Quantitative Lipid Droplet Proteome Analysis Identifies Annexin A3 as a Cofactor for HCV Particle Production

Article

Quantitative Lipid Droplet Proteome Analysis Identifies Annexin A3 as a Cofactor for HCV Particle Production Graphical Abstract

Authors Kathrin Ro¨sch, Marcel Kwiatkowski, Sarah Hofmann, ..., Marcus Wurlitzer, € ter, Eva Herker Hartmut Schlu

Correspondence [email protected]

In Brief Cytoplasmic lipid droplets are vital to hepatitis C virus particle production. Ro¨sch et al. report a quantitative lipid droplet proteome analysis and identify annexin A3 as a regulator of HCV particle maturation and egress.

Highlights d

Quantitative lipid droplet proteome analysis of HCV-infected cells is presented

d

ANXA3 is recruited to lipid-rich fractions in HCV-infected cells

d

ANXA3 functions as a host factor required for efficient HCV particle production

d

ANXA3 promotes virion maturation by facilitating incorporation of apolipoprotein E

Ro¨sch et al., 2016, Cell Reports 16, 3219–3231 September 20, 2016 ª 2016 The Author(s). http://dx.doi.org/10.1016/j.celrep.2016.08.052

Accession Numbers PXD004707

Cell Reports

Article Quantitative Lipid Droplet Proteome Analysis Identifies Annexin A3 as a Cofactor for HCV Particle Production € ttner,1 Marcus Wurlitzer,2 Kathrin Ro¨sch,1 Marcel Kwiatkowski,2 Sarah Hofmann,1 Anja Scho¨bel,1 Cordula Gru € ter,2 and Eva Herker1,3,* Hartmut Schlu 1Heinrich

Pette Institute, Leibniz Institute for Experimental Virology, 20251 Hamburg, Germany Facility Mass Spectrometric Proteomics, Institute of Clinical Chemistry, University Medical Center Hamburg-Eppendorf, 20246 Hamburg, Germany 3Lead Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2016.08.052 2Core

SUMMARY

Lipid droplets are vital to hepatitis C virus (HCV) infection as the putative sites of virion assembly, but morphogenesis and egress of virions remain ill defined. We performed quantitative lipid droplet proteome analysis of HCV-infected cells to identify co-factors of that process. Our results demonstrate that HCV disconnects lipid droplets from their metabolic function. Annexin A3 (ANXA3), a protein enriched in lipid droplet fractions, strongly impacted HCV replication and was characterized further: ANXA3 is recruited to lipid-rich fractions in HCV-infected cells by the viral core and NS5A proteins. ANXA3 knockdown does not affect HCV RNA replication but severely impairs virion production with lower specific infectivity and higher density of secreted virions. ANXA3 is essential for the interaction of viral envelope E2 with apolipoprotein E (ApoE) and for trafficking, but not lipidation, of ApoE in HCV-infected cells. Thus, we identified ANXA3 as a regulator of HCV maturation and egress.

INTRODUCTION Hepatitis C virus (HCV) infection is one of the leading causes of liver-related morbidity and mortality worldwide, accounting for approximately 0.5 million deaths every year (Wedemeyer et al., 2015). The true number of HCV infections is unknown, but recent estimates suggest 80 (64–103) million viraemic HCV infections, and the disease burden will likely continue to rise in most countries (Gower et al., 2014). No vaccine exists, but the recently developed direct-acting antivirals dramatically increase therapeutic responses compared to the standard interferon-based therapy. However, worldwide, the treatment of patients will likely be restricted due to the extremely high costs of the new therapeutics.

HCV belongs to the family of Flaviviridae. The enveloped viral particles contain a single positive-stranded RNA genome of 9.6 kb in length and are associated with lipoproteins and neutral lipids and accordingly named lipoviroparticles. After receptormediated endocytosis, fusion, and uncoating, the viral genome is translated into one polyprotein precursor (reviewed in Lindenbach and Rice, 2005). Host and viral proteases process the viral polyprotein, releasing the three structural proteins (nucleocapsid core, E1, and E2), the viroporin p7, and six non-structural (NS) proteins (NS2, 3, 4A, 4B, 5A, and 5B). Multi-protein RNA replication complexes containing minimally NS3–5B proteins replicate the viral RNA within ER (endoplasmic reticulum)-derived structures termed the membranous web. The membranous web contains single-, double-, and multi-membrane vesicles (Romero-Brey et al., 2012), as well as cytosolic lipid droplets that may serve as viral assembly sites (reviewed in Lindenbach and Rice, 2013). After encapsidation of newly synthesized viral RNA at ER membranes near lipid droplets, the virus is thought to exit the cell via the secretory pathway, thereby maturing to low-density lipoviroparticles (reviewed in Lindenbach and Rice, 2013). In the absence of full viral replication, only two viral proteins localize to lipid droplets, core and NS5A (Barba et al., 1997; Shi et al., 2002). All other viral proteins are found in close proximity to lipid droplets in infected cells (Miyanari et al., 2007), but they lack intrinsic lipid-droplet-targeting features, as they fail to localize to lipid droplets when expressed individually (Camus et al., 2013). Translocation of both core and NS5A to lipid droplets requires triglyceride biosynthesis as inhibitors of diacylglycerol acyltransferase-1 (DGAT1) impair trafficking to lipid droplets and subsequent HCV assembly (Camus et al., 2013; Herker et al., 2010). Proper processing of core is a prerequisite for core’s ability to traffic to lipid droplets (Targett-Adams et al., 2008), and mutations in either core or NS5A that disrupt lipid droplet binding suppress HCV assembly (Boulant et al., 2007; Miyanari et al., 2007). Trafficking of core to lipid droplets additionally requires cytosolic phospholipase A2 activity (Menzel et al., 2012), and lipid-droplet-binding proteins act as host factors with, e.g., PLIN3/TIP47 required for both HCV RNA replication and release of virions (Ploen et al., 2013a, 2013b; Vogt

Cell Reports 16, 3219–3231, September 20, 2016 ª 2016 The Author(s). 3219 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

A

Jc1 NS5AB-mKO2-BSD E2

p7 NS2

NS3

NS5AB Cleavage Sites NS4A

Core E1

NS4B

NS5A

mKO2 BSD

B

Lysate

LDs

C

M

Figure 1. Quantitative Lipid Droplet Proteome Analysis of HCV-Infected versus Uninfected Control Cells

LDs

NS5B

PLIN2/ADRP PLIN3/TIP47

Heavy AA

Light AA

Calreticulin Tubulin

6 passages 6–10 days +/- blasticidin selection

D

MnSOD

Lysis

Li L Lipid Droplets D s

3x SDS- MS Wash PAGE

E

METTL7A *** tRNA splicing, via endonucleolytic cleavage and ligation BP DDX3X ACACA ** C14orf166 * membrane-bounded vesicle CC ACSL4 * DDX1 * tRNA-splicing ligase complex CC PNPLA3 ** IGF2BP1 * cytoplasmic stress granule CC APOB RTCB ** endoplasmic reticulum lumen CC VPS13A PABPC1 ** FASN * PAPSS2 ** cytoplasmic part CC LSS * DPP4 * RNA binding MF ABHD5 ** ANXA3 ** chylomicron-mediated lipid transport REAC LMNB1 ** SERPINH1 * negative regulation of sequestering of triglyceride BP NUS1 ARL8B * triglyceride metabolic process BP PNPLA2 * VIM * HSD17B7 * CTSD long-chain fatty-acyl-CoA biosynthetic process BP ACSL3 * VDAC2 cholesterol biosynthetic process BP PSAT1 MGST1 endoplasmic reticulum CC DHRS1 *** MTTP * lipid particle CC CHP1 *** GGH STMN1 * oxidoreductase activity MF ATP1A1 FABP5 * SSR1 steroid biosynthesis KEGG IPO5 SFPQ fatty acid biosynthesis KEGG PRKDC * VDAC1 fatty acid metabolism KEGG GBF1 VAT1 metabolic pathways KEGG SCCPDH * P4HB * HSPB1 PPAR signaling pathway KEGG MYDGF * LDHA ** ERP29 * ChREBP activates metabolic gene expression REAC SQLE * ** ARF4 fatty Acyl-CoA biosynthesis REAC LDAH EPHX1 acyl chain remodeling of DAG and TAG REAC ELMOD2 DDX5 metabolism of lipids and lipoproteins REAC AHCY ATL3 NSDHL RAB32 DHRSX * 0 2 FAM213A HSP90AB1 * -log10 p-value RPS13 PHGDH DDOST RPN2 CAPRIN1 ATP5A1 DPP4 PCYOX1 RPN1 naive TMED10 MTTP HSP90AB1STMN1 shNT ERP29 ATP5B PRKDC FABP5 ANXA3 ARL8B PHB2 LDHA ACACA LMAN2 shARF4 SERPINH1 DHRS1 PAPSS2 CANX FASN DHRSX shANXA3 ARF4 VDAC3 SCCPDH P4HB shVIL1 TPRG1L PNPLA2 shPABPC1 RHOC CHP1 VIM DDX1 SRSF1 shLMAN2 LMNB1 ETFA PABPC1 shHMGCS1 ACSL3 ABHD5 RTCB HM13 PNPLA3 METTL7A shRAB32 SQLE RPS11 LSS ACSL4 IGF2BP1 shDPP4 RPS4X C14orf166 HSD17B7 CALR shDDX1

