|Home | About | Journals | Submit | Contact Us | Français|
Hepatitis C virus (HCV) is a major cause of liver disease worldwide with steatosis, or “fatty liver”, being a frequent histologic finding. In previous work, we identified sequence polymorphisms within domain 3 (d3) of genotype 3 HCV Core protein that correlated with steatosis and in vitro lipid accumulation. In this study, we investigated the sufficiency of d3 to promote lipid accumulation, the role of HCV genotype in d3 lipid accumulation and the subcellular distribution of d3.
Stable cell lines expressing green fluorescent protein (GFP) fusions with HCV Core d3 from genotype 3 steatosis (d3S), non-steatosis (d3NS) and genotype 1 (d3G1) isolates were analyzed by immunofluorescence (IF), Oil Red O (ORO) staining and triglyceride (TG) quantitation
Cells expressing d3S had significantly more ORO than d3NS or d3G1 cells (p values: 0.02 and <0.0001 respectively) as well as TG (p=0.03 and 0.003 respectively). IF analysis showed domain 3 does not co-localize to lipid droplets but partially co-localizes to the Golgi.
Our results suggest that HCV Core d3 is sufficient to mediate the accumulation of lipid by a mechanism that is independent of domains 1 and 2. Our results also suggest that altered lipid trafficking may be involved.
Hepatitis C virus (HCV) is a major cause of liver disease globally and the leading indication for adult liver transplantation in the United States. Hepatic steatosis is present in approximately 50% of HCV infected patients, however the rates differ according to HCV genotype[2–4]. Patients infected with HCV genotype 3 have steatosis more often and of higher grade than patients infected with genotype 1[5, 6]. Up to 80% of genotype 3 infected patients have steatosis, and several studies have indicated that steatosis in these patients correlates closely with measures of viral activity[7–11]. This is in contrast to genotype 1 patients, where major risk factors for steatosis are obesity, diabetes and alcohol use[12–15]. Steatosis has proven to be a major risk factor for fibrosis progression and carcinogenesis[8, 16–18].
Studies on the pathogenesis of HCV-associated steatosis have frequently implicated the Core protein. HCV Core forms the nucleocapsid of the virus, and has been associated with altering a diverse range of intracellular pathways[19–21]. The most frequent pathway implicated is VLDL synthesis and the Microsomal Triglyceride Transfer protein (MTP)[22, 23]. Lipid accumulation now appears to be an integral part of the viral lifecycle[24, 25]. Recent studies have shown that HCV uses the VLDL synthesis pathway for virus release and Core protein binding to lipid droplets is required for efficient virus assembly and production[24–27]. These findings are consistent with data demonstrating that in vitro expression of Core protein from a variety of genotypes leads to lipid accumulation[12, 28, 29].
In previous work examining viral factors associated with lipid accumulation in HCV genotype 3 infection, we identified specific polymorphisms at amino acid residues 182 and 186 within domain 3 (d3) of the Core protein that correlated with the presence or absence of steatosis in a small group of patients. We recapitulated these differences in an in vitro system of hepatocyte lipid accumulation, demonstrating that Core carrying these polymorphisms induced more lipid accumulation.
We report here the results of follow up studies we performed to test the hypothesis that d3 alone is sufficient to cause intracellular lipid accumulation. We demonstrate that HCV Core d3 green fluorescent protein (GFP) fusions promote more lipid droplets of greater size and increased triglyceride content. We show that d3 sequence differences within genotype 3 and between genotypes 3 and 1 determine the amount of lipid accumulation and that d3 does not exert its action by direct interaction with lipid droplets but appears to partially co-localize with the golgi apparatus.
HCV Core sequences used as templates for subsequent experiments included: steatosis associated domain 3 (d3S) (HCV1 DCRI, GenBank ID#EU099414), non-steatosis associated (d3NS) (HCV11 DCRI, GenBank ID#EU099415), genotype 1 (d3G1) (HCV-N, GenBank ID#AF139594, gift from Dr. Steve Weinman).
The plasmid named “78”, (gift from Dr. Brian Doehle), which contained the backbone of pcDNA3 ligated with an NdeI-EcoRI fragment from pEGFP-C, containing part of the CMV promoter, the GFP ORF and part of the multiple cloning site was used for generating GFP fusion proteins. We designed custom oligonucleotides listed in Table 1 to generate the d3S, d3NS, d3G1 and domain 2–3 constructs. We used EcoRI and XbaI and subsequently cloned into digested and purified vector. Ligated products were transformed into E. coli and colonies were selected after overnight growth on Luria-Bertani agar containing Ampicillin. Recombinant plasmids were purified and sequenced to verify code and frame.
