In this study we took advantage of a human liver cell line expressing mouse in place of human CD81 to select for a population of HCV that utilizes this entry factor with high efficiency for infection of host cells. Employing reverse genetics we showed that the ability to utilize mouse CD81 was conferred by a combination of three mutations within the viral surface proteins E1 and E2.
In principle, change of receptor usage mediated by these mutations may reflect an adaptation of the viral CD81 binding site to accommodate the mouse entry factor and to enhance binding affinity. However, neither these residues nor the domains in which they are located in have previously been implicated in the virus-CD81 interaction. Two of the identified mutations (V388G and M405T) are part of the so called hypervariable region 1 (HVR1). This highly variable domain at the N-terminus of E2 has been implicated in the interaction with SR-BI 
and is dispensable for binding of soluble E2 to CD81 
. In fact, deletion of HVR1 enhances binding of soluble E2 to soluble or cell surface-bound CD81 
. In addition viruses lacking HVR1 are only moderately impaired in entry and remain susceptible to neutralization by CD81-specific antibodies 
, thus strongly suggesting that the HVR1 is dispensable for the interaction with CD81. Consequently, it is unlikely that the HVR1 is part of the viral CD81 binding site and that the adaptive mutation within this domain directly enhance the binding to mouse CD81.
Alternatively, adaptive changes may act indirectly by eliminating obstructions that otherwise sterically prevent access of mouse CD81 to the viral CD81 binding site. In the absence of structural information of the viral E1/E2 complex we probed the influence of these mutations on the interaction with CD81 using soluble human CD81 and CD81-specific antibodies. To phenotypically distinguish adaptation to mouse CD81 from cell culture adaptation we compared the mouse CD81-adapted virus to a recently described cell culture-adaptive mutation within E2 (G451R) that is known to modulate SR-BI and CD81 receptor usage and virus neutralization 
. Interestingly, both mutant viruses interacted more efficiently with soluble human CD81-LEL and they were both more susceptible to neutralization by this protein (). Previously, Flint et al. were unable to detect binding of HCV E2 protein to soluble mouse CD81 
. Our attempts to detect changes of HCV binding to soluble mouse CD81 did not yield conclusive results precluding a definite answer if the affinity to mouse CD81 was changed by the adaptation (data not shown). Therefore, it remains to be determined if parental or adapted HCV binds mouse CD81. Nevertheless, these results indicate that both virus mutants have a more accessible binding site for human CD81 which likely facilitates interaction and neutralization. Such a more open conformation may also assist interaction with mouse CD81. Notably, also the mouse CD81 adapted virus remains dependent on CD81 since it does not infect cells devoid of CD81. Therefore, either the adapted HCV indeed does interact with mouse CD81 or it utilizes an essential function of mouse CD81 as entry factor even in the absence of direct binding. As CD81 is a member of the protein family of tetraspanins, which assemble multi-protein complexes within membranes, such a function could for instance be the coordination of additional HCV entry factors into a functional receptor complex. In line with this notion, a direct interaction between CD81 and CLDN1 has been reported 
Interestingly, the mouse CD81 adapted virus was more resistant to neutralization by CD81-specific antibodies and infected cells carrying a given abundance of human CD81 at their surface with increased efficiency compared to wildtype and the Jc1/G451R variant HCV (). These findings suggest that only the mouse CD81-adaptive changes have enhanced human CD81 receptor usage for infection. Notably, this change however was not accompanied by a reduction of the minimal threshold CD81 density needed for viral entry (). It is worth mentioning here that Grove et al. noted decreased neutralization of JFH1 carrying the G451R mutation by CD81-specific antibodies 
. Although we do not know the reason for this discrepancy it is possible that the phenotype of this mutation at least with regard to CD81 usage may differ dependent on the strain of glycoproteins (J6 and JFH1, respectively).
Receptor competition experiments with antibodies against SR-BI and CLDN1 revealed a similar behaviour of the mutant viruses, both exhibiting a decreased SR-BI dependence, while CLDN1-dependence was indistinguishable between both mutant viruses and wildtype HCV (). Finally, when reducing the number of available OCLN molecules by RNA interference, both wild type and the G451R adapted virus were clearly more inhibited than Jc1/mCD81. These data indicate that both cell culture adaptation as well as adaptation to mouse CD81 caused reduced SR-BI dependence, however only the latter procedure at the same time also reduced dependence on human CD81 and OCLN.