F

G

0 -6

-4

-2

0

2

4

4

6

Enrichment at lipid droplets (log2 HCV-infected/control)

et al., 2013). Despite recent advances, the mechanistic details of the late stages of HCV replication are still ill defined. Here, we performed an extensive quantitative lipid-droplet proteome analysis of HCV-infected cells to identify host factors for HCV particle production. One of the highly enriched proteins was annexin A3 (ANXA3), a member of the annexin family of calcium-dependent, phospholipid-binding proteins. All annexins show cytosolic and membrane localizations, and some including ANXA3 are also secreted. They are involved in endo- and exocytosis and trafficking, and they serve as membrane scaffolds organizing specific membrane microdomains (reviewed in Gerke et al., 2005). Membrane recruitment is most likely regulated by calcium influx, and, depending on the phospholipid-binding specificity, annexins may target different cellular membranes (reviewed in Gerke et al., 2005; Gerke and Moss, 2002). We found that ANXA3 is specifically recruited to lipid-rich fractions in cells infected with HCV and that it participates in HCV maturation and release. Thus, by performing quantitative lipid droplet

3220 Cell Reports 16, 3219–3231, September 20, 2016

0.5

1

1.5

HCV replication (shTarget/shNT)

(A) Huh7.5 cells cultured in media containing either heavy or light amino acids were infected with an HCV reporter virus. Lipid droplets were isolated by two sequential sucrose gradient centrifugations, washed, and subjected to LC-ESI-MS/MS. (B) Western blot of the lipid droplet (LD) fraction shows enrichment of the lipid-droplet-binding proteins PLIN2/ADRP and PLIN3/TIP47. Other subcellular compartment markers as calreticulin (ER), MnSOD (mitochondria), or b-tubulin were undetectable in lipid droplet fractions. (C) Coomassie-stained SDS-PAGE of a purified lipid droplet (LD) fraction (M, molecular weight marker). (D) Heatmap depicting enriched or depleted proteins (1.5-fold cutoff). *p < 0.05; **p < 0.01; ***p < 0.001. (E) Gene ontology (GO) enrichment analysis of annotations for molecular function (MF), cellular compartment (CC), and biological process (BP), as well as biological pathways through KEGG and Reactome (REAC) annotations. Shown are annotations significantly enriched and hierarchically filtered for enriched (red) and depleted (blue) lipid droplet proteins. Analysis was performed using the gProfileR package in R. (F) Protein interaction networks of dysregulated lipid-droplet-associated proteins. Shown is the interaction network of significantly enriched (red) or depleted (blue) proteins visualized using the cisPath package in R. (G) Huh7.5 cells were transduced with lentiviral stocks expressing the indicated shRNAs, followed by infection with HCV Jc1NS5AB-EGFP. 6 days postinfection, cells were fixed and analyzed by flow cytometry of EGFP to measure HCV infection rates. Shown is the relative infection normalized to the non-targeting shRNA (shNT) control (mean ± SD, n = 2). Red indicates proteins that are enriched at lipid droplets. See also Figure S1 and Tables S1, S2, S3, S4, S5, S6, and S7.

proteomics, we revealed that HCV disconnects lipid droplets from their metabolic function and identified ANXA3 as a host factor for HCV replication. RESULTS HCV Infection Profoundly Changes the Lipid Droplet Proteome To identify regulators of HCV replication, we performed quantitative lipid droplet proteome analysis of HCV-infected versus uninfected control cells, using stable isotope labeling by amino acids in cell culture (SILAC) (Figure 1A). We labeled HCVpermissive Huh7.5 cells with heavy amino acids for at least six passages until near-complete (>95%) incorporation of the heavy amino acids into the cellular proteins was achieved. Cells cultured in media containing the normal light amino acids were infected with HCV carrying a fluorescence reporter (monomeric Kusabira Orange 2; mKO2) to monitor infection rates,

followed by a blasticidin resistance gene (blasticidin S deaminase [BSD]) in between a duplicated NS5A-NS5B cleavage site described previously (Webster et al., 2013). After 2 to 3 weeks, HCV-infected cells that displayed infection rates higher than 95% were used for the experiments. Of note, we used selected HCV-infected cells in half of the experiments. One day prior to lipid droplet isolation, HCV-infected and heavy-amino-acid-labeled control cells were plated at equal cell densities. For lipid droplet isolation, 5 3 107 cells were mixed and lysed in hypotonic lysis buffer with a Dounce homogenizer. Lipid droplets were isolated by two sequential sucrose density centrifugations of post-nuclear supernatant, followed by three washing steps in isotonic buffer. The isolated lipid droplet fractions were subsequently analyzed by western blotting to confirm enrichment of lipid-droplet-binding proteins (PLIN2/ADRP and PLIN3/TIP47) (Bickel et al., 2009) and depletion of markers of other cellular compartments (Figure 1B). SDS-PAGE followed by Coomassie staining was used to separate the lipid droplet fractions (Figure 1C), and after tryptic digest, peptides were analyzed by liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/ MS). As a control, we switched labeling conditions in one experiment, thus culturing the HCV-infected cells with heavy amino acids to exclude isotope-specific effects. In four independent experiments (on two different MS platforms), we identified around 1,500 proteins, of which 316 were detected through multiple peptides in each experiment and selected for further analysis (datasets are listed in Tables S1, S2, S3, and S4). To correct for different cell numbers or numbers of lipid droplets, we next centered the detection ratios of light over heavy peptides (L/H) or vice versa in swapped labeling conditions (H/L) by dividing through the median of the identified proteins (Table S5). Correlation analysis over all identified proteins revealed reproducible protein quantification independent of the labeling conditions, blasticidin selection, or MS platform used, with a mean Pearson correlation coefficient of r = 0.54 for all experiments (Figure S1). Detection ratios of the identified proteins were then ranked according to their mean enrichment at lipid droplets in the four independent experiments. For further analysis, we used cutoff ratios of greater than 1.5-fold enrichment or depletion and selected the proteins with significantly altered abundance (Figure S1). This generated a list of 16 proteins that are enriched significantly in lipid droplet fractions of HCV-infected cells and 21 proteins that are depleted; these proteins showed similar ratios of lipid droplet localization in all four independent experiments (Figure 1D). Next, we conducted gene ontology (GO) enrichment analysis and further characterized the lipid droplet proteome of HCV infection (Figure 1E; Tables S6 and S7). Recruited proteins are enriched for proteins usually not associated with lipid droplet function, such as RNA-binding proteins and proteins associated with cytoplasmic stress granules and vesicles. Intriguingly, the depleted proteins are enriched for lipid metabolic annotations such as lipid particle as the cellular compartment and fatty acid and triglyceride metabolism for biological processes; thus, proteins generally associated with lipid droplets are depleted from lipid droplets in HCV-infected cells. Protein

