PCSK9 impedes hepatitis C virus infection in vitro and modulates liver CD81 expression

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Human PCSK9 is known to enhance the degradation of membrane-bound receptors such as the hepatocyte low-density lipoprotein receptor (LDLR), ApoER2, and very low-density lipoprotein receptor. Because the LDLR is suspected to be involved in hepatitis C virus (HCV) entry, we also tested whether PCSK9 can affect the levels of CD81, a major HCV receptor. Interestingly, stable expression of PCSK9 or a more active membrane-bound form of the protein (PCSK9-ACE2) resulted in a marked reduction in CD81 and LDLR expression. Therefore, we analyzed the antiviral effect of PCSK9 in vitro using the HCV genotype 2a (JFH1) virus. The results clearly demonstrated that cells expressing PCSK9 or PCSK9-ACE2, but not the ACE2 control protein, were resistant to HCV infection. Furthermore, addition of purified soluble PCSK9 to cell culture supernatant impeded HCV infection in a dose-dependent manner. As expected, HuH7 cells expressing PCSK9-ACE2 were also resistant to infection by HCV pseudoparticles. In addition, we showed that CD81 cell surface expression is modulated by PCSK9 in an LDLR-independent manner. Finally, in the liver of single Pcsk9 and double (Pcsk9 + Ldlr) knockout mice, both LDLR and/or CD81 protein expression levels were significantly reduced, but not those of transferrin and scavenger receptor class B type 1. Conclusion: Our results demonstrate an antiviral effect of the circulating liver PCSK9 on HCV in cells and show that PCSK9 down-regulates the level of mouse liver CD81 expression in vivo. Therefore, we propose that the plasma level and/or activity of PCSK9 may modulate HCV infectivity in humans. (HEPATOLOGY 2009.)

Hepatitis C virus (HCV) is a worldwide leading cause of chronic hepatitis, cirrhosis, and hepatocellular carcinoma.1 In the absence of a prophylactic vaccine or a specific antiviral agent, the best treatment currently available for HCV infection is the combination therapy of pegylated interferon and ribavirin.2 HCV is a positive-strand RNA enveloped virus classified as a Hepacivirus within the Flaviviridae family.3

Viral entry within target cells, primarily human hepatocytes, is not very well understood. Several cell surface proteins have been suggested to play a role in the binding of HCV to hepatocytes and/or be critical for viral entry.4, 5 Among the acknowledged receptors or coreceptors of HCV are: CD81,6 scavenger receptor class B type I,7 low-density lipoprotein receptor (LDLR),8 and the recently identified claudin-1, claudin-6, claudin-9,9 and occludin.10 However, it is unclear which ones are required for viral propagation in vivo. Indeed, HCV particles recovered from infected plasma migrated at two distinct densities (1.25 g/mL and 1.06 g/mL).11, 12 It has been suggested that the most infectious virus is located in the low-density fraction that corresponds to lipoviroparticles of HCV coated with apolipoprotein B and E.11 Because LDLR controls the uptake of circulating cholesterol principally through internalization of apolipoprotein B– and E–containing lipoproteins, such as LDL and very low-density lipoprotein, the implication of LDLR as a nonspecific receptor for HCV entry may be significant, at least in vivo.

PCSK9, the latest member of the proprotein convertase family,13, 14 has been extensively studied for its ability to regulate the cell surface levels of LDLR.14–16 Individuals with PCSK9 gain-of-function mutations exhibit hypercholesterolemia, whereas loss-of-function mutations result in hypocholesterolemia.17, 18 Recently, it has been demonstrated that PCSK9 interaction with cell surface LDLR in hepatocytes results in the enhanced degradation of the receptor both ex vivo16, 19 and in vivo,20, 21 thereby reducing the uptake of circulating LDL cholesterol by cells. In humans and mice, PCSK9 is found as a soluble protein, which is secreted in the bloodstream mostly by hepatocytes.21–23 It has been shown that fusion of a transmembrane/cytosolic tail segment at the C-terminus of wild-type PCSK9 (PCSK9-wt), such as that of the cell surface metalloproteinase ACE2 (PCSK9-ACE2) or of the lysosomal protein Lamp1 (PCSK9-Lamp1), results in a membrane-bound nonsecreted PCSK9, which is much more potent in enhancing the degradation of LDLR and its two closest family members, very low-density lipoprotein receptor and ApoER2.24 We used these chimeras to reduce LDLR expression and to evaluate their effects on HCV JFH1 infection of HuH7 cells.