Rat derived 5H cells were transfected using JetPEI reagent (Polyplus, Illkirch, FRANCE) according to the manufacturer’s recommended protocol. After 24 hours, cells were transferred to a 150mm dish and incubated with media containing 2 mg/ml G418 for selection. Subsequent colonies were isolated and maintained in media containing 1 mg/ml G418.
The GFP-Core construct cell lines were passaged into 4 well chamber slides and grown overnight. Cells were then fixed and stained with Oil Red O (ORO) stain as described previously. Briefly, cells were washed twice in PBS, fixed with 4% paraformaldehyde for 30 minutes at 37°C, permeabilized with 0.1% Triton-X in PBS for 5 minutes, washed and stained with DAPI (1:1000) in methanol for 3 minutes at 25°C. Cells were washed with PBS twice and with propylene glycol three times for 5 minutes each. ORO stain in propylene glycol (Newcomer Supply, Madison,WI) was applied to cells for 7 minutes and then cells were washed in 85% propylene glycol for 3 minutes. Cells underwent distilled water rinse twice and mounted using PBS/glycerol 1:1. Slides were examined using Axiovert 200 microscope (Carl Zeiss) with epifluorescent illumination, and images recorded using an AxioCam HRC camera (Carl Zeiss) and Axiovision 4.4 software (Carl Zeiss) using the same settings for all photographs.
Twenty contiguous high power (100x) fields (HPF) were photographed for each stable cell line for analysis. Using methods we have described previously, we used Metamorph software to compare the amount of oil red content in each cell line. We analyzed the entire field by first applying a blue threshold to measure the percentage of the field with DAPI fluorescence, and then we analyzed the corresponding brightfield image by applying a red threshold to analyze the percentage of ORO staining. We divided %ORO stain by % DAPI fluorescence to control for the number of cells in a given field. This was done for each stable cell line, averages and standard deviations were calculated and the results were compared using a two-tailed t test (www.openepi.com).
Cells were maintained in a 150mm dish until confluence then collected on the same day of the experiment. After washing with PBS twice, 400ul PBS was added, collected them with cell scraper, 10ul were saved for measuring protein concentration. For each sample, 100ug total protein was used for detecting TG level. To extract lipid, 1.5ml Chloroform/Methanol (1:2 mixture) was added and incubated at room temperature for 2 minutes, followed by vortex for 30 seconds. 500ul Chloroform was added, followed by vortex for 30 seconds, and then 500ul PBS was added, followed by another vortex for 30 seconds. Samples were then centrifuged at 1000g for 10 minutes at room temperature and the lower phase was carefully collected. This was allowed to air dry overnight, and the next day 50ul mixture (t-butanol/Triton X-114/Methanol by 9:4:2) was added to re-dissolve the dried lipids. TG was measured at 540 nm using Serum Triglyceride Determination Kit (Sigma-Aldrich Inc, St. Louis, MO) according to the manufacturer’s instructions.
Antibodies to Adipose differentiation-related protein (ADRP), Calnexin, Golgi 58k protein and Catalase (Abcam, Cambridge, MA) were used for immunofluorescence experiments. Staining was performed as described above where primary antibody staining was performed for 30 minutes at 37°C, followed by 2 PBS washes and then stained with a secondary goat anti-mouse antibody conjugated to Alexa Fluor 594 (Molecular Probes-Invitrogen, Carlsbad, CA) for 30 minutes at 37°C. After 2 PBS washes, staining procedure proceeded as described above with DAPI stain.
Untransfected 5H cells were used as a negative control for ER stress and untrasfected cells incubated with tunicamycin 10 µg/ml for 24 hours were used as a positive control. All cell lysates were prepared using Passive Lysis Buffer (Promega, Madison, WI) and analyzed by immunoblot using the following primary antibodies: Green Fluorescent protein, Glucose-Regulated Protein (GRP) 78, CHOP (a.k.a GRP94) (Abcam, Cambridge, MA), β-actin (GenScript, Piscataway, NJ); and the following secondary antibodies: goat anti-mouse-HRP and donkey anti-goat-HRP (GenScript, Piscataway, NJ).
Huh7.5 cells (a gift from Dr. Shelton Bradrick) were transfected using JetPEI reagent (Polyplus, Illkirch, FRANCE) in a 4 well chamber slide. After 24 hours, cells were stained using the Immunofluorescence protocol described above.