network analysis highlights the protein clusters identified both for enriched and for depleted proteins (Figure 1F). ANXA3 Is Required for Efficient HCV Spreading Infection To validate our approach of identifying regulators of HCV replication, we used short hairpin RNA (shRNA)-mediated knockdown of a subset of proteins detected at lipid droplets, some of which were enriched in HCV-infected cells. We transduced Huh7.5 cells with lentiviral particles encoding the shRNAs and simultaneously expressing mCherry to monitor transduction efficacy. Subsequently, cells were infected with a fluorescently labeled HCV reporter virus (Jc1NS5AB-EGFP) to identify proteins that impact HCV replication. Among them, knockdown of ANXA3 and ARF4 (ADP-ribosylation factor 4) showed the strongest effect on HCV replication (Figure 1G). As ARF4 is a brefeldin A (BFA)-sensitive ARF, and previous studies had investigated the effect of BFA on HCV replication, we focused on ANXA3 in the present study. In infection experiments, Jc1NS5AB-EGFP efficiently spread in mock- and control shRNA-transduced cells, while knockdown of ANXA3 severely impaired viral spreading (Figures 2A–2D). Knockdown of ANXA3 was confirmed by qRT-PCR and western blotting (Figures 2B and 2C). Previous studies had identified a different annexin, ANXA2, as a host factor for HCV RNA replication or virion assembly (Backes et al., 2010; Saxena et al., 2012), while others only found a minor effect (Dreux et al., 2012). We detected ANXA2, -3, -4, and -5 in lipid droplet fractions, but only ANXA3 was enriched in HCV-infected cells (Table S5). We wanted to clarify the impact of knocking down each of the ANXA proteins on viral replication and found that, in our experimental setup, only ANXA3 knockdown efficiently blocked spreading of Jc1NS5AB-EGFP reporter viruses (Figures 2E and 2F). To independently confirm a role for ANXA3 in HCV replication, we created clonal knockout cell lines using the CRISPR (clustered regularly interspaced short palindromic repeats)/Cas9 system (Figure 2G). Huh7.5 cells were transiently transfected with a Cas9 expression plasmid encoding single guide RNA (sgRNA) targeting the ANXA3 locus right after the translation start codon. After single-cell cloning, indels were verified by sequencing (Figure S2), and ANXA3 knockout was confirmed by western blotting (Figure 2H). We generated three cell lines harboring ANXA3 deletions and one cell line that transiently expressed the Cas9 construct but still harbored the wild-type alleles. When we infected these cell lines with Jc1NS5AB-mKO2 reporter viruses, spreading infection was nearly completely blocked in the ANXA3-knockout cell lines as compared to the wild-type control clone or the parental Huh7.5 cells (Figure 2I). We also analyzed this phenotype by fluorescence microscopy, and, indeed, Jc1NS5AB-mKO2 was impaired from spreading in DANXA3 cells, as indicated by the very low number of infected cells per foci of infection as compared to the control cells (Figure 2J). ANXA3 Is Recruited to Lipid-Rich Fractions in HCV-Infected Cells In order to confirm the re-localization of ANXA3 to lipid droplet fractions in HCV infection, we isolated lipid droplets of naive and HCV-infected Huh7.5 cells by sucrose gradient centrifugation and analyzed ANXA3 levels via western blotting. Despite

Cell Reports 16, 3219–3231, September 20, 2016 3221

100

Mock shNT shANXA3

80

E

*

EGFP-positive cells (%)

EGFP-positive cells (%)

NS5B

Flow cytometry

Huh7.5

D

EGFP

60 40 20 0

day 2

day 4

p7 NS2

NS3

day 3

F Mock shNT shANXA2 shANXA3 shANXA4 shANXA5

60 40 20 0

day 2

day 4

NS5A

mKO2

1.2 1.0 0.8 0.6 0.4 0.2 0

day 6

H NS4B

*

Tubulin

day 5

Parental

E2

C

ANXA3

80

day 6

NS4A

Core E1

shNT shANXA3

100

Jc1 NS5AB-mKO2

G

1.2 1.0 0.8 0.6 0.4 0.2 0

shANXA3

NS5A

NS5B

Cas9 sgRNA

CRISPR Clones ΔANXA3 #1

#2

#3

#4 *

ANXA3

Huh7.5

ANXA2 ANXA3 ANXA4 ANXA5

NS4B

shNT

B

NS3

ANXA expression

p7 NS2

WT

E2

ANXA3 expression

Core E1

Mock

Jc1 NS5AB-EGFP

shRNA (mCherry)

NS4A

A

J ** *

** * *

15 10

* 5 0

50

** **

**

40 30

WT

20

Parental WT ΔANXA3

20 10 0

day 2

day 4

day 6

Parental

WT

ΔANXA3

ΔANXA3

25

Cell number / foci

mKO2-positive cells (%)

I

Parental

Tubulin

Figure 2. ANXA3 Controls HCV Replication (A) Scheme of the experimental design using ANXA3-knockdown cells. (B) Knockdown efficacy was verified by qRT-PCR. (C) Western blot analysis of shANXA3 (shRNA targeting ANXA3)-transduced and control cells. Asterisk indicates an unspecific band recognized by the ANXA3 antibody. (D) Huh7.5 cells were transduced with lentiviral stocks expressing shRNAs targeting ANXA3 (shANXA3) or non-targeting (NT) control (shNT) and infected with Jc1NS5AB-EGFP. 2, 4, and 6 days post-infection, cells were fixed and analyzed by flow cytometry of EGFP to measure spreading infection (mean ± SEM, n = 3). *p < 0.05. (E) HCV spreading infection in ANXA2–5 knockdown cells. Shown is one experiment performed in triplicate (mean ± SD). (F) Knockdown efficacies were verified by qRT-PCR. (G) Scheme of the experimental design using ANXA3-knockout cells. (H) Western blot of ANXA3-knockout and wild-type clones as well as parental Huh7.5 cell lines. Asterisk marks an unspecific band. (I) ANXA3-knockout (clone #3), wild-type (clone #4), and parental Huh7.5 cells were infected with Jc1NS5AB-mKO2 viral stocks and analyzed by flow cytometry of mKO2 to measure spreading infection (mean ± SEM, n = 3). *p < 0.05; **p < 0.01. (J) Confluent ANXA3-knockout (clone #3), wild-type (clone #4), and parental Huh7.5 cells were infected with a very low MOI of Jc1NS5AB-mKO2 and fixed 3 days post-infection for microscopic analysis. Cells were stained with NS5A antibodies and counterstained with Hoechst (scale bar, 10 mm). The number of infected cells per infection focus was counted in two independent experiments in two to three different wells (mean ± SEM). **p < 0.01. See also Figure S2.

3222 Cell Reports 16, 3219–3231, September 20, 2016

Figure 3. ANXA4 Translocates to Lipid Droplet Fractions in HCV-Infected Cells

A B

C

E

D

F

loading the same amount of protein as indicated by similar PLIN2/ADRP levels, ANXA3 was nearly undetectable in lipid droplet fractions from control cells but readily detectable in lipid droplet fractions from HCV-infected cells (Figure 3A). Therefore, ANXA3 is localized to lipid droplet fractions in cells that are infected with HCV. This increase in ANXA3 levels could either reflect an overall increased expression or a specific recruitment of ANXA3 to lipid droplets. Therefore, we analyzed ANXA3 expression following HCV infection by western blotting. Overall, ANXA3 protein expression was unchanged in infected cells, as compared to control cells, pointing to enhanced recruitment of

(A) Huh7.5 cells were infected with Jc1NS5AB-EGFP-BSD and selected for infected cells with blasticidin. Lipid droplets (LDs) were isolated by sucrose gradient centrifugation and analyzed by western blotting. (B) Western blot analysis of ANXA3 expression in Huh7.5 cells during HCV infection. (C) Huh7.5 cells were transfected with Con1 subgenomic replicon (Con1 SGR) RNA and kept under selective pressure with G418. Lipid droplets were isolated by sucrose gradient centrifugation and analyzed by western blotting. (D) Membrane flotation assays using iodixanol gradients of replicon-transfected and control cells. Fractions were collected top (1) to bottom (8) and analyzed by western blotting. The asterisk marks an unspecific band. (E) Lipid droplet fractions of Huh7.5 cells transduced with lentiviral expression constructs for FLAG-core or NS5A-FLAG were analyzed by western blotting (arrow indicates ANXA3, asterisk marks an unspecific band). (F) Huh7.5 cells were electroporated with expression constructs for ANXA3-HA and NS5A-FLAG or FLAG-core, fixed, and stained with HA and FLAG antibodies, followed by lipid droplet staining with LD540. Samples were analyzed by confocal microscopy; single channels are shown in black and white, the merged images are pseudocolored as indicated (scale bars, 10 mm). Co-localization of the viral proteins (VP) core and NS5A with ANXA3-HA was analyzed using coloc2 in Fiji (Schindelin et al., 2012). We analyzed individual cells of three independent experiments and calculated the degree of co-localization using Manders’ co-localization coefficient and Pearson’s correlation coefficient (mean ± SEM).

