HCV infection causes a redistribution of L1ORF1p to lipid droplets

Lipid droplets are of major importance for HCV infection, serving as putative assembly sites for progeny virions, and host factors are recruited to lipid droplets upon infection providing a favorable environment for viral replication. In a recently reported quantitative lipid droplet proteome analysis of HCV-infected and uninfected hepatoma cells, we found a number of host proteins recruited to lipid droplets upon infection [53]. Interestingly, we identified peptides of the L1-encoded ORF1 protein (L1ORF1p) in purified lipid droplet fractions from HCV-infected Huh7.5 cells in three out of four independent experiments, but we never detected them in lipid droplet fractions from uninfected control cells (Fig 1A–1C).

Fig 1

HCV infection re-localizes L1ORF1p to lipid droplets.

(A) Experimental setup to identify lipid droplet–associated proteins. HCV Jc1 reporter constructs encode fluorescent proteins (XFP) flanked by duplicated NS5A-NS5B cleavage sites (S1 Fig). NS, non-structural protein. HCV-infected and uninfected cells were lysed and lipid droplets were isolated by sucrose density centrifugation. Proteins from the floating lipid droplet fraction were analyzed by mass spectrometry or immunoblotting. (B) Scheme depicts the organization of a functional endogenous L1 harboring ORF0, ORF1 (red box), and ORF2 encoding the RNA-binding protein L1ORF1p and L1ORF2p, a protein harboring endonuclease (EN) and reverse transcriptase (RT) activities, and a cysteine-rich domain (C). UTR, untranslated region; A_n, polyA tail; arrows indicate transcriptional start sites. (C) Identification of L1ORF1p in lipid droplet fractions isolated from HCV-infected Huh7.5 cells by tandem mass spectrometry. Dataset from Rösch et al. [53]. Heatmap depicts L1ORF1p levels normalized to PLIN2 measured in four independent quantitative mass spectrometry (MS) experiments. L1ORF1p-specific peptides identified by mass spectrometry are highlighted in red as part of the presented L1ORF1p protein sequence. (D) Immunoblot analysis of lipid droplet fractions isolated from Jc1^NS5AB-EGFP-infected Huh7.5 cells (multiplicity of infection (MOI) 0.02) at 9 days post infection (dpi) demonstrates localization of endogenous L1ORF1p to lipid droplets in HCV-infected cells. Tubulin and PLIN2 served as loading controls for post-nuclear supernatants (PNS) and lipid droplets (LDs), respectively. Shown is one representative experiment (n = 3). (E) Analysis of endogenous L1ORF1p localization in HCV-infected and uninfected Huh7.5 cells by confocal microscopy. HCV Jc1^FLAG-E2 and mock-electroporated cells were fixed at 3 days post electroporation (dpe) and stained using L1ORF1p and FLAG antibodies. Lipid droplets (LDs) were visualized with BODIPY655/676. Shown are representative images (scale bar 10 μm). Colocalization of endogenous L1ORF1p with lipid droplets was analyzed by calculating the Pearson’s correlation coefficients (PCC) of individual cells from 2 independent experiments (# of cells: mock = 119, Jc1^FLAG-E2 = 109; mean ± SEM, ***p< 0.001, Welch’s t-test). (F) HCV infection causes partial colocalization of ^HAL1ORF1p with HCV core at lipid droplets. HCV Jc1 (MOI 0.03), JFH1 (MOI 0.02), and mock-infected Huh7 cells were transfected with the ^HAL1ORF1p expression plasmid at 7 dpi. 2 days post transfection (dpt), cells were fixed and stained using HCV core and HA antibodies; lipid droplets were visualized with BODIPY493/503. Shown are representative images (scale bar 10 μm). Colocalization analysis of ^HAL1ORF1p with lipid droplets or HCV core was performed by calculating the PCC of individual cells from 2 independent experiments (# of cells: mock = 42, Jc1 = 26, JFH1 = 52; mean ± SEM, *p< 0.05, **p< 0.01, ***p< 0.001, Welch’s t-test).

In order to confirm the localization of L1ORF1p to lipid droplet fractions, we infected hepatoma cells with an HCV Jc1 EGFP reporter virus (Jc1^NS5AB-EGFP, viral constructs are listed in S1 Fig) [58] and isolated lipid droplets by sucrose gradient centrifugation followed by immunoblot analysis. Consistent with the results of the mass spectrometry analysis [53], we detected L1ORF1p in lipid droplet fractions from HCV-infected cells, but not from mock-infected controls, although they also expressed endogenous L1ORF1p (Fig 1D). Next, we visualized and compared endogenous L1ORF1p localization in HCV-infected and mock-infected Huh7.5 cells using confocal microscopy (Fig 1E). In mock-infected cells, L1ORF1p displayed a predominantly dotted cytoplasmic localization (Fig 1E). In cells infected with an HCV Jc1^FLAG-E2 strain [59], we observed a partial redistribution of L1ORF1p to half-ring-shaped patterns surrounding individual lipid droplets (Fig 1E). We then quantified colocalization of L1ORF1p with lipid droplets by calculating the Pearson’s correlation coefficient (PCC) (Fig 1E) as well as the Manders’ colocalization coefficient (MCC) (S2A Fig). The MCCs M1 and M2 can be defined as the fraction of channel 1 that overlaps with channel 2 and vice versa. In contrast, PCC calculates the overall correlation of signal intensities between two channels. HCV infection slightly increased the colocalization of L1ORF1p and lipid droplets compared to mock-infected cells (Fig 1E). However, the MCC analysis was inconclusive due to the low signal intensity of endogenous L1ORF1p staining (S2A Fig). Therefore, we overexpressed an HA-tagged L1ORF1p (^HAL1ORF1p) (S1 Fig) in HCV-infected cells and additionally performed colocalization analysis of lipid droplet-localized HCV core with ^HAL1ORF1p. As observed for endogenous L1ORF1p, ^HAL1ORF1p predominantly localized to cytoplasmic foci in mock-infected cells (Fig 1F), which is consistent with the previously reported subcellular localization of L1ORF1p overexpressed from transiently transfected L1 reporter elements [18,19,23]. For infection of Huh7 cells, we chose two different HCV strains: the chimeric cell culture–adapted strain Jc1 [60], and JFH1, a molecular clone of an HCV isolate [61,62]. JFH1 spreads less efficiently between cells compared to Jc1, but infected cells display more lipid droplet–localized HCV capsid protein core in the perinuclear region [63]. ^HAL1ORF1p localized to lipid droplets in both Jc1 and JFH1-infected cells (Fig 1F). While ^HAL1ORF1p displayed a more punctate staining in lipid droplet–rich areas in Jc1-infected cells, we observed a more defined, even half-ring-shaped localization of ^HAL1ORF1p surrounding lipid droplets in JFH1-infected cells that correlated with the core protein levels located at lipid droplets (Fig 1F). PCC and MCC analyses indicate that colocalization of ^HAL1ORF1p with lipid droplets was slightly increased in Jc1-infected cells and more pronounced in JFH1-infected cells compared to uninfected controls (Figs 1F and S2B). However, we observed only a partial colocalization of HCV core and ^HAL1ORF1p, with a mean M1 (^HAL1ORF1p) of approximately 0.2 for Jc1 and JFH1, and M2 (core) of 0.3 for Jc1 and 0.4 for JFH1.

HCV replication increases endogenous L1ORF1p levels

In the initial experiments, we noted a slight increase of endogenous L1ORF1p in post-nuclear supernatants of HCV-infected Huh7.5 cells (Fig 1D). Following this observation, we performed time-course experiments to assess endogenous L1ORF1p expression upon infection with HCV. L1ORF1p levels steadily increased after infection reaching a ~4-fold accumulation of L1ORF1p in HCV-infected relative to mock-infected cells at 12 dpi (Fig 2A). Using primers that bind in the ORF1 and ORF2 region of L1, we quantified endogenous full-length L1 mRNA expression levels and observed only a transient ~2-fold increase of L1 mRNA levels relative to mock-infected cells that declined again after 6 dpi (Fig 2B). An explanation for these results could be that the L1ORF1p immunoblot analysis measures expression from intact genomic L1 loci encoding functional L1ORF1p, while the qRT-PCR analysis detects L1 transcripts expressed from both intact and mutated L1 loci, which do not express functional L1ORF1p. Expression of intact genomic L1 loci is regulated by methylation of CpG islands located in the L1 5’ UTRs and CpG hypomethylation was previously detected in HCV-related cirrhosis and HCC [44]. Furthermore, HCV infection has been described to induce persistent epigenetic changes that are associated with an increased HCC risk in patients [64]. However, in our infection system using Huh7-derived cells, analysis of the methylation state of L1 5’ UTR regions did reveal only partial methylation in Huh7.5 cells without any changes after HCV infection (S3 Fig). These findings suggest that the elevated L1ORF1p levels in our system do not result from an overall upregulation of transcription of full-length L1 elements. L1ORF1p levels were also increased by ~2 fold in stable Con1 subgenomic replicon (Con1-SGR) [65] Huh7.5 cells, expressing a bicistronic RNA that lacks the structural HCV proteins as well as p7 and NS2 and encodes a neomycin resistance gene for selection (Huh7.5-Con1-SGR cells) (Fig 2C). This indicates that active HCV RNA replication or expression of the non-structural HCV proteins account for increased L1ORF1p levels.

