Hepatitis C virus (HCV) is a highly variable, Isolate-Dependent Use of Claudins for Cell Entry by Hepatitis C Virus VIRAL HEPATITIS

Transcription

1 VIRAL HEPATITIS Isolate-Dependent Use of Claudins for Cell Entry by Hepatitis C Virus Sibylle Haid, 1 Christina Grethe, 1 Michael T. Dill, 2 Markus Heim, 2 Lars Kaderali, 3 and Thomas Pietschmann 1 Hepatitis C Virus (HCV) entry involves at least four cellular factors, including CD81, the scavenger receptor class B type I (SCARB-1), occludin (OCLN), and claudin-1 (CLDN1). In addition, CLDN6 and CLDN9 have been shown to substitute for CLDN1 as HCV entry factors in human nonliver cells. We examined the role of different CLDN proteins during HCV entry by using cell lines expressing either predominantly CLDN1 (Huh-7.5) or CLDN6 (HuH6). Huh-7.5 cells were susceptible to all tested HCV isolates, whereas HuH6 cells were only permissive to some viral strains. Silencing of CLDN6 in HuH6 cells revealed that these cells are infected in a CLDN6-dependent fashion, and ectopic expression of CLDN1 or CLDN6 in 293T cells lacking endogenous CLDN expression confirmed that only some HCV strains efficiently use CLDN6 for infection. CLDN1-specific neutralizing antibodies (Abs) fully abrogated infection of Huh-7.5 cells by isolates that use CLDN1 only, whereas viruses with broad CLDN tropism were only partially inhibited by these Abs. Importantly, infection by these latter strains in the presence of anti-cldn1 Ab was further reduced by silencing CLDN6, suggesting that viruses with broad CLDN usage escape CLDN1-specific Abs by utilization of CLDN6. Messenger RNA (mrna) levels of HCV entry factors in liver biopsies of HCV patients infected with different genotype and with variable degree of liver fibrosis were determined. Uniformly high levels of CD81, SCARB-1, OCLN, and CLDN1 mrna were detected. In contrast, abundance of CLDN6 mrna was highly variable between patients. Conclusion: These findings highlight differential CLDN usage by HCV isolates, which may evolve based on variable expression of CLDN proteins in human liver cells. Broad CLDN tropism may facilitate viral escape from CLDN1-specific therapeutic strategies. (HEPATOLOGY 2014;59:24-34) Hepatitis C virus (HCV) is a highly variable, plus-strand RNA virus of the family Flaviviridae and a leading cause of liver disease, including fibrosis, cirrhosis, and hepatocellular carcinoma. 1 The pronounced variability of HCV facilitates viral immune evasion and is attributable to enormous replication rate and error-prone RNA replication. Seven genotypes (GTs) and multiple subtypes are known, with genetic diversity being in the order of more than 30% between individual viral GTs. 2 Although the basic genome structure is conserved among HCV GTs, there are remarkable genotypedependent differences with regard to treatment response and pathophysiology of the infection. For instance, GTs 1 and 4 exhibit inferior response rates, when compared with GTs 2 and 3, in interferon-based therapy regimens, and GT3 virus infection shows a particularly strong association with liver steatosis. 3,4 Abbreviations: Abs, antibodies; CHC, chronic hepatitis C; CLDN, claudin; FACS, fluorescence-activated cell sorting; GTs, genotypes; HCV, hepatitis C virus; HCVcc, cell culture-derived HCV; HCVpp, HCV pseudoparticles; IgG, immunoglobulin G; mir-122, microrna 122; MLV, murine leukemia virus; mrna, messenger RNA; OCLN, occludin; RT-PCR, reverse-transcriptase polymerase chain reaction; SCARB-1, scavenger receptor class B type I; SD, standard deviation; sirna, small interfering RNA From the 1 Institute of Experimental Virology, TWINCORE-Center for Experimental and Clinical Infection Research; a joint venture of the Medical School Hannover (MHH) and the Helmholtz-Centre for Infection Research (HZI), Hannover, Germany; 2 Department of Biomedicine, Hepatology Laboratory, University Hospital Basel, Basel, Switzerland; 3 Institute for Medical Informatics and Biometry, Medical Faculty, Technische Universit at Dresden, Germany. Received April 11, 2013; accepted May 28, This work was supported by a grant from the Helmholtz Association (SO-024; to T.P.). The authors are grateful to Takaji Wakita for the gift of the JFH1 isolate, to Jens Bukh for the different HCV chimeras, to Charles Rice for Huh-7.5 cells and 9E10 antibody, to Brent E. Korba for providing HuH6 cells, and to Matthew Evans for the mir122 expression construct. The authors are grateful to Isidro H otzel and GENENTCH for providing the CLDN1-specific monoclonal antibody. 24

