Cyanovirin-N Inhibits Hepatitis C Virus Entry by Binding to Envelope Protein Glycans*

Inhibition of viruses at the stage of viral entry provides a route for therapeutic intervention. Because of difficulties in propagating hepatitis C virus (HCV) in cell culture, entry inhibitors have not yet been reported for this virus. However, with the development of retroviral particles pseudotyped with HCV envelope glycoproteins (HCVpp) and the recent progress in amplification of HCV in cell culture (HCVcc), studying HCV entry is now possible. In addition, these systems are essential for the identification and the characterization of molecules that block HCV entry. The lectin cyanovirin-N (CV-N) has initially been discovered based on its potent activity against human immunodeficiency virus. Because HCV envelope glycoproteins are highly glycosylated, we sought to determine whether CV-N has an antiviral activity against this virus. CV-N inhibited the infectivity of HCVcc and HCVpp at low nanomolar concentrations. This inhibition is attributed to the interaction of CV-N with HCV envelope glycoproteins. In addition, we showed that the carbohydrate binding property of CV-N is involved in the anti-HCV activity. Finally, CV-N bound to HCV envelope glycoproteins and blocked the interaction between the envelope protein E2 and CD81, a cell surface molecule involved in HCV entry. These data demonstrate that targeting the glycans of HCV envelope proteins is a promising approach in the development of antiviral therapies to combat a virus that is a major cause of chronic liver diseases. Furthermore, CV-N is a new invaluable tool to further dissect the early steps of HCV entry into host cells.

More than 170 million people worldwide are chronically infected by hepatitis C virus (HCV) 4 (1). This virus is a major cause of chronic hepatitis, cirrhosis, and hepatocellular carcinoma (2). In addition, chronic HCV infection has become the most common indication for liver transplantation. Current antiviral therapy is based on the use of polyethylene glycolmodified interferon-␣ in combination with ribavirin. However, this treatment is expensive, relatively toxic, and effective in only half of the treated patients (3). Furthermore, there is as yet no vaccine against HCV. Therefore, more efficacious and better tolerated anti-HCV treatments are sorely needed to combat this major pathogen.
HCV is an enveloped virus that belongs to the Hepacivirus genus in the Flaviviridae family (4). Because of difficulties in propagating HCV in cell culture, many gaps remain in our understanding of the HCV life cycle. A major advance in the investigation of HCV entry was the development of pseudoparticles (HCVpp), consisting of native HCV envelope glycoproteins E1 and E2 assembled onto retroviral core particles (5)(6)(7). This system is potentially very powerful to identify and characterize molecules that block HCV entry. Furthermore, data obtained with HCVpp can also now be confirmed with the help of the recently developed cell culture system that allows efficient amplification of HCV (HCVcc) (8 -10).
During their biogenesis, the two envelope glycoproteins E1 and E2 assemble as a noncovalent heterodimer, which is retained in the endoplasmic reticulum (11). In this compartment, four or five potential glycosylation sites on E1 and up to 11 sites on E2 are modified by N-linked glycosylation (12,13). Analysis of the glycans bound to intracellular HCV envelope glycoproteins has shown that only high-mannose-type oligosaccharides are associated with these proteins (14). Nevertheless, some of these glycans are probably modified after virus assembly and release as suggested by the characterization of the envelope glycoproteins associated with HCVpp (15)(16)(17). It is worth noting that most glycosylation sites on HCV envelope glycoproteins are well conserved (12). Furthermore, glycans associated with HCV envelope glycoproteins play an essential role in protein folding and/or HCV entry (13,18), suggesting that N-linked glycans might be a potential target for the development of new antiviral molecules against HCV.