The data revealed that expression of PCSK9 or its chimeras in otherwise susceptible HuH7 cells blocks infection by HCV JFH1. Surprisingly, PCSK9 reduces not only the surface expression of LDLR but also of CD81, a major HCV receptor. Because high levels of CD81 expression are required for HCV infection of HuH7 cells in vitro,25 we suggest that PCSK9 may prevent cellular HCV entry through both LDLR and CD81.

Abbreviations

HCV, hepatitis C virus; HCVpp, hepatitis C virus pseudoparticles; LDLR, low-density lipoprotein receptor; siRNA, small interfering RNA.

Methods and Results

PCSK9 Expression in Stable Cell Lines.

HuH7 cells stably expressing PCSK9-wt, PCSK9-ACE2, ACE2 (Fig. 1A), or the empty vector phCMV3 were sorted via fluorescence-activated cell sorting (FACS) to isolate the ≈10% highest expressors of cell surface PCSK9 (Fig. 1B), as confirmed by immunobloting of cell lysates (Fig. 1C). Although PCSK9-wt is mostly secreted in the media with ≈1%-5% remaining bound to the cell-surface LDLR,26 In contrast, a large amount of total PCSK9-ACE2 remains attached to the cell surface24 through the transmembrane domain of ACE2 (Fig. 1D). Immunocytochemistry of these stable cells revealed that just like PCSK9-wt, the chimeric protein PCSK9-ACE2 is detected in perinuclear locations (reminiscent of the endoplasmic reticulum), but is also abundantly concentrated at the cell surface (Fig. 1E).

Figure 1.

Stable PCSK9-wt and PCSK9-ACE2 expression in HuH7 cells. (A) Recombinant proteins used were: signal peptide (sp); prosegment (pro); cysteine and histidine-rich domain (CHRD); transmembrane domain (TM); cytoplasmic tail (CT). (B) Cell sorting of PCSK9-ACE2–positive cells, whereupon the ≈10% higher expressors were collected. (C) Western blots showing the levels of stably expressed PCSK9-wt, PCSK9-ACE2, or ACE2 in HuH7 cells. (D) FACS analysis of PCSK9-wt and PCSK9-ACE2 cell surface expression on HuH7 cells. (E) Immunofluorescence confocal analysis of PCSK9-wt and PCSK9-ACE2 protein in HuH7 cells. Arrows point to the reticular or cell surface expression of PCSK9 or PCSK9-ACE2, respectively.

PCSK9 Expression Removes both LDLR and CD81 from the Cell Surface.

To verify if the only HCV receptor affected by PCSK9 was LDLR, we analyzed the fate of the cell surface expression of CD81. Thus, cell surface CD81 levels were compared in HuH7 cells expressing PCSK9-wt, PCSK9-ACE2, ACE2, or an empty vector. FACS analysis revealed that PCSK9-wt and PCSK9-ACE2 not only reduced the LDLR protein level but also that of CD81 (Fig. 2). As controls, we showed that the levels of cell surface transferrin receptor (CD71) and the HCV receptor claudin-1 were not affected by PCSK9 (Supporting Fig. 1).

Figure 2.

The cell surface expression of the LDLR and CD81 is markedly decreased in HuH7 cells stably expressing PCSK9-wt and PCSK9-ACE2. Cell surface expression of LDLR and CD81 was analyzed by FACS using monoclonal antibodies for the detection of LDLR and CD81. These are representative data of at least three independent experiments.

CD81 Expression in Cells Expressing PCSK9.