2 HEPATOLOGY, Vol. 59, No. 1, 2014 HAID ET AL. 25 Though the viral factors responsible for these dissimilarities are poorly defined, it is likely that genotypedependent host-factor usage by HCV may, at least in part, account for these differences. HCV cell entry of all tested HCV isolates requires at least four host-derived entry factors, including scavenger receptor class B type I (SCARB-1), CD81, and the tight junction proteins, claudin-1 (CLDN1) and occludin (OCLN). 5 Besides this, the low-density lipoprotein receptor, Niemann-Pick C1-like-1, as well as receptor tyrosine kinases, such as epidermal growth factor receptor and ephrin receptor A2, modulate cell entry. 5 Finally, CLDN6 and CLDN9, two members of the CLDN protein family, render human cells lacking CLDN1 permissive to HCV, suggesting that they can substitute for lack of CLDN1 during HCV infection. 6,7 However, whether the ability to use alternative CLDN family members is common to all HCV isolates, and whether alternative CLDNs are expressed in the liver or other tissues, was incompletely explored. Our results reveal that CLDN usage is variable between HCV strains. For those viruses with broad CLDN tropism, coexpression of CLDN1 and CLDN6 in human hepatoma cells permits viral escape from CLDN1-specific antibodies (Abs) through use of CLDN6. Furthermore, we observed highly variable levels of endogenous CLDN6 expression in liver biopsies of HCV patients. These findings suggest that availability of CLDN6 may select for viruses with broader CLDN tropism, which may escape CLDN1- specific therapeutics through use of CLDN6. Materials and Methods Abs. CLDN1 (Life Technologies, Woburn, MA), CLDN6 (Santa Cruz, Darmstadt, Germany), and b-actin Abs (Sigma-Aldrich, Steinheim, Germany) were used for western blotting analyses. For neutralization experiments, the anti-cd81 Ab, JS-81 (BD, Heidelberg, Germany), the anti-cldn1 Ab, 5.16v4 (Genentech, San Francisco, CA), and the control immunoglobulin G (IgG), Hu5B6 (Genentech), were used. Preparation of Retroviral Pseudotypes. Murine leukemia virus (MLV)-based retroviral particles were created essentially as previously described. 8 Briefly, 293T cells were transfected with envelope protein expression construct pcz VSV-G, pcdna3 DcE1E2 of the different HCV isolates or an empty vector control, MLV Gag-Pol expression construct phit60, and firefly transducing vector prv-f-luc. Preparation of Cell Culture-Derived HCV Particles (HCVcc). HCVcc particles were collected 48 to 72 hours after electroporation of Huh-7.5 cells with 5 mg of in vitro transcribed RNA of given chimeric HCV constructs. 9,10 Transfections and preparation of in vitro transcripts were performed as described previously. 8 To obtain high-titer reporter virus stocks, virus preparations were 10-fold concentrated on a 20% sucrose cushion using ultracentrifugation. Infection Assays. Preparations of chimeric HCVcc viruses were titrated on HuH6 and Huh-7.5 cells using a limiting dilution infection assay, as described previously. 8 Infectivity of Renilla luciferase reporter viruses and HCV pseudoparticles (HCVpp) particles transducing a firefly luciferase gene were evaluated as reported previously. 9 Statistical Analysis. For statistical analysis, graphs were plotted to show the mean 6 standard deviation (SD) of at least three biological replicates. Statistical analyses were performed using Welch s two-sample t test, Kolmogorov-Smirnoff s test, and, alternatively, Wilcoxon s test (for more than five biological replicates). P values <0.1 were considered marginal significant, <0.05 was considered statistically significant, whereas <0.01 was considered highly significant. Liver Biopsies and Informed Consent. HCV viral load, HCV genotype, and liver biopsies from patients with chronic hepatitis C (CHC; n 5 24) were obtained in the context of routine diagnostic workup. Grading and staging of CHC was performed according to the Metavir classification. All patients gave written informed consent in accord with local ethical committees. RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA), according to the manufacturer s instructions. Additional methods can be found in the Supporting Materials. Address reprint requests to: Thomas Pietschmann, Ph.D., Institute of Experimental Virology, TWINCORE-Center for Experimental and Clinical Infection Research, Feodor-Lynen-Strasse 7-9, Hannover, Germany. thomas.pietschmann@twincore.de; fax: Copyright VC 2013 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Nothing to report. Additional Supporting Information may be found in the online version of this article.