A third possible mechanism that may permit utilization of mouse CD81 for entry is through facilitating essential conformational changes of the glycoprotein complex. This could for instance be mediated by an opening of the overall structure of the viral surface proteins. Our neutralization experiments with E2-specific antibodies revealed a much enhanced susceptibility to neutralization for both mutant viruses. Interestingly, susceptibility to domain C-specific antibodies was more pronounced for the mouse CD81 adapted virus than for Jc1 with the G451R mutation, indicating distinct conformations of the virus particle-resident E2 proteins at this domain. These findings are both suggestive of gross conformational changes and of increased exposure of key neutralizing epitopes. Since these changes were not accompanied by an altered distribution of these viruses in density gradients, we believe that they are not due to differential interaction with lipoproteins. As additional indicator for altered surface protein conformation and as a measure for viral responsiveness to triggers of cell membrane fusion, we compared the ability of wildtype and mutant HCV to infect host cells stimulated by a low pH wash. In agreement with previous findings, wildtype HCV was refractory to a low pH wash immediately after binding to the cell surface at 4°C and only infected cells after an incubation of 1 h at 37°C 
. In contrast, both mutant viruses could be induced to infect cells by low pH directly after binding. This suggests that the requirements for induction of pH-dependent fusion have changed for both viruses.
Since the structure of the HCV E1/E2 complex is unknown it is difficult to finally conclude which of the above mentioned mechanisms that are certainly not mutually exclusive, is responsible for the efficient utilization of mouse CD81. Our data support a model where this combination of mutations leads to an opening of the glycoprotein complex. This “unlocked” structure increases exposure of the CD81 binding site and likely in turn interaction with soluble CD81. Moreover, it also facilitates conformational changes which may permit the virus to utilize a weak CD81 binding (e.g. due to interaction with CD81 from non-human species which bind with low affinity) for its entry process. In addition, such a conformational change may contribute to decreased dependence on SR-BI and OCLN and a lower threshold for pH-triggered cell entry. This interpretation may also explain why these mutations selected on mouse CD81 increase the usage of rat and hamster CD81 to a comparable degree (approximately 100-fold). Given that these forms of CD81 differ from each other by a number of amino acids within the viral binding site 
, this would have been unlikely, if the viral binding site had specifically adjusted to mouse CD81.
Finally, we show that HCV J6-derived glycoproteins adapted to mouse CD81 permit infection of mouse cells in the absence of human factors. Thus, three adaptive mutations within E1 and E2 are sufficient to overcome the barrier for cell entry into mouse cells. In our view, this finding lends further support to the notion that these mutations may facilitate receptor-dependent conformational changes of the glycoproteins thus permitting the use of mouse CD81 and ultimately infection of cells via solely mouse factors. Previous reports have established that besides CD81 at least OCLN poses a strong species barrier to HCV infection 
. In addition, evidence was reported that sub-optimal usage of mouse CLDN1 and SR-BI may also limit efficient infection of mouse cells 
. Therefore, poor entry of HCV into mouse cells seems to be determined by multiple entry factors, making it difficult to envision that mere adaptation of the viral CD81 binding site would be sufficient to allow usage of mouse SR-BI, OCLN, and CLDN1. Notably, while the G451R mutation only reduced SR-BI-dependence, the mouse CD81 adaptive mutations at the same time also reduced CD81- and OCLN-dependence which may be an important prerequisite for the virus to utilize the foreign entry factors which possibly interact with lower affinity.
Likely due to very low RNA replication of HCV in mouse cells we were unable to observe productive infection of mouse cells by Jc1/mCD81 (data not shown). In turn these results highlight that additional host factors limit propagation of HCV in mouse cells downstream of virus cell entry. Nevertheless, our findings are proof of concept that a limited number of mutations can be sufficient to adapt the virus for usage of foreign host factors or possibly even host factor complexes and therefore encourage future efforts to adapt HCV for growth in non-human cells. Combined with genetic approaches and xenotransplantation 
such efforts may ultimately lead to the development of immune competent small animal models for research of HCV pathogenesis, immune control as well as vaccine and drug development.