To further examine the effects of PCSK9 expression, we determined the intracellular distribution of CD81 upon transient transfection of HuH7 cells with PCSK9-wt, PCSK9-ACE2, and PCSK9-Lamp1.24 We opted to perform transient transfections, as opposed to the use of stable cell lines, in order to compare side-by-side HuH7 cells with similar expression levels of the different PCSK9 forms. Under these conditions, the various forms of PCSK9 were visualized using our PCSK9 antibody at 48 hours post transfection (Fig. 3), which represent ≈30% of all HuH7 cells, a number compatible with the known low transfectability of these cells.24 In agreement with our previous data above, CD81 was completely absent in a large proportion of cells expressing either forms of PCSK9 (Fig. 3F,I,L). The loss of CD81 is dependent on PCSK9 expression because all cells transfected with the control plasmid do express CD81 (Fig. 3B). Interestingly, whereas the secretable PCSK9-wt did diminish the levels of CD81 in some neighboring cells not expressing PCSK9 (Fig. 3F), the membrane-bound forms did not (Fig. 3I,L).

Figure 3.

PCSK9 expression induced the disappearance of CD81 from HuH7 cells. PCSK9 and CD81 expression was analyzed via immunocytochemistry on a confocal microscope. HuH7 cells were transiently transfected with complementary DNAs coding for an empty vector, PCSK9-wt, PCSK9-ACE2, or PCSK9-Lamp1. HuH7 cells expressing (A,B,C) pIRES plasmid, (D,E,F) PCSK9-wt, (G,H,I) PCSK9-ACE2, and (J,K,L) PCSK9-Lamp1.

PCSK9 Inhibits Infection by HCV JFH1.

We evaluated the effect of PCSK9 on HCV propagation 3 days post infection. Immunofluorescence revealed that cells expressing PCSK9-ACE2, and to a lesser extent PCSK9-wt, had a reduced number of foci expressing NS5A compared with HuH7 cells expressing ACE2 alone or the empty vector (Fig. 4A). HuH7 cells stably expressing PCSK9-ACE2 are more resistant to HCV JFH1, because only very rare single cells were stained for NS5A, and no colony could be detected. To determine if the continuous expression of high levels of PCSK9-wt and PCSK9-ACE2 could interfere with HCV propagation, we infected different HuH7 cell populations with HCV JFH-1 for 15 days. Following normalization of the emitted chemiluminescense of the immunoreactive NS5A protein with respect to β-actin levels (Fig. 4B, inset), the data revealed that at 15 days post infection of HuH7 cells expressing PCSK9-ACE2 or PCSK9-wt, there was either no detection (<1%) or a marked reduction (<8%) of the viral nonstructural protein NS5A, respectively (Fig. 4B). Using quantitative reverse transcription polymerase chain reaction, we showed that expression of PCSK9-wt and PCSK9-ACE2 protein reduce by >90% and >99.9% the quantity of HCV genome, respectively (Fig. 4C). As a negative control, we used the full-length ACE2 protein, and show that the transmembrane domain/cytoplasmic tail domain of ACE2 in PCSK9-ACE2 was not responsible for the observed inhibition of infection (Fig. 4C). The above results demonstrate that PCSK9-wt and PCSK9-ACE2 strongly inhibit the HCV viral burden resulting from both short and long infection periods.

Figure 4.

HuH7 cells expressing PCSK9 are resistant to HCV JFH1 infection. (A) Immunofluorescence showing foci of infected cells at 72 hours post infection. Colonies were visualized via immunostaining of the NS5A protein and the nuclei were stained with DAPI. (B) Inset, western blotting showing the HCV NS5A protein and β-actin in the infected cells (15 days post infection). The quantification of the remaining NS5A protein expression in the infected HuH7 cells was calculated from the blot (inset). Infected normal HuH7 cells were arbitrary set as 100% and the NS5A expression level were normalized to the chemiluminescence of β-actin. (C) Intracellular JFH-1 RNA at 15 days post infection was calculated via quantitative polymerase chain reaction. Input RNAs were normalized to RNA 18S.

Soluble PCSK9 Inhibits HCV Infection in a Dose-Dependent Manner.

