudin-1 complexes inhibits HCV entry, suggesting a critical role in the viral entry process. Furthermore, we have recently identified amino acid residues in the first extracellular loop (EC1) of Claudin-1 that define association with CD81 and viral co-receptor activity. Biophysical and structural characterization will identify the oligomeric status and conformation of the proteins used by HCV, and ultimately aid the design of therapeutic agents targeting viral infection.


This work was supported by the European Commission [via contract LSHG-CT-2006–037793 (OptiCryst)], the Biotechnology and Biological Sciences Research Council (via a Research Equipment Initiative grant to R.M.B.), and by Wellcome Trust and Medical Research Council grants to J.A.M.


  • 4th European Conference on Tetraspanins: An Independent Meeting held at the University of Birmingham, Birmingham, U.K., 8–10 September 2010. Organized by Fedor Berditchevski (Birmingham, U.K.), Jane McKeating (Birmingham, U.K.), Peter Monk (Sheffield, U.K.), Lynda Partridge (Sheffield, U.K.), Mike Tomlinson (Birmingham, U.K.) and Annemiek van Spriel (University Medical Centre, Nijmegen, The Netherlands). Edited by Mike Tomlinson (Birmingham, U.K.).

Abbreviations: 2D, two-dimensional; EC1, extracellular loop 1; EC2, extracellular loop 2; FRET, Förster resonance energy transfer; HCV, hepatitis C virus; mAb, monoclonal antibody; TEM, tetraspanin-enriched microdomain; TJ, tight junction; TM, transmembrane


View Abstract