ANXA3 to lipid-rich fractions following infection with HCV (Figure 3B). Next, we investigated ANXA3 localization in cells transfected with a bicistronic Con1 subgenomic replicon (Con1 SGR) that lacks all structural proteins, including core, as well as p7 and NS2, and only expresses NS3 to NS5B, as well as a neomycin resistance gene (Choi et al., 2004). After selection for cells that replicate HCV RNA, we isolated lipid droplets of replicon and control cells and analyzed them by western blotting. As observed before, ANXA3 was barely detectable in the lipid droplet fraction of control cells (Figure 3C). In contrast, in cells selected for expression of the subgenomic HCV replicon and expressing the NS3-5B RNA replicase, ANXA3 was recruited to the lipid droplet fraction (Figure 3C). The HCV RNA is replicated within the membranous web. PLIN3/TIP47 is a lipid-dropletbinding protein that acts as a host factor for HCV RNA replication and is re-localized to low-density lipid-rich membranes in membrane floatation assays (Vogt et al., 2013). Hence, we analyzed ANXA3 localization in HCV RNA—replicating cells by

Cell Reports 16, 3219–3231, September 20, 2016 3223

membrane floatation assay. Post-nuclear cellular lysates were separated by a discontinuous iodixanol gradient and analyzed by western blotting. ANXA3 localized to high-density membranes but additionally co-fractionated with NS5A and PLIN2/ ADRP in low-density lipid-rich membranes in cells actively replicating HCV RNA (Figure 3D). Thus, expression of NS3NS5B and active viral RNA replication are sufficient to re-localize ANXA3 to lipid-rich membranes. To analyze which viral protein re-localizes ANXA3, we ectopically expressed the two viral proteins that traffic to lipid droplets independently of viral replication, core and NS5A, and performed lipid droplet isolations. In control cells, ANXA3 was barely detectable in lipid droplet fractions. In contrast, ANXA3 traffics to lipid-rich fractions in cells expressing either core or NS5A (Figure 3E). We next assessed colocalization of both proteins together with hemagglutinin (HA)-tagged ANXA3 in immunofluorescence microscopy. We could not analyze endogenous ANXA3, as all antibodies tested failed to specifically detect endogenous ANXA3 in immunofluorescence staining. Overexpressed ANXA3-HA mainly showed cytoplasmic staining pattern. Co-localization analysis through quantifying co-localization according to Manders and Pearson revealed that ANXA3-HA partially colocalized both with NS5A and with core protein (Figure 3F) but did not display clear ring-like lipid droplet staining patterns. Taken together, the data indicate that ANXA3 is specifically recruited to lipid-rich membrane fractions by core and NS5A during HCV infection. HCV RNA Replication Is Independent of ANXA3 Next, we dissected which step of viral replication depends on ANXA3 expression. As ANXA3 is recruited to lipid-rich membrane fractions in cells replicating a subgenomic replicon RNA, we first used the Con1 SGR system to analyze viral RNA replication in cells harboring ANXA3 shRNA (Figure 4A). Con1 SGR RNA was transfected into Huh7.5 cells transduced with the different shRNAs. After 3 weeks in selection media, growing cell colonies were visualized by crystal violet staining. We observed slightly fewer colonies in ANXA3-knockdown cells as compared to control cells (Figure 4A). In order to verify this result, we used a monocistronic genotype 2a replicon system containing a firefly luciferase in between a duplicated NS5A-NS5B cleavage site and partially deleted in E1 and E2 to prevent viral spreading (Jc1DE1E2NS5AB-Fluc; Figure 4B) (Webster et al., 2013). In contrast to the subgenomic replicon, this system is very sensitive and allows the examination of viral RNA replication in short-term experiments. In-vitro-transcribed Jc1DE1E2NS5AB-Fluc RNA was transfected into ANXA3-knockdown or control cells and subjected to luciferase assays at different time points post-electroporation. To normalize to the transfection efficacy, all values are expressed as fold increase of luciferase activity per microgram of protein over a 4-hr time point before viral RNA replication occurs. Compared to the control, we could not detect any differences in viral RNA replication rates in ANXA3-knockdown cells (Figure 4B). In addition, we analyzed the viral protein expression in ANXA3-knockdown and control cells after electroporation of wild-type JFH1 RNA by western blotting. ANXA3-knockdown and control cells showed comparable levels of core and NS5A,

3224 Cell Reports 16, 3219–3231, September 20, 2016

indicating that viral RNA replication and translation are not affected by knockdown of ANXA3 (Figure 4C). ANXA3 Is Required for Efficient HCV Particle Production We next asked whether ANXA3 influences the later stages of HCV replication. We transfected full-length HCV Jc1NS5AB-EGFP genomes into ANXA3-knockdown or control cells and confirmed equal transfection efficacy by flow cytometry of EGFP. We then isolated total cellular RNA and viral RNA from the culture supernatant and measured HCV copy numbers by qRT-PCR at day 3 and day 6 post-transfection. In line with the previous experiments, intracellular HCV copy numbers did not differ significantly between ANXA3-knockdown and control cells, indicating equal viral RNA replication rates (Figure 4D). In contrast, ANXA3 knockdown significantly decreased HCV RNA copy numbers as well as core protein released in the culture supernatant both on day 3 and day 6 post-transfection (Figures 4D and 4E). As HCV RNA copy numbers only poorly reflect the amount of infectious particles, we determined infectious viral titers by measuring the 50% tissue culture infective dose (TCID50) (Lindenbach et al., 2005), using the same experimental setup as described earlier. Viral titers were severely reduced at day 3 and day 6 post-transfection in ANXA3-knockdown cells in comparison to control cells (Figure 4F), supporting the model that ANXA3 influences HCV particle production. Core Translocation to Lipid Droplets and Capsid Envelopment Is Independent of ANXA3 No accumulation of infectious viral particles was observed inside ANXA3-knockdown cells; instead, we detected slightly reduced intracellular viral titers pointing to HCV viral assembly or maturation as the steps in the viral life cycle requiring proper ANXA3 function (Figure 5A). A prerequisite for HCV assembly is the trafficking of the capsid protein core and NS5A to lipid droplets. Therefore, we first analyzed the subcellular localization of the two viral proteins by immunofluorescence microscopy and lipid droplet isolations. Core and NS5A strongly localized to lipid droplets in both control and ANXA3-knockdown cells, excluding that disturbed trafficking of these proteins was responsible for the defect in virus production observed in ANXA3-knockdown cells (Figures 5B and 5C). Next, we examined the formation of high-molecular-mass (HMM) core complexes and envelopment of the capsids as described previously (Gentzsch et al., 2013). Cell lysates were analyzed by 2D blue native SDS-PAGE, and results indicated that formation of HMM core complexes is independent of ANXA3 expression for Jc1 as well as JFH1 strains and independent on the ANXA3-inactivation method (Figure 5D). For envelopment assays, cell lysates were treated with proteinase K to determine the amount of core that is enveloped and, thus, protected from proteinase K digestion. As a control, lysates were incubated with Triton X-100, which disrupts all membranes. The amount of core protein that was protected was similar in control and in ANXA3-knockdown cells, demonstrating that the envelopment of core was not dependent on ANXA3 expression (Figure 5E). As control, the membrane was probed with NS5A antibodies, and, as expected, NS5A is not protected against proteinase K digestion, similar to core in envelope-deleted viral strains (Figure 5E). As annexins had been previously described

NS3

Intracellular HCV RNA

NS4B

NS5A

EGFP

HCV RNA copy number (GE x 109 / ml)

HCV RNA copy number (GE x 108 / μg total RNA)

0.6 0.4 0.2 0

day 3

day 6

Core

10

NS5A Tubulin 48 h

72 h

NS5B

1.0

0.8

JFH1

NS5B

Secreted HCV RNA

1.4

1.0

Luc

ANXA3

24 h

shNT shANXA3 1.2

NS5A

100

shNT shANXA3 ** **

E 0.8

shNT shANXA3 ** **

F

0.7 0.8

HCV core protein (fmol x 103 / ml)

p7 NS2

NS4B

1 NS4A

E2

NS3

shNT shANXA3

shANXA3

Jc1 NS5AB-EGFP Core E1

p7 NS2

shANXA3

Core

0.6 0.4 0.2

Infectivity (TCID50 x 104 / ml)

D

C

Jc1ΔE1E2 NS5AB-Fluc

NS5B

Control

NS5A

HCV RNA replication (RLU over 4 h)

shNT

NS4B

NS4A

NS3

NeoR

shNT

B

Con1 SGR

EMCV IRES

HCV IRES

NS4A

A

0.6 0.5 0.4 0.3 0.2

shNT shANXA3 ** * 3

2

1

0.1 0

day 3

day 6

0

0

day 3

day 6

day 3

day 6

Figure 4. ANXA3 Regulates HCV Progeny Virion Production (A) Huh7.5 cells transduced with lentiviral shRNAs were electroporated with Con1 SGR and kept under selective pressure for 3 weeks, and surviving colonies were stained with crystal violet. Shown is one representative experiment. shNT, non-targeting shRNA; shANXA3, shRNA targeting ANXA3. (B) Luciferase activity of Jc1DE1E2NS5AB-Fluc-transfected cells was measured 4, 24, 48, and 72 hr post-electroporation. Shown is the luciferase activity (RLU, relative light units) per microgram of protein standardized to the 4-hr time point (mean ± SEM, n = 3). (C) JFH1-electroporated cells were lysed and analyzed by western blotting. (D–F) Huh7.5 cells transduced with lentiviral shRNAs were electroporated with full-length HCV Jc1NS5AB-EGFP RNA. Equal transfection rates were verified by flow cytometry of EGFP 3 days post-electroporation. (D) Intra- and extracellular HCV RNA copy numbers were determined by qRT-PCR. Shown is the absolute quantification of the HCV copy number (GE, genome equivalents) per microgram of total RNA (intracellular) or milliliter of culture supernatant (extracellular) (mean ± SEM, n = 3). **p < 0.01. (E) Quantification of HCV core protein released in the supernatants (mean ± SEM, n = 3) **p < 0.01. (F) The infectious titers of the culture supernatants were determined by TCID50 titration on naive Huh7.5 cells (mean ± SEM, n = 3–5). *p < 0.05; **p < 0.01.