Since all commercially available HCV Core antibodies all recognize epitopes within domains 1–2, we made d3-GFP fusion constructs so that we could independently monitor for the presence and localization of our domain 3 constructs. It is important to indicate that there are very few amino acid changes between the constructs and they are at fixed positions (Figure 1a).
We generated stable cell lines using rat liver 5H cells, which have low baseline lipid levels making them a good system for our prior and ongoing experiments. Huh7 and HepG2 contain too much lipid at baseline to allow for analysis of lipid accumulation. As demonstrated in figure 1b, stably transfected cells that expressed any of the constructs: control GFP, steatosis-associated domain 3 (d3S), genotype 1 domain 3 (d3G1) or non-steatosis domain 3 (d3NS), all had equivalent levels of protein expression.
We used fluorescence microscopy to examine the alterations in lipid metabolism after d3 fusion protein expression. Control GFP cells had small scattered lipid droplets, which was consistent with our previous experiments in transiently transfected cells (Figure 2a). Cells expressing the d3S construct had GFP fluorescence throughout the cytoplasm and large lipid droplets. These droplets usually had a perinuclear distribution. When examining cells expressing the d3G1 and d3NS constructs, they too had these large lipid droplets, but they were less frequently in number.
We sought to analyze the apparent differences in lipid content between the domain 3 constructs using two alternative approaches. The first was based on image analysis using Metamorph software. We have used Metamorph previously and have validated its use for the quantitation of ORO content in a high-power field. After statistical analysis of 20 HPF/construct, the steatosis-associated domain 3 cells had almost 2 fold more ORO content than the d3NS and 40% more than the d3G1 construct (Figure 2b).
The second method of lipid quantitation was direct measurement of triglyceride content in the stable cells. The results mirrored the differences observed using the MetaMorph analysis. The d3S expressing cells had 2 fold more triglyceride content than d3NS cells and approximately 30% more than d3G1 (Figure 3).
Given that d3 is the “signal peptide” and has been shown to localize to the ER membrane, we sought to evaluate the contribution of ER stress to lipid accumulation caused by d3[31–33]. We evaluated two markers of ER stress, GRP78 (BiP) and CHOP (GRP94), in our stable cell lines expressing domain 3 constructs. We incubated 5H cells with tunicamycin 10 µg/ml as a positive control. As shown in Figure 4, GRP78 expression was increased in tunicamycin treated cells but equivalent between control and all domain 3 expressing cells. The same was true for CHOP.
Since we established that d3 was sufficient for lipid accumulation, we next sought to determine the intracellular compartment where the lipids were accumulating. Our previous work as well as that of others has shown that Core protein localizes to lipid droplets and the residues that are responsible for this localization lie within domain 2[28, 34]. An unresolved issue is how lipid droplet localization and lipid accumulation are related, if at all. Using ADRP as a marker for lipid droplets, we found that d3 did not co-localize with these large droplets regardless of which d3 construct was used, as illustrated by the lack of fluorescent overlap (Figure 5), yet it was able to cause significant lipid to accumulate. We compared the d3 construct to a construct where domain 2 was retained with d3 (d2-3), which should restore the ability to localize to lipid droplets. We observed almost complete overlap of GFP and ADRP staining, indicating co-localization of this d2-3 construct with lipid droplets.
After observing the initial results of significant lipid accumulation with d3 and the localization away from the lipid droplet, we sought to determine where in the cell d3 may be mediating this effect. Given their importance in various phases of lipid metabolism, we chose to analyze markers for endoplasmic reticulum, Golgi apparatus and peroxisome. Analysis of stable cells indicated a partial co-localization of d3 with the Golgi apparatus based on the overlap of fluorescent staining (Figure 6a). There was little or no overlap observed with ER or peroxisomal markers. These were no differences in localization of the d3S compared to the d3G1 or d3NS constructs (data not shown).
To confirm this finding, we repeated this experiment using Huh7.5 cells transiently transfected with the d3S construct. Again we observed partial co-localization of the d3 construct with the Golgi apparatus (Figure 6b).
In this study we have shown that HCV Core d3 is sufficient for intracellular lipid accumulation. GFP fusion constructs of d3 alone led to large lipid containing droplets or aggregates, and the amount of ORO staining was highest in the d3S construct and lowest in the d3NS construct. These results were substantiated by the significantly higher triglyceride accumulation that was observed with the d3S construct compared to the d3NS. d3 does not co-localize with lipid droplets to mediate this effect as judged by our ADRP staining, but rather exhibits partial co-localization with the golgi apparatus in both the stable rat-derived liver cells and human derived Huh7.5 cells.