Next, we sought to confirm that the effect observed on viral replication was due to a bona fide PCSK9 protein activity and not a cell defect that appeared following the selection of the stable cellular pool. Accordingly, prior to infection, naïve HuH7 cells were preincubated with various concentrations of purified soluble PCSK9 up to 15 μg/mL (Fig. 5).27 The number of foci revealed that PCSK9 can inhibit HCV infection up to 2.5-fold (Fig. 5A) in a dose-dependent manner, demonstrating that PCSK9 is the mediator of the observed viral inhibition.

Figure 5.

Soluble purified PCSK9 inhibits HCV JFH1 infection in HuH7 cells. HuH7 cells were plated and treated with specified amounts of purified human PCSK9 (0-15 μg/mL) for 4 hours prior to infection with J6/JFH1 in the presence PCSK9. At 3 days post infection, HCV NS5A protein was stained with a rabbit polyclonal antibody. The activity of soluble PCSK9 was calculated from the colony-forming units (CFU) observed in the treated HuH7 cells as compared with mock control. These are representative data of two independent experiments performed in duplicate. Inset: Purity of soluble recombinant PCSK9 (Coomassie blue).

PCSK9-ACE2 Inhibits HCV Pseudoparticle Infection.

To confirm that CD81 is critical for the loss of infectivity in PCSK9-positive cells, we used HCV pseudoparticles (HCVpp) and VSV pseudoparticles as a control. HCVpp are produced in HEK293 cells and are devoid of lipoproteins on their surface, preventing their interaction with the LDLR.28 Therefore, HCVpp infection is thought to be mainly mediated through the CD81 receptor.5 The results presented in Fig. 6A show that HCVpp can infect normal HuH7 cells or the ACE2-expressing control cells, but not cells expressing PCSK9-ACE2 (>80% inhibition). Therefore, with HCVpp, the reduction of infectivity correlates with CD81 expression.

Figure 6.

PCSK9 blocks HCVpp infection and CD81 modulation by PCSK9 is LDLR-independent. (A) HuH7 or stable HuH7 ACE2 and HuH7 PCSK9-ACE2 cells were infected with HCVpp or VSV pseudoparticles. Luciferase activity (RLU) was calculated 72 hours post infection, and pseudoparticle infectivity was calculated relative to the infection of HuH7 cells by HCVpp, arbitrarily set at 100%. VSV pseudoparticle infections were used as positive controls. The values are the average of five independent experiments. (B) HuH7 cells were transfected in duplicate with a nonspecific siRNA as negative control (NC) or siRNAs directed against LDLR (siRNA-A and siRNA-B). Expression was verified 48 hours post transfection via western blotting. (C) HuH7 and PCSK9-ACE2 cells were transfected with siRNA-NC or siRNA-A. Expression of CD81 was verified 48 hours post transfection via western blotting. Percentages represent the remaining expression of the proteins after normalization to β-actin.

PCSK9 Removes CD81 in an LDLR-Independent Fashion.

To define the implication of the LDLR in the CD81-enhanced degradation by PCSK9, we sought to reduce the level of LDLR in PCSK9-ACE2–expressing HuH7 cells using a knockdown approach. Accordingly, two siRNAs (A and B) directed against LDLR were tested, and the results showed that only siRNA-A significantly reduced LDLR expression (>78%) (Fig. 6B). To investigate the molecular interaction between PCSK9, LDLR, and CD81, we incubated naïve HuH7 cells and HuH7 cells stably expressing PCSK9-ACE2 with siRNA-A and quantitated the levels of CD81 via western blotting (Fig. 6C). In our hands, the CD81 detected by the widely used 5A6 monoclonal antibody appeared as two bands migrating with molecular masses below ≈21 kDa, as has been observed in HuH7 cells.29, 30 Accordingly, knockdown of LDLR messenger RNA in HuH7 cells expressing PCSK9-ACE2 results in a significant reduction of immunoreactive CD81 levels (>79%), but does not affect the level of CD81 in normal HuH7 cells (Fig. 6C). These data show that the reduction of CD81 protein levels by PCSK9 does not require the presence of LDLR.

Analysis of CD81 Expression in the Livers of PCSK9 Knockout Mice.