3 26 HAID ET AL. HEPATOLOGY, January 2014 Results Isolate-Dependent Infection of HuH6 Cells. Recently, we reported that the human hepatoblastoma cell line, HuH6, supports efficient HCV RNA replication and production of infectious virus particles. 8 However, these cells were refractory to infection with the GT2a/2a chimeric virus, Jc1. The poor permissiveness of HuH6 cells to GT2a infection was linked to low endogenous CLDN1 expression in these cells, because CD81, SCARB-1, and OCLN were highly expressed and because ectopic expression of CLDN1 rendered the cells susceptible to Jc1. 8 To analyze whether the resistance of HuH6 cells was limited to HCV GT2a viruses, we challenged these cells with HCVpp bearing the glycoproteins of different HCV GTs on their surface and transducing a luciferase-expressing lentiviral vector (Fig. 1A). Huh- 7.5 cells, highly permissive to HCV infection of multiple HCV isolates, were used as control. As expected, Huh-7.5 cells were infected by all tested HCVpp, as evidenced by high expression of luciferase 72 hours postinoculation (Fig. 1A). Moreover, we confirmed that HuH6 cells were resistant to GT2a pseudoparticles (J6 and JFH1). Surprisingly, however, these cells were readily infected with GT1a- (H77 isolate) and GT1b-derived (Con1 isolate) HCVpp (Fig. 1A). To extend this finding, we challenged both cell lines with HCVcc of different JFH1-based infectious chimeras representing HCV genotypes 1 through 7 10 (Fig. 1B). Despite low CLDN1 expression in HuH6 cells (Fig. 1C), 8 these cells were permissive to infection by H77c (GT1a), Con1 (GT1b) and J4 (GT1b), J8 (GT2b), S52 (GT3a), ED43 (GT4a), and HK6a (GT6a) particles (Fig. 1B). In contrast, Jc1 carrying J6-derived glycoproteins (GT2a), JFH1 (GT2a), SA13 (GT5a), and QC69 (GT7a) viruses did not infect these cells, suggesting that absence of CLDN1 expression in HuH6 cells limits infection by these strains. To quantify strain-specific differences between susceptibility of Huh-7.5 and HuH6 cells, we calculated the fold difference of the infectious titer for each virus chimera toward these two cell lines (Supporting Fig. 1). As expected, susceptibility of HuH6 cells for HCVcc particles was generally lower, compared to highly permissive Huh-7.5 cells, likely reflecting a lower permissiveness of HuH6 cells for HCV RNA replication. 8 However, the infectious titer of H77c (GT1a) and S52 (GT3a) was only approximately 100- fold lower in HuH6, compared with Huh-7.5, cells. In contrast, susceptibility of HuH6 cells toward infection by SA13 (GT5a) and Jc1 (GT2a) was approximately 1- to 10-million-fold lower, compared with Huh-7.5 cells, respectively. Because these virus chimeras share the same viral replicase encoding nonstructural proteins (JFH1-derived NS3-NS5B), these strong differences likely reflect different permissiveness of Huh-7.5 and HuH6 cells to the cell entry steps of these viruses. Expression profiling confirmed similar levels of CD81, SCARB-1, and OCLN between these cells (data not shown). 8 In contrast, CLDN1 and CLDN6 abundance was variable between Huh-7.5 and HuH6 cells: CLDN1 protein expression was well detectable in lysates of Huh-7.5 cells, but undetectable by western blotting in HuH6 cells, correlating with a more than 20-fold lower messenger RNA (mrna) level in the latter cells. Notably, HuH6 cells expressed detectable amounts of CLDN6 protein, whereas expression in Huh-7.5 cells was below the detection limit of our western blotting analysis, despite comparable CLDN6 mrna levels in both cell lines (Fig. 1C). This difference may reflect dissimilar post-transcriptional regulation of CLDN6 expression between these cell lines. Collectively, these results highlight that Huh-7.5 cells predominantly express CLDN1, whereas HuH6 cells primarily produce CLDN6. Combined with our observation that all tested HCV strains readily infect Huh-7.5 cells, but only some strains enter HuH6 cells, these results suggest that all tested HCV isolates readily use CLDN1 for cell entry, whereas only some strains also utilize CLDN6. HCV Isolate-Dependent Usage of CLDN1 and CLDN6 in 293T Cells. To confirm the isolatedependent usage of these CLDNs as HCV entry factors, we ectopically expressed cherry-tagged CLDN1 or CLDN6 in the human embryonic kidney cell line, 293T (Fig. 2A), which has very low endogenous expression of these proteins (Fig. 1C). Comparable expression level of cherry-tagged CLDN proteins was confirmed by fluorescence-activated cell sorting (FACS) analysis (Fig. 2A). Subsequently, these cells were challenged with HCVpp carrying different HCV envelope proteins, and infection was quantified using luciferase assays. Importantly, H77 (GT1a) and Con1 (GT1b) glycoprotein carrying HCVpp readily infected 293T cells with cherry-tagged CLDN1 and CLDN6 (Fig. 2A). In contrast, pseudoparticles with JFH1- and J6-derived viral glycoproteins selectively infected CLDN1-expressing 293T cells. Therefore, these results confirm that HCV isolates differ with regard to CLDN tropism. Some strains, such as, for example, H77 (GT1a) and Con1 (GT1b), efficiently use both CLDN1 and 6, whereas others, such as, for example, JFH1 and J6 (GT2a), solely use CLDN1 to access cultured cells.

