PLK1-ELAVL1/HuR-miR-122 signaling facilitates hepatitis C virus proliferation

Significance

Millions of people suffer from chronic infection with hepatitis C virus (HCV), which causes severe liver disease, including hepatocellular carcinoma (HCC). While it is well known that the proliferation of HCV in host cells is highly dependent on the liver-specific microRNA, miR-122, the regulatory mechanisms are largely unknown. We herein identified ELAVL1/HuR as a critical regulator of miR-122, HCV proliferation, and potentially HCV-associated diseases. Our chemical library screening further revealed that the anticancer agent, rigosertib, exerts strong anti-HCV activity by targeting PLK1 to modulate ELAVL1/HuR-miR-122 signaling. Our work demonstrates that the PLK1-ELAVL1/HuR-miR-122 signaling axis is significant for HCV proliferation and suggests that this host cellular signaling could be targeted for treating HCV and its associated diseases.

Abstract

The liver-specific microRNA, miR-122, plays an essential role in the propagation of hepatitis C virus (HCV) by binding directly to the 5′-end of its genomic RNA. Despite its significance for HCV proliferation, the host factors responsible for regulating miR-122 remain largely unknown. In this study, we identified the cellular RNA-binding protein, ELAVL1/HuR (embryonic lethal-abnormal vision-like 1/human antigen R), as critically contributing to miR-122 biogenesis by strong binding to the 3′-end of miR-122. The availability of ELAVL1/HuR was highly correlated with HCV proliferation in replicon, infectious, and chronically infected patient conditions. Furthermore, by screening a kinase inhibitor library, we identified rigosertib, an anticancer agent under clinical trials, as having both miR-122-modulating and anti-HCV activities that were mediated by its ability to target polo-like kinase 1 (PLK1) and subsequently modulate ELAVL1/HuR-miR-122 signaling. The expression of PLK1 was also highly correlated with HCV proliferation and the HCV positivity of HCC patients. ELAVL1/HuR-miR-122 signaling and its mediation of PLK1-dependent HCV proliferation were demonstrated by performing various rescue experiments and utilizing an HCV mutant with low dependency on miR-122. In addition, the HCV-inhibitory effectiveness of rigosertib was validated in various HCV-relevant conditions, including replicons, infected cells, and replicon-harboring mice. Rigosertib was highly effective in inhibiting the proliferation of not only wild-type HCVs, but also sofosbuvir resistance-associated substitution-bearing HCVs. Our study identifies PLK1-ELAVL1/HuR-miR-122 signaling as a regulatory axis that is critical for HCV proliferation, and suggests that a therapeutic approach targeting this host cell signaling pathway could be useful for treating HCV and HCV-associated diseases.

MicroRNAs (miRNAs) are small noncoding RNAs that regulate gene expression by binding to complementary sites in the 3′ untranslated regions (UTRs) of target mRNAs to promote their degradation and translational repression (1, 2). miRNA-122 (miR-122) is a representative liver-specific miRNA. It is the most abundant miRNA in hepatocytes (comprising ~70% of the total miRNAs) and plays essential roles in liver development, liver metabolism, and hepatocellular carcinoma (HCC) (3). miR-122 has also been shown to be critically required for the life cycle of hepatitis C virus (HCV); this is regarded as a representative noncanonical function of miRNAs (4).

HCV is a major cause of human liver disease, and HCV infection is a major global health concern, with an estimated 130 to 170 million people infected worldwide. Chronic HCV infections are known to frequently cause liver cirrhosis and ultimately lead to HCC (5, 6). HCV belongs to the Hepacivirus genus in the Flaviviridae family and contains a single-stranded positive-sense RNA genome. The HCV genome is composed of a single open reading frame (ORF) flanked with highly structured 5′ and 3′UTRs that are important for the translation, replication, and stability of the viral RNAs. Upon entry of the virus into host cells, viral RNA is released into the cytoplasm to serve as a template for the synthesis of viral proteins (6). An internal ribosomal entry site (IRES) element located in the 5′UTR directs the translation of a single ORF to produce a ~3,000 amino acid polyprotein that is further processed into 10 individual proteins. As viral proteins (e.g., the RNA-dependent RNA polymerase, NS5B) accumulate in the host cell, negative-sense RNAs are generated and subsequently utilized to amplify the positive-sense RNAs (7, 8).

Regarding the critical involvement of miR-122 in the HCV life cycle, it is notable that, whereas conventional miRNAs bind to the 3′UTR of target mRNAs to promote their degradation and translational repression (1, 2), miR-122 binds to two adjacent target sites at the very end of the 5′UTR of the HCV genome and stimulates the accumulation of viral RNAs (3, 4). Several mechanisms have been proposed to explain this regulation. First, miR-122 reportedly stabilizes the HCV genome by protecting it from degradation mediated by the 5′-3′ exonucleases, Xrn1 and Xrn2 (9, 10). Second, some studies have suggested that miR-122 can stimulate the IRES-dependent translation of HCV RNAs (11, 12). Third, miR-122 was shown to promote the switching of HCV RNAs from translation to replication, presumably by displacing PCBP2 from the 5′UTR, and consequently enhancing the replicating pool of viral RNAs (13). Regardless of the underlying mechanism(s), efficient miR-122-mediated HCV RNA accumulation likely requires the argonaute proteins, especially Ago2, suggesting that the canonical RNA-induced silencing complex (RISC) is involved (14). However, other cellular factors required for this regulation remain unknown.

In this report, we identify the RNA-binding protein, ELAVL1 (embryonic lethal-abnormal vision-like 1)/human antigen R (HuR), as a cellular protein that interacts with the 3′-end of miR-122 and modulates its cellular level, thereby positively affecting the accumulation of HCV RNAs. We further reveal that the anticancer agent, rigosertib, is a chemical modulator whose anti-HCV effect occurs via the impact of polo-like kinase 1 (PLK1) on ELAVL1/HuR-regulated miR-122. Based on these lines of experimental evidence, we propose PLK1-ELAVL1/HuR-miR-122 signaling as a key regulatory axis for HCV proliferation.

Results

Identification of ELAVL1/HuR as a Cellular Protein with Strong Binding to miR-122.

To improve our mechanistic understanding of how miR-122 is regulated in various liver diseases, including HCV infection and HCV-associated HCC, we sought to identify cellular proteins capable of binding to miR-122. Accordingly, we developed an affinity assay for mature miR-122. As shown in Fig. 1A, the 5′-end of mature miR-122 was conjugated with a 12-carbon spacer linker and further biotinylated to enable it to be captured on an affinity column. The biotinylated miR-122 was incubated with homogenates of Huh7 cells (which exhibited relatively high-level expression of miR-122; SI Appendix, Fig. S1), and the captured cellular proteins were analyzed by streptavidin-magnetic bead pull-down.

Fig. 1.