to modulate phospholipase A2 (PLA2) activity (Gerke and Moss, 2002) and cPLA2 is required for HCV virion production (Menzel et al., 2012), we probed cPLA2 activity in HCV-infected control and ANXA3-knockdown cells and did not detect any difference (Figure S3). Knockdown of ANXA3 Affects the Density and Specific Infectivity of HCV Particles To address whether ANXA3 influences the specific infectivity of the particles released, we divided the amount of infectious particles (focus-forming units; FFUs) by the amount of the viral capsid protein core or the number of HCV genomes. Intriguingly, particles released by cells lacking ANXA3 had a lower specific infectivity than the particles released by control cells (Figure 6A). To analyze whether ANXA3 directly influences particle infectivity, we analyzed whether ANXA3 is incorporated into the virions. Therefore, we

purified E2-FLAG tagged viral particles from culture supernatant by subsequent precipitation, ultracentrifugation, and affinity purification, but we did not detect ANXA3 attached to the purified HCV particles (Figure S3). In addition, antibodies directed against ANXA3 failed to neutralize HCV (Figure S3). One major determinant of the specific infectivity of HCV particles is the density of the lipoviroparticles. When we compared the densities by density gradient centrifugation, we observed the peak of infectivity and secreted core protein at lower densities in culture supernatant from cells expressing ANXA3 (Figure 6B); secreted HCV RNA displayed only a slight shift (indicated by arrows). Of note, the higher plateau at lower densities most likely reflects that there is less viral RNA in ANXA3-knockdown cell supernatant; thus, this plateau seems higher than in the control cells when shown as percentage of total RNA. The change in density during HCV maturation critically depends on

Cell Reports 16, 3219–3231, September 20, 2016 3225

A

B

C

D

E

Figure 5. Core Trafficking and Capsid Envelopment Is Independent of ANXA3 (A) Equal amounts of HCV Jc1NS5AB-EGFP-electroporated cells were lysed, and intracellular infectivity was determined by TCID50 titration on naive Huh7.5 cells (mean ± SEM, n = 5). shNT, non-targeting shRNA; shANXA3, shRNA targeting ANXA3. (B) Huh7.5 cells transduced with lentiviral shRNAs were electroporated with Jc1NS5AB-EGFP-BSD and selected for infected cells with blasticidin. Lipid droplets (LDs) were isolated by sucrose gradient centrifugation and analyzed by western blotting. The asterisk marks an unspecific band. (C) Huh7.5 cells transduced with lentiviral shRNAs and infected with Jc1 were stained with core or NS5A antibodies and BODIPY and analyzed by confocal microscopy (scale bar, 10 mm). (D) 48 hr post-electroporation of HCV RNA, cell lysates were prepared by freeze-thaw cycles and analyzed by 2D blue native followed by SDS-PAGE and western blotting to detect core in high-molecular-mass (HMM) and low-molecular-mass (LMM) complexes. (E) Cell lysates were subjected to proteinase K (PK) treatment and analyzed by western blotting. As control, lysates were left untreated or incubated with Triton X-100 prior to proteolytic digestion. Shown is one representative western blot experiment and the densitometric quantification of core levels as the percentage of protected core (+PK) compared to untreated control (mean ± SEM, n = 6).

very low-density lipoprotein secretion (Gastaminza et al., 2008; Huang et al., 2007; Jiang and Luo, 2009). Therefore, we investigated the activity of the microsomal transfer protein (MTP) and could not detect dependence on ANXA3 in the presence or absence of HCV replication (Figure 6C). Secretion of apolipoproteins ApoE and ApoB into the culture supernatant of ANXA3knockdown and control cells was also similar in uninfected cells (Figure 6D; Figure S4). In contrast, ApoE, but not ApoB, secretion was significantly reduced in ANXA3-knockdown cells that replicated either subgenomic or genomic HCV RNA (Figure 6D; Figure S4). Of note, even in HCV-infected cells, the density of total cellular and secreted ApoE was not affected by ANXA3 knockdown (Figure 6B; Figure S4). Thus, in uninfected cells, lipoprotein secretion functions independently from ANXA3, but in HCV-infected cells, ANXA3 contributes to ApoE secretion but not lipidation.

3226 Cell Reports 16, 3219–3231, September 20, 2016

ANXA3 Promotes Viral Particle Maturation by Facilitating Incorporation of ApoE During HCV maturation, ApoE likely attaches to HCV particles via its interaction with the envelope protein E2 (Lee et al., 2014). We probed this interaction using a fully infectious HCV construct expressing FLAG-tagged E2. In cells expressing ANXA3, we were able to detect the ApoE-E2 interaction. In contrast, in ANXA3-knockdown or -knockout cells, this interaction was nearly abolished (Figures 7A and 7B). ANXA3 was additionally required for the interaction between E2 and the capsid core, implicating that incorporation of the envelope proteins, as well as ApoE, requires ANXA3 (Figures 7A and 7B). Importantly, the interactions between E2 and ApoE, as well as core, were only observed in the context of full viral replication, as co-expressed core and E1/E2 expression constructs containing the signal peptide of core to ensure correct

60 50 40 30 20 10

40 30 20 10 0 1.00

HCV core (% / fraction)

1.12

50 40 30 20 10 0 1.00

1.04

40 30 20 10 0 1.04

1.08

30 20 10 0 1.00

1.12

1.04

Density D

1.5

shNT shANXA3 Sup.

2.0

ApoB ApoE

Cells

1.0 0.5

140

ApoE

ApoB NS5A

1.08

1.12

Density

shNT

100 80

(A) To calculate the specific infectivity of released HCV particles, infectivity (focus forming units; FFU) was determined by TCID50 titration and divided by the amount of HCV core protein or the number of genome equivalents (GE) (mean ± SEM, n = 3). *p < 0.05. shNT, non-targeting shRNA; shANXA3, shRNA targeting ANXA3. (B) To determine the density of released viral particles, culture supernatants were analyzed by iodixanol gradient centrifugation and viral RNA copy numbers, infectivity (as TCID50), and core and ApoE protein levels were determined (mean ± SEM, n = 3). (C) MTP activity was analyzed in shRNA-transduced and Con1 SGR-transfected Huh7.5 cells and depicted as picomoles of transferred fluorescent substrate per microgram of protein (mean ± SEM, n = 10). (D) Lipoprotein secretion was assessed by western blotting with ApoE and ApoB antibodies of culture supernatant (Sup.) of control, Con1 SGR-transfected, or Jc1NS5AB-EGFP-BSD-transfected Huh7.5 cells. Quantification of ApoE secretion by densitometric analysis of western blots (mean ± SEM, n = 3–9). **p < 0.01. See also Figures S3 and S4.

shANXA3

120

ApoE secretion (% of control)

0

C

1.12

40

50

-10 1.00

1.08

Density

shNT shANXA3

Specific infectivity (FFU / 105 GE)

1.08

60

8

MTP activity (pmol transferred / μg protein)

1.04

60

Density

0

4

Figure 6. ANXA3 Is Required for Efficient HCV Maturation

shNT shANXA3

ApoE (% / fraction)

70

B

Infectivity (% TCID50 / fraction)

shNT shANXA3 80 *

HCV RNA copy number (% GE / fraction)

Specific infectivity (FFU / fmol core)

A

**

**

In summary, our results suggest that ANXA3 acts as a host factor for HCV particle maturation, with subsequent effects on the number and infectivity of particles released (Figure 7G).

60 40

DISCUSSION

Here, we performed an extensive quantitative lipid droplet proteome analysis of 0 0 HCV-infected versus uninfected control SGR Jc1 Mock SGR Jc1 cells to reveal the perturbations caused by HCV infection and to identify regulators subcellular localization did not interact with each other or with of HCV replication. We compiled a list of 316 proteins that were endogenous ApoE (Figure 7C). Prior to interaction with E2, reliably identified in the lipid droplet fractions, with 16 proteins ApoE interacts with NS5A, an interaction that is thought to re- being recruited to lipid droplets and 21 proteins being discruit ApoE to virion assembly sites. We used a viral strain that placed from lipid droplets in HCV-infected cells. Strikingly, encodes an HA-tagged NS5A to study this interaction. Intrigu- some of the recruited proteins are annotated for RNA-binding ingly, we detected no change in the level of ApoE interacting proteins, which fits with the present model that part of the with NS5A in ANXA3-knockdown cells (Figure 7D). Therefore, membranous web, where the HCV RNA replicates, as well as ANXA3 is required for virion maturation steps in the ER. To viral translation centers, are found in close proximity of lipid verify our immunoprecipitation results, we analyzed the sub- droplets. Two of our top hits were DEAD box proteins 1 and cellular localization of E2/core and ApoE. We observed that, 3 (DDX1 and DDX3), which both have been described to in uninfected cells, ApoE mainly localizes to the Golgi com- localize to lipid droplets in cells expressing the HCV protein partment, while in HCV-infected cells, ApoE is scattered in a core (Sato et al., 2006), thus validating our SILAC-based punctuate pattern throughout the cell co-localizing with E2 (Fig- approach. DDX3 was subsequently shown to bind to core ure 7E). This pattern is reverted in ANXA3-knockdown cells and to interact with the 30 UTR of the HCV RNA activating a where ApoE is, again, in the Golgi co-localizing with Golgi ma- non-canonical IkB kinase (IKK)-alpha pathway that, in turn, intrix protein GM130 (Figure 7F). We observed a similar pattern of duces lipogenic genes, resulting in more lipid droplets, less co-localization in ANXA3-knockdown cells when we enhanced core-lipid droplet interaction, and virion assembly (Ariumi et al., 2007; Li et al., 2013). probed for E2 and core (Figure 7E). 20

Tubulin

Cell Reports 16, 3219–3231, September 20, 2016 3227

1.0

***

***

0.0

shNT shANXA3 ApoE (M1)

E2 (M2)

Manders' coloc. coefficients

shNT shANXA3

0.0

ApoE (M1)

Core (M2)

Merged

0.5 0.6

FLAG-E2

ApoE/ GM130

1.0

** ***

*** ***

0.5

0.0

shNT shANXA3 ApoE GM130 ApoE GM130 (M1) (M2) (M1) (M2)

Jc1

Mock

G Manders' coloc. coefficients

shNT

shNT shANXA3

1.0

***

***

0.5

shANXA3

ApoE NS5A-HA ApoE

ApoE/ NS5A:

GM130

Merged

0.5

FLAG-E2

ApoE

NS5A-HA

0.5

1.0

Core

IP:HA Input

Input IP:FLAG

Input

F

Manders' coloc. coefficients

Core

FLAG-E2 ApoE Core FLAG-E2 ApoE Core

Merged

shANXA3

ApoE

Jc1 NS5A-HA shNT shANXA3 shNT shANXA3

WT ΔANXA3 WT ΔANXA3

FLAG-E2

FLAG-E2 ApoE Core FLAG-E2 ApoE * Core

Manders' coloc. coefficients

ApoE

D

– – + + E1/FLAG-E2 – + – + core

IP:FLAG

IP:FLAG

Input

FLAG-E2 ApoE Core FLAG-E2 ApoE * Core

shNT

E

C

Jc1 FLAG-E2

shNT

B

shANXA3

Jc1 FLAG-E2 shNT shANXA3 shNT shANXA3

A

0.0

shNT shANXA3 Core (M1)

Entry & Uncoating Envelope Proteins Core NS5A ApoE ANXA3 ANXA3-independent ANXA3-dependent

Translation & Trafficking RNA Replication NS5A-ApoE Capsid Assembly

ANXA3

E2-ApoE Maturation & Release

E2 (M2)

Figure 7. ANXA3 Is Required for Subversion of ApoE Secretion (A and B) ANXA3-knockdown or -knockout cells were transfected with Jc1Flag-E2, and 5 days post-transfection, cells were lysed in high-stringency buffer for co-immunoprecipitation (IP) of FLAG-E2 with ApoE and core. shNT, non-targeting shRNA; shANXA3, shRNA targeting ANXA3. (C) Huh7.5 cells transduced with lentiviral expression constructs for core and E1/FLAG-E2 were lysed in high-stringency buffer for co-immunoprecipitation of FLAG-E2 with ApoE and core. (D) shRNA-transduced cells were transfected with Jc1Flag-E2 NS5A-HA RNA and lysed in low-stringency buffer for co-immunoprecipitation of NS5A with ApoE. (E and F) shRNA-transduced cells were infected with Jc1Flag-E2, fixed, and stained with ApoE, FLAG, core, and GM130 (Golgi) antibodies. Samples were analyzed by confocal microscopy; single channels are shown in black and white, and the merged images are pseudocolored as indicated (scale bars, 10 mm). Co-localization was analyzed using coloc2 in Fiji of individual cells of two to three independent experiments by calculating the degree of co-localization using the Manders’ co-localization coefficient (mean ± SEM). *p < 0.05; **p < 0.01; ***p < 0.001. (G) Model: HCV infection causes re-localization of ANXA3 to lipid droplet fractions to ensure efficient HCV maturation and egress.

Proteins depleted from lipid droplets were mainly annotated for lipid metabolic processes, indicating that HCV perturbs the protein composition to disconnect the lipid droplets from their normal metabolic function. This phenomenon might contribute to the development of steatosis in patients and is line with our

3228 Cell Reports 16, 3219–3231, September 20, 2016

previous data that lipid droplets are stabilized in cells expressing the capsid protein core (Harris et al., 2011). One of the proteins highly enriched in lipid droplet fractions was the phospholipid-binding protein ANXA3 that was studied in greater molecular detail. This enrichment was not due to an

overall increased expression but due to a specific re-localization to lipid-rich fractions in HCV-infected cells. As overexpressed ANXA3 failed to localize to ring-like structures, ANXA3 is likely trafficked to lipid-rich membrane structures and not to the lipid droplet surface. We found that expression of single viral proteins, either core or NS5A, is sufficient to re-localize ANXA3. However, we were unable to detect a direct interaction by co-immunoprecipitation experiments (data not shown) indicating that ANXA3 is recruited by interaction partners or by changes in the membrane composition that enable binding of ANXA3. Alternatively, HCV infection could also cause an efflux of calcium from the ER at ER-lipid droplet contact sites that, in turn, could enable the binding of ANXA3 to membranes. Interestingly both core and NS5A can affect calcium homeostasis in the ER (Dionisio et al., 2009). Thus, locally elevated calcium levels could trigger enhanced ANXA3 binding. We observed that ANXA3 is required for a step in HCV particle production after envelopment of the viral capsids. Viral particles released from cells lacking ANXA3 showed a higher density and less specific infectivity pointing to a defect in proper virion assembly or maturation. ANXA3 expression was required for ApoE secretion, but not ApoB secretion or MTP activity, in cells infected with HCV but not in uninfected cells. Interestingly, ApoE subcellular localization changes in response to HCV infection, a phenotype reverted by ANXA3 knockdown. Through interaction studies, we show that the recruitment of ApoE via NS5A to virion assembly sites is independent of ANXA3, while ApoE/E2 interaction requires ANXA3. Thus, ANXA3 mediates ApoE trafficking (but not lipidation) and HCV maturation steps within the ER. Interestingly, it was recently reported that ANXA3 is secreted and that it functions in a positive-feedback loop that promotes cancer stem cell selfrenewal and tumor growth in hepatocellular carcinoma (Tong et al., 2015). ANXA3 expression is approximately 10-fold higher in cancer cell lines as Huh7 cells; thus, ANXA3 effect on HCV replication might contribute to the higher replication rates in Huh7 cells. However, further studies have to explore in greater detail the mechanistic role of ANXA3 in the late stages of the HCV life cycle. EXPERIMENTAL PROCEDURES For detailed experimental procedures, see the Supplemental Information.

Knockdown and Knockout of ANXA3 To knock down ANXA3, we used lentiviral shRNA constructs (pSicoR-MS1) as described previously (Herker et al., 2010; Wissing et al., 2011). We created clonal ANXA3-knockout cell lines using the CRISPR/Cas9 system as described previously (Ran et al., 2013). SILAC Labeling, HCV Infection, and Lipid Droplet Isolation for Mass Spectrometry For isotope metabolic protein labeling, we used the SILAC Protein Quantitation Kit according to the manufacturer’s instructions (Thermo Scientific). After the incorporation was confirmed, light or heavy Huh7.5 cells were infected with Jc1 reporter strains and cultured for 2 to 3 weeks. Lipid droplets were isolated by two sequential sucrose density centrifugations of post-nuclear supernatant, followed by three washing steps in isotonic buffer. The lipid-droplet-associated proteins were separated by SDS-PAGE. After tryptic in-gel digestion, LC-ESI-MS/MS analyses were performed on a quadrupole-time-of-flight (Q-TOF) mass spectrometer (Q-TOF Premier, Micromass/Waters) or on a linear trap quadrupole (LTQ) orbitrap mass spectrometer (Orbitrap Fusion, Thermo Scientific). Both instruments were coupled with an ESI source to a nano-UPLC (ultra-performance liquid chromatography) system (nanoACQUITY, Waters; Dionex UltiMate 3000 RSLCnano, Thermo Fisher Scientific). Data from LC-ESIQ-TOF-MS/MS analysis were analyzed using the open-source software framework OpenMS (Sturm et al., 2008) and the OpenMS Proteomic Pipeline (TOPPAS) (Kohlbacher et al., 2007). For peptide and protein identification, LC-MS/MS raw data were processed as described previously (Kwiatkowski et al., 2015). SILAC pairs were detected and quantified using SILACAnalyzer (Nilse et al., 2010). The LC-MS/MS data from orbitrap analysis were processed with MaxQuant version 1.5.2.8 (Cox and Mann, 2008). The MS proteomics data have been deposited into the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (Vizcaı´no et al., 2013) with the dataset identifier PXD004707. Biochemical and Cell Biological Methods Lipid droplets were isolated by sucrose gradient centrifugation (Herker et al., 2010). Iodixanol gradient centrifugations for membrane flotation and to determine HCV particle density were performed as described previously (Catanese et al., 2013; Vogt et al., 2013). 2D blue native SDS-PAGE and proteinase K digestion protection assay were performed as described (Gentzsch et al., 2013). Co-immunoprecipitation, determination of ApoE/B secretion, and immunofluorescence microscopy were performed as described previously (Herker et al., 2010). Bioinformatics and Statistical Analysis For bioinformatics and statistical analysis, we used R (R Core Team, 2015), RStudio (RStudio Team, 2015), and GraphPadPrism (GraphPad Software). Statistical analysis was performed using unpaired two-tailed Student’s t test and, in the case of normalized data, one-sample t test.

Cell Lines and Culture Conditions HEK293T cells obtained from the American Type Culture Collection and Huh7.5 cells obtained from C.M. Rice were grown under standard cell culture conditions.

ACCESSION NUMBERS

HCV Infection and Replicon Assays HCV Jc1 reporter constructs encoding fluorescent proteins, selection markers, or the firefly luciferase between a duplicated NS5A-NS5B cleavage site and Jc1FLAG-E2 were described previously (Eggert et al., 2014; Webster et al., 2013). For HCV infection and transfection experiments, in-vitro-transcribed HCV RNA was electroporated into Huh7.5 cells. Culture supernatant was harvested, filtered, and concentrated by polyethylene glycol 8000 precipitations. For subgenomic replicon assays, replicon RNA was electroporated into Huh7.5 cells and viral RNA replication was measured by either survival under G418 treatment or luciferase activity as described previously (Vogt et al., 2013).

SUPPLEMENTAL INFORMATION

The accession number for the data reported in this paper is PRIDE: PXD004707.

Supplemental Information includes Supplemental Experimental Procedures, four figures, and seven tables and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2016.08.052. AUTHOR CONTRIBUTIONS K.R. designed, conducted, and analyzed the experiments. M.K., M.W., and H.S. performed the MS experiments for proteome analysis and analyzed the data. S.H., A.S., and C.G. provided reagents and performed some experiments. E.H.

Cell Reports 16, 3219–3231, September 20, 2016 3229

designed and analyzed the experiments and supervised the project. The manuscript was written by E.H. and K.R., with input from all authors.

RNA from infected cells to plasmacytoid dendritic cells triggers innate immunity. Cell Host Microbe 12, 558–570.

ACKNOWLEDGMENTS

Eggert, D., Ro¨sch, K., Reimer, R., and Herker, E. (2014). Visualization and analysis of hepatitis C virus structural proteins at lipid droplets by super-resolution microscopy. PLoS ONE 9, e102511.

We thank R. Bartenschlager (University of Heidelberg) for Jc1 constructs; C.M. Rice (Rockefeller University) for Huh7.5 cells and RFP-NLS-IPS; J. McLauchlan (Medical Research Council Virology Unit) for the JFH1 construct; T. Wakita (National Institute of Infectious Diseases, Japan) for JFH1; B. Webster and W.C. Greene (Gladstone Institute of Virology and Immunology) for the HCVcc reporter constructs; J.H. Ou (University of Southern California) for the pUC Con1 subgenomic replicon; M. Spindler for pSicoR-MS1; F. Zhang (Broad Institute) for pSpCas9 (BB)-2A-Puro (PX459, Addgene plasmid # 48139); B. Fehse (University Clinic Hamburg Eppendorf) for LeGO vectors; and C. Thiele (University of Bonn) for LD540. This work was supported by funds from the DFG (HE 6889/2-1 to E.H. and INST 337/15-1 2013 and INST 337/16-1 2013 to H.S.). 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. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Received: November 12, 2015 Revised: May 20, 2016 Accepted: August 16, 2016 Published: September 20, 2016 REFERENCES Ariumi, Y., Kuroki, M., Abe, K., Dansako, H., Ikeda, M., Wakita, T., and Kato, N. (2007). DDX3 DEAD-box RNA helicase is required for hepatitis C virus RNA replication. J. Virol. 81, 13922–13926. Backes, P., Quinkert, D., Reiss, S., Binder, M., Zayas, M., Rescher, U., Gerke, V., Bartenschlager, R., and Lohmann, V. (2010). Role of annexin A2 in the production of infectious hepatitis C virus particles. J. Virol. 84, 5775–5789. Barba, G., Harper, F., Harada, T., Kohara, M., Goulinet, S., Matsuura, Y., Eder, G., Schaff, Z., Chapman, M.J., Miyamura, T., and Bre´chot, C. (1997). Hepatitis C virus core protein shows a cytoplasmic localization and associates to cellular lipid storage droplets. Proc. Natl. Acad. Sci. USA 94, 1200–1205. Bickel, P.E., Tansey, J.T., and Welte, M.A. (2009). PAT proteins, an ancient family of lipid droplet proteins that regulate cellular lipid stores. Biochim. Biophys. Acta 1791, 419–440. Boulant, S., Targett-Adams, P., and McLauchlan, J. (2007). Disrupting the association of hepatitis C virus core protein with lipid droplets correlates with a loss in production of infectious virus. J. Gen. Virol. 88, 2204–2213. Camus, G., Herker, E., Modi, A.A., Haas, J.T., Ramage, H.R., Farese, R.V., Jr., and Ott, M. (2013). Diacylglycerol acyltransferase-1 localizes hepatitis C virus NS5A protein to lipid droplets and enhances NS5A interaction with the viral capsid core. J. Biol. Chem. 288, 9915–9923. Catanese, M.T., Uryu, K., Kopp, M., Edwards, T.J., Andrus, L., Rice, W.J., Silvestry, M., Kuhn, R.J., and Rice, C.M. (2013). Ultrastructural analysis of hepatitis C virus particles. Proc. Natl. Acad. Sci. USA 110, 9505–9510. Choi, J., Lee, K.J., Zheng, Y., Yamaga, A.K., Lai, M.M., and Ou, J.H. (2004). Reactive oxygen species suppress hepatitis C virus RNA replication in human hepatoma cells. Hepatology 39, 81–89. Cox, J., and Mann, M. (2008). MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372. Dionisio, N., Garcia-Mediavilla, M.V., Sanchez-Campos, S., Majano, P.L., Benedicto, I., Rosado, J.A., Salido, G.M., and Gonzalez-Gallego, J. (2009). Hepatitis C virus NS5A and core proteins induce oxidative stress-mediated calcium signalling alterations in hepatocytes. J. Hepatol. 50, 872–882. Dreux, M., Garaigorta, U., Boyd, B., De´cembre, E., Chung, J., Whitten-Bauer, C., Wieland, S., and Chisari, F.V. (2012). Short-range exosomal transfer of viral