Fig 2

HCV infection increases L1ORF1p protein levels.

(A) L1ORF1p protein levels in mock and Jc1^NS5AB-EGFP-infected Huh7.5 cells (MOI 0.2) were analyzed by immunoblotting at 3–12 dpi and quantified by densitometry. Shown is the relative L1ORF1p level normalized to tubulin expression (mean ± SEM, n = 6–9, *p< 0.05, ***p< 0.001, Welch’s t-test) that demonstrates a strong increase of L1ORF1p expression in HCV-infected cells. (B) HCV infection transiently increases expression of endogenous full-length L1 elements. Time course of L1 expression after infection of Huh7.5 cells with Jc1^NS5AB-EGFP (MOI 0.2). Endogenous L1 mRNA levels were determined by qRT-PCR using RNA isolated at the indicated time points and primer pairs binding within ORF1 or ORF2 (mean ± SEM, n = 3–4, *p< 0.05, Welch’s t-test). Primer binding sites within L1 sequences are marked in the scheme. (C) L1ORF1p levels in stable polyclonal Huh7.5-Con1-SGR cells were analyzed by immunoblotting and quantified. Shown is the relative L1ORF1p level normalized to tubulin (mean ± SEM, n = 3).

HCV infection restricts L1 retrotransposition

Although L1 activity during early development and its role in human disease were investigated in numerous studies, little is known about L1 mobilization in the context of viral infections. Using an L1 retrotransposition reporter assay that is based on dual-luciferase activity [66] (Fig 3A), we compared L1 retrotransposition rates in mock-infected and HCV-infected Huh7.5 cells (Fig 3B). Following infection with Jc1^NS5AB-EGFP, we transfected the cells with the L1 retrotransposition reporter plasmid L1_RP-FLuc (pYX017) [66] and quantified L1 retrotransposition activity at 5 days post transfection. Strikingly, we observed a significant decrease in marked L1 retrotransposition frequency in HCV Jc1-infected cells by ~75% relative to mock-infected cells (Fig 3C). To assess whether the L1 retrotransposition frequencies obtained from the engineered L1-FLuc reporter assay correlate with the number of marked genomic L1-FLuc de novo insertions, we performed a second set of L1 reporter assays, and this time isolated genomic DNA to quantify the number of L1-FLuc de novo retrotransposition events by qRT-PCR (S4 Fig). Consistent with the reduction of L1 retrotransposition frequency measured by luciferase activity, the number of genomic L1-FLuc de novo insertions were reduced by ~75% in Jc1-infected cells compared to mock-infected cells (S4 Fig), confirming that the reduced firefly luciferase signal in HCV-infected cells reflects a reduced number of L1-FLuc retrotransposition events.

Fig 3

HCV infection decreases L1 retrotransposition frequency.

(A) Scheme of the dual-luciferase L1 retrotransposition reporter assay expressing Renilla luciferase (RLuc) for transfection normalization and firefly luciferase (FLuc) as reporter for L1 retrotransposition (L1_RP-FLuc). The FLuc gene is in antisense orientation and interrupted by an intron in sense orientation flanked by splice donor (SD) and acceptor (SA) sites ensuring that FLuc is expressed only after splicing of its pre-mRNA, reverse transcription, and integration of an intact Fluc cDNA copy. Thus, FLuc activity is directly proportional to the number of L1_RP-FLuc de novo retrotransposition events. (B) Scheme of the experiments; the indicated cells were transfected with the L1_RP-FLuc reporter and luciferase assays were performed. (C) Mock or Jc1^NS5AB-EGFP-infected Huh7.5 cells (MOI 0.2) were transfected with L1_RP-FLuc at 7 dpi. Depicted is the relative L1 retrotransposition frequency 5 days after duplicate transfections from 3 independent experiments (mean ± SEM, n = 6, ***p< 0.001, Welch’s t-test). (D) JFH1-infected Huh7 cells were cultured for ~6 months. Mock and JFH1-infected cells with low and high infection rates were transfected with L1_RP-FLuc. Shown is the relative L1 retrotransposition frequency 6 dpt of duplicate transfections from 2 independent experiments (mean ± SEM, n = 4, **p< 0.01, Welch’s t-test). Immunoblot analysis of HCV core expression confirmed differences in infection rates.

We additionally validated our results using an alternative L1 retrotransposition reporter assay, in which EGFP expression instead of firefly luciferase activity served as reporter for L1 retrotransposition [67] (S5 Fig). To this end, we transfected Huh7.5 cells 2 and 7 days post HCV Jc1 infection, and again observed a marked decrease in L1 retrotransposition events in HCV-infected cells by ~75% compared to mock infected cells at late time points after infection (S5 Fig). While HCV strain Jc1 is a partially cell-culture adapted J6/JFH1 chimera that spreads rapidly, the original isolate JFH1 displays much slower infection kinetics. To model the impact of chronic HCV infection on L1 retrotransposition frequency, we analyzed L1 retrotransposition of marked L1 elements in long-term JFH1-infected Huh7 cells (Fig 3B and 3D). HCV-infected cells were transfected with the L1 dual-luciferase reporter plasmid at ~6 months post electroporation with JFH1 RNA. Interestingly, we observed an HCV infection level–dependent decrease in L1 retrotransposition with ~30% decrease in Huh7 cells with low level JFH1 infection, and a ~50% decrease in Huh7 cells with a high level JFH1 infection as indicated in the corresponding immunoblots of the cell lysates (Fig 3D).

HCV infection does not enhance expression of L1 restricting innate immune response genes

As multiple host-encoded proteins involved in the innate antiviral response negatively regulate L1 retrotransposition [35,68], we investigated if HCV replication induces the expression of L1 restriction factors, including interferon-stimulated genes (ISGs) such as MOV10, or members of the APOBEC3 protein family [24,68–71]. However, HCV infection of Huh7 and Huh7-derived hepatoma cell lines only marginally induces ISGs [72,73]. Accordingly, we did not observe the induction of APOBEC3 protein family members in HCV-infected or Huh7.5-Con1-SGR cells (S6 Fig). Further, expression of the known L1 restriction factors ADAR1 [74], MOV10 [69,75,76], and TRIM5α [77] was decreased in HCV-infected cells and unchanged in Huh7.5-Con1-SGR cells (S6 Fig), and HCV infection did not change MOV10 protein levels (Fig 5B), indicating that L1 mobilization is not restricted by ISG induction in HCV-infected cells.

Lipid droplet localization of L1ORF1p is linked to HCV core trafficking

As we observed the distinct re-localization of L1ORF1p to lipid droplets in HCV-infected cells (Fig 1), we addressed if this redistribution can be pinpointed to a specific step in the HCV replication cycle or to specific HCV proteins. To investigate a possible effect of active HCV RNA replication and expression of NS5A, we used Huh7.5 cells electroporated with JFH1-SGR^BSD RNA (Fig 4A). In contrast to HCV-infected cells, we did not observe an enrichment of L1ORF1p in lipid droplet fractions of JFH1-SGR^BSD-transfected cells compared to control cells (Fig 4B), indicating that neither active HCV RNA replication nor the expression of the HCV non-structural proteins NS3–NS5B are involved in L1ORF1p re-localization. However, replication of an envelope-deleted HCV RNA replicon (Jc1ΔE1E2^{NS5AB-EGFP-BSD}) caused a redistribution of endogenous L1ORF1p to lipid droplet fractions (Fig 4A and 4C), arguing that the capsid protein core or the function of p7 and NS2 play a role in this phenotype. Two of the HCV proteins, core and NS5A, are known to strongly localize to lipid droplets in HCV-infected cells as well as upon single protein expression [48,52]; therefore, we expressed FLAG-tagged core or NS5A in Huh7.5 cells by lentiviral transduction (Fig 4A). As expected, we detected only negligible amounts of endogenous L1ORF1p in isolated lipid droplet fractions of NS5A or control lentivirus-transduced cells. In contrast, L1ORF1p was strongly enriched in lipid droplet fractions from core-expressing cells (Fig 4D). To confirm these data, we co-transfected Huh7 cells with ^HAL1ORF1p and ^FLAGcore expression plasmids and performed immunofluorescence microscopy to visualize the distribution of ^HAL1ORF1p, lipid droplets, and HCV core. ^HAL1ORF1p re-localized from cytoplasmic foci to core-containing lipid droplets and quantification revealed a significantly increased colocalization of ^HAL1ORF1p with lipid droplets and a partial direct colocalization with core (Fig 4E and 4F). To exclude HCV genotype-specific effects, we performed the same experiments with genotype 1b (gt 1b) constructs and found again that L1ORF1p was not enriched in lipid droplet fractions isolated from Con1-SGR-electroporated cells, whereas HCV core (gt 1b) expression caused a redistribution of L1ORF1p to lipid droplets (S7A–S7C Fig).