In order to assess whether the observed ex vivo effect of PCSK9 on CD81 in HuH7 cells would be the same in vivo, we analyzed the levels of CD81 in liver sections of Pcsk9−/− mice that we recently reported.21 In Fig. 7A, western blot analysis showed that CD81 expression is substantially higher in mice that do not express PCSK9, which argues in favor of a down-regulation of liver CD81 protein levels by PCSK9. Next, we analyzed the level of both CD81 and LDLR via immunofluorescence of 10 μm-thick liver cryosections from two different mice for each genotype. The results revealed that the labeling of CD81 appeared noticeably higher over the basolateral membrane of hepatocytes, facing sinusoids, in the Pcsk9−/− mice, as compared with wild-type mice (Fig. 7B, middle panels). To further confirm that the effect of PCSK9 on CD81 is LDLR-independent, we analyzed the CD81 expression in the livers of double knockout mice lacking both PCSK9 and LDLR.21 Here also, the level of CD81 was distinctly higher in the livers of double knockout mice when compared with wild-type animals, suggesting that PCSK9 down-regulates CD81 independently from LDLR (Fig. 7B, right panels). These experiments were also repeated with a different antibody against CD81 (EAT1, 1/20, Santa Cruz Biotechnology, CA), and generated similar results (data not shown). As positive controls, we visualized LDLR in the same mice and noted that LDLR is highly expressed in the livers of Pcsk9−/− mice (Fig. 7B, left panels), as reported.21 We also showed that the liver immunoreactivity of the hepatocyte growth factor receptor and scavenger receptor class B type 1, another HCV receptor, were not significantly different in wild-type, PCSK9 knockout, and transgenic mice overexpressing PCSK9 in hepatocytes (PCSK9-Tg) (Fig. S2).21 These data extend the ex vivo effects of PCSK9 in hepatocytes toward the realm of its regulation of the level of hepatic CD81 in vivo.

Figure 7.

CD81 expression in the liver of Pcsk9 and Ldlr KO mice. (A) Representative set of data from two independent experiments for the immunodetection of CD81 in liver extracts of wild-type or PCSK9−/− mice. The CD81 expression was normalized using glyceraldehyde 3-phosphate dehydrogenase as a control. (B) LDLR and CD81 expression were analyzed via immunohistochemistry (green) in liver cryosections of wild-type, single knockout (Pcsk9−/−), and double knockout (Pcsk9−/− Ldlr−/−) mice. For each condition, at least 10 fields were compared. These experiments were repeated twice. Nuclei were stained with TO-PRO-3 (blue). Bars = 20 μm.

Discussion

PCSK9-wt, or its more active form PCSK9-ACE2, reduces the levels of membrane-bound receptor proteins such as LDLR, ApoER2, and very low-density lipoprotein receptor from the cell surface, due to its enhanced sorting toward the degradative pathway of endosomes/lysosomes.19, 24, 26 It should be noted that the enzymatic activity of PCSK9 is not required in this process, as PCSK9 remains inactive because it is tightly bound to its inhibitory pro-domain.13–16 Our data show that expression of PCSK9 reduces cell surface levels of two alleged receptors of HCV, namely CD81 and LDLR (Fig. 2). In HuH7 cells, co-immunostaining for CD81 and PCSK9 clearly indicated that the appearance of PCSK9 coincides with the complete disappearance of CD81 in most expressing cells, and that only upon overexpression of soluble, but not membrane-bound, PCSK9 are neighboring cells affected (Fig. 3), presumably by the secreted overexpressed PCSK9, which can act extracellularly.14 This result agrees with the notion that at physiological levels, endogenous PCSK9 mostly acts in an autocrine and/or paracrine fashion in liver hepatocytes.21 To our knowledge, this is the first report demonstrating that PCSK9 can down-regulate the cell surface expression of both LDLR and CD81. It is worth mentioning that the level of purified PCSK9 needed to significantly decrease the HCV titre is ≈7 μg/mL (Fig. 5), a value well above some of the reported circulating levels of PCSK9 in human (0.05-0.6 μg/mL),23 though using a different quantitation technique, a mean value of plasma PCSK9 that is ≈50-fold higher was also reported.31

