In hepatocytes, PLIN2 is the major protein coating lipid droplets (LDs), an organelle the hepatitis C virus (HCV) hijacks for virion morphogenesis. We investigated the consequences of PLIN2 deficiency on LDs and on HCV infection. Knockdown of PLIN2 did not affect LD homeostasis, likely due to compensation by PLIN3, but severely impaired HCV particle production. PLIN2-knockdown cells had slightly larger LDs with altered protein composition, enhanced local lipase activity and higher β-oxidation capacity. Electron micrographs showed that, after PLIN2 knockdown, LDs and HCV-induced vesicular structures were tightly surrounded by ER-derived double-membrane sacs. Strikingly, the LD access for HCV core and NS5A proteins was restricted in PLIN2-deficient cells, which correlated with reduced formation of intracellular HCV particles that were less infectious and of higher density, indicating defects in maturation. PLIN2 depletion also reduced protein levels and secretion of ApoE due to lysosomal degradation, but did not affect the density of ApoE-containing lipoproteins. However, ApoE overexpression in PLIN2-deficient cells did not restore HCV spreading. Thus, PLIN2 expression is required for trafficking of core and NS5A proteins to LDs, and for formation of functional low-density HCV particles prior to ApoE incorporation.
Long considered as inert lipid storage organelles, lipid droplets (LDs) have gained interest due to their critical involvement not only in lipid metabolism and metabolic disorders, but also in intracellular trafficking pathways, inflammatory responses and host–pathogen interaction. LDs have an organic core of triglycerides (TGs) and sterol esters and are surrounded by an amphipathic phospholipid monolayer with proteins embedded on the surface (Fujimoto and Parton, 2011). The most abundant proteins associated to LDs belong to the perilipin (PLIN) protein family (PLIN1–PLIN5) (Kimmel et al., 2010). PLIN proteins are important regulators of cellular lipid metabolism directly controlling how and when cells (and tissues) store, mobilize and utilize lipids. PLIN2 (also known as ADRP), the main LD coat protein in hepatocytes, is constitutively found on the LD surface and, if unbound, is rapidly degraded by the proteasome (Takahashi et al., 2016). PLIN2 stabilizes and protects LDs from degradation by lipases or by the autophagic machinery (Kaushik and Cuervo, 2015; Listenberger et al., 2007). PLIN2 as well as PLIN3 (also known as Tip47) are substrates for chaperone-mediated autophagy; degradation of both is a prerequisite for breakdown of LDs by lipolysis or macrolipophagy (Kaushik and Cuervo, 2015, 2016). In line with this, overexpression of PLIN2 in cell lines causes an accumulation of neutral lipids and LDs owing to reduced TG turnover (Imamura et al., 2002; Listenberger et al., 2007). In mice, reduced PLIN2 levels are associated with lower levels of TGs and protection against diet-induced steatosis (Chang et al., 2006; Libby et al., 2016; McManaman et al., 2013; Tsai et al., 2017a).
One leading cause of liver diseases, such as liver cirrhosis and hepatocellular carcinoma, is hepatitis C virus (HCV) infection. Approximately 71 million people are viraemic and 0.4 million people die each year from HCV-related complications. Direct-acting antivirals (DAAs) induce viral clearance in ∼95% of the patients, a major improvement over interferon-based therapies, but the treatment is extremely costly and the accessibility in high-prevalence countries is limited. Hepatic steatosis is frequently observed in patients suffering from chronic HCV infection, and virus replication is intertwined with lipid metabolism of the liver (Negro, 2014; Paul et al., 2014). The structural capsid protein core and the non-structural protein 5A (NS5A) of HCV localize to cytosolic LDs when expressed as single proteins in uninfected cells. In infected cells, they initiate HCV assembly at the close-by membranes of the ER (Barba et al., 1997; Miyanari et al., 2007; Shi et al., 2002). Trafficking of core and NS5A proteins to LDs is mediated by diacylglycerol acyltransferase 1 (DGAT1), which catalyzes the final step in TG biosynthesis; core protein additionally requires the cytosolic phospholipase A2 (cPLA2) enzyme (Camus et al., 2013; Herker et al., 2010; Menzel et al., 2012). Inhibition of either enzyme leads to strong reduction in HCV particle production. Upon infection, all components of the viral replication machinery are found in close proximity to LDs (Miyanari et al., 2007). PLIN3 directly interacts with HCV NS5A and has been shown to be vital for HCV RNA replication (Ploen et al., 2013a; Vogt et al., 2013). Infectious viral particles are lipoviroparticles containing neutral lipids and apolipoproteins (Andre et al., 2002; Merz et al., 2011). Mobilization of lipids from LDs for lipoviroparticle formation is mediated by the lipase co-activator comparative gene identification 58 (CGI-58, also known as ABHD5) (Vieyres et al., 2016) and intact lipoprotein synthesis and secretion are essential for production of infectious particles (Gastaminza et al., 2008; Huang et al., 2007; Jiang and Luo, 2009; Lee et al., 2014). In addition, factors recruited to LDs upon HCV infection participate in virion morphogenesis, such as the phospholipid-binding protein annexin A3, which mediates re-routing of ApoE (Rösch et al., 2016). Similar to what is seen upon PLIN2 overexpression, HCV core protein causes lipid accumulation by inhibiting adipose triglyceride lipase (ATGL; also known as PNPLA2)-mediated lipid mobilization (Camus et al., 2014; Harris et al., 2011). But core can also displace PLIN2 from the LD surface (Boulant et al., 2008). Regarding HCV replication, overexpression of PLIN2 leads to an increase in HCV replication while its knockdown using siRNAs has inconsistent effects (Branche et al., 2016; Zhang et al., 2016).
Here, we revisited and investigated the consequences of PLIN2 knockdown on lipid metabolism and on HCV replication.
PLIN2 is required for efficient HCV infection
LDs are essential cellular organelles for HCV infection (Miyanari et al., 2007; Paul et al., 2014). The most prominent protein on the surface of LDs in hepatocytes and hepatoma cells permissive for HCV is PLIN2 (Brasaemle et al., 1997; Fujimoto et al., 2004). Previous reports on the interaction of PLIN2 with HCV are contradictory (Branche et al., 2016; Zhang et al., 2016). To investigate whether PLIN2 itself or lipid metabolic processes regulated by PLIN2 are required for HCV infection, we generated lentiviral constructs encoding six different shRNAs targeting PLIN2 and a non-targeting control (shNT) to transduce the hepatoma cell line Huh7.5 (Fig. 1A). The shRNAs induced variable PLIN2 knockdown levels and shRNA #3 and #4 reduced and increased cell growth of the transduced cells, respectively (Fig. 1B,C). Next, cells transduced with the four shRNAs that did not affect cell growth were inoculated with a low multiplicity of infection (MOI) of an HCV Jc1 reporter strain carrying an EGFP fluorescent reporter inserted between NS5A and NS5B with a duplicated protein cleavage site (Jc1NS5AB-EGFP) (Webster et al., 2013), and analyzed by flow cytometry at up to 6 days post infection (dpi). HCV spreading was impaired in shPLIN2-transduced cells and correlated with the PLIN2 knockdown levels (Fig. 1D). Therefore, PLIN2 expression is required for efficient HCV infection. As shPLIN2 #2 slightly impacted cell viability, we chose shPLIN2 #1 (from now on referred to as shPLIN2) for mechanistic studies. shPLIN2 induced a stable knockdown for at least 15 days (Fig. 1E). To avoid adaptation, we performed most of the experiments at 5 days after transduction with the lentiviral constructs.