Fig 4

HCV core mediates L1ORF1p redistribution from cytoplasmic foci to lipid droplets.

(A) Scheme of the experimental setup. Huh7.5 cells were either electroporated with JFH1 subgenomic replicon (JFH1-SGR^BSD, gt 2a) RNA encoding NS3–NS5B (see B) or with a partially envelope-deleted Jc1 replicon Jc1ΔE1E2^{NS5AB-EGFP-BSD} (gt 2a, see C), or transduced with lentiviral expression constructs for HCV ^FLAGcore or NS5A^FLAG (gt 2a, see D), and lipid droplets were isolated by sucrose density centrifugation. (B–D) Immunoblot analysis of lipid droplet fractions isolated from Huh7.5 cells electroporated with JFH1-SGR^BSD (n = 3) (B), Jc1ΔE1E2^{NS5AB-EGFP-BSD} (n = 1) (C), or transduced with lentiviral expression constructs for ^FLAGcore or NS5A^FLAG (n_FLAGcore = 4, n_NS5AFLAG = 2) (D). Shown are representative experiments. Tubulin and PLIN2 served as loading controls for post-nuclear supernatants (PNS) and lipid droplets (LDs), respectively. (E) Confocal microscopy of Huh7 cells transfected with ^HAL1ORF1p and ^FLAGcore (gt 2a) expression plasmids and stained with anti-core and anti-HA tag antibodies and BODIPY493/503 for lipid droplets. Shown are representative images (scale bar 10 μm). (F) Quantification of colocalization of ^HAL1ORF1p with lipid droplets and ^HAL1ORF1p with core as presented in (E). Shown are Manders’ colocalization coefficients (MCC) M1 and M2 and Pearson’s correlation coefficients (PCC) of individual cells from 3 independent experiments (# of cells: control = 91, HCV core = 61; mean ± SEM, ***p< 0.001, Welch’s t-test).

One pre-requisite of HCV core trafficking to lipid droplets is its C-terminal processing by the cellular signal peptide peptidase [78]. To determine if L1ORF1p and HCV core concomitantly traffic to lipid droplets, we took advantage of a core mutant with an unprocessable signal peptide (SPMT) [78] that is retained at the ER (S8A Fig). After subcellular fractionation, we mainly detected HCV core^SPMT in microsomal membrane (MM) fractions, where also the majority of L1ORF1p accumulated (S8B Fig). In contrast, core^WT expression led to an enrichment of both core^WT and endogenous L1ORF1p at lipid droplets. We confirmed these findings by immunofluorescence microscopy (S8C Fig) and observed a decreased localization of ^HAL1ORF1p to lipid droplets in cells expressing the core SPMT mutant compared to cells expressing wild-type core (S8D Fig). Taken together, our results indicate that HCV core mediates redistribution of L1ORF1p from cytoplasmic foci to lipid droplets and that this effect is independent of viral RNA replication and expression of NS5A.

HCV core associates with an L1ORF1p-containing ribonucleoprotein complex

As our results suggested an HCV replication-independent effect of the HCV core protein on subcellular L1ORF1p localization, we hypothesized that both proteins directly interact with each other. To test this hypothesis, we performed co-immunoprecipitation experiments from Huh7.5 cells stably expressing ^FLAGcore or NS5A^FLAG (gt 2a). As L1ORF1p was reported to interact with many host-encoded proteins in an RNA-dependent manner [22–24,26] and both core and NS5A can bind RNA [79–81], we also investigated whether a putative interaction with endogenous L1ORF1p is RNA-dependent. Our data show that L1ORF1p co-purified with ^FLAGcore but not with NS5A^FLAG (Fig 5A). RNAse A treatment of the cell lysate prior to immunoprecipitation resulted in a decreased L1ORF1p signal (Fig 5A), indicating an RNA-dependent interaction of HCV core with L1ORF1p. In line with these results, L1ORF1p also co-purified with ^FLAGcore (gt 1b) in an RNA-dependent manner (S7D Fig). As mentioned above, L1ORF1p is known to interact with many other RNA-binding proteins in an RNA-dependent manner and is associated with proteins found in cytoplasmic stress granules and P-bodies. Comparing our previously published lipid droplet proteome analysis [53] with a recent L1ORF1p interactome study [22], we found that some L1ORF1p-interactors were also enriched in lipid droplet fractions from HCV-infected cells (22 out of 32 overlapping proteins were enriched, and 10 were decreased in lipid droplet fractions of HCV-infected cells, S9 Fig). Hence, we additionally probed the immunoprecipitates for two jointly identified proteins: the polyadenylate-binding protein 1 (PABPC1) and the RNA helicase MOV10. Similar to L1ORF1p, both proteins were found to co-precipitate with ^FLAGcore but not with NS5A^FLAG (Fig 5A). To confirm our results, we ectopically expressed ^HAL1ORF1p in HCV-infected or mock-infected Huh7.5 cells and performed the reciprocal co-immunoprecipitation by pulling down HA-tagged L1ORF1p. We detected HCV core in ^HAL1ORF1p immunoprecipitates and to a lower extent in samples precipitated from RNAse A-treated cell lysates, substantiating the RNA-dependent interaction of HCV core and L1ORF1p (Fig 5B). Immunoblot analyses confirmed the recently reported RNA-dependent interaction of PABC1 and MOV10 with L1ORF1p [22] and showed that this was independent of HCV infection (Fig 5B). Thus, we conclude that in HCV-infected cells, core is part of cellular RNP complexes that are associated with L1ORF1p explaining the observed concerted co-precipitation of HCV core with L1ORF1p, PABPC1, and MOV10.

Fig 5

L1ORF1p interacts with HCV core in an RNA-dependent manner.

(A) Co-immunoprecipitation analysis of ^FLAGcore or NS5A^FLAG with endogenous L1ORF1p and its interaction partners MOV10 and PABPC1. Lysates of cells transduced with lentiviral ^FLAGcore or NS5A^FLAG expression constructs were incubated either with RNaseOUT or RNase A followed by FLAG-specific immunoprecipitation and immunoblotting or total RNA isolation and subsequent agarose gel electrophoresis and ethidium bromide (EtBr) staining (bottom panel). Successful RNase A treatment of cell lysates was demonstrated by the absence of 18S and 28S rRNA. Shown is one representative experiment (n = 3). (B) Co-immunoprecipitation analysis of ^HAL1ORF1p in HCV Jc1^NS5AB-EGFP-infected or mock-infected Huh7.5 cells. Jc1^NS5AB-EGFP-infected (MOI 0.2) or mock-infected Huh7.5 cells were transfected 7 dpi with the ^HAL1ORF1p expression construct and lysed 2 dpt. RNAse treatment, total RNA isolation, and immunoprecipitation with HA-specific agarose beads were performed as described in (A). One representative experiment is presented (n = 3). (C) HCV RNA co-precipitates with ^HAL1ORF1p. Relative quantification of HCV RNA copies in immunoprecipitates obtained from Jc1^NS5AB-EGFP-infected Huh7.5 cells transiently expressing ^HAL1ORF1p. Cells were lysed 2 dpt and 9 dpi and HA-specific immunoprecipitation was performed. Untransfected cells served as control and successful infection and expression of ^HAL1ORF1p and HCV core as well as immunoprecipitation were confirmed by immunoblotting. HCV copy numbers were quantified by qRT-PCR using a serial dilution of in vitro transcribed HCV RNA as a standard (mean ± SEM, n = 6, *p< 0.05, Mann-Whitney U test). (D) Comparison of HCV RNA copies and interaction with HCV core in immunoprecipitates of endogenous L1ORF1p and YB-1 obtained from from Jc1^NS5AB-EGFP-infected Huh7.5 cells. Expression and immunoprecipitation of L1ORF1p and YB-1 as well as HCV core were analyzed by immunoblotting. Immunoprecipitation without specific antibodies (-AB) served as control. HCV RNA copy numbers were determined as described in (C). Shown are relative HCV RNA copy numbers as fold over control (n = 2).