These results have several implications. The first is to expand the previously attributed role of d3. Most studies have focused on the cleavage events that occur between domains 2 and 3 and between d3 and the E1 glycoprotein[31, 32, 35]. Specific investigations have previously mutated these cleavage sites and stressed the importance of proper cleavage for virus assembly and release[36–38]. Our previous work defining polymorphisms within domain 3 and subsequently using site-specific mutagenesis provided cause for reconsideration. This work provides further evidence that domain 3 appears to mediating much more than just cleavage events.
The second implication is to further separate the phenomena of lipid droplet binding and lipid accumulation. Many of the previous studies have focused on regions within domain 2 being important for lipid droplet binding[25, 34]. In our previous study, we hypothesized that these 2 phenomena may be mediated by 2 distinct regions of the Core protein. This work provides specific support of this hypothesis by showing that domain 3 constructs alone caused the significant lipid to accumulate in the absence of co-localization to the lipid droplet but co-localized to lipid droplets when combined with domain 2. These results are in contrast to previous papers describing the importance of residue 164 and the Phenylalanine that is a hallmark of genotype 3 constructs in causing the upregulation of Fatty Acid Synthase[39, 40]. Our previous data made us skeptical of the importance of this result given that steatosis and non-steatosis isolates both had the F164 residue but yet had significant differences in the amount of lipid accumulating. We hypothesize that this residue may serve to maintain the lipid content of droplets already formed but is not likely to be the primary cause of lipid accumulation.
The third implication is the apparent importance of seemingly subtle sequence differences within domain 3. The non-steatosis construct, which is a genotype 3 isolate that differs only by a single amino acid at position 186 (Isoleucine instead of Valine) still causes lipid to accumulate but at approximately half the amount seen with the steatosis construct. The genotype 1 construct was in between but closer to the non-steatosis construct. We interpret these results as supporting the selection over time of sequences that maintain the hydrophobic helix formation required for proper cleavage at the proximal and distal ends which also allow for enough lipids to accumulate enabling viral replication and assembly to occur. Alanine scanning through this region of Core had minimal effect on viral titers using the JFH-1 system, which supports this hypothesis since alanine substitution would preserve the helix structure. Some sequences clearly exceed that minimum requirement, which we hypothesize to lead to the viral induced steatosis observed in patients with chronic infection.
Based on the results of previous studies and d3’s role as the “signal peptide”, the lack of co-localization with the ER was surprising to us. However, the partial co-localization of d3 with the Golgi apparatus may provide insights about a possible mechanism. We would hypothesize the d3 alters the trafficking of triglycerides and other lipids by a hitherto unidentified mechanism and allows them to accumulate. More specific studies on the natural lifecycle of domain 3 after cleavage and specific studies of lipid trafficking will help to address this issue.
Our results should be analyzed in the light of certain limitations. Though we are using GFP fusion constructs which might raise concern about the context of our results, our domain 3 expressing cells resemble the appearance of Huh7.5 cells expressing a mutant Core protein with an altered domain 2–3 cleavage site. These cells had increased lipid and had multiple large vacuolar structures similar to those seen in our domain 3 expressing cells. Second, this system lacks the context of other viral proteins and the entire viral life cycle. We would agree that analysis of lipid accumulation in that setting would be definitive and it is our goal to perform such an experiment. These results however still are consistent with previous results given the accumulation of free cholesterol with domain 3 constructs and the genotype differences.
In conclusion, we have demonstrated that HCV Core d3 is sufficient for intracellular lipid accumulation and further underlines the importance of subtle sequence changes in d3’s ability to cause lipid accumulation. These data provide further evidence for an expanded role of domain 3 in the viral life cycle beyond simple cleavage events, possibly indicating altered trafficking as the most likely or dominant role. Elucidation of the specific mechanisms of lipid accumulation may serve to uncover a pan-genotype drug target that would serve to disrupt an early step in the viral life cycle.
We would like to thank members of the Diehl lab for their collaborative effort and support. We would like to also thank Pat Seed, Puneet Seth, Shelton Bradrick and Alex Thompson for their scientific input and critical review of this manuscript
Financial Support: RJ-NIH K08DK76598, American Liver Foundation/American Association of the Study of Liver Diseases Sheila Sherlock Translational Research Award, Duke Children’s Miracle Network Research Grant, Duke Center for AIDS Research
Conflicts-No conflicts to report for any of the authors
Portions of this work were presented at the following meetings:
45th Annual Meeting of the Infectious Diseases Society of America, Oct 4–7, 2007; San Diego, CA.
2008 American Society of Virology Annual Meeting, Jul 12–16, 2008; Ithaca, NY, Abstract #W6-1