4 HEPATOLOGY, Vol. 59, No. 1, 2014 HAID ET AL. 27 Fig. 1. Isolate-dependent infection of Huh-7.5 and HuH6 cells. (A) Huh-7.5 and HuH6 cells were infected with given HCVpp or pseudotypes carrying the G protein of vesicular stomatitis virus (VSV-G). Infection efficiency was determined by luciferase assay (RLU, relative light unit). Mean values of 13 independent experiments and the SD are given. (B) Huh-7.5 and HuH6 cells were infected with the same dose of given HCVcc preparations. Seventy-two hours postinoculation, cells were fixed and the virus titer was determined by immunohistochemistry (TCID 50, 50% tissue culture infection dose). Mean values of quadruple measurements and the SD are given. (C) CLDN1 and CLDN6 mrna expression were determined by quantitative RT-PCR analysis and normalized to the amount of total RNA using a GAPDH-specific RT-PCR reaction. Protein expression was determined by western blotting analysis using Abs specific for CLDN1 and CLDN6, respectively. b-actin expression was monitored as loading control. GAPDH, glyceraldehyde 3-phosphate dehydrogenase. The dashed bar represents the detection limit of the assay.

5 28 HAID ET AL. HEPATOLOGY, January 2014 Fig. 2. Isolate-dependent usage of CLDN1 and CLDN6 in 293T cells. (A) 293T cells were transfected with expression constructs encoding given cherry-tagged CLDN proteins or a control vector. At time of infection, CLDN expression was determined by FACS analysis. Percentage of cherry-positive cells is given. (B) Forty-eight hours after transfection of 293T cells, cells were infected with HCVpp of different genotypes. Seventy-two hours postinfection, cells were lysed and luciferase activity was measured. Mean values and SD of six independent experiments are shown. Statistical significance of differences was calculated relative to the corresponding control (no env). CLDN6-Dependent HCV Infection of HuH6 Cells. Next, we explored whether those HCV strains capable of infecting HuH6 cells do so in a CLDN6- dependent fashion. To this end, we silenced CLDN1, CLDN6, or CD81 entry factors in HuH6 cells and as a reference in Huh-7.5 cells (Fig. 3A). To improve the sensitivity of our infection assay in HuH6 cells, we created a derivative cell line expressing high levels of the liver-specific microrna 122 (mir-122), which is known to increase HCV translation and replication (data not shown). 11 Subsequently, these cells were challenged with HCVcc chimeras Con1/1b/R2a, Jc1/2a/ R2a, and S52/3a/R2a, expressing viral structural proteins of the Con1 (GT1b), J6 (GT2a), and S52 (GT3a) viral isolates and a Renilla luciferase reporter gene 9 (Fig. 3B). As expected, transient transfection of these cell lines with small interfering RNAs (sirnas) specific to CD81, CLDN1, or CLDN6 selectively repressed the cognate mrnas in both Huh-7.5 and HuH6 cells, whereas the irrelevant sirna control did not affect any of these mrnas (Fig. 3A). In both cell lines, silencing of CD81 strongly reduced HCV cell entry for all viral strains tested, thus confirming CD81-dependent infection for both cell lines and for all viral strains tested. In Huh-7.5 cells, knockdown of CLDN1 inhibited infection of all three virus isolates to between 20% and 60% of control cells, whereas silencing of CLDN6 had little effect (Fig. 3B). In contrast, infection of HuH6 mir- 122 cells with Con1/1b/R2a and S52/3a/R2a viruses was strongly repressed to 10%-20% of control cells by silencing of CLDN6, but not by knock down of CLDN1. As described above, HuH6 mir122 cells were refractory to infection by the GT2a reporter virus, Jc1/ 2a/R2a (data not shown). Collectively, these results indicate that Huh-7.5 cells are primarily infected by CLDN1, the dominant CLDN protein in