The localization of cellular RNP complexes and P-body or stress granule proteins to HCV replication sites in infected cells has been described before [53,55–57,82]. Indeed, a unique virus-host protein complex has been suggested that is collectively redistributed to lipid droplets in HCV JFH1-expressing cells [55]. Noteworthy, the presence of HCV core seems to be required but not sufficient for redistribution of cellular RNP complexes to lipid droplets [55,56]. Based on the concerted interaction between L1ORF1p, PABPC1, MOV10, and HCV core, we speculated that HCV core not only mediates L1ORF1p redistribution but induces the re-localization of L1ORF1p-associated RNP complexes to lipid droplets in HCV-infected cells. In line with our hypothesis, we detected PABPC1 and MOV10 in lipid droplet fractions isolated from HCV core-expressing Huh7.5 cells, but not from control lentivirus-transduced or NS5A-expressing cells (S10A Fig). To substantiate our findings, we subsequently transduced Huh7.5 cells with expression constructs for HCV core and either HA-tagged wild-type L1ORF1p (^HAL1ORF1p^WT) or an RNA-binding mutant (^HAL1ORF1p^Mut; RR_261-262AA mutant) [8,16,17] that was reported earlier to have a severely reduced ability to localize to L1 RNPs [17,18]. Immunoblot analysis of the isolated lipid droplet fractions revealed lower amounts of ^HAL1ORF1p^Mut relative to ^HAL1ORF1p^WT in lipid droplet fractions (S10B Fig), confirming that L1ORF1p is translocated to lipid droplets in an RNA-dependent manner as part of a larger RNP. Yet, we did not detect L1ORF2p in our lipid droplet proteome analysis [53]. This might be due to very low expression levels of endogenous L1ORF2p compared to L1ORF1p, because even L1 expression constructs in transfected cells were reported to produce L1ORF1p at up to 1000–10000-fold higher levels than L1ORF2p and expression might be restricted to a subset of cells within a population [83–86].

The presence of RNA-binding proteins or a cellular RNP complex at HCV replication sites poses the possibility that the HCV RNA genome is part of this complex. Thus, we again performed HA-specific co-immunoprecipitation from HCV-infected cells transiently expressing ^HAL1ORF1p and determined the HCV RNA copy number in the precipitates by quantitative RT-PCR (qRT-PCR). We detected a significant enrichment of HCV genome copies in ^HAL1ORF1p samples, suggesting that the HCV RNA is indeed part of the L1ORF1p-containing RNP complex (Fig 5C). However, from our experimental setup, we cannot conclude if L1ORF1p directly binds the viral genome or if the binding is facilitated by any of the other proteins that are part of the co-immunoprecipitated RNP complexes. The latter is not unlikely, as other P-body or stress granule-associated proteins that were previously found to localize to HCV replication sites, such as DDX3 or YB-1, also associate with the viral RNA [56,87,88]. To compare the relative amount of HCV RNA co-precipitating with L1ORF1p in relation to other RNA-binding proteins, we performed immunoprecipitation of endogenous L1ORF1p and YB-1 from HCV-infected Huh7.5 cells and determined HCV RNA copy numbers (Fig 5D). We confirmed that HCV core co-precipitated with both endogenous proteins. Again, we detected HCV RNA in L1ORF1p precipitates but the HCV RNA copy numbers were 10-fold higher in YB-1 precipitates which can be explained by the strong cis preference of L1ORF1p for its own mRNA. Still, L1 proteins have been reported to associate with other cellular RNAs in trans [13,15,22,32,33], raising the possibility that L1ORF1p binds directly to HCV RNA.

Presence of HCV RNA 3’ UTRs impairs L1 mobilization

Previous studies have shown that the formation of intact L1 RNPs is an important prerequisite for L1 retrotransposition [17]. Thus, we reasoned that HCV infection might disturb L1 RNP integrity or interferes with the transport of intact L1 RNPs to the nucleus, thereby causing a reduced L1 retrotransposition frequency. To test the hypothesis that HCV core–mediated re-localization of L1ORF1p to lipid droplets contributes to the reduced L1 retrotransposition we observed in HCV-infected cells, we first transduced Huh7.5 cells for overexpression of HCV core or NS5A, and subsequently transfected these cells with the L1 dual-luciferase reporter plasmid (Fig 6A and 6B). Unexpectedly, stable expression of HCV core slightly increased L1 retrotransposition frequencies relative to the NS5A and control lentivirus-transduced Huh7.5 cells (Fig 6B). Thus HCV-core mediated redistribution of L1ORF1p to lipid droplets is not sufficient to impair L1 retrotransposition in HCV-infected cells.

Fig 6

An HCV RNA-induced mechanism but not viral protein expression impairs L1 retrotransposition.

(A) Scheme of the experiments; the indicated cells were transfected with the L1_RP-FLuc reporter and luciferase assays were performed. (B) Huh7.5 cells were transduced with lentiviral expression constructs for HCV ^FLAGcore, NS5A^FLAG, or control lentivirus-transduced. Stable expression of HCV ^FLAGcore, but not of NS5A^FLAG increased L1 retrotransposition frequency in Huh7.5 cells relative to control lentivirus-transduced cells. Shown is the relative L1 retrotransposition frequency at 6 dpt of duplicate transfections from 3 independent experiments (mean ± SEM, n = 6, *p< 0.05, Welch’s t-test). (C) L1 retrotransposition frequency in HCV SGR cells relative to naive Huh7.5 cells. Huh7.5 JFH1-SGR^BSD (gt 2a), Huh7.5-Con1-SGR (gt 1b) or Huh7.5 cells were transfected with L1_RP-FLuc. Shown is the relative L1 retrotransposition frequency 5–6 dpt of duplicate transfections from 3 independent experiments (mean ± SEM, n = 6, **p< 0.01, Welch’s t-test). (D) Plasmid-based expression of HCV ^FLAGcore and the JFH1 NS3–NS5B (gt 2a) polyprotein does not affect L1 retrotransposition frequency in Huh7.5 cells relative to control cells. Cells were transfected with the respective plasmids 24 hours prior to transfection with the L1_RP-FLuc reporter. Shown is the relative L1 retrotransposition frequency at 5 dpt of duplicate transfections from 2 independent experiments (mean ± SEM, n = 4). (E) Transfection with HCV 3’ UTR RNA reduces L1 retrotransposition. Huh7.5 cells were transfected with the L1_RP-FLuc reporter plasmid and ~6 hours later, cells were additionally transfected with 0.5 μg of in vitro transcribed HCV 3’ UTR RNA per well. Mock-transfected cells served as control and RNA transfection was repeated 48 h later. Shown is the relative L1 retrotransposition frequency at 5 dpt of duplicate transfections from 4 independent experiments (mean ± SEM, n = 8; ***p< 0.001, Welch’s t-test).

We then investigated if HCV RNA replication itself negatively affects L1 retrotransposition and transfected Huh7.5 cells harboring a JFH1 subgenomic replicon (JFH1-SGR^BSD, gt 2a; S1 Fig) or Huh7.5-Con1-SGR cells with the L1 dual-luciferase reporter plasmid and performed luciferase reporter assays in presence of ongoing HCV RNA replication (Fig 6A and 6C). Strikingly, L1 retrotransposition was reduced by more than 50% in JFH1-SGR^BSD and in Con1-SGR cells compared to naïve Huh7.5 cells (Fig 6C), indicating that HCV RNA replication, the expression of the non-structural proteins of the replicase complex (NS3–NS5B), the formation of HCV replication compartments, or host-responses towards viral infection interfere with L1 retrotransposition. We next sought to distinguish if non-structural proteins or viral RNA are responsible for suppressing L1 mobilization. Ectopic expression of the HCV polyprotein from NS3–NS5B as well as plasmid-based expression of ^FLAGcore did not change L1 retrotransposition frequencies compared to control cells (Fig 6D), indicating that the decreased L1 retrotransposition in HCV-SGR cells is linked to presence or replication of HCV RNA. As the 3’ UTR of HCV contains a pathogen-associated molecular pattern that is sensed by HCV-infected cells [88,89], we transfected L1_RP-FLuc-expressing cells with in vitro-transcribed HCV JFH1 3’ UTR RNA. Strikingly, we observed ~75% reduction in L1 retrotransposition frequencies compared to control cells (Fig 6E). This result suggests that HCV RNA-induced mechanisms restrict retrotransposition of engineered L1 reporter elements.