Identification of ELAVL1/HuR as a cellular RNA-binding protein that has a strong affinity to mature miR-122. (A) Primary structure of 5′-biotinylated miR-122, experimental scheme of the affinity chromatography, and liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Protein bands specific to miR-122 were dissected and analyzed by LC-MS/MS. The identity, PSM, and molecular weight of major proteins were listed in the table (Bottom). (B) Proteome captured on 5′-biotin-miR-122 affinity column was visualized by Ponceau S staining. (C) Western blotting also confirmed that the ELAVL1/HuR was highly enriched in miR-122 affinity column. Ago2 was used as a positive binding control. Dicer, β-actin (ACTB), and GAPDH were used as negative binding controls. (D) Association between ELAVL1/HuR and miR-122 was confirmed with ribonucleoprotein-immunoprecipitation (RNP-IP) northern blotting using ELAVL1/HuR-specific antibody in Huh7 and Huh7.5.1 cells. 5S rRNA and U6 spliceosomal RNA (U6) were used as controls. Immunoprecipitated endogenous mature miR-122 was indicated by an arrow (lanes α-HuR-IP). (E) Sequence of four major miR-122 isomiRs and miR-21. (F) Binding affinity of the purified recombinant ELAVL1/HuR (GST-HuR) to the miR-122 isomiRs was analyzed by REMSA. 5′-³²P-end-labeled RNA-protein complexes were separated on nondenaturing 5% polyacrylamide gel electrophoresis, and the signals were quantified by phosphor imaging. Signal intensities of REMSA from two independent experiments were averaged and shown as % (100% in lane 5). ND, not determined; Free, free probes.

Our RNA pull-down assay and subsequent proteomic analysis enabled us to identify a handful of cellular proteins that appeared to bind miR-122 (listed in Fig. 1A). Among them, ELAVL1/HuR had the highest peptide-spectrum match (PSM) value, suggesting that it had the strongest association with miR-122. The strong association of ELAVL1/HuR with miR-122 was confirmed by western blotting of the sample used in mass spectrometry analysis, which showed that ELAVL1/HuR was highly enriched on the miR-122-affinity column (Fig. 1 B and C). In assay-validating experiments, we confirmed that Ago2, a core protein of RISC, also bound to miR-122, as previously reported (14). ELAVL1/HuR is a member of the essential RNA-binding proteins and is responsible for regulating various posttranscriptional processes, including RNA stability and mRNA translation, via binding to A/U-rich elements. Several reports proposed that there may be a functional link between ELAVL1/HuR and miRNAs (15), but the direct impact of ELAVL1/HuR on the functions of miR-122 had not previously been explored. We, therefore, focused on analyzing ELAVL1/HuR in this context.

As shown in Fig. 1 B and C, ELAVL1/HuR was found to be physically associated with miR-122 in cell lysates. The association between endogenous ELAVL1/HuR and miR-122 was also confirmed by performing ribonucleoprotein-immunoprecipitation (RNP-IP) northern blotting with Huh7 and Huh7.5.1 cell extracts and an ELAVL1/HuR-specific antibody. As shown in Fig. 1D, RNP-IP with an ELAVL1/HuR-specific antibody, but not control normal mouse IgG, coprecipitated notable amounts of miR-122 from both cell lines. Furthermore, a significant amount of miR-122 was coprecipitated by overexpressed FLAG-HuR, whereas control beads generated only a marginal signal (SI Appendix, Fig. S2A).

Based on these findings, we next investigated whether there is a direct interaction between ELAVL1/HuR and miR-122. Recent studies showed that there are many different isomiRs of miR-122 present in different proportions (Fig. 1E). We chose four major isomiRs of human miR-122 with different 3′-ends and performed RNA electrophoresis mobility shift assay (REMSA) with radiolabeled isomiR probes and purified recombinant GST-HuR proteins. Our results revealed that the 22-nucleotide (nt) miR-122, which is the most abundant isomiR, generated a strongly shifted band that most likely corresponded to 5′-³²P-end-labeled-miR-122-bound GST-HuR, and the band intensity increased with the concentration of GST-HuR (Fig. 1F). The intensity of the shifted band was gradually decreased by an increasing amount of cold miR-122 (SI Appendix, Fig. S3), indicating that the latter competed against radiolabeled-miR-122 for binding to GST-HuR. Together, these results clearly demonstrated that ELAVL1/HuR directly binds to miR-122. The overall intensities of the 23-nt 3′ A- or 3′ U-tailed isomiR probes and miR-21, which was used as a negative control, bound to GST-HuR were much lower than that of the 22-nt miR-122, and the isomiR lacking the 3′-end G had no GST-HuR-bound probe signal (Fig. 1F). These results indicate that the 3′-end of mature miR-122 is crucial for its binding to ELAVL1/HuR.

ELAVL1/HuR Positively Regulates the Expression of miR-122 at the Posttranscriptional Level.

To determine the effect of ELAVL1/HuR on miR-122, we examined the fate of miR-122 when ELAVL1/HuR was depleted from cells using two different short hairpin RNAs (shRNAs). We first confirmed the efficient depletion of ELAVL1/HuR in Huh7 cells by western blotting (Fig. 2B). Total RNAs were then extracted, and the levels of miR-122 and pri-miR-122 were analyzed using northern blotting and qRT-PCR, respectively. Intriguingly, depletion of ELAVL1/HuR dramatically decreased miR-122 (by ~90% relative to the control), but only marginally altered the level of pri-miR-122 (Fig. 2 A and C). This suggested that ELAVL1/HuR positively regulates miR-122 biogenesis, likely after the transcription of pri-miR-122. Similar results were obtained when small-interfering RNA (siRNA) was used to deplete ELAVL1/HuR (SI Appendix, Fig. S4). The downregulation of miR-122 following depletion of ELAVL1/HuR was also observed in another hepatoma cell line, PLC/PRF/5, which exhibited a moderate expression of miR-122 (SI Appendix, Figs. S1 and S5). This indicates that the effect of ELAVL1/HuR on miR-122 biogenesis is not cell-line specific.

Fig. 2.

ELAVL1/HuR modulates the expression of miR-122 at the posttranscriptional level. (A) Huh7 cells were infected with lentivirus expressing ELAVL1/HuR-specific shRNAs, and total RNAs were prepared and subjected to northern blotting. Mature miR-122 was quantified by phosphor imaging of the northern blotting in triplicate. Error bars in the graph represent ± SD, and the P-values compare each shRNA of ELAVL1/HuR (shHuR-1 and shHuR-2) to the control (shCon-1). quant., quantification; ***P < 0.001. (B) The level of ELAVL1/HuR was analyzed by western blotting after specific knockdown of ELAVL1/HuR in Huh7 cells. Eukaryotic initiation factor 4GI (4GI), symplekin (Sym), ACTB, and Lin28B were used as loading controls. (C and F) Expression level of pri-miR-122 RNA and control ACTB mRNA was analyzed by qRT-PCR (Left) and semi-quantitative RT-PCR (sqRT-PCR, Right). exp., expression; NS, not significant. (D) Overexpression of ELAVL1/HuR enhanced the level of mature miR-122. GFP-HuR was overexpressed in Huh7 cells, and mature miR-122 was quantified via northern blotting in triplicate. **P < 0.01. (E) The expression of GFP and GFP-HuR was monitored by western blotting.

The effect of ELAVL1/HuR on miR-122 biogenesis was further examined under the overexpression condition. For this purpose, we constructed a plasmid encoding GFP-HuR and used it to transiently transfect Huh7 cells. As shown in Fig. 2 D and E, overexpression of GFP-HuR markedly enhanced the level of mature miR-122 compared with that of the GFP control, and thus had an effect opposite to that of ELAVL1/HuR depletion. Similar results were observed when GFP-HuR was overexpressed with a pri-miR-122 minigene in HEK293T cells (SI Appendix, Fig. S6). This positive effect of GFP-HuR on the level of miR-122 appeared to occur at the posttranscriptional level, since overexpression of ELAVL1/HuR had no appreciable effect on the level of pri-miR-122 in the same RNA samples (Fig. 2F).

In addition to the results obtained from our ELAVL1/HuR depletion and overexpression experiments, those from miR-122 sensor-based experiments (SI Appendix, Figs. S7–S10) also indicated that ELAVL1/HuR positively impacts on miR-122 biogenesis at the posttranscriptional level. A proposed mechanism by which ELAVL1/HuR may affect miR-122 stability is described in the Discussion section.

ELAVL1/HuR Regulates HCV Proliferation by Modulating miR-122 Biogenesis.

Previous reports demonstrated that miR-122 is critically required for HCV proliferation (3, 4, 10–14, 16, 17), and the present work shows that ELAVL1/HuR plays an essential role in miR-122 biogenesis. We, therefore, speculated that ELAVL1/HuR could be linked to HCV proliferation via miR-122. To examine this possibility, we utilized HCV subgenomic replicon (HIRL-2Aneo-NSrep)-harboring Huh7 cells that enabled us to easily measure HCV proliferation. This HCV replicon contains the Renilla luciferase gene where viral structural genes would normally reside, and thus its activity represented the extent of viral proliferation in Huh7 cells (Fig. 3A) (18). Intriguingly, the shRNA-mediated depletion of ELAVL1/HuR yielded a dramatic reduction of Renilla luciferase activity (Fig. 3 B, Left). Similar decreases of HCV replicon gRNAs and viral proteins (NS3, NS5A, and NS5B) were observed by qRT-PCR analysis and western blotting, respectively (Fig. 3 B, Right and Fig. 3C).

Fig. 3.

ELAVL1/HuR-miR-122 signaling facilitates HCV proliferation. (A) Structure of HCV subgenomic replicon (HIRL-2Aneo-NSrep, genotype 1b) that was used for quantitatively measuring HCV proliferation in Huh7 cells. The Renilla luciferase (Rluc.) is directly fused with foot-and-mouth disease virus 2A peptide (2A)—neomycin-resistant gene (Neo) cassette. (B) ELAVL1/HuR was specifically depleted with shRNAs in HIRL-2Aneo-NSrep Huh7 cells and the Renilla luciferase activities were measured (Left). Levels of HCV genomic RNA (HCV gRNA) and ELAVL1/HuR mRNA (HuR mRNA) in ELAVL1/HuR-depleted cells were also monitored by qRT-PCR (Right). Data represent the mean values of three independent experiments. Error bars in the graph represent ± SD, and the P-values compare each shRNA of ELAVL1/HuR (shHuR-1 and shHuR-2) to the control (shCon-1). ***P < 0.001. (C) Expression of HCV nonstructural proteins (NS3, NS5A, and NS5B) and ELAVL1/HuR protein was monitored in ELAVL1/HuR-depleted HIRL-2Aneo-NSrep Huh7 cells by western blotting. (D) The level of miR-122 was analyzed in ELAVL1/HuR-depleted HIRL-2Aneo-NSrep Huh7 cells. Mature miR-122 was quantified by phosphor imaging of the northern blotting in triplicate. Error bars in the graph represent ± SD, and the P-values compare each shRNA of ELAVL1/HuR (shHuR-1 and shHuR-2) to the control (shCon-1). ***P < 0.001. (E) Inhibitory effect of ELAVL1/HuR knockdown on HCV proliferation was rescued by supplementing miR-122. ELAVL1/HuR was depleted by ELAVL1/HuR-specific shRNA (shHuR-1) in HIRL-2Aneo-NSrep/FL (SI Appendix, Fig. S18A) Huh7 cells. The relative luciferase activities were quantitatively measured after transfecting control (NC-mimic) or miR-122 mimic (122-mimic) RNA oligonucleotides. The knockdown of ELAVL1/HuR was also confirmed by western blotting. Fluc., ACTB, GAPDH, and H3 were used as loading controls. The data represent the mean values of three independent experiments. For the evaluation of statistical significance, the one-way analysis of variance (ANOVA) with Tukey’s multiple comparison test was performed, and the P-value was determined by comparison of activities among groups. NS, not significant; *P < 0.05, ***P < 0.001. (F) Schematic depiction of subgenomic replicon (HIFL-NSrep) of wild-type (WT) and G28A adaptive mutant HCV (genotype 2a) that were used for quantitatively measuring the HCV proliferation. Position of G28 residue is indicated by solid arrow. (G) The proliferation of low miR-122-dependent G28A mutant of HCV was less affected by the depletion of ELAVL1/HuR than that of WT-HCV. WT and G28A mutant replicon RNAs (HIFL-NSrep) were primarily transfected into Huh7.5.1 cells. Luciferase activities were measured after sequential transfection of control or ELAVL1/HuR-specific siRNA. The knockdown of ELAVL1/HuR was confirmed by western blotting. Sym, PTB-associated splicing factor (PSF), and H3 were used as loading controls. Data represent the mean values of three independent experiments. The P-values compare firefly luciferase activity of each siRNA for ELAVL1/HuR (siHuR) to control (siCon), and the siHuR transfections between WT and G28A. ***P < 0.001. (H) Two ELAVL1/HuR-KO Huh7 cells (KO #14 and #18) were established using CRISPR/Cas9 system and confirmed via western blotting. (I) HCV replicons were transfected into the ELAVL1/HuR-KO Huh7 cells, and HCV proliferation was monitored. The firefly luciferase activities from the WT and ΔGDD mutant of HIFL-NSrep (genotype 2a) (12) were monitored after transfecting replicon RNAs into parental and two ELAVL1/HuR-KO (KO #14 and #18) Huh7 cells. Luciferase activities are presented as log10 scale. (J) Parental and two ELAVL1/HuR-KO Huh7 cells were infected with cell culture-derived HCV (HCVcc). Intracellular HCVcc RNA levels were analyzed by qRT-PCR. intra., intracellular; Un, uninfected; ND, not detected. (K) Control or ELAVL1/HuR-specific siRNA was transfected into Huh7.5.1 cells, and sequential infection with WT or G28A mutant of HCVcc was performed. Intracellular HCVcc RNA levels were analyzed by qRT-PCR. ND, not detected; **P < 0.01, ***P < 0.001. (L) Expression of ELAVL1/HuR proteins in HCV replicon-transfected Huh7 cell in Fig. 3I was monitored by western blotting at indicated time points (9, 24, and 48 h). NS5A expression represents HCV proliferation rate. PSF and ACTB were used as loading controls. (M) ELAVL1/HuR protein was highly expressed in liver biopsy of HCC patients with chronic HCV infection. Lanes 1-2, HCV-negative; lanes 3-4, HCV-positive HCC. GAPDH was used as a loading control. (N) The TMA slide including HCV-positive HCC (9 cases in duplicate) as well as normal tissues (10 cases in duplicate) was immunohistologically stained with ELAVL1/HuR-specific antibody (3A2). Strong staining patterns of ELAVL1/HuR were observed in HCV-positive HCC cores and the representative images are shown (Left). (Scale bar, 20 µm.) In the graph (Right), blue bars represent total percentage of each group in the TMA slide, and red bars represent percentage of ELAVL1/HuR-positively stained cores.

Consistent with the abovementioned results (Fig. 2A), ELAVL1/HuR depletion significantly reduced miR-122 in HCV replicon-harboring Huh7 cells (Fig. 3D). Importantly, the extent of ELAVL1/HuR depletion was highly correlated with the levels of miR-122, viral RNAs, viral proteins, and Renilla luciferase activity (compare Fig. 3 B, C, and D). shHuR-1, which yielded a larger reduction of ELAVL1/HuR than shHuR-2, more severely decreased the miR-122 level and viral indices. Similar reductions of miR-122 and HCV replicon activity were also observed when ELAVL1/HuR was depleted by siRNAs (SI Appendix, Fig. S12). These results strongly suggest that ELAVL1/HuR plays a positive and miR-122-mediated role in HCV proliferation.

To clarify the ability of miR-122 to mediate the positive effect of ELAVL1/HuR on HCV proliferation, we examined the effect of miR-122 supplementation on HCV proliferation in ELAVL1/HuR-depleted Huh7 cells stably harboring both HCV replicon and the firefly luciferase (HIRL-2Aneo-NSrep/FL). When miR-122-mimic RNA oligonucleotide (as a supplement) was transfected into ELAVL1/HuR-depleted cells, the Renilla luciferase activity that had been decreased by ELAVL1/HuR depletion was significantly restored to that of control cells, indicating that miR-122 supplementation considerably rescued viral proliferation (Fig. 3E). Similar results were observed when ELAVL1/HuR was depleted by siRNAs (SI Appendix, Fig. S14).

To further validate the miR-122 dependency of ELAVL1/HuR on HCV proliferation, we introduced a G28A mutation, which has been reported to have low miR-122 dependency (19), into the HCV replicon (HIFL-NSrep, genotype 2a) (Fig. 3F). We then examined the effect of siRNA-mediated ELAVL1/HuR depletion in HCV replicon-harboring Huh7.5.1 cells, which are highly permissive to HCV proliferation. ELAVL1/HuR depletion led to a strong suppression of WT-HCV replicon activity (firefly luciferase) and a relatively moderate, still significant suppression of G28A mutant (Fig. 3G). Substantial difference in the response to ELAVL1/HuR depletion between WT and G28A-HCV may result from miR-122-dependency on HCV proliferation. Together, these results clearly demonstrated that miR-122 plays a significant role in mediating the positive effect of ELAVL1/HuR on HCV proliferation.

This positive effect was further confirmed in ELAVL1/HuR-null Huh7 cells generated using a CRISPR/Cas9-mediated knockout (KO) approach. We used ELAVL1/HuR-specific guide RNAs and established two ELAVL1/HuR-null Huh7 clones (KO #14 and #18). The KO of ELAVL1/HuR was confirmed by western blotting (Fig. 3H) and genomic DNA sequencing (SI Appendix, Fig. S15). Parental or ELAVL1/HuR-null Huh7 cells were transfected with HCV replicon RNAs encoding firefly luciferase (as an indicator of HCV proliferation), and firefly luciferase activity was measured at various time points posttransfection. To clarify the HCV replication activity, an HCV replicon containing a glutamate-aspartate-aspartate (GDD) deletion mutation in the catalytic core of NS5B replicase (ΔGDD), which was defective in replication, was used as a negative control (Fig. 3I) (12).

In parental Huh7 cells, the HCV replicon RNAs showed the expected pattern of firefly luciferase activity over time, reflecting mostly viral translation at the early stage and viral replication after 9 h posttransfection. In contrast, the firefly luciferase activity from the ΔGDD-mutant replicon was dramatically decreased after 9 h posttransfection, indicating a lack of viral replication. Comparison between WT and ΔGDD-mutant replicons allowed us to differentiate the replication and translation of viral RNAs. To our surprise, the replication of WT-HCV replicon RNAs was completely abrogated in both of the ELAVL1/HuR-null Huh7 cell clones (WT in KO #14 and #18 cells in Fig. 3I), which exhibited luciferase activity patterns that were almost identical to those of the ΔGDD mutant in the same cells (see WT and ΔGDD in KO #14 and #18 cells) and much lower than those in parental cells. The luciferase activity during the early phase was also low (0.5~3-fold increase between time 0 and 3 h, log10 scale in graph) in the null cells compared to that of the parental cells (~30-fold increase), further indicating that the translation of replicon RNAs was severely impaired. Similar results were obtained in HCV-infected Huh7 cells. Parental or ELAVL1/HuR-null Huh7 cells were infected with HCVcc (20), and qRT-PCR was performed to monitor intracellular viral RNAs. Negligible HCV RNAs were detected in ELAVL1/HuR-null cells, whereas considerable levels were detected in parental cells (Fig. 3J), demonstrating that ELAVL1/HuR plays a crucial role in HCV proliferation. The miR-122 dependency of the positive effect of ELAVL1/HuR on HCV proliferation was also confirmed in Huh7.5.1 cells infected with WT- and G28A-HCVcc (Fig. 3K and SI Appendix, Fig. S29).

Moreover, we observed that transfection of HCV replicon RNAs could up-regulate the expression of ELAVL1/HuR in Huh7 cells (Fig. 3L). The time-dependent increase of ELAVL1/HuR coincided well with that of the HCV NS5A protein; the levels of both proteins were markedly increased at 48 h posttransfection. Importantly, significantly higher levels of ELAVL1/HuR protein were found in the liver tissues of HCV-positive HCC patients compared with those of HCV-negative individuals (Fig. 3M). Similar results were obtained when we performed immunohistochemistry (IHC) on an HCV-positive HCC liver tissue microarray (TMA) (Fig. 3N). More than 60% of the HCV-positive HCC liver tissue samples in the microarray were ELAVL1/HuR-positive, whereas no normal sample was positive. These data collectively demonstrate that ELAVL1/HuR acts as an essential factor for the HCV proliferation through regulating miR-122, and also suggest that it may be involved in HCV-associated diseases.

The PLK1 Inhibitor, Rigosertib, Exerts Anti-HCV Activity by Modulating miR-122.

In an effort to identify chemical agent(s) that inhibit HCV by modulating miR-122 biogenesis and examine the signaling pathway responsible for regulating miR-122, we screened a kinase inhibitor library in Huh7 cells dually harboring the HCV replicon and the miR-122 sensor (HIRL-2Aneo-NSrep/FL-5×122) (Fig. 4A). In this HCV/miR-122 dual-sensing system, a compound that negatively affects miR-122 biogenesis should increase miR-122 sensor (firefly luciferase) activity and decrease HCV replicon (Renilla luciferase) activity. This HCV/miR-122 dual-sensing system properly responded to the miR-122 mimic, anti-miR-122, and siHuR RNA oligonucleotides (SI Appendix, Fig. S17), and thus appeared to be a reliable assay system. For the kinase inhibitor library screen, 278 well-characterized kinase inhibitors were applied to the dual-sensing system.

Fig. 4.

Rigosertib, a specific inhibitor of PLK1, suppresses HCV proliferation via modulating ELAVL1/HuR-miR-122 signaling. (A) Schematic depiction of HCV/miR-122 dual-sensing system (HIRL-2Aneo-NSrep/FL-5×122) that monitors HCV proliferation and miR-122 biogenesis simultaneously. In this system, HCV proliferation and miR-122 biogenesis are monitored by luciferase activities of Renilla and firefly, respectively. (B) Screening of the kinase inhibitor library using the dual-sensing system to identify modulator(s) of HCV proliferation as well as miR-122 biogenesis. Firefly luciferase (miR-122 sensor; Top) and Renilla luciferase (HCV replicon; Bottom) activities of cells treated with #62, #63, and #65 compounds were shown as the green- and orange-colored columns, respectively. Danoprevir (Dano.) was used as a positive control. (C) The dual-sensing system was treated with rigosertib; Renilla luciferase (Left) and firefly luciferase (Right) activities were dose-dependently down-regulated and up-regulated, respectively. Each activity was normalized by luciferase activity of DMSO-treated cells. (D) The levels of NS5A and ELAVL1/HuR proteins were monitored by western blotting after treatment of rigosertib. Sym was used as a loading control. (E) The expression of miR-122 was analyzed by qRT-PCR after treatment of rigosertib on the dual-sensing system. ***P < 0.001. (F) HCVcc-infected Huh7 cells were treated with rigosertib in various doses (0, 0.125, 0.25, 0.5, and 1 µM). Intracellular HCVcc gRNA levels were analyzed by qRT-PCR. ND, not detected. (G) The effect of rigosertib on HCV proliferation was monitored in vivo mouse model. The IVIS luminescence images of Renilla luciferase activities represent HCV proliferation (HIRL-2Aneo-NSrep). Relative quantification results of those activities were shown in graphs. **P < 0.01.

As shown in Fig. 4B, we identified a group of kinase inhibitors that significantly decreased HCV replicon activity and concomitantly increased miR-122 sensor activity in our system. They were all PLK1 inhibitors; among them, compound #65 had the strongest (and somewhat inversely correlated) effects on HCV replicon and miR-122 sensor activity. This compound corresponded to the chemical inhibitor, rigosertib, which is an anticancer drug candidate currently in clinical trials (SI Appendix, Fig. S18B) (21). Rigosertib exhibited a strong HCV-inhibiting activity with an estimated half-maximal inhibitory concentration (IC₅₀) of 0.1 to 0.5 μM; this is comparable to that of sofosbuvir, an HCV NS5B-selective inhibitor with FDA approval, which is currently used as an anti-HCV drug (Fig. 4C and SI Appendix, Fig. S19) (22). Notably, rigosertib significantly and dose-dependently stimulated the miR-122-dependent sensor as well (Fig. 4C), while sofosbuvir did not increase the activity of the miR-122 sensor (SI Appendix, Fig. S23C). The potent effect of rigosertib was further confirmed in Huh7 cells harboring HCV replicon and firefly luciferase (HIRL-2Aneo-NSrep/FL) (SI Appendix, Fig. S18C). In an effort to correlate the effect of rigosertib with the ELAVL1/HuR-mediated regulation of miR-122, we examined the impact of rigosertib treatment on the cellular levels of ELAVL1/HuR and miR-122. As expected, rigosertib at effective doses considerably decreased the levels of not only ELAVL1/HuR and HCV NS5A (Fig. 4D), but also mature miR-122 (Fig. 4E and SI Appendix, Fig. S20).

The HCV-inhibitory activity of rigosertib was further examined in HCVcc-infected Huh7 cells. Rigosertib significantly and dose-dependently decreased the level of intracellular HCVcc RNAs (Fig. 4F), with an inhibitory efficacy quite comparable to those obtained from HCV replicon activity, and the modulations of ELAVL1/HuR and miR-122 (Fig. 4 C–E). Moreover, the effectiveness of rigosertib was validated in HCV replicon-harboring mice. As adopted in other studies (23–25), HCV replicon-harboring Huh7 cells were subcutaneously implanted into immunocompromised mice, and rigosertib or vehicle (DPBS) was intraperitoneally administered. The bioluminescence images were acquired prior to and after rigosertib treatment. The Renilla luminescence (reflecting HCV proliferation) from mice treated with rigosertib was considerably decreased by ~40% of the control activity, while that from vehicle-treated mice was slightly increased (Fig. 4G). The firefly luminescence (reflecting the miR-122 activity) was increased, indicating that the level of miR-122 was decreased (SI Appendix, Fig. S21). Similarly, endogenous miR-122 levels in rigosertib-treated mouse liver were also lower (~40%) than that in control (SI Appendix, Fig. S22). Together, these results demonstrated that rigosertib is effective as an anti-HCV agent and suggested that its effect might be mediated by ELAVL1/HuR-regulated miR-122.

PLK1 Regulates HCV Proliferation via Targeting ELAVL1/HuR-miR-122 Signaling.

In addition to rigosertib, compounds #62 and #63 also showed strong HCV-inhibiting and miR-122 sensor-stimulating effects, respectively (Fig. 4B). As all the three compounds are PLK inhibitors that majorly target PLK1 (26), we speculated that PLK1 may play a positive role in HCV proliferation through regulating the ELAVL1/HuR-miR-122 axis. To test this hypothesis, we generated four shRNAs specific for PLK1 and used them to deplete PLK1 from assayable HCV replicon-harboring Huh7 cells. As shown in Fig. 5 A and B, depletion of PLK1 dramatically decreased the luciferase activity and HCV NS5A protein level. Under the same conditions, the levels of ELAVL1/HuR and miR-122 were also significantly reduced (Fig. 5 B and D). Importantly, the extent of PLK1 depletion was highly correlated with the decreases seen in the levels of luciferase activity, NS5A, ELAVL1/HuR, and miR-122. We next examined the role of PLK1 in HCV and the ELAVL1/HuR-miR-122 axis in overexpression conditions. As expected, based on the results of our PLK1 depletion experiments, overexpression of WT-PLK1 resulted in noticeable increases of ELAVL1/HuR and HCV NS5B (Fig. 5E) and considerable decrease of miR-122 sensor activity (firefly luciferase) and increased HCV activity (Renilla luciferase) in dual sensor-harboring Huh7 cells (SI Appendix, Fig. S28). Moreover, PLK1 depletion had little or no effect on the cellular level of ELAVL1/HuR mRNAs (Fig. 5C). Overexpressed FLAG-PLK1 proteins were shown to physically associate with endogenous ELAVL1/HuR proteins (Fig. 5F). A similar interaction was observed when FLAG-HuR was overexpressed (SI Appendix, Fig. S24). Together, these data strongly indicated that PLK1 is an essential factor for HCV proliferation, suggesting that it acts upstream of the ELAVL1/HuR-miR-122 axis and regulates ELAVL1/HuR at the posttranscriptional and/or posttranslational level.

Fig. 5.

PLK1 regulates HCV proliferation via targeting ELAVL1/HuR-miR-122 signaling. (A) Huh7 cells harboring HCV replicon (HIRL-2Aneo-NSrep/FL) were infected with lentivirus expressing PLK1-specific shRNAs and the luciferase activities were measured. Data represent the mean values of three independent experiments. The P-values compare each shRNA for PLK1 to control (shCon-2). **P < 0.01, ***P < 0.001. (B) Expression of PLK1, NS5A, and ELAVL1/HuR protein was monitored in PLK1-depleted HIRL-2Aneo-NSrep/FL Huh7 cells by western blotting. PSF and ACTB were used as loading controls. (C) ELAVL1/HuR mRNA levels were also analyzed by qRT-PCR. The results were normalized with those of ACTB mRNA. (D) The level of miR-122 was analyzed with northern blotting in PLK1-depleted HIRL-2Aneo-NSrep/FL Huh7 cells. (E) WT and catalytically inactive mutant (K82M) of FLAG-PLK1 were overexpressed in HIRL-2Aneo-NSrep/FL Huh7 cells; ELAVL1/HuR and NS5B proteins were analyzed by western blotting. (F) PLK1 associates with ELAVL1/HuR. FLAG-PLK1 was overexpressed in HEK293T cells, and the physical interaction with ELAVL1/HuR was analyzed by immunoprecipitation. TUT4, STAT3, β-tubulin, and ACTB were used as negative binding controls. (G) Inhibitory effect of PLK1 knockdown on HCV proliferation was rescued by supplementing miR-122. PLK1 was depleted with PLK1-specific shRNA (shPLK1-1) in HIRL-2Aneo-NSrep/FL Huh7 cells. The relative luciferase activities were quantitatively measured after transfecting control (NC-mimic) or miR-122 mimic (122-mimic) RNA oligonucleotides. The knockdown of PLK1 was also confirmed by western blotting. Fluc. and H3 were used as loading controls. The data represent the mean values of three independent experiments. The ANOVA with Tukey’s multiple comparison test was performed to compare activities among groups. NS, not significant; **P < 0.01. (H) The proliferation of low miR-122-dependent G28A mutant of HCV was less affected by the depletion of PLK1 than that of WT-HCV. WT and G28A mutant replicon RNAs (HIFL-NSrep) were primarily transfected into Huh7.5.1 cells. Luciferase activities were measured after sequential transfection of control (siNC) or PLK1-specific siRNAs. The PLK1 knockdown was confirmed by western blotting. Sym and H3 were used as loading controls. Data represent the mean values of three independent experiments. The P-values compare firefly luciferase activity of each siRNA for PLK1 (siPLK1) to control (siNC), and the siPLK1 transfections between WT and G28A. **P < 0.01, ***P < 0.001. (I) PLK1-depleted Huh7 cells were infected with HCVcc and intracellular HCVcc gRNA levels were analyzed by qRT-PCR. ND, not detected. (J) Protein expression of PLK1 in HCV replicon-transfected Huh7 cell in Fig. 3I was monitored by western blotting at indicated time points (9, 24, and 48 h). H3 was used as a loading control. (K) The same TMA slide used in Fig. 3N was also immunohistologically stained with PLK1-specific antibody. (Scale bar; 50 µm.) In the graph (Right), blue bars represent total percentage of each group in the TMA slide, and red bars represent percentage of PLK1-positively stained cores.

To further clarify the PLK1-ELAVL1/HuR-miR-122 signaling axis, we examined the effect of miR-122 supplementation on HCV proliferation in PLK1-depleted HCV replicon-harboring Huh7 cells. Our results revealed that supplementation of an miR-122-mimic RNA oligonucleotide almost completely rescued the viral proliferation impairment induced by shRNA-mediated PLK1 depletion (Fig. 5G). Similar results were obtained when PLK1 was depleted by siRNAs (SI Appendix, Fig. S25). This HCV proliferation-rescuing effect was also observed when GFP-HuR was supplemented to PLK1-depleted cells (SI Appendix, Fig. S26). To further validate the miR-122 dependency of the positive effect of PLK1 on HCV proliferation, we also utilized a G28A mutant HCV replicon and examined the effect of siRNA-mediated PLK1 depletion in HCV-permissive Huh7.5.1 cells. We observed that PLK1 depletion led to a strong suppression of WT-HCV replicon activity (firefly luciferase) and a relatively moderate, still significant suppression of G28A mutant (Fig. 5H). Together, these results show that ELAVL1/HuR and miR-122 play significant roles in mediating the essential effect of PLK1 on HCV proliferation. The crucial effect of PLK1 was further confirmed by our observation that intracellular viral RNAs were barely detected in PLK1-depleted Huh7 cells infected with HCVcc (Fig. 5I).

We also found that transfection of HCV replicon RNAs could up-regulate the expression of PLK1 together with ELAVL1/HuR (Fig. 5J). Prominent increases of both proteins were observed at 48 h posttransfection. Moreover, similar to our findings for the ELAVL1/HuR protein (Fig. 3N), the PLK1 protein was highly expressed in the liver tissues of HCV-positive HCC patients compared to those of normal controls in liver TMA IHC results (Fig. 5K). About 78% of HCV-positive HCC liver tissue samples in the microarray were PLK1 positive, whereas no normal sample was PLK1 positive. These data collectively demonstrated that PLK1 acts as an essential factor for HCV proliferation through regulating the ELAVL1/HuR-miR-122 axis and also suggested that it may play a role in HCV-associated diseases.

Discussion

In this study, we sought to identify cellular proteins that are crucial for the regulation of miR-122, which is a significant factor for HCV proliferation and normal functions of liver. Using the unbiased methodology of an miRNA-capture pull-down assay and proteomic analysis, we identified ELAVL1/HuR as the major binding protein of miR-122 (Fig. 1). Further experiments revealed that there was a tight correlation between ELAVL1/HuR protein availability and the miR-122 level (Fig. 2), indicating that they are strongly linked. We then provided several lines of compelling evidence suggesting that ELAVL1/HuR critically regulates the stability of miR-122.

First, we found that ELAVL1/HuR directly bound to mature miR-122 in vitro, and the 3′-end region was critical for this binding. In a comparative analysis of various miR-122 isomiRs with different ends, only 22-nt miR-122 with a 3′-end G allowed the efficient binding of ELAVL1/HuR (Fig. 1F). Second, depletion of ELAVL1/HuR facilitated the decay of miR-122 in vitro and in vivo (SI Appendix, Fig. S11), indicating that ELAVL1/HuR contributes to the stability of miR-122. The ability of ELAVL1/HuR to bind the 3′-end of miR-122 and its positive impact on the stability of miR-122 lead us to speculate that ELAVL1/HuR may physically mask the 3′-end of miR-122 to protect it from attack by cellular 3′-5′ exoribonuclease(s), such as poly(A)-specific ribonuclease (27).

It is well known that miR-122 critically contributes to HCV proliferation by directly binding to two consecutive target sites at the very 5′-end of the HCV genome. Therefore, we expected that ELAVL1/HuR, which we herein identified as a crucial regulator of miR-122 biogenesis, would have a significant effect on HCV proliferation. Indeed, we observed a tight correlation between the availability of ELAVL1/HuR and the activity of HCV proliferation (Fig. 3 B and C). The level of miR-122 was also highly correlated with that of ELAVL1/HuR under the same condition (Fig. 3D), confirming that ELAVL1/HuR regulates miR-122 biogenesis. Importantly, we further showed that miR-122 plays an important role in mediating the pro-HCV effect of ELAVL1/HuR by observing that the proliferation of an HCV replicon, which was suppressed by ELAVL1/HuR depletion, was considerably rescued by exogenously supplemented miR-122-mimic RNA oligonucleotides (Fig. 3E). We observed that the proliferation of a G28A mutant HCV replicon (which could be miR-122 independent or less dependent) was less affected by ELAVL1/HuR depletion than WT-HCV, offering further evidence supporting our contention (Fig. 3 F and G). Moreover, we observed that ELAVL1/HuR formed a complex with miR-122 and the HCV RNA (SI Appendix, Fig. S16). Together, these results supported the idea that ELAVL1/HuR facilitates HCV proliferation by stabilizing and increasing the level of miR-122.

Beyond the ability of ELAVL1/HuR to directly regulate miR-122 biogenesis, ELAVL1/HuR may have additional effects on miR-122-mediated HCV RNA replication, translation, and/or stabilization. The only published study to examine the function of ELAVL1/HuR with respect to the 5′UTR of the HCV RNA found that ELAVL1/HuR positively regulates the IRES-dependent translation of the HCV polyprotein, and this does not require direct binding of the protein to the IRES element (28). Thus, we speculate that ELAVL1/HuR in complex with miR-122 might be an effector for translational control. This scenario is actually evidenced by the observation that ELAVL1/HuR moderates the miR-122-mediated suppression of CAT-1 mRNA translation at stressed conditions (29).

Another published study reported a 3′UTR-mediated function of ELAVL1/HuR. Shwetha and colleagues showed that ELAVL1/HuR directly bound to the 3′UTR of the HCV RNA and modulated the initiation of viral replication, possibly through an antagonistic interplay with other trans-acting factors (e.g., PTB and La proteins) (30). In this context, the involvements of miR-122 and ELAVL1/HuR may be explained by a circularized HCV RNA model in which miR-122 and ELAVL1/HuR RNP draw the ends of the HCV RNA together to improve the efficiency of replication. Another intriguing observation is that the 3′ region of miR-122 (all nucleotides aside from the seed sequence) are required for its ability to enhance the translation of the HCV RNA (11). A possible explanation for this phenomenon is that a cellular protein that interacts with the 3′-terminal part of miR-122 (e.g., ELAVL1/HuR) may recruit canonical or noncanonical translational initiation factors to the IRES element. Ago2, which binds predominantly in the vicinity of the miR-122-binding sites on the HCV RNA as well as to miR-122 itself, could cooperate with ELAVL1/HuR. In this regard, redistribution of ELAVL1/HuR from the nucleus to the cytoplasm upon HCV infection is noteworthy, given that a similar phenomenon has been reported for other miR-122-associated core proteins, such as Ago2 (30, 31).

Despite the several lines of evidence presented herein demonstrating that miR-122 plays a crucial role in the pro-HCV function of ELAVL1/HuR, it should be noted that supplementation of miR-122-mimic RNA oligonucleotides did not achieve a complete rescue (Fig. 3E) and the proliferation of G28A-HCV was still affected, albeit weakly, by ELAVL1/HuR depletion (Fig. 3G). We thus could not exclude the possibility that an miR-122-independent mechanism contributes to the pro-HCV function of ELAVL1/HuR. As a possible mechanism, ELAVL1/HuR might exert its positive effect on HCV proliferation through binding to AU-rich sequences on the HCV RNA (e.g., the 3′UTR) and thereby regulating its stability, or through indirectly regulating the expression of other cellular genes. In that sense, the expression change of several proviral and antiviral genes identified from our RNA sequencing of ELAVL1/HuR-depleted cells should also be considered (SI Appendix, Fig. S13).

Consistent with our findings, another group recently highlighted the functional significance of ELAVL1/HuR in the HCV life cycle. Using a genome-wide screen based on CRISPR/Cas9-mediated KO, the authors identified several host genes crucial for the life cycle of Flaviviruses, which include HCV and Dengue virus (32). ELAVL1/HuR was found to act as an essential gene, especially for the HCV life cycle. However, the molecular details underlying the impact of ELAVL1/HuR on HCV remained largely unexplored. The present study substantially improves our understanding of the action mechanism of ELAVL1/HuR in this context by uncovering ELAVL1/HuR-miR-122 as a key regulatory axis for HCV proliferation. Our data further strengthen the idea that ELAVL1/HuR is necessary for the HCV life cycle, as we report that the protein level of ELAVL1/HuR is prominently up-regulated in several HCV-relevant conditions. A considerable increase of ELAVL1/HuR protein was observed at 48 h after the transfection of HCV replicon RNA into Huh7 cells (Fig. 3L). More importantly, a significantly higher level of ELAVL1/HuR was detected in the livers of HCV-positive HCC patients compared with tissues from HCV-negative individuals (Fig. 3M). Likewise, a similar pattern was observed in HCV-positive HCC patient tissues compared to normal controls in TMA analysis (Fig. 3N).

Another important finding of our study began with the screening of a kinase inhibitor library to identify chemicals with both miR-122-modulating and anti-HCV activities. The three chemicals with the strongest activity were all PLK1 inhibitors, which strongly suggested that there could be a link between PLK1 and miR-122-mediated HCV modulation.

Indeed, experiments performed using four PLK1-specific shRNAs with different extents of depletion showed that PLK1 exhibited strong positive correlations with miR-122 and HCV proliferation (Fig. 5 A, B, and D), and the level of ELAVL1/HuR protein was also highly correlated with these parameters under the same conditions (Fig. 5B). Similar results were obtained when PLK1 was pharmacologically inhibited by rigosertib, which was the strongest hit identified from our kinase inhibitor library screening (Fig. 4B). Together, these results strongly suggest that PLK1 regulates the ELAVL1/HuR-miR-122 axis and these interactions play a critical role in HCV proliferation. The significance of miR-122 and ELAVL1/HuR in the pro-HCV effect of PLK1 was supported by two key experimental evidence. First, the PLK1 depletion-suppressed proliferation of HCV was considerably rescued by exogenously supplemented miR-122-mimic RNA oligonucleotides and GFP-HuR (Fig. 5G and SI Appendix, Figs. S25 and S26). Second, PLK1 depletion had less effect on the proliferation of a G28A mutation-containing HCV replicon than that on WT-HCV (Fig. 5H and SI Appendix, Fig. S27). The significance of PLK1 in the HCV life cycle was further strengthened by our observation that the protein level of PLK1 was prominently up-regulated in several HCV-relevant conditions. The protein up-regulation of PLK1 and ELAVL1/HuR was concomitantly observed in HCV RNA-transfected Huh7 cells (Fig. 5J) and, importantly, the PLK1 protein was highly expressed in the tissue samples of HCV-positive HCC patients (Fig. 5K).

We might then ask: How does PLK1 affect ELAVL1/HuR in cells? Several lines of evidence support the idea that ELAVL1/HuR is regulated by PLK1 at the posttranscriptional level. First, PLK1 depletion had little or no effect on the cellular level of ELAVL1/HuR mRNAs (Fig. 5C), while the level of ELAVL1/HuR protein was severely affected under the same condition (Fig. 5B). Second, overexpression of WT-PLK1 but not a catalytically inactive mutant (K82M) increased the ELAVL1/HuR protein level, indicating that PLK1 activity regulates the ELAVL1/HuR protein level (Fig. 5E). Third, PLK1 coprecipitated ELAVL1/HuR (Fig. 5F and SI Appendix, Fig. S24), indicating that there is a physical association between these two proteins. A more precise mechanism will be elucidated in future research.

Although a previous report suggested a potential link between PLK1 and HCV replication (33), the present study substantially improves our understanding of this research area by revealing that ELAVL1/HuR-miR-122 mediates PLK1-dependent HCV proliferation and proposing the PLK1-ELAVL1/HuR-miR-122-HCV signaling axis (Fig. 6).

Fig. 6.

Proposed signaling axis and interplay among PLK1, ELAVL1/HuR, and miR-122 on the HCV proliferation (red-colored letters and lines). The miR-122-binding region of HCV genome is highlighted as red-dotted circle. Gray-dotted lines represent potential alternative regulation mechanism of HCV proliferation other than proposed signaling axis in this study. Exo., 3′-5′ exoribonuclease.

Our identification of rigosertib as a relevant agent primarily contributed to the discovery of PLK1-ELAVL1/HuR-miR-122 signaling as a regulatory axis for HCV proliferation; the strong HCV-inhibitory activity of rigosertib itself is also of great interest. Rigosertib exhibited a strong HCV-inhibitory activity under all tested conditions, including in replicons (Fig. 4C), virus-infected cells (Fig. 4F), and replicon-harboring mice (Fig. 4G). Its inhibitory efficacy was estimated to have an IC₅₀ of 0.1 to 0.5 μM from various cellular assays (Fig. 4 C and F), and thus was quite comparable to that of the FDA-approved anti-HCV drug, sofosbuvir (SI Appendix, Fig. S19). We also observed that rigosertib showed strong inhibitory efficacies against a sofosbuvir resistance-associated substitution (RAS)-bearing HCV proliferation (SI Appendix, Fig. S30). Despite its high effectiveness, the side effects of rigosertib observed in clinical trial and the existence of sufficiently effective anti-HCV drug limit its application for treating HCV. Therefore, considering that rigosertib targets a host cellular protein/pathway (PLK1-ELAVL1/HuR-miR-122), it could be useful for studying miR-122-dependent viral pathogenesis and liver functions.

In the present study, we found that ELAVL1/HuR critically contributes to miR-122 biogenesis through binding to the 3′-end of the mature miRNA and regulating its stability. Moreover, identification of rigosertib as a potent HCV inhibitor and chemical modulator of miR-122 revealed a PLK1-ELAVL1/HuR-miR-122-HCV axis and suggested that this signaling axis could be targeted as a therapeutic approach for treating HCV and HCV-associated diseases. A more detailed understanding of the PLK1-ELAVL1/HuR-miR-122 axis will be pursued in future studies.

Materials and Methods

Small RNA Analysis by Northern Blotting.

Total RNAs were isolated with TRIzol (Thermo Fisher Scientific) and northern blotting was carried out N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC; Sigma-Aldrich) cross-linking method as described previously (34, 35). See SI Appendix for details.

Luciferase Assay of HCV Replicons and HCV/miR-122 Dual-Sensing System.

Various assayable HCV subgenomic replicons and HCV/miR-122 dual-sensing system-bearing cells were lysed with Passive Lysis Buffer or Renilla-Glo Luciferase Assay Buffer (Promega) (35). The aliquots of lysates were analyzed by Dual-Luciferase Reporter Assay System or Renilla-Glo Luciferase Assay System (Promega) according to manufacturer’s direction.

Human Liver Biopsy Specimens.

The liver biopsy specimens used in this study include the following: anti-HCV-negative patients, n = 2; anti-HCV-positive patients with chronic HCV, n = 2. Collection of liver biopsy samples and the study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Seoul St. Mary’s Hospital, Korea (clinical trial number: KC12SIS0207). Written informed consent was obtained from all subjects involved in the study.

In Vivo Imaging of HCV Proliferation on Subcutaneous Xenograft Mouse Model.

To verify the effect of rigosertib treatment on HCV proliferation in vivo mouse model, Huh7 cells-harboring HCV replicon (HIRL-2Aneo-NSrep) were used. HIRL-2Aneo-NSrep Huh7 cells (5 × 10⁵) were subcutaneously injected into the right flank area of mice. After 3 days, the level of basal luminescence (0 h, baseline) was monitored through intraperitoneal administration of Renilla luciferase substrate (ViviRen, Promega). A few hours later, control vehicle (DPBS) or rigosertib (150 mg/kg) (26) was administered through an intraperitoneal route, and luminescence signal was acquired at 24 h posttreatment and compared with basal luminescence in the same mice. All animal studies were reviewed and approved by the IACUC of NCCRI (approval number: NCC-22-725).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. The RNA-seq data generated in this study are available at the NCBI Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra) (BioProject accession number: PRJNA906319).