The depletion of CD81 and LDLR from HuH7 cells by PCSK9 had a direct effect on the cells' susceptibility to infection by HCV (Fig. 4). It is interesting to note that the membrane-bound PCSK9-ACE2, which accumulates at the cell surface,24 was a more potent inhibitor of infection than the secreted PCSK9-wt (Fig. 4). To validate the fact that the inhibition of infection observed with the stable cell lines expressing PCSK9 can be transposed to susceptible HuH7 cells, we incubated naïve HuH7 cells with increasing doses of purified active soluble PCSK9. The results showed that soluble PCSK9, when added prior to infection, was able to block infection in a dose-dependent manner (Fig. 5). It is tempting to speculate that, if transposed to humans, elevated circulating plasma PCSK9 concentrations and/or activity due to gain of function mutations could in theory modulate liver susceptibility to HCV infection.

We next tried to determine if the presence of LDLR is required for the reduction of CD81 by PCSK9. Using an siRNA knockdown approach, we showed that the reduction of LDLR levels actually decreased, rather than increased, the level of CD81 in PCSK9-ACE2 cells, but had no effect on naïve HuH7 cells (Figs. 6B,C). This indicates that the PCSK9 effect on CD81 was not dependent on the presence of LDLR. This conclusion was further strengthened by the immunocytochemical analysis of mouse liver double knockout for PCSK9 and LDLR (Fig. 7). Indeed, hepatic CD81 expression in those mice was markedly higher than in normal mice (Fig, 7B, right panels), suggesting that the in vivo regulation of CD81 by PCSK9 is independent of LDLR. We also believe that the up-regulation of CD81 observed in Pcsk9−/− mice (Fig. 7A) established the relevance of PCSK9 as an antiviral target that requires confirmation in humans.

Recently, it has been demonstrated that the level of CD81 expression in HuH7 cells is central to infectivity and that even a moderate reduction in CD81 influences the cell susceptibility to infection by cell culture–adapted HCV.25, 32 The plasma concentration of PCSK9, which originates primarily from hepatocytes,21 greatly varies from one individual to another, due in part to genetic factors.17, 33 As a result, during the acute phase of HCV infection, the likelihood of eliminating the virus could be greater in individuals with a higher concentration and/or activity of circulating PCSK9. Furthermore, a large-scale community-based study demonstrated an association between HCV viremia and cholesterol levels in HCV patients.34 This correlation is in favor of our assumption suggesting that higher circulating levels of PCSK9 may correlate with lower viremia, a hypothesis that needs experimental validation. It must be noted, however, that high levels and/or activity of PCSK9 may result in hypercholesterolemia in view of the resulting lower hepatic levels of LDLR. Thus, a combination therapy aimed at up-regulating PCSK9 and lowering LDL cholesterol may be appropriate. This may include the combination of statins, which are known to up-regulate PCSK935 and down-regulate LDL cholesterol, and a mimic of the domain of PCSK9 that interacts with CD81.

In conclusion, this study demonstrates that hepatic PCSK9 can down-regulate in vitro two of the putative in vivo liver receptors of HCV, namely CD81 and LDLR. The fate of other suspected in vivo HCV receptors or coreceptors has not yet been determined. Because higher plasma LDL cholesterol (known to be positively regulated by PCSK9) in HCV patients correlates with better response to treatment,36 we postulate that the presumably elevated PCSK9 expression in some of these responders contributes to the elimination of the virus. Therefore, it would be quite informative to measure the levels of PCSK9 via enzyme-linked immunosorbent assay in the serum of these high responders and to correlate these values with the degree of responsiveness to antiviral drugs. Thus, measurements of plasma PCSK9 in infected individuals may guide future refinements of clinical and drug development strategies for HCV infections.

Acknowledgements

We appreciate the gifts of JFH1 and J6/JFH1 from Dr. Takaji Wakita and Dr. Ralf Bartenschlager, respectively. We thank Josée Hamelin and Edwidge Marcinkiewicz for technical assistance and Brigitte Mary for secretarial help. We also thank Drs. Rex Parker and Franck Duclos for purified PCSK9 and Dr. Francois-Loïc Cosset for pseudoparticle constructs.

Ancillary