Cyanovirin-N (CV-N) is a 101-amino acid protein (11 kDa) that was isolated from an aqueous extract of the cyanobacterium Nostoc ellipsosporum (19). CV-N was originally identified as an active agent against HIV-1 and HIV-2 (19) and later as an antiviral agent against some other enveloped viruses (20). The antiviral activity of CV-N seems to involve specific interactions with Man-8 and Man-9 N-linked glycans found on viral envelope glycoproteins, which prevent viral entry into target cells (21)(22)(23)(24)(25). Because HCV envelope glycoproteins are highly glycosylated and potentially contain oligomannose glycans (15)(16)(17), we sought to determine whether CV-N has antiviral activity against this virus. Here, we demonstrate that CV-N inhibits HCV entry into host cells at low nanomolar concentrations. In addition, we show that the binding of CV-N to N-linked glycans associated with HCV glycoproteins blocks the interaction between E2 and CD81, a cell surface molecule involved in HCV entry (reviewed in Ref. 26). These data demonstrate that targeting the glycans of HCV envelope proteins is a new and promising approach in the development of antiviral therapies to combat hepatitis C.
Production of HCVpp and Infection Assay-HCVpp were produced as described previously (5,17) with plasmids kindly provided by B. Bartosch and F. L. Cosset (INSERM U758, Lyon, France). With the exception of studies comparing envelope proteins of different genotypes, HCVpp were produced using a plasmid encoding the envelope glycoproteins of genotype 1a (H strain). The following plasmids encoding HCV envelope glycoproteins of different genotypes, kindly provided by J. Ball (Nottingham University, Nottingham, United Kingdom), were used in this work: UKN1B-5.23 (genotype 1b); UKN2B-1.1 (genotype 2b); UKN3A-1.28 (genotype 3a); UKN4-11.1 (genotype 4); UKN5-14.4 (genotype 5); and UKN6-5.340 (genotype 6) (32). In addition, the plasmid encoding HCV envelope glycoproteins from genotype 2a (strain JFH-1) was kindly provided by T. Pietschamnn and R. Bartenschlager (University of Heidelberg, Heidelberg, Germany). For the production of vesicular stomatis virus-Gpp, a plasmid encoding the vesicular stomatis virus glycoprotein G (33) was used. Supernatants containing the pseudotyped particles were harvested 48 h after transfection, filtered through 0.45-m pore-sized membranes. HCVpp were added to Huh-7 cells seeded the day before in 24-well plates and incubated for 3 h at 37°C. The supernatants were then removed, and the cells were incubated in Dulbecco's modified essential medium, 10% fetal bovine serum at 37°C. At 48 h postinfection, luciferase assays were performed as indicated by the manufacturer (Promega, Madison, WI).
Production of HCVcc and Infection Assay-The plasmid pJFH1, containing the full-length cDNA of JFH1 isolate (genotype 2a) and kindly provided by T. Wakita (Tokyo Metropolitan Institute for Neuroscience, Tokyo, Japan), was used to generate HCVcc as described previously (8). In brief, the pJFH1 plasmid was linearized and used as a template for in vitro transcription with the MEGAscript kit from Ambion. In vitro transcribed RNA was delivered to Huh-7 cells by electroporation, and viral stocks were obtained by harvesting cell culture supernatants at 3 to 4 days post-transfection. Secondary viral stocks were obtained by additional amplifications on naïve Huh-7 cells (34).
Immunoprecipitation Assay-HCVpp were incubated at 4°C for 2 h in the presence or absence of CV-N (0.6 g/ml). HCVpp were then lysed using 1% Triton X-100. CV-N was immunoprecipitated using the anti-CV-N polyclonal antibody. Immunoprecipitates were extensively washed with phosphate-buffered saline containing 0.2% Triton X-100, eluted in Laemmli buffer, and analyzed by Western blotting using anti-E1 (A4) and anti-E2 (H47) mAbs.
Protein Deglycosylation-HCVpp were concentrated on a 20% sucrose cushion and denatured in denaturation buffer (0.5% SDS and 1% 2-mercaptoethanol) by boiling for 15 min at 95°C. The protein samples were then divided into two equal portions: one for digestion with peptide N-glycosidase F and one for an undigested control. Digestion was carried out for 2 h at 37°C in the buffer provided by the manufacturer (New England Biolabs, Ipswich, MA).