these cells. In contrast, HuH6 cells are infected by CLDN6, albeit only by those HCV strains capable of efficient utilization of CLDN6 for cell entry. Mapping of CLDN Determinants Relevant to HCV Isolate-Specific Usage. To examine which domains of CLDN1 are required to render CLDN6 permissive to HCV strains that otherwise are unable to use CLDN6 for cell entry, we constructed a set of cherry-tagged CLDN6/CLDN1 chimeric proteins. In each case, a subdomain of the first extracellular loop of CLDN6 (EL1; amino acids 29-81) was replaced with the homologous CLDN1 sequence (Fig. 4A). 293T cells were transiently transfected with expression constructs encoding these proteins, and FACS analysis revealed comparable expression of wild-type CLDN6 and the CLDN6/CLDN1 chimeras (Fig. 4C). Subsequently, these cells were challenged with HCVpp carrying H77 (GT1a), Con1 (GT1b), J6 (GT2a), and JFH1 (GT2a) envelope proteins. Interestingly, domain shuffling between CLDN6 and CLDN1 within the region of the extracellular loop 1 did not grossly influence permissiveness toward the GT1-derived strains, H77 and Con1, with broad tropism toward CLDN1 and CLDN6. In contrast, presence of CLDN1 residues 29 through 48 in hcldn6/el1a rendered the chimeric CLDN protein permissive for J6 and JFH1 HCVpp, which display a narrow tropism toward CLDN1 only (Fig. 4D). Therefore, these results suggest that CLDN determinants for HCV isolate-specific usage are located between residues 29 through 48. HCV Strains With Broad CLDN Tropism Escape From CLDN1-Specific Abs by Using CLDN6. The results presented above suggested that HCV strains with broad CLDN tropism may escape therapeutic strategies selectively targeting CLDN1 by using CLDN6 as an

6 HEPATOLOGY, Vol. 59, No. 1, 2014 HAID ET AL. 29 Fig. 3. Silencing of HCV entry factors in Huh-7.5 and HuH6 cells. (A) mrna expression of CLDN1, CLDN6, and CD81 was determined by quantitative RT-PCR analysis 48 hours post-transfection of Huh-7.5 and HuH6 mir122 cells with the given sirnas. Entry factor mrna levels were normalized to the amount of total RNA determined by a GAPDH-specific RT-PCR. (B) sirna-silenced Huh-7.5 (black bar) and HuH6 mir122 cells (gray bar) were infected with the indicated Renilla reporter viruses. Forty-eight hours postinoculation, cells were lysed and luciferase activity was measured. Infectivity is expressed relative to control cells pretreated with the irrelevant sirna. Infection of HuH6 mir122 cells with GT2a reporter viruses did not result in a productive infection (data not shown). Mean values of three independent experiments, SD, and statistical significance of differences relative to the control (irr) are given. GAPDH, glyceraldehyde 3-phosphate dehydrogenase.

7 30 HAID ET AL. HEPATOLOGY, January 2014 Fig. 4. Mapping of CLDN determinants responsible for isolate-specific usage. (A) Schematic drawing of CLDN6/CLDN1 chimeric constructs. CLDN6 is depicted in gray, whereas CLDN1 is shown with open boxes. The amino acid positions of the junctions between the chimeras are given. (B) Sequence alignment between CLDN1 and CLDN6 within subdomain A of the first extracellular loop (EL1A). Residues critical for CLDN1 usage, as determined by Evans et al., 14 are highlighted by an asterisk. (C) 293T cells were transfected with chimeric CLDN6/CLDN1 constructs, and CLDN expression was determined by FACS analysis 48 hours post-transfection at the time point of HCVpp infection. (D) Sveenty-two hours postinfection, cells were lysed and luciferase activity was measured. Mean values of quadruple experiments, SD, as well as the statistical significance of differences relative to the control (no env) are given. The dashed bar represents the detection limit of the assay.