Production of infectious HCV particles depends on PLIN2 expression
Next, we investigated the impact of PLIN2 silencing on different steps of HCV replication. To probe viral RNA replication, we transfected shRNA-transduced cells with in vitro transcribed RNA of an envelope-deleted Jc1 strain encoding a firefly luciferase reporter (Jc1ΔE1E2NS5AB-Luc). Knockdown of PLIN2 only slightly reduced HCV RNA replication early after transfection of viral RNA (Fig. 1F). Therefore, HCV RNA replication and translation are mostly independent of PLIN2 expression. To investigate the production of infectious virions, we electroporated full-length Jc1NS5AB-EGFP RNA, verified equal transfection rates 3 days post electroporation (dpe), and harvested the supernatant to determine the HCV RNA copy number, release of the viral capsid protein core and the infectivity in the supernatant (50% tissue culture infectious dose, TCID50). The amount of HCV RNA, of the capsid protein, and infectivity were all significantly reduced in the supernatant of cells lacking PLIN2 (Fig. 1G). In rescue experiments, production of infectious viral progeny was partially restored in PLIN2-knockdown cells upon expression of an shRNA-resistant mutant but not upon expression of wild-type PLIN2 (Fig. S1). Hence, expression of the LD protein PLIN2 is required for efficient production of HCV progeny.
PLIN3 and ATGL levels increase at LDs in PLIN2-knockdown cells
PLIN2 is the most abundant protein coating LDs in the liver (Brasaemle et al., 1997; Fujimoto et al., 2004). In contrast, the exchangeable PLIN3 predominantly localizes to LDs under conditions of LD biogenesis (Bulankina et al., 2009; Wolins et al., 2001). PLIN3 protein levels increase at LDs in oleate-loaded PLIN2-knockdown cells (Bell et al., 2008; Sztalryd et al., 2006) and in lipid-rich fractions of HCV-replicon cells owing to direct interaction with the viral NS5A protein (Vogt et al., 2013). We investigated how loss of PLIN2 affects the protein composition of LDs under steady-state conditions and during HCV infection. We transduced HCV-infected and uninfected cells with lentiviral shRNAs, isolated LDs by density-gradient centrifugations and analyzed the protein levels of LD-associated proteins (Fig. 2A). To verify equal protein levels for LD fractions, which are devoid of reliable markers, we performed silver staining prior to western blot analysis. The PLIN2 shRNA reduced PLIN2 protein levels in input and in LD fractions (Fig. 2B). In the input, we did not detect changes in PLIN3 protein levels. In contrast, LDs isolated from PLIN2-knockdown cells had more PLIN3 compared to the control in both HCV-infected and uninfected cells. This indicates that PLIN3 is recruited to LDs lacking PLIN2 even in the absence of oleate loading. Both PLIN2 and PLIN3 can counteract the breakdown of TGs (Kaushik and Cuervo, 2015), and simultaneous knockdown of both proteins leads to increased recruitment of the TG lipase ATGL and its co-activator CGI-58 (Bell et al., 2008). ATGL, and less consistently CGI-58, increased in LD fractions after knockdown of PLIN2 in both uninfected and HCV-infected cells without concomitant changes in the input fractions (Fig. 2B). Therefore, knockdown of PLIN2 is sufficient to cause relocalization of the lipase complex even though PLIN3 levels increase in LD fractions.
Lack of PLIN2 increases local lipolysis rates of isolated LDs and increases mitochondrial β-oxidation capacity
To investigate whether increased levels of LD-localized ATGL increases local lipolysis, we performed in vitro self-digestion experiments with isolated LDs (Fig. 2C). After isolation and quantification of TG and protein content, LDs were incubated in the presence of fatty-acid-free bovine serum albumin (BSA) to accelerate lipolysis rates. After self-digestion, levels of released free fatty acids (FFA) were determined. Indeed, we detected higher amounts of fatty acids released from LDs isolated from PLIN2-knockdown cells compared to LDs isolated from control cells (Fig. 2C). To examine whether the increased local lipolysis rate affects steady-state lipid content, we measured the TG content of the cells, but did not detect significant differences (Fig. 2D). To address how PLIN2-deficient cells handle excess fatty acids, cells were incubated with oleate. TG levels increased in both shNT- and shPLIN2-expressing cells after oleate loading, as expected, but PLIN2-knockdown cells had slightly less TGs indicating a reduced capacity to store TGs (Fig. 2D). To analyze whether the excess fatty acids are degraded by β-oxidation, we incubated cells with 14C-labeled palmitate and measured the production of 14CO2 and 14C-containing acid-soluble metabolites (ASMs). Intriguingly, cells depleted of PLIN2 had significantly increased maximal levels of β-oxidation (Fig. 2E). To confirm that all fatty acids are channeled to β-oxidation, cells were treated with a DGAT1 inhibitor (DGAT1i) to prevent esterification of palmitate. β-oxidation rates were essentially the same with and without DGAT1i and elevated in PLIN2-knockdown cells. In conclusion, PLIN2-deficient cells have increased LD-associated lipase activities and preferentially degrade excess fatty acids rather than store them as TGs.
PLIN2-deficient cells have slightly larger LDs without changes in total LD content
Overexpression of PLIN2 in Huh7 cells alters LD morphology (Branche et al., 2016; Zhang et al., 2016). We investigated the impact of PLIN2 depletion on LDs by performing 3D spinning disk confocal microscopy (Fig. 3A). To knockdown PLIN2 we used an shRNA construct expressing a puromycin-resistance gene that efficiently suppressed PLIN2 expression (Fig. 3B). In addition, cells were transduced with the HCV RFP–NLS–IPS reporter to monitor HCV infection (Jones et al., 2010). This reporter indicates uninfected cells through mitochondrially localized RFP–NLS–IPS. After infection, the viral protease NS3–4A cleaves the IPS target site and RFP–NLS translocates to the nucleus. We recorded z-stacks and determined individual LD volumes as well as the number of LDs per cell (Fig. 3C). PLIN2 knockdown caused a significant increase in average LD volumes (mean, shNT=0.132 µm3 versus shPLIN2=0.156; median, shNT=0.088 versus shPLIN2=0.103). In non-infected cells, this was accompanied by a slight decrease of the numbers of LDs per cell, resulting in no change in total LD volume per cell. PLIN2 deficiency led to a similar increase in LD volumes in HCV-positive as in HCV-negative cells (mean, shNT=0.139 µm3 versus shPLIN2=0.152; median, shNT=0.090 versus shPLIN2=0.100) without significant changes in the number or total LD volume per cell. Therefore, even though PLIN2 deficiency enhances local fatty acid flux, LDs are only slightly affected, indicating compensation through enhanced recruitment of PLIN3.