CV-N Inhibits HCVcc and HCVpp Infectivity-
To determine the effect of CV-N on HCV infectivity, we first tested the capacity of HCVcc to infect Huh-7 cells in the presence of increasing amounts of CV-N. HCVcc infectivity was measured by analyzing the levels of expression of HCV glycoprotein E2. As shown in Fig. 1A, E2 expression decreased in a dose-dependent manner in the presence of CV-N, indicating that CV-N inhibits HCVcc infectivity. Importantly, CV-N was active at low nanomolar concentrations. Furthermore, adding CV-N at 3 h postinfection had no effect on HCVcc infectivity, indicating that CV-N blocks an early step of the HCV life cycle (Fig. 1A).
To confirm the effect of CV-N on virus entry, we analyzed its antiviral effect on HCVpp. As shown in Fig. 1B, HCVpp infectivity was reduced in a dose-dependent manner in the presence of CV-N (EC 50 ϭ 1.6 Ϯ 0.1 nM corresponding to 17.6 ng/ml). In addition, CV-N had only a slight effect on HCVpp infectivity when added at 3 h postinfection, confirming that it blocked the entry step (Fig. 1C). CV-N had no antiviral activity against par-ticles pseudotyped with the envelope glycoprotein G of vesicular stomatitis virus at the highest concentration tested (Fig. 1D), confirming that the antiviral activity of HCVpp is specific for the presence of HCV envelope glycoproteins on the pseudotyped particles. It is worth noting that CV-N had similar antiviral activity on HCVpp generated with envelope glycoproteins from different genotypes (Fig. 2), indicating that the anti-HCV activity of CV-N is broad. It is important to note that no increase in cell mortality was observed in the presence of CV-N at the highest concentration tested. Because CV-N had a similar effect on both HCVcc and HCVpp entry, the antiviral activity of CV-N was further studied with the HCVpp system, which is a more convenient tool for the characterization of HCV entry.
CV-N Interacts with E1E2 Heterodimers-The inhibitory effect of CV-N on HCVpp and HCVcc may potentially be attributed to CV-N interaction with either target cells or viral particles. To investigate whether CV-N inhibits HCVpp entry by affecting the cell, Huh-7 cells were incubated with CV-N prior to infection. However, preincubation of Huh-7 cells with CV-N (0.1 g/ml) had only a minor effect on HCVpp entry (Fig. 3A), suggesting that the anti-HCV activity of CV-N is probably due to an effect on viral particles. The slight effect of CV-N added to the cells before infection might be due to the binding of CV-N to a cell surface molecule involved in HCV entry. To test whether the anti-HCV activity of CV-N is due to an effect on viral particles, CV-N and HCVpp were incubated before contact with target cells. In these conditions, CV-N should have more time to bind to HCVpp and hence have a stronger antiviral activity. In contrast, the antiviral activity of CV-N should not be modified in these conditions if CV-N does not interact with the particles. As shown in Fig. 3B, the anti-HCV activity of CV-N was greater when CV-N was preincubated with HCVpp, suggesting that CV-N interacts directly with HCVpp and thereby reduces their infectivity.
We suspected that, because of its capacity to bind to N-linked glycans, the anti-HCV effect of this protein might be explained by an interaction between CV-N and HCV envelope glycoproteins. To determine whether CV-N interacts with E1E2 heterodimers, the presence of complexes involving CV-N and E1E2 was analyzed in coimmunoprecipitation experiments. After immunoprecipitation with an anti-CV-N antibody, the presence of E1E2 heterodimers in the immunoprecipitates was analyzed by Western blotting with specific mAbs. As shown in Fig. 4, HCV envelope glycoproteins were detected after immunoprecipitation with the anti-CV-N antibody, indicating that HCV envelope glycoproteins interact with CV-N. In contrast, neither E1 nor E2 was detected in the absence of CV-N. These results allowed us to demonstrate that CV-N specifically interacts with E1E2 heterodimers.   SEPTEMBER 1, 2006 • VOLUME 281 • NUMBER 35

CV-N Interacts with N-linked Glycans of HCV Envelope
Glycoproteins-Because CV-N interacts with N-linked highmannose oligosaccharides Man-8 and Man-9 (21-25) and E1E2 heterodimers contain this type of glycan, we sought to determine whether such interactions can be responsible for the anti-HCV activity of CV-N. As a first approach, we treated HCVpp envelope glycoproteins with peptide N-glycosidase F to remove the glycans associated with HCV envelope glycoproteins. Interestingly, interaction between CV-N and HCV envelope proteins was abolished after this enzymatic deglycosylation (Fig.  5A), indicating that CV-N interacts with the N-linked glycans of HCV envelope glycoproteins. We also investigated the capacity of Man-9 to reduce the anti-HCV activity of CV-N. To this end, we analyzed whether the infectivity of HCVpp would be modified in the presence of CV-N (0.1 g/ml) incubated with increasing amounts of Man-9. Interestingly, Man-9 restored HCVpp infectivity in a concentration-dependent manner (Fig. 5B). It is worth noting that, at the highest concentration tested, Man-9 alone had no effect on HCVpp infectivity (Fig. 5B). In addition, preincubation of CV-N with Man-9 also reduced the interaction between CV-N and HCV envelope glycoproteins in our coimmunoprecipitation assay (Fig. 5C).
Unfortunately, our previously reported glycosylation mutants (13) did not allow us to identify the specific glycan(s) on HCV envelope glycoprotein that bind CV-N. This is probably due to the fact that CV-N binds to several glycans on HCV envelope glycoproteins. Altogether, our results indicate that the anti-HCV activity of CV-N is the result of its interaction with N-linked glycans associated with HCV envelope glycoproteins.
CV-N Inhibits E2-CD81 Interaction-Because CV-N interacts with E1E2 heterodimers via their glycans, this compound might interfere with critical interactions between viral envelopes and cell-surface molecules that are required for virus entry into target cell. The best characterized molecule involved in HCV entry is CD81 (26), which has been shown to interact with HCV glycoprotein E2 (36). Therefore, we sought to determine whether CV-N is able to prevent the interaction of E1E2 heterodimers with this cellular protein.
To this end, HCVpp were incubated for 2 h with increasing amounts of CV-N, and the interactions between E1E2 heterodimers and CD81 were analyzed in a pulldown assay using a glutathione S-transferase protein fused to the large extracellular loop of human CD81 (CD81-LEL). As shown in Fig. 6, CV-N reduced E1E2-CD81-LEL interaction in a dosedependent manner, indicating that CV-N inhibits the binding of E1E2 to one of its potential receptors. It is worth noting that CV-N did not interact with CD81-LEL preadsorbed onto glutathione-Sepharose beads (data not shown). These results indicate that CV-N interacts with E1E2 heterodimers and prevents the interaction between E2 and the large extracellular loop of CD81, suggesting that CV-N blocks HCV  entry by inhibiting the contact between HCV envelope glycoproteins and CD81.

DISCUSSION
Inhibition of viruses at the stage of viral entry provides a route for therapeutic intervention, as evidenced by the recent development of inhibitors of HIV entry (37). Because of difficulties in propagating HCV in cell culture, entry inhibitors have not yet been reported for this virus. However, thanks to the recent development of HCVpp and HCVcc systems, studying HCV entry is now possible, and these systems are essential tools for the identification and characterization of molecules that block HCV entry. Here, we investigated the potential anti-HCV activity of CV-N. Our data show that CV-N interacts with N-linked glycans associated with HCV envelope glycoproteins, which leads to an inhibition of interaction between HCV glycoprotein E2 and CD81. Importantly, CV-N is active against HCV at low nanomolar concentrations and has a broad effect, because it is able to inhibit entry of various genotypes. This report is a proof-of-concept that HCV glycans are an interesting target for the development of new antiviral molecules that block HCV entry. Furthermore, CV-N is a new invaluable tool to further dissect the early steps of HCV entry into host cells.