8 HEPATOLOGY, Vol. 59, No. 1, 2014 HAID ET AL. 31 Fig. 5. Neutralization of Huh-7.5 cell infection by HCV CLDN1-specific Abs and CLDN6-specific sirnas. Huh-7.5 cells were preincubated with the indicated concentrations of (A) CLDN1-specific Abs 5.16v4 or (B) CD81-specific Abs JS-81, respectively. Subsequently, cells were challenged with given reporter viruses and cells were lysed 48 hours later. Mean values of three independent experiments and SD are given. (C) Huh-7.5 cells were transfected with CLDN6 or irrelevant sirna 48 hours before neutralization with 20 mg/ml of anti-cldn1 Ab or control IgG (Hu5B6). After addition of the Abs (2.5 hours), given reporter viruses were added. Forty-eight hours after virus inoculation, cells were lysed and luciferase activity was measured. Mean values of triplicate experiments, including the SD, are given. Infectivity is expressed relative to cells treated with the irrelevant sirna and the control Ab. Statistical analysis was only performed for the 25-mg/mL values by using an analysis of variance test, followed by a paired t test. alternative entry factor, provided it is coexpressed in the HCV target cells. To test this hypothesis, we determined the capacity of a CLDN1-specific Ab 12 to neutralize infection of HCV strains with narrow or broad CLDN tropism in Huh-7.5 cells. Notably, these cells predominantly express CLDN1, but also express modest levels of endogenous CLDN6 mrna (Fig. 1C). Remarkably, the CLDN1-specific Ab potently repressed infection of Huh-7.5 cells by HCVcc particles carrying J6-derived glycoproteins in a dose-dependent fashion. Notably, these glycoproteins only use CLDN1, and infectivity of this virus was inhibited by approximately 98% at the highest Ab dose tested (Fig. 5A). In contrast, at this Ab dose (25 mg/ml), only approximately 80% of the GT1b chimera (Con1), which is able to use both CLDN1 and CLDN6, was neutralized. Even more strikingly, infection by the GT3a chimeric virus, which also uses both CLDN proteins for cell entry, was highly resistant (only 38% neutralization) to these Abs, even at a dose of 25 mg/ml (Fig. 5A). In contrast to differential neutralization of these viruses by the anti- CLDN1 Ab, neutralization with anti-cd81 was comparable between the strains and reduced infectiousness to less than 1% at a dose of 6.25 mg/ml (Fig. 5B). Next, we explored whether endogenous CLDN6 expression in Huh-7.5 cells permits escape from CLDN1-specific Abs selectively for those strains that are able to use this alternative entry factor. To this end, we combined anti-cldn1 Ab treatment with silencing of CLDN6 mrna. As expected, the high dose of anti-cldn1 Ab used (20 mg/ml) strongly repressed infection by Jc1/2a/R2a and pretreatment of the target cells with the CLDN6-specific sirnas did not further reduce virus infection, compared to cells pretreated with irrelevant sirna (Fig. 5C). In contrast, silencing of CLDN6 mrna further reduced infectiousness of the Con1/1b/R2a virus in the presence of anti-cldn1 Abs, yielding a residual infectivity of only 5%, compared with approximately 20% in the case of cells pretreated with the irrelevant sirna.

9 32 HAID ET AL. HEPATOLOGY, January 2014 Finally, infection of the S52/3a/R2a virus was only slightly affected by the anti-cldn1 Ab combined with the irrelevant sirna (75% residual infectivity; Fig. 5C). However, pretreatment of the cells with CLDN6-specific sirnas reduced infectiousness of this virus to approximately 14% of the control. Collectively, these observations suggest that Con1 (GT1b) and S52 (GT3a) viral strains partially escape anti- CLDN1 Abs by using endogenous levels of CLDN6 expressed in Huh-7.5 cells. Highly Variable Abundance of CLDN6 mrna in Liver Biopsies of HCV Patients. To obtain first insights into possible mechanisms that may select for viruses with differential CLDN-tropism and to confirm endogenous expression of CLDN6 in vivo, we determined the abundance of HCV entry factors SCARB-1, CD81, OCLN, CLDN1, and CLDN6 in liver biopsies of HCV patients (see Supporting Table 1 for patient data). Of note, we designed strictly mrna-specific quantification systems by selecting hydrolysis probe-based reverse-transcriptase polymerase chain reaction (RT-PCR) strategies across intron-exon boundaries for each gene to exclude contamination of our quantitative PCR with residual DNA. Using this approach, we observed highly abundant mrna encoding SCARB-1, CD81, OCLN, and CLDN1 in all biopsies tested, indicating that these mrnas are highly expressed irrespective