One peculiar observation was that HCV infection induced expression of endogenous L1, but it also drastically reduced the retrotransposition frequency of marked L1 reporter elements (Figs 2 and 3). Interestingly, endogenous but not ectopically expressed L1ORF1p protein levels were decreased in Huh7.5 cells stably expressing the HCV core protein, but not in NS5A expressing cells (S11 Fig). This reduction of endogenous L1ORF1p levels again inversely correlated with a slightly increased L1 retrotransposition frequency (Fig 6B). Our observation is consistent with the recently reported decrease in L1 retrotransposition frequency of engineered L1 retrotransposition reporter constructs as consequence of an increased expression of truncated or full-length L1ORF1p in trans [90]. Therefore, endogenous L1ORF1p levels in HCV-infected vs stable HCV core-expressing cells might contribute to restricted or enhanced mobilization of engineered L1 reporter elements, respectively.

HCV-induced stress granule formation correlates with restriction of L1 retrotransposition

Among the various mechanisms described to control L1 activity, it was reported that the SAM and HD domain containing protein 1 (SAMHD1) blocks L1 retrotransposition by promoting stress granule formation and enhancing sequestration of functional L1 RNPs and endogenous L1ORF1p in these stress granules [20]. As HCV RNA replication induces stress granule formation in vitro [54,57,91,92], we reasoned that L1 RNPs might be stalled in HCV-induced stress granules, hence not being able to complete the retrotransposition cycle. To explore this hypothesis, we performed immunofluorescence staining of HCV-infected cells for nucleolysin TIA-1 isoform p40/ T-cell-restricted intracellular antigen-1 (TIA1) and Ras GTPase-activating protein-binding protein 1 (G3BP1) as stress granule markers followed by confocal microscopy (Fig 7A–7C). HCV infection triggered cytoplasmic TIA1/G3BP1 positive stress granule formation in ~20% of HCV Jc1^FLAG-E2-positive cells, whereas only 1.4% of HCV-negative cells displayed stress granules (Fig 7A and 7D). Likewise, ~18% of Huh7.5 cells transfected with HCV subgenomic replicon RNA (JFH1-SGR^tagBFP-NLS) harbored TIA1/G3BP1 positive granules, compared to ~5% of control-electroporated cells (Fig 7B and 7D). In line with previous reports [57], the presence of HCV JFH1 3’ UTR RNA was sufficient to trigger stress granule formation (S13D Fig). Co-staining of endogenous L1ORF1p and TIA1 revealed localization of L1ORF1p to TIA1-positive granules and lipid droplets in Jc1^FLAG-E2-positive cells, but only to TIA1-positive granules in JFH1-SGR^tagBFP-NLS-positive cells (Fig 7A and 7B), indicating that L1ORF1p is indeed partially sequestered in stress granules. In contrast, ectopic expression of HCV core alone did not induce stress granule formation, although we observed a ring-like re-localization of L1ORF1p and to a lower extend a re-localization of TIA1 to HCV core–positive lipid droplets (Fig 7C and 7D). The redistribution of specific stress granule proteins to lipid droplets during HCV infection has been described by several groups [54,55,57,91]. Comparing our lipid droplet proteome dataset [53] to annotated stress granule and P-body proteins (downloaded from g:Profiler) [93] revealed that only a fraction of stress granule and P-body markers is found in lipid droplet fractions during HCV infection (S12 Fig), indicating that HCV infection alters and re-distributes stress granules. We next compared G3BP1 and TIA1 protein levels by immunoblot analysis and observed a pronounced increase in G3BP1 levels in HCV-infected Huh7.5 compared to the control (Figs 7E and S13A). G3BP1 levels were also slightly increased in Huh7.5-Con1-SGR cells compared to naïve Huh7.5 cells (Fig 7F), whereas stable expression of ^FLAGcore and NS5A^FLAG did not affect expression levels of the two stress granule markers TIA1 and G3BP1 (S13B Fig).

Fig 7

HCV RNA replication triggers the formation of stress granule which colocalize with L1ORF1p.

(A–C) HCV RNA replication induces the formation of TIA1/G3BP1 positive stress granules. Confocal microscopy analysis of stress granule formation in HCV-infected and HCV replicon cells. Huh7.5 cells were electroporated with full-length HCV Jc1^FLAG-E2 (A) or subgenomic JFH1-SGR^tagBFP-NLS (B) RNA; mock-electroporated cells were transfected with the ^FLAGcore expression plasmid or vector control (C). Cells were fixed at 3 days post electroporation and 2 days post transfection and stained with antibodies against TIA1, G3BP1, L1ORF1p, and FLAG as indicated. HCV infection was confirmed by staining of FLAG-E2 (A). The nuclear BFP signal marks HCV replicon positive cells (B). ^FLAGcore expression was visualized using a FLAG-specific antibody (C). Shown are representative images (scale bar 10 μm). (D) Quantification of stress granule (SGs) positive cells from (A–C). Fields were randomly selected and cells showingTIA1/G3BP1 positive granules were counted as SG positive (# of cells from 2 independent experiments: (A) mock = 142, Jc1^FLAG-E2 = 130; (B and C) mock/vector control = 250, JFH1-SGR^tagBFP-NLS = 182, ^FLAGcore = 68)). (E–F) Immunoblot analysis of TIA1 and G3BP1 expression in HCV Jc1^NS5AB-EGFP-infected Huh7.5 cells (E) and stable Huh7.5 Con1-SGR cells (F). Note that the immunoblot analyses presented in (F) were performed with the same samples as the immunoblots shown in Fig 2C. HCV core and NS3 served as markers for HCV infection and subgenomic replicons and tubulin was used as loading control. Shown are representative immunoblot analyses. (G) Knockdown of G3BP1 partially restores L1 retrotransposition frequency in HCV SGR cells. Stable Huh7.5 JFH1-SGR^BSD cells were transduced with lentiviral shRNAs targeting TIA1, G3BP1, or a non-targeting control (shNT). Knockdown efficiency was analyzed by immunoblotting using G3BP1- and TIA1-specific antibodies. HCV NS3 expression served as control for HCV replication and tubulin served as loading control. Shown is one representative immunoblot (n = 4). (H) shTIA1, shG3BP1, and shNT-transduced JFH1-SGR^BSD cells were transfected with the L1_RP-FLuc reporter plasmid and luciferase assays were performed. Shown is the relative L1 retrotransposition frequency at 5 dpt of duplicate transfections from 4 independent experiments (mean ± SEM, n = 8).

Finally, to mechanistically confirm our hypothesis that HCV restricts L1 retrotransposition by inducing stress granule formation, we transduced Huh7.5 JFH1-SGR^BSD cells with lentiviral shRNAs targeting G3BP1 and TIA1 and performed L1 retrotransposition assays. The selected shRNAs efficiently reduced G3BP1 and TIA1 protein levels in subgenomic replicon and in Huh7.5 cells (Figs 7G and S13C). Importantly, relative L1 activity was increased ~2.5 fold in G3BP1-knockdown cells compared to shNT control cells (Fig 7H). In contrast, knockdown of TIA1 did not restore L1 retrotransposition frequency. This might be due to the fact that downregulation of TIA1 concomitantly increased G3BP1 levels in JFH1-SGR^BSD cells (Fig 7G). In line, immunoblot analysis of HCV-infected cells showed increased G3BP1 level in TIA1-knockdown cells compared to shNT control cells and a complete failure of our shRNAs to reduce G3BP1 expression in HCV-infected cells (S13C Fig), suggesting compensatory upregulation of G3BP1 in HCV-infected cells. Taken together, our data indicate that L1 retrotransposition is restricted in HCV RNA-replicating cells due to sequestration of L1-encoded proteins in HCV RNA-induced stress granules.

L1ORF1p expression does not affect HCV replication

As L1ORF1p is redistributed to HCV replication sites, we also assessed a potential function of L1ORF1p in HCV infection and replication by performing overexpression and knockdown experiments in Huh7.5 cells. As overexpressed ^HAL1ORF1p resides in a complex with HCV RNA (Fig 5C), we analyzed the impact of ^HAL1ORF1p overexpression on HCV infection. Huh7.5 cells stably expressing ^HAL1ORF1p from a lentiviral vector were infected with a Jc1^NS5AB-EGFP reporter virus and analyzed for viral spreading by flow cytometry (Fig 8A). ^HAL1ORF1p overexpression had no effect on HCV spreading (Fig 8B) despite strong overexpression of ^HAL1ORF1p (Fig 8C). Conversely, we infected transduced cells expressing an shRNA targeting the L1 5’ UTR [94,95] or a non-targeting control with a Jc1^{p7-GLuc-2A-NS2} reporter virus [96] and determined GLuc activity as a measure for HCV infection kinetics (Fig 8D). Of note, the selected L1 shRNA (endo453) was described to perfectly match to the bidirectional promoter region of the 5’ UTR of functional L1Hs elements [94]. Again, we did not observe any changes in HCV infection rates (Fig 8E), regardless of the substantial downregulation of endogenous L1ORF1p protein levels (Fig 8F). Taken together, our data indicate that L1ORF1p does not function as host factor affecting HCV replication and its observed re-localization is rather a side effect than a requirement for a favorable HCV replication environment.

Fig 8

L1ORF1p does not affect HCV replication.

(A) Experimental setup to test for the effect of L1ORF1p overexpression on HCV infectivity in Huh7.5 cells. ^HAL1ORF1p was overexpressed by lentiviral transduction followed by infection with Jc1^NS5AB-EGFP (MOI 0.04). (B) HCV spreading kinetics were analyzed by flow cytometry of the EGFP reporter (mean ± SEM, n = 4). (C) Overexpression of ^HAL1ORF1p was confirmed by immunoblot analysis. (D) Experimental setup to test for the effect of the downregulation of endogenous full-length L1 expression on HCV infectivity. Huh7.5 cells were transduced with shRNAs targeting the L1 5’ UTR (shL1) or a non-targeting control (shNT) for a minimum of 10 days prior to infection with the Gaussia luciferase (GLuc) reporter strain Jc1^{p7-GLuc-2A-NS2} (MOI 0.5). Luciferase reporter assays were performed to determine HCV infection kinetics. (E) Luciferase reporter assays to determine the effect of the knockdown of endogenous L1 expression on HCV replication. Results are shown as relative light units (RLU) per ml supernatant (mean ± SEM, n = 3). (F) Efficiency of L1 knockdown using shRNAs shL1 or shNT was confirmed by immunoblot analysis of endogenous L1ORF1p expression in cell lysates isolated at 5 dpi. Shown is one representative experiment.

Reagents

The following antibodies, beads, and dyes were obtained commercially: HCV core (clone 7–50, sc-57800), HA (clone Y-11, sc-805-G), PABPC1 (clone 10E10, sc-32318), MOV10 (clone B-3, sc-515722), calnexin (c-20, sc-6465), TIA1 (c-20, sc-1751) (all Santa Cruz Biotechnology), G3BP1 (Clone 23/G3BP, 611127, BD Biosciences), PLIN2 (GP40, Progen; ab52356, abcam), HCV NS3 (ab65407, abcam), HCV NS5A (HCM-131-5, IBT), FLAG (F7425), FLAG (F1804), HA (H6908), tubulin (clone B5-1-2, T6074), anti-FLAG M2 affinity gel (A2220), anti-HA affinity gel (HA-7, A2095), recombinant protein A (15918–014) and protein G (15920–010) agarose beads (all Sigma), L1ORF1p (clone 4H1, MABC1152, Merck), YB-1 (ab12148, abcam), HRP-labelled secondary antibodies (Jackson ImmunoResearch), HRP-labelled TrueBlot secondary antibodies (Rockland Immunochemicals), Alexa488-, Alexa555-, and Alexa647-conjugated secondary antibodies (all donkey, IgG (H+L)), BODIPY493/503 (D-3922), BODIPY 655/676 (B-3932) (all Life Technologies), Hoechst33342 (Thermo Fisher). L1ORF1p antibody #984 was described previously [33]. Oligonucleotides and PCR primer were purchased from Sigma. Restriction enzymes for molecular cloning were obtained from NEB, other enzymes from Thermo Fisher. Unless stated otherwise, chemicals were purchased from Sigma or Applichem and cell culture reagents from Gibco/Thermo Fisher.

Cell lines

Huh7 cells were provided by Ralf Bartenschlager, Huh7.5 cells [117] and Huh7.5.1 [118] cells were obtained from Apath, LLC, and HEK293T cells from the American Type Culture Collection. All cell lines were grown in DMEM supplemented with 10% FBS (Biochrom Superior or Gibco) under standard cell culture conditions. All plasmid transfection experiments were performed using FuGENE 6 (Promega). In vitro transcribed HCV JFH1 3’ UTR RNA was transfected using the TransIT-mRNA Transfection Kit (Mirus Bio LLC). For transfection of HCV replicon RNA (Con1-SGR, JFH1-SGR^tagBFP-NLS, JFH1-SGR^BSD, or Jc1ΔE1E2^{NS5AB-EGFP-BSD}) or full-length HCV RNA, cells were electroporated with in vitro-transcribed RNA as described [53]. Huh7.5-Con1-SGR cells were established by selection of Con1-SGR RNA-electroporated Huh7.5 cells with 1 mg/ml G418. Huh7.5 cells electroporated with the JFH1-SGR^BSD subgenomic replicon RNA were cultured in medium containing 10 μg/ml blasticidin.

Plasmids

Schemes of plasmids and constructs are presented in S1 Fig. The following plasmids were described previously: HCV JFH1 wild-type [53,62], Jc1 wild-type (J6/JFH1 chimera) [60], Con1 subgenomic replicon [65], full length or envelope-deleted HCV Jc1 reporter strains encoding fluorescent proteins and selection markers between a duplicated NS5A-NS5B cleavage site (Jc1^NS5AB-EGFP, Jc1^NS5ABmKO2, Jc1ΔE1E2^{NS5AB-EGFP-BSD}) [58], HCV Jc1^FLAG-E2 [59], HCV Jc1^{p7-GLuc-2A-NS2} [53], lentiviral LeGO-iCer2 vectors encoding FLAG-tagged HCV JFH1 core or NS5A [53], LeGOCer2 [119], lentiviral vectors and expression plasmids encoding HCV core^WT and core^SPMT (genotype 1b), the lentiviral vector encoding a FLAG-tagged HCV core (genotype 1b) and the FLAG-tagged HCV core (genotype 2a) expression plasmid [51,120], lentiviral HCV RFP-NLS-IPS expression construct [121], the L1_RP-FLuc dual-luciferase reporter plasmid pYX017 [66], and the EGFP-based L1 reporter (pLRE3-EF1-mEGFPI and pLRE3-EF1-mEGFP(Δintron)) [67].

The HCV JFH1 subgenomic replicon SGR^tagBFP-NLS (gt 2a) was constructed by replacing core-NS2 from pBR322 JFH1 [53] with a tagBFP-NLS marker generated through overlap extension PCR using MXS_TagBFP [122] as a template (primers: JFH1-5’NTR fw ACCTGCCCCTAATAGGGGCGA, core-tagBFP rev CTCCTTAATCAGCTCGCTCATGGCGCGCCGGTTGGTGTTTCT, core-tagBFP fw AGAAACACCAACCGGCGCGCCATGAGCGAGCTGATTAAGGAG, BFPtag-NLS-P2A rev GAAGTTTGTGGCGCCGCTGCCGCCAACTTTTCTTTTCTTTTTTGGCATGTATCTGGCCACTGCCACCTC, P2A-NS3 fw GGCAGCGGCGCCACAAACTTC, NS3 rev CCCAACGACGTGGCCCCTAGGGCAGAGCAC) and Age I and Avr II restriction sites. For JFH1-SGR^BSD (gt 2a), core–NS2 was replaced with a blasticidin resistance gene generated through overlap extension PCR using Jc1^{NS5AB-mKO2-BSD} [58] as a template (primers: JFH1-5’NTR fw ACCTGCCCCTAATAGGGGCGA, core-BSD rev TTGAGACAAAGGCTTGGCCATGGCGCGCCGGTTGGTGTTTCT, core-BSD fw AGAAACACCAACCGGCGCGCCATGGCCAAGCCTTTGTCTCAA, BSD-P2A rev GAAGTTTGTGGCGCCGCTGCCGCCCTCCCACACATAACCAGA, P2A-NS3 fw GGCAGCGGCGCCACAAACTTC, NS3 rev CCCAACGACGTGGCCCCTAGGGCAGAGCAC) and Age I and Avr II restriction sites.

For ectopic expression of the HCV JFH1 NS3-NS5B polyprotein, the MXS chaining Kit [122] was used to construct an MXS EF1α::MCS-bGHpA pGK::Neo^R-bGHpA vector, using MluI, SalI and XhoI restriction sites (oligonucleotides: MCS sense CGCGTAACTCGAGACGTATGCGGCCGCGGCAGTACAGGATCCGGATACCCATACGACGTACCAGATTACGCTTGAG; MCS as TCGACTCAAGCGTAATCTGGTACGTCGTATGGGTATCCGGATCCTGTACTGCCGCGGCCGCATACGTCTCGAGTTA). JFH1 NS3-NS5B was amplified from pBR322 JFH1 [53] (primers: LIC NotI HCV NS3 fw CGCTGTCGAGACGTATGCGGCCATG GCTCCCATCACTGCTTATG, LIC BamHI HCV_NS5B rev ACGTCGTATGGGTATCCGGATC CTACCGAGCGGGGAGTAGG) and cloned into the MXS EF1α::MCS-bGHpA pGK::Neo^R -bGHpA vector by ligation independent cloning using BamHI and NotI restriction sites.

Lentiviral shRNA constructs were cloned into pSicoR-MS1 as described [51,67] using the following target sequences: shL1 CCAGGCTTGCTTAGGTAAACA (endo 453 described in [94,95]), shTIA1 TCCTGGCTCATCTCTTTATTC, shG3BP1 TTAGTCTTTCACTTCCAATTT, shNT GCGCGATAGCGCTAATAATT. For the pEF1α-^HAL1ORF1p expression vector L1ORF1 was cloned into pEBB [123] using CMV L1-RP [32] as a template and the BamHI restriction site (primers: pEBB^HAL1ORF1 fw GATAGGATCCGCTAGCATGTACCCATACGATGTTCCAGATTACGCTCTCGAGATGGGGAAAAAACAGAAC, pEBB^HAL1ORF1 rev GATAGGATCCTTACATTTTGGCATGATTTTG (gift from S. Wissing)). For stable expression, ^HAL1ORF1p^WT and ^HAL1ORF1p^Mut were cloned into the lentiviral pSicoR-MS1 lacking the U6 promoter (pSicoR-MS1 ΔU6) by overlap extension PCR using pEF1α-^HAL1ORF1p and pSicoR-MS1 as template and NheI and EcoRI restriction sites (primers: ^HAL1ORF1p fw TGGATCCGCTAGCATGTACCCATACGAT, L1ORF1p AAA rev CTCTGCTGCGGCTTGTAGGGTTTCTGCCGAGAG, AAA fw CAAGCCGCAGCAGAGTGGGGGCCAATATTC, ^HAL1ORF1p-2A-rev TCGACGTCTCCCGCAAGCTTAAGAAGGTCAAAATTCATTTTGGCATGATT, 2A-mCherry fw TGCGGGAGACGTCGAGTCCAACCCTGGGCCAGTGAGCAAGGGCGAG, EcoRI-mCherry rev CTCGACGAATTCTTACTTGTACAG).

Immunofluorescence and confocal microscopy

For immunofluorescence analysis, cells seeded on coverslips were fixed in 4% PFA, permeabilized 5 min in 0.1% Triton-X-100/PBS and incubated in blocking solution (5% BSA, 1% fish skin gelatin, 50 mM Tris in PBS). Following overnight incubation with primary antibodies in blocking solution, cells were incubated with Alexa Fluor-coupled secondary antibodies, lipid droplets were stained with BODIPY493/503 or BODIPY655/676, and coverslips were embedded in Mowiol mounting media. Confocal microscopy was performed on a Nikon C2+ or on a Leica TCS SP5 II confocal laser scanning microscope. For colocalization analysis, individual cells were analyzed using the Coloc2 function of Fiji [124] to calculate the Manders’ colocalization coefficient (MCC) and the Pearson’s correlation coefficient (PCC).

Immunoblot and co-immunoprecipitation

Cells were lysed in NP-40 lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet-P40) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) and 1x protease inhibitor cocktail (Sigma) for 30–60 min. Nuclei and cell debris were removed by centrifugation. Clarified lysates were subjected to SDS-PAGE followed by blotting onto a nitrocellulose membrane (GE Healthcare). In general, samples were run on the same gel, transferred to one membrane and probed with the respective antibodies. Bands were detected by chemiluminescence using Lumi-Light substrate (Roche), SuperSignal West Femto (Thermo Fisher), and ECL hyperfilm (Amersham) or Image Lab (BioRad). Band signal intensities were quantified using the quantification function of Image Lab or the densitometric quantification function of Fiji [124].

For immunoprecipitation, cells were lysed in NP-40 lysis buffer as described above. For analysis of RNA-based interactions, lysates were pre-incubated with 100 μg/ml RNAse A (Thermo Fisher) or 100 U/ml RNaseOUT (Invitrogen) for 45 min at 4°C, rotating. Prior to immunoprecipitation, successful RNAse A treatment was confirmed on an agarose gel after RNA isolation using Tri Reagent (Sigma). To capture HA-tagged or FLAG-tagged proteins, lysates were incubated with anti-HA or anti-FLAG M2 affinity gel (Sigma) for 1 h at 4°C, rotating. Subsequently, beads were washed four times in cold NP-40 lysis buffer. Precipitated proteins were eluted in Laemmli buffer and analyzed by immunoblotting.

For immunoprecipitation of endogenous proteins, cells were lysed as described above. Prior to clarification, lysates were passed 8 x through a 23 G needle. Clarified lysates were precleared for 30 minutes at 4°C, rotating, using protein A or protein G agarose beads (Sigma). Incubation with the respective antibodies was performed overnight at 4°C, rotating. Equilibrated protein A (for rabbit antibodies) or protein G (for mouse antibodies) agarose beads were added and immunoprecipitation was performed for ~6 hours at 4°C, rotating. Samples were processed as described above and analyzed by immunoblotting. All immunoprecipitation steps for endogenous proteins were performed using 0.5% NP 40 lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.5% Nonidet-P40) in presence of 100 U/ml RNAseOUT (Invitrogen).

To quantify HCV RNA copy numbers in cellular and L1ORF1p-associated RNPs, immunoprecipitation was performed in presence of 100 U/ml RNAseOUT, followed by RNA isolation and quantitative RT-PCR, and immunoblotting of aliquots.

Lipid droplet isolation and subcellular fractionation

Lipid droplets were isolated as previously described with minor modifications [125]. In brief, cells were harvested in cold PBS, resuspended in hypotonic sucrose buffer (0.25 M sucrose, 1 mM EDTA supplemented with 1 mM DTT, 1x protease inhibitor cocktail and 1 mM PMSF) and lysed mechanically in a Dounce homogenizer. Post-nuclear supernatants (PNS) were overlaid with isotonic potassium phosphate buffer (0.1 M potassium phosphate pH 7.4, 100 mM KCl, 1 mM EDTA, supplemented with 1 mM PMSF) and centrifugation was performed for 2 h at 100 000 x g, 4°C in an SW60 rotor (Beckman Coulter). Floating lipid droplets were harvested using a bent canula, and PNS and lipid droplet fractions were analyzed by immunoblotting.

For subcellular fractionation, lipid droplets were separated from the microsomal fraction by differential centrifugation as previously described with minor modifications [126,127]. Briefly, cells were resuspended in hypotonic sucrose buffer, lysed mechanically in a Dounce homogenizer, and the resulting cell homogenate was centrifuged twice for 5 min at 600 x g, 4°C to separate the nuclei. To separate the mitochondria, the post-nuclear supernatant (PNS fraction) was centrifuged twice for 10 min at 10 000 x g, 4°C. The supernatant was transferred to an ultracentrifugation tube and overlaid with isotonic potassium phosphate buffer. Ultracentrifugation was performed in an SW60 rotor (Beckman Coulter) for 1 h at 100 000 x g, 4°C. The floating lipid droplet fraction was harvested as described above. The pelleted microsomal fraction was resuspended in RIPA lysis buffer (50 mM Tris, pH 7.4, 1 mM EDTA, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with 1x protease inhibitor cocktail, 100 mM PMSF, and 2% Triton-X-100 to ensure complete solubilization of the ER-retained HCV core^SPMT protein. For each fraction, equal amounts of total protein were subjected to immunoblot analysis.

L1 retrotransposition reporter assays

L1 retrotransposition was analyzed by transient transfection of target cells with a functional L1 reporter element containing a retrotransposition reporter gene that is interrupted by an intron. The reporter gene is only functional after splicing of the intron and integration of a cDNA copy of the reporter gene–harboring L1 element into the genome, thus indicating a full cycle of L1 retrotransposition [16,66,101].

The luciferase-based L1 retrotransposition reporter assay was performed using the dual-luciferase reporter plasmid pYX017 [66]. The encoded L1_RP element under the control of a CAG promoter/enhancer element includes a firefly luciferase reporter cassette to quantify marked L1 retrotransposition events, and the plasmid backbone additionally encodes a Renilla luciferase for normalization of transfection rates. Huh7.5 cells were transfected with pYX017 and lysed 5–6 days post transfection (dpt) using 1x passive lysis buffer (Promega). Transfected cells were selected with puromycin, except for experiments with Jc1^NS5AB-EGFP-infected Huh7.5 cells shown in Fig 3B that were performed without selection. Luciferase activity was determined with the Dual-Luciferase Reporter Assay System (Promega) using a Centro LB 960 luminometer (Berthold Technologies). Inactive L1 control reporter elements (pYX015) [66] yielded firefly luciferase values below the limit of detection thus validating the assay.

For flow cytometry-based retrotransposition analysis, Huh7.5 cells were transfected with the L1 reporter plasmid pLRE3-EF1-mEGFPI or pLRE3-EF1-mEGFP(Δintron), a transfection control plasmid lacking the intron in the EGFP reporter cassette [67]. At 6 dpt, cells were fixed in 2% PFA and analyzed for EGFP expression via flow cytometry on a BD LSR Fortessa (BD Bioscience). A Jc1^NS5AB-mKO2 reporter virus was used to discriminate HCV-infected cells from mock-infected cells by mKO2 expression. To confirm EGFP expression as marker for retrotransposition, cells transfected with L1_LRE3-EGFP were treated with 10 μM reverse transcriptase inhibitor abacavir that inhibits L1 retrotransposition [128] or DMSO as vehicle control 1 dpt. Treatment was renewed every other day. Cells were fixed in 2% PFA at 6 dpt and analyzed by flow cytometry. Flowjo (Treestar) was used for flow cytometry data analysis.

HCV infection assays

For generation of HCV viral stocks, in vitro-transcribed HCV RNA was prepared and electroporated into Huh7.5 or Huh7.5.1 cells as described previously [51,53]. Viral titers (TCID₅₀) were determined by serial dilution on Huh7.5 cells stably expressing the RFP-NLS-IPS HCV reporter [53,121]. To measure HCV spreading in ^HAL1ORF1p-overexpressing Huh7.5 cells, lentiviral transduced cells were infected with Jc1^NS5AB-EGFP and analyzed at the indicated time points by flow cytometry with a BD LSR Fortessa (BD Bioscience). Data were analyzed with Flowjo (Treestar). To determine HCV infection rates in L1 knockdown cells, shRNA-transduced Huh7.5 cells were infected with Jc1^{p7-GLuc-2A-NS2} as described [127] and Gaussia luciferase activity in the supernatant was measured using the Renilla Luciferase Assay System (Promega) and a Centro LB 960 luminometer (Berthold Technologies).

Lentivirus production

Lentiviral particles were produced using HEK293T cells as described before [53,129]. Lentiviral stocks were titrated on Huh7.5 cells. All lentiviral transductions were performed in medium containing 4 μg/ml polybrene.

RNA isolation and quantitative RT-PCR

Total cellular RNA was isolated from cells, clarified cell lysates, or immunoprecipitation samples using Tri Reagent (Sigma). Glycogen (RNA grade, Thermo Fisher) was added as carrier to samples in which low RNA yield was expected. RNA isolated from cells and clarified lysates was treated with rDNAseI (DNA-free Kit, Invitrogen). RNA was reversely transcribed using Superscript III reverse transcriptase (Invitrogen), random hexamer primers (Qiagen), and RNAseOut (Invitrogen). Maxima SYBR Green Mastermix (Thermo Fisher) or the Luna Universal qPCR Master Mix (NEB) were used for qRT-PCR analysis on a 7500HT Fast Real-time PCR System or a StepOne Plus Real time PCR System (both Applied Biosystems). The following qRT-PCR primers were used: L1ORF1p fw TCAAAGGAAAGCCCATCAGACTA; L1ORF1p rev TTGGCCCCCACTCTCTTCT [130]; L1ORF2p fw GAGAGGATGCGGAGAAATAGGA; L1ORF2p rev GGATGGCTGGGTCAAATGGT [131]; HCV fw CGGGAGAGCCATAGTGG; HCV rev AGTACCACAAGGCCTTTCG [51]; 18s rRNA fw GTAACCCGTTGAACCCCATT; 18s rRNA rev CCATCCAATCGGTAGTAGCG; ADAR1 fw ATCAGCGGGCTGTTAGAATATG; ADAR1 rev AAACTCTCGGCCATTGATGAC; APOBEC3A fw TGGCATTGGAAGGCATAAGAC; APOBEC3A rev TTAGCCTGGTTGTGTAGAAAGC; APOBEC3B fw CGCCAGACCTACTTGTGCTAT; APOBEC3B rev CATTTGCAGCGCCTCCTTAT; APOBEC3C fw CTTGGTTCTGCGACGACATAC; APOBEC3C rev TCCTGGTAACATGGATACTGGAA; APOBEC3D fw CTTTCGAGGCCCGGTACTAC; APOBEC3D rev GTGATCTGGAAGCGCCTGTTA; APOBEC3F fw GGCCCGCGTGAAGATTATG; APOBEC3F rev GAGTGGTGCTTTACAACTTCCA; APOBEC3G fw GCATCGTGACCAGGAGTATGA; APOBEC3G rev GTCAGGGTAACCTTCGGGT [132]; TRIM5α fw CTGGAGATGCTGAGGCAGAAGC; TRIM5α rev GTCCAGGATGTCTCTCAGTTGC (Origene # HP2162777); APOBEC3H fw CCCGCCTGTACTACCACTGG; APOBEC3H rev GGGTTGAAGGAAAGCGGTTT [133].

In vitro transcription of HCV JFH1 3’ UTR RNA

For in vitro transcription of the HCV JFH1 3’ UTR, PCR was performed using pBR322 JFH1 [53] as template (primers: T7 JFH1 3 UTR fw TAATACGACTCACTATAG AGCGGCACACACTAGGTAC, JFH1 3 UTR rev ACATGATCTGCAGAGAGACCA), adding a T7 promoter to the 3’ UTR sequence. The resulting PCR product served as template for in vitro transcription using the MEGAscript T7 Transcription Kit (Thermo Fisher).

Extraction of DNA and quantitative real time PCR

To quantify L1-Fluc de novo insertions in cells that were subjected to the L1 dual-luciferase reporter assay, DNA was isolated using the QuickExtract DNA Extraction kit (Lucigen via Biosearch Technologies) according to the manufacturer’s instructions. Briefly, cells were harvested, counted, and equal cell numbers were used for DNA isolation. Samples were analyzed for the presence of marked L1-FLuc de novo insertions by quantitative real-time PCR (qRT-PCR) using the Luna Universal Probe qPCR Master Mix (NEB) and the following primers and probes [66]: FLuc exon-exon junction probe CTTCCCACCTGCCACC; FLuc fw GCAAAAGAAGCTACCGATCATACA; FLuc rev GAAGCTCTCGGGCACGAA. To compare plasmid transfection efficiencies, we performed conventional SYBR green qPCR using the Luna Universal qPCR Master Mix (NEB) and the following primers binding to the puromycin resistance gene (PuroR) or Renilla luciferase gene encoded by the L1 dual-luciferase reporter plasmid pYX017: RLuc fw GGCAAGAGCGGGAATGGCT; RLuc rev CAGTCGTGGCCCACAAAGAT; PuroR fw AGCAACAGATGGAAGGCCTC; PuroR rev GGCGCTGCCCAGACCCTT. Of note, the assay was validated using an inactive L1-FLuc dual reporter pYX015 [66] that cannot retrotranspose and consequently was undetectable using the spliced FLuc probe and primers.

L1 promoter methylation analysis

To compare L1 promoter methylation, genomic DNA was isolated from HCV Jc1^NS5AB-EGFP-infected and uninfected Huh7.5 cells at different time points using the Quick DNA Miniprep Kit (Zymo Research) according to the manufacturer’s instructions. DNA methylation was analyzed by methylation-sensitive restriction enzyme digestion and PCR with gene-specific primers (MSRE-PCR) using the OneStep qMethyl-PCR Kit (Zymo Research) according to the manufacturer’s protocol. The following primers were designed to amplify a 363 bp fragment previously used for bisulfite sequencing analysis of a constellation of L1 loci, which included both young Ta-1 and older subfamilies of the L1Hs/L1PA1 family such as Ta-0 due to the high degree of L1 sequence conservation [134]: CpG analysis fw AAGGGGTCAGGGAGTTCCCTT; CpG analysis rev TGTCTGTGCCCTGCCCCCA.

Statistical analysis

For statistical analysis, we used R and RStudio [135,136]. Statistical analysis was performed using an unpaired two tailed t-test with unequal variance (Welch t-test), as indicated in the Figure legends. qRT-PCR data of HCV RNA copies in immunoprecipitated samples were tested for significance using the Mann-Whitney U test. Samples size (n) represents independent experiments if not stated otherwise in the Figure legends.