3230 Cell Reports 16, 3219–3231, September 20, 2016

Gastaminza, P., Cheng, G., Wieland, S., Zhong, J., Liao, W., and Chisari, F.V. (2008). Cellular determinants of hepatitis C virus assembly, maturation, degradation, and secretion. J. Virol. 82, 2120–2129. Gentzsch, J., Brohm, C., Steinmann, E., Friesland, M., Menzel, N., Vieyres, G., Perin, P.M., Frentzen, A., Kaderali, L., and Pietschmann, T. (2013). hepatitis c Virus p7 is critical for capsid assembly and envelopment. PLoS Pathog. 9, e1003355. Gerke, V., and Moss, S.E. (2002). Annexins: from structure to function. Physiol. Rev. 82, 331–371. Gerke, V., Creutz, C.E., and Moss, S.E. (2005). Annexins: linking Ca2+ signalling to membrane dynamics. Nat. Rev. Mol. Cell Biol. 6, 449–461. Gower, E., Estes, C., Blach, S., Razavi-Shearer, K., and Razavi, H. (2014). Global epidemiology and genotype distribution of the hepatitis C virus infection. J. Hepatol. 61 (1, Suppl), S45–S57. Harris, C., Herker, E., Farese, R.V., Jr., and Ott, M. (2011). Hepatitis C virus core protein decreases lipid droplet turnover: a mechanism for core-induced steatosis. J. Biol. Chem. 286, 42615–42625. Herker, E., Harris, C., Hernandez, C., Carpentier, A., Kaehlcke, K., Rosenberg, A.R., Farese, R.V., Jr., and Ott, M. (2010). Efficient hepatitis C virus particle formation requires diacylglycerol acyltransferase-1. Nat. Med. 16, 1295–1298. Huang, H., Sun, F., Owen, D.M., Li, W., Chen, Y., Gale, M., Jr., and Ye, J. (2007). Hepatitis C virus production by human hepatocytes dependent on assembly and secretion of very low-density lipoproteins. Proc. Natl. Acad. Sci. USA 104, 5848–5853. Jiang, J., and Luo, G. (2009). Apolipoprotein E but not B is required for the formation of infectious hepatitis C virus particles. J. Virol. 83, 12680–12691. Kohlbacher, O., Reinert, K., Gro¨pl, C., Lange, E., Pfeifer, N., Schulz-Trieglaff, O., and Sturm, M. (2007). TOPP–the OpenMS proteomics pipeline. Bioinformatics 23, e191–e197. Kwiatkowski, M., Wurlitzer, M., Omidi, M., Ren, L., Kruber, S., Nimer, R., Rob€ter, H. (2015). Ultrafast extraction ertson, W.D., Horst, A., Miller, R.J., and Schlu of proteins from tissues using desorption by impulsive vibrational excitation. Angew. Chem. Int. Ed. Engl. 54, 285–288. Lee, J.Y., Acosta, E.G., Stoeck, I.K., Long, G., Hiet, M.S., Mueller, B., Fackler, O.T., Kallis, S., and Bartenschlager, R. (2014). Apolipoprotein E likely contributes to a maturation step of infectious hepatitis C virus particles and interacts with viral envelope glycoproteins. J. Virol. 88, 12422–12437. Li, Q., Pe`ne, V., Krishnamurthy, S., Cha, H., and Liang, T.J. (2013). Hepatitis C virus infection activates an innate pathway involving IKK-a in lipogenesis and viral assembly. Nat. Med. 19, 722–729. Lindenbach, B.D., and Rice, C.M. (2005). Unravelling hepatitis C virus replication from genome to function. Nature 436, 933–938. Lindenbach, B.D., and Rice, C.M. (2013). The ins and outs of hepatitis C virus entry and assembly. Nat. Rev. Microbiol. 11, 688–700. Lindenbach, B.D., Evans, M.J., Syder, A.J., Wo¨lk, B., Tellinghuisen, T.L., Liu, C.C., Maruyama, T., Hynes, R.O., Burton, D.R., McKeating, J.A., and Rice, C.M. (2005). Complete replication of hepatitis C virus in cell culture. Science 309, 623–626. Menzel, N., Fischl, W., Hueging, K., Bankwitz, D., Frentzen, A., Haid, S., Gentzsch, J., Kaderali, L., Bartenschlager, R., and Pietschmann, T. (2012). MAP-kinase regulated cytosolic phospholipase A2 activity is essential for production of infectious hepatitis C virus particles. PLoS Pathog. 8, e1002829. Miyanari, Y., Atsuzawa, K., Usuda, N., Watashi, K., Hishiki, T., Zayas, M., Bartenschlager, R., Wakita, T., Hijikata, M., and Shimotohno, K. (2007). The lipid droplet is an important organelle for hepatitis C virus production. Nat. Cell Biol. 9, 1089–1097.

Nilse, L., Sturm, M., Trudgian, D., Salek, M., Sims, P.G., Carroll, K., and Hubbard, S. (2010). SILACAnalyzer - a tool for differential quantitation of stable isotope derived data. In Computational Intelligence Methods for Bioinformatics and Biostatistics, F. Masulli, L. Peterson, and R. Tagliaferri, eds. (Springer), pp. 45–55. Ploen, D., Hafirassou, M.L., Himmelsbach, K., Sauter, D., Biniossek, M.L., Weiss, T.S., Baumert, T.F., Schuster, C., and Hildt, E. (2013a). TIP47 plays a crucial role in the life cycle of hepatitis C virus. J. Hepatol. 58, 1081–1088. Ploen, D., Hafirassou, M.L., Himmelsbach, K., Schille, S.A., Biniossek, M.L., Baumert, T.F., Schuster, C., and Hildt, E. (2013b). TIP47 is associated with the hepatitis C virus and its interaction with Rab9 is required for release of viral particles. Eur. J. Cell Biol. 92, 374–382. R Core Team (2015). R: a language and environment for statistical computing (Vienna, Austria: R Foundation for Statistical Computing). http://www. R-project.org. Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A., and Zhang, F. (2013). Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8, 2281– 2308. Romero-Brey, I., Merz, A., Chiramel, A., Lee, J.Y., Chlanda, P., Haselman, U., Santarella-Mellwig, R., Habermann, A., Hoppe, S., Kallis, S., et al. (2012). Three-dimensional architecture and biogenesis of membrane structures associated with hepatitis C virus replication. PLoS Pathog. 8, e1003056. RStudio Team (2015). RStudio: Integrated Development Environment for R. http://www.rstudio.com. Sato, S., Fukasawa, M., Yamakawa, Y., Natsume, T., Suzuki, T., Shoji, I., Aizaki, H., Miyamura, T., and Nishijima, M. (2006). Proteomic profiling of lipid droplet proteins in hepatoma cell lines expressing hepatitis C virus core protein. J. Biochem. 139, 921–930. Saxena, V., Lai, C.K., Chao, T.C., Jeng, K.S., and Lai, M.M. (2012). Annexin A2 is involved in the formation of hepatitis C virus replication complex on the lipid raft. J. Virol. 86, 4139–4150. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682.

Shi, S.T., Polyak, S.J., Tu, H., Taylor, D.R., Gretch, D.R., and Lai, M.M. (2002). Hepatitis C virus NS5A colocalizes with the core protein on lipid droplets and interacts with apolipoproteins. Virology 292, 198–210. Sturm, M., Bertsch, A., Gro¨pl, C., Hildebrandt, A., Hussong, R., Lange, E., Pfeifer, N., Schulz-Trieglaff, O., Zerck, A., Reinert, K., and Kohlbacher, O. (2008). OpenMS - an open-source software framework for mass spectrometry. BMC Bioinformatics 9, 163. Targett-Adams, P., Hope, G., Boulant, S., and McLauchlan, J. (2008). Maturation of hepatitis C virus core protein by signal peptide peptidase is required for virus production. J. Biol. Chem. 283, 16850–16859. Tong, M., Fung, T.M., Luk, S.T., Ng, K.Y., Lee, T.K., Lin, C.H., Yam, J.W., Chan, K.W., Ng, F., Zheng, B.J., et al. (2015). ANXA3/JNK signaling promotes self-renewal and tumor growth, and its blockade provides a therapeutic target for hepatocellular carcinoma. Stem Cell Reports 5, 45–59. ^ te´, R.G., Csordas, A., Dianes, J.A., Fabregat, A., Foster, J.M., Vizcaı´no, J.A., Co Griss, J., Alpi, E., Birim, M., Contell, J., et al. (2013). The PRoteomics IDEntifications (PRIDE) database and associated tools: status in 2013. Nucleic Acids Res. 41, D1063–D1069. Vogt, D.A., Camus, G., Herker, E., Webster, B.R., Tsou, C.L., Greene, W.C., Yen, T.S., and Ott, M. (2013). Lipid droplet-binding protein TIP47 regulates hepatitis C Virus RNA replication through interaction with the viral NS5A protein. PLoS Pathog. 9, e1003302. Webster, B., Ott, M., and Greene, W.C. (2013). Evasion of superinfection exclusion and elimination of primary viral RNA by an adapted strain of hepatitis C virus. J. Virol. 87, 13354–13369. Wedemeyer, H., Dore, G.J., and Ward, J.W. (2015). Estimates on HCV disease burden worldwide - filling the gaps. J. Viral Hepat. 22 (Suppl 1), 1–5. Wissing, S., Montano, M., Garcia-Perez, J.L., Moran, J.V., and Greene, W.C. (2011). Endogenous APOBEC3B restricts LINE-1 retrotransposition in transformed cells and human embryonic stem cells. J. Biol. Chem. 286, 36427– 36437.

Cell Reports 16, 3219–3231, September 20, 2016 3231