of HCV genotype, disease duration, and degree of liver fibrosis (Fig. 6 and data not shown). In contrast, abundance of CLDN6 mrna in liver biopsies was generally lower, compared to the above-mentioned transcripts. Nevertheless, the average expression of CLDN6 mrna across all liver biopsies tested was comparable to the mrna level in Huh-7.5 and HuH6 cells, suggesting that these cell lines may reflect a level of CLDN6 mrna corresponding to the one in hepatocytes in vivo. Notably, expression of the CLDN6 mrna was highly variable between patients differing more than 50-fold between individuals (Fig. 6E). Stratification of biopsies according to HCV genotype, degree of liver fibrosis, disease duration, or gender did not reveal an overt correlation between any of these parameters and degree of CLDN6 expression (Supporting Fig. 2). In summary, these results confirm high expression of SCARB-1, CD81, OCLN, and CLDN1 mrna in liver biopsies and highlight largely variable expression of CLDN6. Discussion In this study, we show that HCV isolates differ with regard to their utilization of CLDN proteins for cell entry into human hepatoma cells. Specifically, all tested viral strains efficiently utilize CLDN1, whereas only some isolates are able to use CLDN6 as well. Moreover, broad CLDN tropism permits escape from CLDN1-specific Abs, provided a modest level of CLDN6 is coexpressed in the same cell (as, for instance, observed in Huh-7.5 cells in our study). Finally, CLDN6 mrna levels are highly variable in liver biopsies of HCV patients. Zheng et al. and Meertens et al. reported previously that besides CLDN1, also CLDN6 and CLDN9 function as HCV entry factors. 6,7 However, these groups did not observe an overt preference of HCV strains for CLDN1, 6, or 9. In this latter regard, our findings differ from these two studies. Use of different host cells may, in part, account for this. In addition, with the exception of J6-derived glycoproteins (GT2a), none of the isolates that we found to use only CLDN1 were included in these previous studies, and a detailed comparative assessment of differential CLDN usage was not performed. We provide several lines of evidence supporting our conclusion of isolate-specific utilization of CLDNs for HCV cell entry. First, we show that HuH6 cells, which predominantly express CLDN6 and very little CLDN1,aresusceptibleforonlysomeHCVstrains both in HCVpp and HCVcc infection assays (e.g., H77 and Con1), whereas they are refractory to infection by other HCV isolates (e.g., J6 and JFH1; Fig. 1). Second, knockdown of endogenous CLDN6 expression in HuH6 cells confirmed that those isolates that infect these cells do so through CLDN6 (Fig. 3). Of note, we previously showed that na ıve HuH6 cells are rendered permissive for viruses with J6-derived envelope proteins upon restoration of CLDN1 expression, 8 thus excluding a general refractoriness of these cells to infection by this HCV type. Third, ectopic expression of CLDN6 and CLDN1 in 293T cells with very low endogenous expression of CLDNs revealed that those strains that infect HuH6 cells (e.g., H77 and Con1) use both CLDN1 and CLDN6, whereas those isolates that are unable to infect HuH6 cells only efficiently use CLDN1 (e.g., J6 and JFH1; Fig. 2). Finally, transfer of the first portion of the CLDN1 extracellular loop into the backbone of CLDN6 rendered cells expressing the chimeric protein partially permissive for isolates with narrow CLDN tropism (Fig. 4). Collectively, these observations strongly support the conclusion of isolate-dependent usage of CLDN1 and CLDN6 by HCV. We did not investigate CLDN9 usage in this work. However, because the respective subdomain is

10 HEPATOLOGY, Vol. 59, No. 1, 2014 HAID ET AL. 33 Fig. 6. Expression of HCV entry factor mrna in human liver biopsies. mrna was prepared from liver biopsies of 24 HCV patients. Abundance of (A) SCARB-1, (B) CD81, (C) OCLN, (D) CLDN1, and (E) CLDN6 mrna was determined. For comparison, mrna expression was also determined in Huh-7.5, HuH6, and 293T cells. GAPDH mrna was measured in parallel to normalize for equal amounts of total RNA. Mean values of three independent measurements are shown. GAPDH, glyceraldehyde 3-phosphate dehydrogenase. almost fully conserved between CLDN6 and CLDN9 (only residue 28 is polymorphic), it is likely that also CLDN9usagewillbestrainspecific.Inthefuture,it will be interesting to map viral determinants responsible for differential CLDN usage, because such signatures may be useful to predict CLDN receptor usage.