The binding of CV-N to N-linked glycans on HCV envelope glycoproteins inhibits E2-CD81 interaction. Because N-linked glycans might not be necessary for E2 binding to CD81 (38), it is likely that the binding of CV-N to the glycans of HCV envelope glycoproteins masks the CD81 binding site on E2 by steric hindrance. This also suggests that one or several high-mannose glycans surround the CD81 binding site on E2. However, due to the absence of a three-dimensional structure for HCV envelope glycoproteins, we do not know the spatial relationship between HCV glycans and the CD81 binding site. Interestingly, the extracellular form of a truncated E2 glycoprotein is less efficient than its intracellular form to interact with CD81, and it has been proposed to be due to the processing of some oligomannoses into complex-type glycans (38,39). However, it is unlikely that these glycans are processed in the context of the E1E2 heterodimer. Indeed, the processing of E2 glycans is different when E1 and E2 do not assemble as a heterodimer (40). Together, FIGURE 5. CV-N interacts with N-linked glycans of HCV envelope glycoproteins. A, CV-N does not interact with deglycosylated HCV envelope proteins. HCVpp were denatured, digested with peptide N-glycosidase F, and incubated with CV-N. CV-N was then immunoprecipitated with the polyclonal antibody against CV-N. Coimmunoprecipitated HCV envelope proteins were analyzed by Western blotting with anti-E1 (A4) and anti-E2 (3/11) mAbs. Deglycosylated proteins are indicated by an asterisk. B, Man-9 inhibits the anti-HCV activity of CV-N. Increasing amounts of Man-9 was preincubated with or without CV-N. Huh-7 cells were then incubated for 3 h with HCVpp in the presence of Man-9-treated CV-N (CV-N concentration: 0.1 g/ml) or Man-9 alone. The inoculum was then removed, and the cells were further incubated. Virus entry was measured as in Fig. 1B. The results are presented as percentages of infectivity relative to infectivity of HCVpp in the absence of CV-N and Man-9. Results are reported as the mean Ϯ S.D. of three independent experiments. C, Man-9 inhibits CV-N interaction with E1E2 heterodimers. HCVpp were incubated in the presence of CV-N (0.1 g/ml) and increasing amounts of Man-9 and lysed with Triton X-100. CV-N was then immunoprecipitated with the polyclonal antibody against CV-N. Coimmunoprecipitated HCV envelope glycoproteins were analyzed by Western blotting with an anti-E2 (3/11) mAb. these observations suggest that a slight increase in steric hindrance can already affect E2-CD81 interaction, indicating that small antiviral molecules targeting HCV glycans can potentially be designed.
As for HIV, the development of resistance against drugs that target the glycans on HCV glycoproteins will probably result in mutations in some glycosylation sites (41,42). However, most of the glycosylation sites are strongly conserved in HCV (12,43) and, contrary to HIV, shifts in terms of the relative position of the glycans are seldom observed for HCV glycans (43). In addition, some HCV glycans have been shown to play a role in virus entry (13). Therefore, adaptation of HCV to the selective pressure of CV-N will be more difficult than it is for HIV.
In the case of HIV, CV-N has been shown to have considerable potential for development as a prophylactic topical microbicide to prevent sexual transmission (44). However, in the context of HCV infection, CV-N would need to reach the systemic circulation for the treatment of HCV-infected patients to have an antiviral activity. In the context of an experimental infection of mice with Ebola virus, CV-N has been shown to reach the systemic circulation and exhibit measurable activity (45), suggesting that this drug can cross several physiological compartments after subcutaneous injection. CV-N might therefore be functional in inhibiting HCV in infected patients; however, further pharmacokinetic analyses in animal models are needed to determine the tissue distribution of CV-N.
Inhibitors of virus entry, such as CV-N, will reinforce the arsenal against HCV. The recent resolution of the three-dimensional structure of various HCV components and the development of in vitro assays to assess the antiviral potency of molecules targeting these elements have made it possible to screen and develop specific small inhibitory molecules (46). Currently, the major targets of new antivirals are the HCV internal ribosome entry site, NS3 serine protease, and NS5B, the RNA-dependent RNA polymerase. Although a small number of compounds have started to show promising results in early-phase clinical trials, preclinical evidence is accumulating, demonstrating that development of resistance will eventually limit the efficacy of these new drugs (47). This is due to the fact that HCV is a highly variable virus with rapid viral kinetics, large population sizes, and a quasispecies distribution (48). Therefore, combinations of multiple drugs with different targets will be required to treat chronic hepatitis C. Blocking virus entry by a compound, such as CV-N, offers a new opportunity to develop a highly active antiviral combination for the treatment of chronic HCV infection.