LDs and HCV-induced vesicular replication organelles are surrounded by double membranes in cells lacking PLIN2
Next, we examined PLIN2-knockdown cells by electron microscopy (EM) to analyze LDs and their surrounding membrane structures. Tight interactions of LDs with ER membranes have been reported in cells overexpressing Rab18 (Ozeki et al., 2005). Mechanistically, this was linked to PLIN2 levels at LDs as the close apposition of LDs with ER cisterna was recapitulated by depleting PLIN2 in 3T3 cells.
Here, we used Huh7.5 cells expressing the EGFP–NLS–IPS HCV reporter that were transduced with shRNAs targeting PLIN2 prior infection with HCV Jc1. HCV-infected and uninfected shRNA-transduced (mCherry-positive) target cells were identified by epifluorescence and the samples were fixed and processed for EM with a modified osmium-thiocarbohydrazide-osmium (OTO) method (Seligman et al., 1966) for enhancing the contrast of membranes and LDs (Fig. 4A, Figs S2–S5). In shNT-transduced cells, HCV infection led to the typical membranous web formation with single-, double- and multi-membrane vesicles in close proximity to LDs (cyan and yellow arrowheads). LDs were either isolated or close to ER cisterna (black arrowheads). Interestingly, in HCV-infected PLIN2-deficient cells, LDs as well as the vesicular structures induced by HCV were often found in tight association with double-membrane structures that formed almost closed sacs connected to ER membranes (red arrowheads). These LD-associated membrane sacs were almost never observed in cells expressing PLIN2. The close association of LDs with double-membrane sacs also occurred in uninfected PLIN2-knockdown cells, but without the vesicular structures induced by infection. The double-membrane sacs contained one or more LDs with tight contact sites between the organelles. For some areas, the inner membrane of these sacs was fused with the phospholipid monolayer of LDs inside the sacs (white arrowheads). Quantification of the LD phenotype revealed that more than 50% of LDs were tightly associated with double-membrane sacs in both HCV-infected cells and uninfected controls (Fig. 4B,C). To analyze the 3D structure of the double-membrane sacs, we performed electron tomography (ET). HCV-infected cells expressing PLIN2 shRNAs were identified by epifluorescence and fixed and processed for ET. As illustrated in the colored 3D reconstruction, LDs (yellow) and the vesicular HCV replication structures (cyan) were tightly enclosed by double membranes (red) that were in continuity with ER cisterna (Fig. 5A; Movies 1–3). Membrane sacs devoid of the vesicular structures displayed an even closer apposition to LDs in uninfected cells.
To further corroborate the origin of the membrane sacs, we performed correlative light and electron microscopy (CLEM) of cells stained with fluorescent ER-Tracker or LysoTracker, or expressing GFP–LC3B (microtubule associated protein 1 light chain 3B) as a marker for autophagosomes. Samples were analyzed by live-cell spinning disk confocal microscopy and directly fixed and processed for EM. ER signals colocalized with and were in close apposition to LDs in PLIN2-knockdown and control cells (Fig. 5B). We then focused on LDs surrounded by double-membrane sacs identified in the electron micrographs. Z-stacks of fluorescence light microscopy images of these LDs revealed cup-shaped ER structures at LD edges (red arrowheads) and close-by ER cisterna (black arrowheads). LDs were in close contact with ER cisterna in cells expressing PLIN2 (black arrowheads).
In contrast, we did not observe colocalization of membrane-surrounded LDs with either LysoTracker or GFP–LC3 (Fig. 5B). As GFP–LC3 might not be representative of endogenous LC3, we additionally investigated LD isolations. We did not detect enrichment of endogenous LC3 in LD fractions of PLIN2-depleted or control cells (Fig. 5C). Therefore, the double-membrane sacs enclosing LDs and HCV replication organelles are likely derived from ER cisterna that stay attached to LDs in cells lacking PLIN2.
HCV protein trafficking to LDs is impaired in cells depleted of PLIN2
HCV core and NS5A proteins localize to LDs to initiate capsid assembly and virion morphogenesis (Paul et al., 2014). Hindering trafficking of core or NS5A to LDs either by mutation or by inhibition of critical host factors results in decreased infectious particle production (Boulant et al., 2007; Herker et al., 2010; Liefhebber et al., 2014; Menzel et al., 2012; Miyanari et al., 2007). Vice versa, cells infected with mutant HCV viruses defective in virion morphogenesis often accumulate core at LDs (Gentzsch et al., 2013; Zayas et al., 2016). To investigate whether LD localization of viral proteins is affected by PLIN2 deficiency, we first analyzed levels of core protein in LD fractions from HCV-infected cells. While we did not observe any changes in the cell homogenate, we detected a striking ∼75% decrease of levels of core in LD fractions from PLIN2-knockdown cells (Fig. 6A). In HCV Jc1-infected cells core localizes to LDs and is readily used for virion morphogenesis (Shavinskaya et al., 2007), limiting the amount of core at LDs. Therefore, we ectopically expressed core and investigated the localization in PLIN2-deficient cells. Core was readily detected in LD fractions from control cells but almost undetectable in LD fractions of PLIN2-knockdown cells (Fig. 6B). Intriguingly there is a clear correlation between core and PLIN2 levels in LD fractions, while PLIN3 and PLIN2 show an inverse correlation (Fig. S6A,B). These results reveal a severe trafficking defect of core protein to LDs in the absence of PLIN2 expression. When we probed LDs from HCV-infected cells for NS5A, we found a similar decrease of protein levels in LD fractions from PLIN2-knockdown cells (Fig. 6A).
To further validate the trafficking defect of the viral proteins, we performed immunofluorescence studies of HCV-infected cells and probed for core and NS5A localization. We quantified LD localization by determining the Manders' overlap coefficients, which indicate the amount of signal of one channel that overlaps with signal from the other channel. While core mainly localizes to LDs, NS5A displays both LD localization and a punctate ER and Golgi staining pattern. Intriguingly, NS5A displayed significantly reduced colocalization with LDs in cells lacking PLIN2 (Fig. 6C). In contrast, although we observed less core at LDs in cell fractionation experiments, we did not observe significantly less colocalization between core and LDs by microscopy (Fig. 6C). However, when we analyzed core signal intensity around LDs, we found slight differences between control and PLIN2-knockdown cells. The normalized signal intensity profiles revealed that core more tightly surrounded LDs in control cells than in PLIN2-deficient cells (Fig. 6D). To analyze whether LD-surrounding membranes sacs are lost during our LD isolation procedure, we analyzed LD fractions by EM and did not observe differences between control and PLIN2-deficient cells, indicating that the membrane sacs are stripped off during isolation (Fig. S6C). These results indicate that core is localizing to the double-membrane sacs that surround LDs in cells lacking PLIN2 while NS5A is not. The difference between core and NS5A might be due to the differences in how core and NS5A interact with membranes and the much stronger LD affinity of HCV core protein. Of note, the resolution limit of confocal microscopy does not allow distinguishing between the localization at the LD surface or the localization to surrounding membrane sacs. As core and NS5A both depend on DGAT1 activity for trafficking to LDs, we probed DGAT1 localization, activity, and interaction with the viral proteins in PLIN2-deficient and control cells (Fig. S7). Of note, core and DGAT1 interact at the ER prior to core trafficking to LDs. Accordingly, we did not detect a defect in core–DGAT1 or NS5A–DGAT1 interaction in co-immunoprecipitation experiments. DGAT1 localization is also independent of PLIN2 expression. We additionally probed DGAT1 and DGAT2 activity in a cell-based assay by using specific inhibitors, and found that the relative activity of DGAT1 and DGAT2 was similar in control and in PLIN2 -knockdown cells. Therefore, it is likely that the membrane sacs surrounding LDs in PLIN2-deficient cells restrict access for the viral proteins.
We next investigated whether the morphological changes around LDs and the restricted LD access for core affect its multimerization and envelopment. First, we performed 2D Blue native PAGE to detect complexes of core ranging from low molecular mass (LMM) to high molecular mass (HMM) of PLIN2-knockdown and control cells transfected with HCV RNA as described previously (Gentzsch et al., 2013; Rösch et al., 2016). Core multimerization was intact in cells lacking PLIN2 (Fig. 6E). Using the same experimental set-up, we analyzed envelopment of core: membrane-protected proteins are resistant to proteinase treatment, while cytosolic unprotected proteins are readily degraded. Core protein was partially resistant against proteinase K treatment in contrast to NS5A, which served as a positive control for protease digestion. When cells were treated with Triton X-100 prior to proteolytic digestion to disrupt all membranes, core protein was completely degraded. Comparing shPLIN2-transduced with control cells did not reveal differences in core envelopment (Fig. 6F). Therefore, core protein can multimerize and is protected by membranes even in the absence of prior LD localization.
We then probed the interaction of the viral envelope protein E2 with ApoE, which occurs after core multimerization and envelopment, as those steps are not impaired in ApoE-deficient cells (Lee et al., 2014). We performed co-immunoprecipitation experiments using a full-length HCV Jc1 construct expressing Flag–E2 and probed for co-precipitation of ApoE (Fig. 6G). Flag–E2 interacted with ApoE in the presence or absence of PLIN2, albeit less efficiently. In uninfected cells, ApoE mainly localizes to the Golgi. We previously observed annexin A3-mediated redistribution of ApoE to non-Golgi compartments in HCV-infected cells (Rösch et al., 2016). When we investigated if ApoE in PLIN2-knockdown cells changes its localization pattern in response to HCV infection, we noted that ApoE staining was weaker in PLIN2-knockdown compared to control cells. Nevertheless, we detected less colocalization between ApoE and GM130 (GOLGA2), a marker for the Golgi compartment, in cells infected with HCV as compared to uninfected cells, independently of PLIN2 expression (Fig. 6H).
To investigate whether multimerized and enveloped core together with an intact E2–ApoE interaction correlates with infectiousness, we determined the intracellular infectivity of PLIN2-knockdown and control cells transfected with HCV RNA encoding a Gaussia luciferase reporter (Jc1p7-Gluc-2A-NS2). PLIN2-knockdown cells produced intracellular particles that were less infectious as compared to controls (Fig. 6I). In addition, virions from PLIN2-depeleted cells had a lower specific infectivity (Fig. 6I). The specific infectivity of HCV lipoviroparticles depends on their density (Bartenschlager et al., 2011). Thus, we determined the density distribution of intracellular and extracellular lipoviroparticles by iodixanol density gradient centrifugations followed by TCID50 titration. The density peak of intracellular infectious particles shifted towards higher densities in PLIN2-deficient cells, indicating less lipidation (Fig. 6J). In addition, the density profile of core-containing particles isolated from PLIN2-knockdown cells showed a lower amount of particles at densities below 1.10 g/ml (Fig. 6J). Strikingly, very-low-density HCV particles (<1.05 g/ml) were almost completely absent from supernatants of PLIN2-deficient cells (Fig. 6K). Overall, the distribution profile resembled the density profile of HCV particles produced in cells lacking ApoE (Lee et al., 2014); thus, we next investigated apolipoproteins.
PLIN2-knockdown cells secrete less ApoE-containing lipoproteins independently of HCV infection
Neutral lipids stored in LDs are a major source for lipidation of lipoproteins. We speculated that the altered LD protein composition and structure might affect lipoprotein metabolism, and could explain the reduced ApoE staining pattern and HCV particle density and infectivity. Lipoprotein secretion starts with co-translational lipidation of ApoB and formation of luminal LDs, as well as translation of ApoE into the ER lumen (Lehner et al., 2012). The formation of mature very-low-density lipoproteins (VLDLs) likely occurs during secretion via the Golgi. Huh7-derived cell lines display a defect in VLDL metabolism; they produce VLDLs that are less lipidated than normal VLDLs and secrete ApoB and ApoE independently (Meex et al., 2011; Schöbel et al., 2018). Interestingly, the formation of infectious HCV particles in Huh7-derived cells depends on ApoE but not ApoB (Jiang and Luo, 2009). To examine the expression and secretion levels of apolipoproteins, we performed western blot analysis of cell lysates and supernatants. Strikingly, ApoE, but not ApoB levels, depended on the expression of PLIN2 both in cell lysates and, were even more pronounced, in culture supernatants (Fig. 7A). We detected reduced protein levels of ApoE in supernatants of cells depleted of PLIN2, correlating with PLIN2-knockdown levels induced by different shRNA constructs (Fig. S6D). HCV-infected cells displayed similar changes in expression and secretion of ApoE upon PLIN2 depletion (Fig. 7B). Of note, ApoB was barely detectable in supernatants of HCV-infected cells, a phenotype that has been described before (Domitrovich et al., 2005; Mancone et al., 2012; Tsai et al., 2017b). The reduction of ApoE levels was not caused by changes in mRNA expression (Fig. 7C). Therefore, we analyzed protein degradation with specific inhibitors; leupeptin for lysosomal pathways and MG132 for proteasomal pathways. Interestingly, inhibition of lysosomal but not proteasomal protein degradation restored ApoE levels in cells with reduced PLIN2 expression, both in control and in HCV-infected cells (Fig. 7D,E). However, while treatment with leupeptin and with an inhibitor of autophagy, 3-methyladenine (3-MA), restored intracellular ApoE levels, we could not rescue the amount of secreted ApoE (Fig. S6E), indicating that ApoE that is destined for degradation cannot be re-routed for secretion. In contrast to HCV particles, secreted ApoE was of similar density in cell lysates and in culture supernatants of PLIN2-knockdown and control cells as determined by density gradient experiments (Fig. 7F). To investigate whether the reduced availability of ApoE explains the HCV phenotype, we performed rescue experiments by overexpressing ApoE. Through lentiviral delivery, we achieved strong expression and secretion of ApoE in both PLIN2-deficient and control cells (Fig. 7G). We then inoculated these cells with HCV Jc1NS5AB-EGFP and followed viral spreading. PLIN2-knockdown reduced HCV spreading in both ApoE-overexpressing and control cells to a similar extent (Fig. 7H). Therefore, PLIN2 expression is required for secretion of ApoE-containing lipoproteins and for morphogenesis of infectious HCV lipoviroparticles. However, HCV morphogenesis cannot be rescued by ApoE overexpression, indicating that the lack of core and NS5A trafficking to LDs and insufficient lipidation are the main reasons for the impaired production of infectious HCV progeny in PLIN2-deficient cells.
During the replication cycle of HCV, the viral proteins core and NS5A traffic onto the LD surface, a process essential for successful viral progeny production (Boulant et al., 2007; Miyanari et al., 2007). Active TG synthesis is vital for this translocation as inhibition of TG biosynthesis results in an HCV assembly block (Camus et al., 2013; Herker et al., 2010; Liefhebber et al., 2014). Proteins present at the LD surface have been previously described as HCV host factors with, for example, PLIN3 and Rab18 required for both successful HCV RNA replication and assembly of virions (Ploen et al., 2013a,b; Salloum et al., 2013; Vogt et al., 2013). Here, we investigated the role of PLIN2, the most abundant protein that coats LDs in hepatocytes. We describe that PLIN2 expression is required for the production of infectious HCV particles (see model in Fig. 8).
Previous publications investigated PLIN2 in HCV infection and had inconsistent findings: knockdown of PLIN2 using siRNAs had either no or variable effects on HCV infection and on LD morphology (Branche et al., 2016; Zhang et al., 2016). In contrast, when overexpressed, PLIN2 enhanced virus infection, by upregulating the expression of the HCV entry factor occludin (OCLN), and increased average LD volumes. The reduction of PLIN2 mRNA and protein levels by siRNAs was only moderate. Our results using the successful and strong shRNA-mediated knockdown clearly point to a pro-viral role for PLIN2 in HCV replication. HCV spreading correlated very well with knockdown levels induced by different shRNAs targeting PLIN2. In line with the function of LDs in particle production, we found that PLIN2 is mainly required for morphogenesis of progeny virions. However, the sequence and timing of morphogenesis is still unknown.
PLIN2 deficiency impaired trafficking of the viral proteins core and NS5A to LDs. Core expressed either in HCV-infected cells or as a single viral protein was nearly undetectable in LD fractions despite similar protein levels in cell homogenates. In line with this, less NS5A was present in LD fractions of PLIN2-knockdown cells. This was the case even though all other probed LD-associated host proteins displayed increased LD localization in PLIN2-knockdown cells. But although NS5A displayed significantly reduced colocalization with LDs in cells lacking PLIN2, we did not observe a lower degree of colocalization of core with LDs. Nevertheless, the analysis of signal intensities revealed that core more tightly surrounded LDs in control cells than in PLIN2-deficient cells. These results indicate that core localizes to the double-membrane sacs that surround LDs in cells lacking PLIN2 while NS5A does not. This mislocalization of core and NS5A correlated with reduced levels of intracellular infectious viral particles. Interestingly, lack of LD localization did not interfere with multimerization and envelopment of core, processes that have been shown to occur in the absence of virion production (Ai et al., 2009). In addition, the envelope glycoprotein E2 still interacted with ApoE. We previously failed to detect an interaction between E2 and ApoE in cells only expressing the structural proteins (Rösch et al., 2016) even though other groups have reported it (Lee et al., 2014) indicating that E2 and ApoE can indeed interact without virion assembly. Taken together, multimerization and membrane protection of core, as well as the interaction of E2 with ApoE and its mislocalization to non-Golgi compartments all occur in the absence of core trafficking to LDs in PLIN2-deficient cells, but do not result in the formation of intracellular and secreted infectious virions. In contrast, core localization at LDs clearly correlates with intracellular infectivity.
HCV particles mature during assembly or secretion by incorporating lipids and apolipoproteins, most importantly ApoE (Chang et al., 2007; Gastaminza et al., 2008). PLIN2-knockdown cells produced intracellular particles that were less infectious with decreased specific infectivity and lower amounts of very-low-density lipoviroparticles. This fits with experiments in ApoE-deficient cells, which supported HCV capsid formation and envelopment but where maturation into lipoviroparticles did not occur (Lee et al., 2014), indicating defective lipidation and/or incorporation of ApoE in PLIN2-knockdown cells. Indeed, we detected a decrease of intracellular and extracellular ApoE levels after PLIN2 knockdown both in uninfected and HCV-infected cells. Inhibiting lysosomal but not proteasomal protein degradation restored intracellular but not secreted ApoE levels. In line with this, lysosomal inhibitors increased ApoE levels in HepG2 cells and macrophages (Deng et al., 1995; Ye et al., 1993). A recent report even suggested that HCV replication itself causes enhanced autophagic degradation of ApoE (Kim and Ou, 2018). Of note, the strongest effect on ApoE degradation was seen with subgenomic replicons that do not produce virions. It is also well known that Huh7-derived cell lines secrete ApoE and ApoB independently (Meex et al., 2011; Schöbel et al., 2018; Takacs et al., 2017). Consistent with earlier reports, we detected less ApoB after infection with HCV (Domitrovich et al., 2005; Mancone et al., 2012; Tsai et al., 2017b), but in our hands ApoE remained stable during infection. However, mechanistically the effects of PLIN2 depletion on ApoE-containing lipoproteins and on HCV morphogenesis clearly differ: while HCV lipoviroparticles are of higher density in PLIN2-deficient cells there is no change in the density of ApoE-containing lipoproteins. And while ApoE levels are higher after inhibition of lysosomal degradation, total HCV core levels are unaffected by PLIN2 knockdown and inhibitor treatment. In addition, the defect in HCV morphogenesis cannot be rescued by ApoE overexpression, indicating that PLIN2 expression is required to allow core and NS5A trafficking to LDs and for the formation of functional low-density HCV particles prior to ApoE incorporation.
Fluorescence microscopy analysis of LDs in PLIN2-deficient cells revealed a slight increase in average LD volumes. These results corroborate findings in oleate-loaded cells showing that only simultaneous PLIN2 and PLIN3 knockdown severely changes LD morphology (Bell et al., 2008; Sztalryd et al., 2006). In EM and ET analyses of PLIN2-depleted cells, we noticed LDs that are surrounded by double-membrane sacs that are likely ER-derived, as indicated by CLEM. Interestingly, overexpression of Rab18 causes dissociation of PLIN2 from LDs concomitant with tight interactions of LDs with ER membranes (Ozeki et al., 2005). These membranes were termed LD-associated membranes (LAMs) and occur after depletion of PLIN2 from LDs in 3T3 cells. The membrane sacs we observed are likely LAMs and, in addition to LDs, enclose vesicular structures induced by HCV infection.
The near-complete PLIN2-deficiency induced by the shRNAs had no major effects on overall TG and LD content, similar to what has been described before (Bell et al., 2008). In western blot analysis of isolated LDs, we detected recruitment of PLIN3 to LDs in PLIN2-knockdown cells. PLIN3 likely compensates for the lack of PLIN2 to shield LDs from degradation. Less PLIN2 at LDs also caused an increased LD localization of the lipase ATGL and elevated LD-associated lipase activity. Previous publications have investigated the role of PLIN2 and PLIN3 proteins in the regulation of ectopic fat deposition after loading of hepatoma cells with fatty acids (Bell et al., 2008). PLIN3 localizes to LDs under conditions of LD biogenesis, and increased levels of PLIN3 at LDs were observed in PLIN2-deficient cells after oleate loading. Our results indicate that, under steady-state conditions, the regular turnover of LDs is sufficient to recruit PLIN3 to LDs. In contrast to our study, Bell et al. (2008) detected recruitment of the lipase complex and increased lipase activity, as measured by determining the amount of FFAs in the medium after oleate loading, only when both PLIN2 and PLIN3 were depleted simultaneously. In the case of single PLIN2 deficiency, liberated fatty acids are likely directly shuttled towards β-oxidation. When challenged with excess amounts of fatty acids, the capacity to esterify fatty acids to TGs was reduced in PLIN2-deficient cells, but degradation via β-oxidation was elevated. These findings corroborate recent data for oleate-loaded cultured PLIN2–/– myotubes revealing reduced TG levels are mostly attributable to increased ATGL activity and fatty acid oxidation rates (Feng et al., 2017).
How does PLIN2 deficiency interfere with ApoE metabolism? Biochemical studies in PLIN2−/− mice indicated elevated luminal TG levels (Chang et al., 2006). In addition, inhibition of ApoB degradation triggers tight association of ApoB-containing ER membranes that form crescents around LDs (Ohsaki et al., 2008). Remarkably, PLIN2 localized to ApoB-opposing sites of LDs and overexpression of PLIN2 abolished, whereas its downregulation augmented, the accumulation of ApoB and membranes around LDs. A defect in ER membrane and LD interaction concomitant with membrane alterations around LDs could cause the ApoE phenotype in PLIN2-deficient cells, where ApoE is degraded. Future studies will need to address the molecular mechanisms underlying ApoE destabilization in PLIN2-depleted cells.
In summary, our study demonstrates that PLIN2 expression is required for proper LD architecture that is needed for trafficking of viral proteins to LDs and for formation of functional low-density HCV particles prior to ApoE incorporation.
MATERIALS AND METHODS
Plasmids and HCV constructs
HCV Jc1 reporter constructs encoding fluorescent proteins or firefly luciferase between a duplicated NS5A–NS5B cleavage site, and Flag-tagged E2 or envelope-deleted versions (Jc1FLAG-E2, Jc1NS5AB-EGFP and Jc1FLAG-E2-NS5AB-EGFP, Jc1ΔE1E2 Jc1NS5AB-Fluc), as well as the secreted Gaussia luciferase reporter (Jc1p7-GLuc-2A-NS2) were as described previously (Eggert et al., 2014; Schöbel et al., 2018; Webster et al., 2013). The lentiviral constructs expressing the JFH1 core and NS5A (Rösch et al., 2016), and RFP-NLS-IPS (Jones et al., 2010) were as described previously. For EGFP-NLS-IPS, the EGFP was cloned into RFP-NLS-IPS via XbaI and BsrGI (the XbaI site was destroyed). PLIN2 and NT shRNAs were cloned into pSicoR-MS1 (mCherry) and pSicoR-Puro (target sequences: shNT 5′-GCGCGATAGCGCTAATAATT-3′, shPLIN2#1 5′-GCTAGAGCCGCAAATTGCA-3′, shPLIN2#2 5′-GGTTCAGAAGCCAAGTTATTA-3′, shPLIN2#3 5′-CAGAAGCTAGAGCCGCAAATT-3′, shPLIN2#4 5′-TGGTTCAGAAGCCAAGTTATT-3′, shPLIN2#5 5′-CAGCCATCAACTCAGATTGTT-3′, shPLIN2#6 5′-TGAAGGATTTGATCTGGTT-3′). PLIN2WT and PLIN2MT were cloned into pSicoR-MS1 replacing mCherry by overlap extension PCR using pCMV6-XL4 PLIN2 (Origene) as a template (primers: PLIN2fw 5′-CTGTGACCGGCGCCTACGATGGCATCCGTTGCAGTT-3′, PLIN2 MTrev 5′-AGGTATTGGCAACTGCAATCTGTGGTTCCAGC-3′, PLIN2 MTfw 5′-GCAGTTGCCAATACCTATGCCT-3′, PLIN2rev 5′-TAGGTCCCTCGACGAATTTTAATGAGTTTTATGC-3′). ApoE was cloned into pSicoR-MS1 replacing mCherry with ApoE3 using pCMV4-ApoE3 (Addgene plasmid #87086; Hudry et al., 2013) as a template (primers: ApoEfw 5′-CGGCGCCTACGCTAGCATGAAGGTTCTGTGGGCT-3′, ApoErev 5′-GTCCCTCGACGAATTCTCAGTGATTGTCGCTGGGCAC-3′).
Cell lines, culture conditions, and viability assays
HEK293T cells were obtained from the American Type Culture Collection and Huh7.5 cells from Charles M. Rice (Rockefeller University, NY) were grown under standard cell culture conditions in high-glucose DMEM supplemented with 10% FBS (Biochrom Superior), 1% GlutaMax (Gibco), and 1% penicillin-streptomycin (Sigma). All cell lines were authenticated by STR fingerprinting and were tested for mycoplasma every 3–6 months. Cell viability was analyzed with CellTiter 96 AQueous One Solution Reagent (Promega).
Antibodies and reagents
All antibodies and reagents were obtained commercially and used as indicated: PLIN2 (ab52355), ApoE (ab52607), ApoB (ab31992), ATGL (ab2138S), CGI-58/ABDH5 (ab73551), PLIN3 (ab47639) [all Abcam, 1:1000 for western blotting (WB), 1:100 for immunofluorescence (IF)], PLIN2 (610102 Progen, 1:250 WB), tubulin (T6074, Sigma, 1:2000 WB), HCV core (clone C7-50, sc-57800), GM130 (sc-16268), DGAT1 (H-255, sc-32661) (all Santa Cruz Biotechnology, 1:250 WB, 1:25 IF), HCV NS5A (HCM-131-5, IBT, 1:500 WB, 1:100 IF), Flag (F7425, Sigma, 1:1000 WB), LC3B (D11, Cell Signaling, 1:1000 WB), Flag agarose (A2220 Sigma), Alexa 488-, Alexa 594-, and Alexa 647-conjugated secondary antibodies [all donkey, IgG (H+L), Life Technologies 1:1000–1:1500 IF], HRP-labeled secondary antibodies (Jackson Laboratories, 1:10,000 WB), BODIPY493/503 (D-3922), BODIPY 655/676 (B-3932) ER-Tracker Green (E34251), LysoTracker Green (L7526) (all Thermo Fisher), DGAT1 inhibitor (PF-04620110) and DGAT2 inhibitor (PF-06424439) (Sigma). Chemicals were purchased from AppliChem, Sigma, and Merck, if not noted otherwise.
Production of lentiviral particles
HCV infection assays
HCV viral stocks were prepared by electroporation of in vitro-transcribed HCV RNA into Huh7.5 cells as described (Herker et al., 2010; Rösch et al., 2016). The infectivity (TCID50) was assessed by serial limiting dilution on Huh7.5 cells stably expressing the HCV reporter RFP-NLS-IPS (Jones et al., 2010; Rösch et al., 2016) or using the Gaussia luciferase reporter strain (Jc1p7-Gluc-2A-NS2). HCV infection or HCV RNA transfection rates were measured with the Fortessa flow cytometer (BD Bioscience) and analyzed using FlowJo (Treestar). Replication of luciferase reporters was determined using Luciferase assay systems (Promega) and a Tecan plate reader.
For LD quantification, cells seeded on µ-dishes (Ibidi) were fixed with 2% paraformaldehyde prior to staining with BODIPY493/503 to visualize the LDs on a custom-made spinning disk confocal microscope (Nikon) equipped with a CFI Apo TIRF 100×1.49 NA objective, a dual-camera Yokogawa W2 spinning disk confocal scan head and Andor iXON 888 cameras. An Andor Borealis System ensured illumination flatness of 405, 488, 561 and 647 nm lasers. Images were deconvolved with 75 iterations and a high noise level with AutoQuant X2 (Media Cybernetics) were used for 3D reconstruction and volumetric analysis with the 3D objects counter function of Fiji (Bolte and Cordelières, 2006; Schindelin et al., 2012) and 3D reconstruction with Imaris (Bitplane). Objects below the resolution of the microscope and the acquisition settings (0.028 µm3) were excluded. Living cells were incubated with ER-Tracker or LysoTracker and BODIPY 655/676 for CLEM. For colocalization studies, cells seeded on coverslips were fixed, permeabilized with 0.1% Triton X-100, incubated in blocking solution (5% BSA, 1% fish skin gelatin, 50 mM Tris in PBS) and stained with the antibodies in blocking solution. The target cells were visualized either with the spinning disk confocal microscope or with an Ti2-A1R-HD plus laser-scanning confocal microscope (Nikon) using a galvano scanner x2 with a LU-NV series laser unit [405 nm, 488 nm, 561 nm and 647 nm; built-in acousto-optic tunable filter (AOTF)] and equipped with a CFI PlanApo 60×1.40 oil objective. For colocalization analysis, we used the coloc2 plugin of Fiji to determine the Manders' colocalization coefficients (Schindelin et al., 2012). Z-stacks of 3D images were deconvolved with the NIS-Elements AR software (Nikon, blind 3D deconvolution, noisy level with 30 iterations and 0 spherical aberration). To determine intensity profiles around LDs, intensities and distances of signals were measured using NIS-Elements AR software (Nikon). Intensity profiles were centered around the peak of the LD signals and normalized to the full width at half maximum (FWHM) signal intensity of LD signals in RStudio (RStudio Team, 2015).
Transmission electron microscopy, electron tomography, correlative light and electron microscopy and negative staining of LDs
Target cells grown in gridded µ-dishes (Ibidi) were visualized via live-cell epifluorescence or spinning disk confocal microscopy (Nikon) and fixed following a modified OTO fixation method (Hofmann et al., 2018; Seligman et al., 1966). Cells were fixed with EM-grade 2.5% glutaraldehyde and 1% osmium tetroxide in PBS for 1–16 h, washed with PBS and then treated with 0.1% thiocarbohydrazide for 30 min, followed by staining with 1% reduced osmium tetroxide in 1.5% potassium hexacyanoferrate for 15 min. Samples were treated with 1% gallic acid for 30 min, dehydrated with ethanol by progressive lowering of temperature, infiltrated with low viscosity epoxy resin (Epon 812), and polymerized at 60°C overnight. The embedded targeted cells were tracked back, trimmed and sectioned at 50–60 nm for transmission electron microscopy (TEM) and 350–450 nm for ET. LDs isolated as described below were subjected to negative staining as described previously (Pogan et al., 2018). TEM samples were visualized on a FEI Tecnai G20 Twin electron microscope at 80 kV. Micrographs were acquired using a Veleta-2K×2K side-mounted TEM CCD camera (Olympus). Representative TEM images were analyzed and assembled using ImageJ. Tomograms were acquired using a bottom-mounted 4K CCD camera (Eagle 4K, FEI) with tilt series from −65° to +65° (1° increments). 3D datasets were aligned and reconstructed with Inspect 3D Xpress software (FEI) using the sequential iterative reconstruction with 20 iterations. Rendering and visualization was performed with Imaris (Bitplane). For CLEM, spinning disk confocal images were deconvolved with NIS-Elements AR (Nikon, blind 3D deconvolution, noisy level with 25 iterations and 0 spherical aberration). Fluorescence and TEM images were aligned by hand in Adobe Photoshop.
Determination of LD-associated proteins and lipase activity
LDs for western blotting and in vitro self-digestion were isolated as described previously (Camus et al., 2014; Rösch et al., 2017; Schweiger et al., 2014). Briefly, cells were detached in PBS and lysed in hypotonic sucrose buffer (0.25 M sucrose, 1 mM EDTA, 1 mM DTT, pH 7.2, and 20 µg/ml leupeptin, 2 µg/ml antipain, and 1 µg/ml pepstatin) using a Dounce homogenizer. Post-nuclear supernatants were overlaid with an isotonic potassium phosphate buffer (0.1 M potassium phosphate pH 7.4, 100 mM KCl, 1 mM EDTA, 20 µg/ml leupeptin, 2 µg/ml antipain, and 1 µg/ml pepstatin) and centrifuged at 36,000 g for 2 h in an SW41 rotor (Beckmann). The LD fraction was harvested using a blunt bended cannula and further analyzed by SDS-PAGE, silver staining or western blotting, or were used to determine LD-associated lipase activity as described previously (Camus et al., 2014; Schweiger et al., 2014). TG concentrations in the isolated LDs were measured using the Infinity Triglyceride Kit (Thermo Fisher). 0.4–1 mM of TGs were used for self-digestion with 1% BSA (fatty acid free, Sigma) at 37°C with continuous shacking for 1 h. To solubilize the released FFAs, Triton X-100 was added to final concentration of 1%, incubated for 10 min at room temperature and, following centrifugation (20,000 g for 30 min), the FFA concentration of the underlying solution was determined with a NEFA kit (WAKO chemicals). The amount of released FFA reflects the LD-associated lipase activity.
Western blot and co-immunoprecipitation analysis
Cells were lysed in RIPA buffer [150 mM NaCl, 50 mM Tris-HCl pH 7.6, 1% NP-40, 0.5% sodium deoxycholate, 5 mM EDTA, protease inhibitor cocktail (Sigma)] for 1 h. To determine ApoE and ApoB secretion, cells seeded in a six-well plate were cultured in serum-free medium (Opti-MEM) overnight and the supernatant was directly used for western blot analysis. Proteins were transferred onto a nitrocellulose membrane (Amersham). For chemiluminescent detection, we used Lumi-Light western blotting substrate (Roche), SuperSignal West Femto (Thermo Fisher), and ECL hyperfilm (Amersham). Band intensities were quantified using the densitometric quantification function of Fiji (Schindelin et al., 2012).
For immunoprecipitation analysis, cells were lysed in NP-40 lysis buffer [50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, protease inhibitor cocktail (Sigma)] for 1 h on ice and, for DGAT1 immunoprecipitation, passed 10 times through a G23 needle. Clarified lysates were incubated with FLAG M2 affinity gel (Sigma) for 2 h at 4°C to capture Flag-tagged proteins. Beads were then washed five times with ice-cold NP-40 lysis buffer and analyzed by western blotting.
2D Blue native PAGE
2D Blue native PAGE was performed as described previously (Gentzsch et al., 2013; Rösch et al., 2016). 106 shRNA-transduced Jc1NS5AB-EGFP-electroporated cells were lysed in 80 µl native PAGE buffer (0.75 M aminocaproic acid, 50 mM Tris-Bis, pH 7.0) supplemented with 15 µl 10% n-dodecyl-β-D-maltopyranoside for 30 min on ice. Post-nuclear supernatants were supplemented with 20 µg/ml leupeptin, 2 µg/ml antipain, and 1 µg/ml pepstatin, mixed with 10 µl 5% Coomassie Brilliant Blue G and the same volume of 2× non-reducing sample buffer (62.5 mM Tris-HCl, pH 6.8, 25% glycerol, 0.1% Bromophenol Blue). 40 µl of each sample was loaded on a 4–20% polyacrylamide precast gradient gel (BioRad). After separation, the gel was incubated in 2× SDS sample buffer (150 mM Tris/HCl, pH 6.8, 1.2% SDS, 30% glycerol, 0.002% Bromophenol Blue, 15% β-mercaptoethanol) for 1 h. Gel stripes were placed horizontally on a 15% polyacrylamide SDS gel and core complexes were analyzed by western blotting.
Proteinase K digestion assay
Envelopment of the core protein was analyzed as described previously (Gentzsch et al., 2013; Rösch et al., 2016). shRNA-transduced cells were electroporated with Jc1NS5AB-EGFP RNA, lysed by freeze–thaw cycles, and treated with 100 µg/ml proteinase K in the absence or presence of 1% Triton X-100 and analyzed by western blotting.
Iodixanol gradient centrifugation
To determine HCV particle and ApoE density through centrifugation experiments, we used linear 6–56% iodixanol (Progen) gradients as described previously (Nielsen et al., 2006; Rösch et al., 2016; Schöbel et al., 2018). Jc1p7-Gluc-2A-NS2 RNA-transfected cells were harvested by trypsinization at 6 days post transduction (dpt), resuspended in 1 ml TNE lysis buffer (10 mM Tris-HCl pH 8, 150 mM NaCl, 2 mM EDTA), and lysed through multiple freeze–thaw cycles. Cell debris was removed by centrifugation (300 g for 5 min). Clarified culture supernatant was concentrated with polyethylene glycol. Cell lysate, unconcentrated (ApoE) or concentrated supernatant (TCID50, core) was loaded on top of a gradient and centrifuged in an SW41 rotor (Beckman) at 204,095 g for 18 h. From top to bottom, 500 µl fractions were harvested and used to determine the infectivity, core and ApoE protein levels (BioCat HCV Core Antigen and MabTech AB ApoE ELISA), and the density using a refractometer (DR 201-95, Krüss).
Measurement of TG content and mitochondrial β-oxidation
Total lipids were extracted with methanol–chloroform (Bligh and Dyer, 1959). The organic phase was supplemented with Triton X-100 in chloroform, dried under nitrogen and solubilized in water (2.5% Triton X-100 final concentration). TGs were measured using the Infinity Triglyceride Kit (Thermo Fisher).
We used 14C-labeled palmitate to measure the β-oxidation as described previously (Huynh et al., 2014). Cells were incubated in 0.3% BSA solution containing 100 µM non-radioactive palmitate, 0.4 µCi/ml 14C-labeled palmitate and 1 mM L-carnitine for 3 h. The cell supernatant was transferred into tubes containing 1 M perchloric acid and a paper disk soaked with 1 M NaOH inside the cap to capture released 14CO2. After 1 h at room temperature the radioactivity captured in the paper disks as well as the acid soluble metabolites were measured on a LS 6500 Scintillation Counter (Beckman).
RNA isolation and qRT-PCR analysis
Total RNA of cell lysates was isolated using TriReagent (Sigma) and cellular DNA digested with TURBO DNA-free DNase kit (Ambion). Viral RNA of the supernatant was isolated using Nucleo-Spin RNA Virus kit (Macherey Nagel). cDNA was synthesized using Superscript III reverse transcriptase (Invitrogen) with random hexamer primers (Qiagen) and RNaseOut (Thermo Fisher). cDNA was subjected to quantitative PCR (qPCR) using the Maxima SYBR Green master mix (Thermo Fisher) on a 7900HT Fast Real-time PCR System (Applied Biosystems). qPCR primers were selected from the Harvard primer bank (Wang et al., 2012).
For statistical analysis, we used R (R Core Team, 2015), RStudio (RStudio Team, 2015) and Prism (GraphPad). Statistical analysis was performed using unpaired two-tailed Welch's t-test, Mann–Whitney U-test, and, in case of normalized data, a one sample t-test, as indicated in the figure legends. Samples size (n) represents independent experiments.
We thank Ralf Bartenschlager (University of Heidelberg, Germany) for Jc1 constructs, Charles M. Rice (Rockefeller University, NY) for Huh7.5 cells and RFP- NLS-IPS, Takaji Wakita (National Institute of Infectious Diseases, Japan) for JFH1, Brian Webster and Warner C. Greene (Gladstone Institutes, CA) for the HCVcc reporter constructs, Matt Spindler (Gladstone Institutes) for pSicoR-MS1, and Boris Fehse for Lego-iCer2 (University Clinic Hamburg Eppendorf, Germany). We thank Rudolph Reimer and Carola Schneider from the Core facility Microscopy & Image Analysis for support.
Conceptualization: S.L., E.H.; Methodology: S.L., V.N.-D., E.H.; Investigation: S.L., C.G., V.N.-.D.; Writing - original draft: S.L., E.H.; Writing - review & editing: S.L., V.N.-D., E.H.; Visualization: S.L., V.N.-D., E.H.; Supervision: E.H.; Funding acquisition: E.H.
This work was in part supported by funds from the Deutsche Forschungsgemeinschaft (HE 6889/2). The Heinrich Pette Institute, Leibniz Institute for Experimental Virology is supported by the Free and Hanseatic City of Hamburg and the Federal Ministry of Health (Bundesministerium für Gesundheit). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The authors declare no competing or financial interests.