11 34 HAID ET AL. HEPATOLOGY, January 2014 Such information could be particularly relevant for future therapeutic strategies aiming to block the interaction between HCV and CLDN1 to prevent HCV infection. Recently, Fofana et al. reported potent neutralization of HCV infection by means of CLDN1- specific Abs. 13 Such Abs could be particularly valuable to prevent infection of the donor liver by HCV in the course of liver transplantation. In such a context, it would be reasonable to assess the CLDN tropism of the circulating virus and/or to confirm that the Abs used prevent both CLDN1- and CLDN6/CLDN9- dependent HCV cell entry. Notably, we report here that HCV strains with broad CLDN tropism (e.g., Con1 and, particularly, the GT3a-derived S52 strain) are capable of escaping CLDN1-specific Abs by using endogenous levels of CLDN6 coexpressed in Huh-7.5 cells (Fig. 5). Therefore, future work should address whether this route of escape is possible also in humanized mice repopulated with primary human hepatocytes. If that is true, Abs that bind both CLDN1 and CLDN6/CLDN9 or a mixture of Abs blocking these CLDN family members could be used to prevent viral escape. Finally, this model could also be used to test whether the endogenous level of CLDN6 (possibly also CLDN9) is critical to permit viral escape from CLDN1-specific Abs. In contrast to all other HCV entry factors tested by us in this study, abundance of CLDN6 mrna was highly variable in liver biopsies among HCV patients (approximately 50-fold differences). Therefore, endogenous levels of CLDN6 protein in the liver may vary, which, in turn, may influence the ability of HCV to escape through usage of CLDN6. Notably, the mrna level of CLDN6 in Huh-7.5 cells was lower than the one in 17 of 24 liver biopsies, suggesting that a number of HCV patients may indeed have sufficient CLDN6 expression to permit viral escape. Therefore, future work should address CLDN6 expression in the liver (and liver-resident cell types) and the relevance of differential CLDN6 abundance for the course of HCV infection. Given possible differences in post-transcriptional regulation of CLDN6 expression, these studies should also include assessment of protein levels. Moreover, it is currently unclear which mechanisms select for viruses with narrow or broad CLDN tropism. Limiting expression of CLDN1 does not seem to be responsible because CLDN1 mrna was consistently high among all biopsies. However, at this point, we cannot exclude that different protein levels or subcellular distribution of CLDN1 between patients may create an environment where abundance of CLDN1 is limiting for HCV, thus selecting variants that also efficiently use CLDN6. Similarly, differential abundance of CLDN6 expression may influence selection of viruses with differential CLDN tropism. Finally, it is unknown whether CLDN tropism modulates the course of HCV infection and/or treatment response by permitting viral interactions with other tissues and host cells. These questions are important areas of future research. Acknowledgment: The authors thank all members of the Institute for Experimental Virology at TWIN- CORE for helpful comments and discussions of this work. References 1. Lemon SM, Walker CM, Alter MJ, Yi M. Hepatitis C virus. In: Knipe DM, Howley PM, eds. Fields Virology. Vol. 5. Philadelphia, PA: Lipincott Williams & Wilkins; 2007: Nakano T, Lau GM, Lau GM, Sugiyama M, Mizokami M. An updated analysis of hepatitis C virus genotypes and subtypes based on the complete coding region. Liver Int 2012;32: Seeff LB, Hoofnagle JH. National Institutes of Health Consensus Development Conference: management of hepatitis C: HEPATO- LOGY 2002;36:S1-S2. 4. Negro F. Hepatitis C virus-induced steatosis: an overview. Dig Dis 2010;28: Zeisel MB, Felmlee DJ, Baumert TF. Hepatitis C virus entry. Curr Top Microbiol Immunol 2013;369: Zheng A, Yuan F, Li Y, Zhu F, Hou P, Li J, et al. Claudin-6 and claudin-9 function as additional coreceptors for hepatitis C virus. J Virol 2007;81: Meertens L, Bertaux C, Cukierman L, Cormier E, Lavillette D, Cosset FL, Dragic T. The tight junction proteins claudin-1, -6, and -9 are entry cofactors for hepatitis C virus. J Virol 2008;82: Haid S, Windisch MP, Bartenschlager R, Pietschmann T. Mouse-specific residues of claudin-1 limit hepatitis C virus genotype 2a infection in a human hepatocyte cell line. J Virol 2010;84: Haid S, Novodomska A, Gentzsch J, Grethe C, Geuenich S, Bankwitz D, et al. A plant-derived flavonoid inhibits entry of all HCV genotypes into human hepatocytes. Gastroenterology 2012;143: e Gottwein JM, Scheel TK, Jensen TB, Lademann JB, Prentoe JC, Knudsen ML, et al. Development and characterization of hepatitis C virus genotype 1-7 cell culture systems: role of CD81 and scavenger receptor class B type I and effect of antiviral drugs. HEPATOLOGY 2009; 49: Jopling CL, Yi M, Lancaster AM, Lemon SM, Sarnow P. Modulation of hepatitis C virus RNA abundance by a liver-specific MicroRNA. Science 2005;309: Hotzel I, Chiang V, Diao J, Pantua H, Maun HR, Kapadia SB. Efficient production of antibodies against a mammalian integral membrane protein by phage display. Protein Eng Des Sel 2011;24: Fofana I, Krieger SE, Grunert F, Glauben S, Xiao F, Fafi-Kremer S, et al. Monoclonal anti-claudin 1 antibodies prevent hepatitis C virus infection of primary human hepatocytes. Gastroenterology 2010;139: , 964.e Evans MJ, von Hahn T, Tscherne DM, Syder AJ, Panis M, Wolk B, et al. Claudin-1 is a hepatitis C virus co-receptor required for a late step in entry. Nature 2007;446: