The antiviral , viperin, localizes to lipid droplets via its N-terminal amphipathic ␣-helix

Ella R. Hinsona and Peter Cresswella,b,1

aDepartment of Immunobiology, and bHoward Hughes Medical Institute, Yale University School of Medicine, New Haven, CT 06520-8011

Contributed by Peter Cresswell, October 15, 2009 (sent for review September 23, 2009) Lipid droplets are intracellular lipid-storage organelles that are lacking this domain have greatly reduced antiviral activity (11, thought to be derived from the (ER). Several 14). Although the precise mechanism by which viperin inhibits pathogens, notably virus, use lipid droplets for repli- HCV is still unknown, viperin was previously shown to inhibit cation. Numerous questions remain about how lipid droplets are virus budding by disrupting plasma membrane lipid generated and used by viruses. Here we show that the IFN-induced raft microdomains, which are sites of influenza virion assembly antiviral protein viperin, which localizes to the cytosolic face of the and budding (14). Independent of viral infection, the N-terminal ER and inhibits HCV, localizes to lipid droplets. We show that the amphipathic ␣-helix of viperin inhibits protein secretion and N-terminal amphipathic ␣-helix of viperin that is responsible for ER appears to induce ER membrane curvature (12). localization is also necessary and sufficient to localize both viperin In this study, we show that the N-terminal amphipathic ␣-helix and the fluorescent protein dsRed to lipid droplets. Point muta- is also necessary and sufficient to localize viperin to lipid tions in the ␣-helix that prevent ER association also disrupt lipid droplets. Removing the N-terminal amphipathic ␣-helix pre- droplet association, and sequential deletion mutants indicate that vents lipid droplet association and placing this domain at the N the same number of helical turns are necessary for ER and lipid terminus of the reporter fluorescent protein dsRed is sufficient droplet association. Finally, we show that the N-terminal amphi- to relocalize dsRed from the nucleus and cytosol to lipid droplets pathic ␣-helix of the hepatitis C viral protein NS5A can localize in oleic acid-treated cells. Furthermore, mutations or trunca- dsRed and viperin to lipid droplets. These findings indicate that the tions in the amphipathic ␣-helix that disrupt the ability of viperin amphipathic ␣-helices of viperin and NS5A are lipid droplet-tar- to bind to the ER also prevent lipid droplet association, indi- geting domains and suggest that viperin inhibits HCV by localizing cating that ER association is necessary for lipid droplet local- to lipid droplets using a domain and mechanism similar to that used ization and that viperin is likely to be associated with lipid by HCV itself. droplets as they form and bud from the outer leaflet of the ER membrane. In addition, we determined that the N-terminal HCV ͉ NS5A ͉ amphipathic ␣-helix of NS5A of HCV is also necessary and sufficient to localize not only the C terminus of viperin but also ipid droplets consist of a core of neutral lipids surrounded by dsRed to lipid droplets. These findings show that the amphi- Lan outer phospholipid monolayer and associated . pathic ␣-helices of both viperin and NS5A are lipid droplet- These organelles are thought to be generated when neutral lipids targeting domains and suggest that viperin may use a mechanism accumulate in the endoplasmic reticulum (ER) bilayer. The mass similar to NS5A to localize to lipid droplets, allowing viperin to of lipids is believed to bulge from the ER membrane into the inhibit HCV replication. cytoplasm, with the cytosolic leaflet of the ER membrane forming the outer phospholipid layer of the lipid droplet (1, 2). Results Recently, lipid droplets have been shown to play a role in several Viperin Localizes to Lipid Droplets in IFN␣- and Oleic Acid-Treated cellular processes, including lipid storage, lipid trafficking, and . Because viperin localizes to the cytoplasmic face of protein storage and degradation. The importance of this or- the ER and lipid droplets are thought to be generated from the ganelle is underscored by the fact that lipid droplets have been cytosolic leaflet of the ER membrane, we examined whether linked to several metabolic diseases, most notably diabetes and viperin localizes to lipid droplets. Bone marrow-derived macro- obesity (1). phages (BMM⌽) derived from C57/B6 wild-type mice were Lipid droplets have been shown to play a critical role in the treated with both IFN␣ and oleic acid to induce viperin expres- replication of several pathogens. One of the most well- sion and lipid droplet formation, respectively. These cells were characterized examples is hepatitis C virus (HCV) (3, 4). Upon examined by immunofluorescence for colocalization with a infection, the HCV core initially localizes to cytosolic face of the fluorescent BODIPY dye, which stains neutral lipids within lipid ER. After maturation, which requires two cleavage events, the droplets, and an antibody to adipocyte differentiation-related core localizes to lipid droplets by a newly exposed domain, D2, protein (ADRP), which is located in the outer phospholipid and which contains two amphipathic ␣-helices (5, 6). The core protein layer of the lipid droplet (2). Viperin appeared to recruits the HCV nonstructural (NS) proteins and HCV repli- surround the neutral lipids that were stained with BODIPY. cation complexes to lipid droplets (4). Several of these NS Furthermore, viperin colocalized with ADRP, which specifically proteins localize to the ER before lipid droplet recruitment and localizes to the lipid droplet delimiting monolayer (Fig. 1A). To ␣ contain amphipathic -helices that are thought to facilitate confirm these findings, lipid droplets were isolated using a interactions with the ER (7–9). In addition, HCV is thought to sucrose gradient from wild-type and viperin knockout BMM⌽. bring ER membranes in close proximity to the lipid droplet to create a localized environment with a high concentration of membranes for viral replication and assembly (4). Author contributions: E.R.H. and P.C. designed research; E.R.H. performed research; E.R.H. Viperin is an IFN-induced antiviral protein that is induced analyzed data; and E.R.H. and P.C. wrote the paper. upon HCV infection and inhibits HCV replication (10, 11). Like The authors declare no conflict of interest. the HCV NS proteins, viperin has been shown to localize to the Freely available online through the PNAS open access option. cytosolic face of the ER through an N-terminal amphipathic 1To whom correspondence should be addressed. E-mail: [email protected]. ␣ ␣ -helix (7, 9, 12, 13). This N-terminal amphipathic -helix is This article contains supporting information online at www.pnas.org/cgi/content/full/ essential for viperin to inhibit HCV and influenza, as mutants 0911679106/DCSupplemental.

20452–20457 ͉ PNAS ͉ December 1, 2009 ͉ vol. 106 ͉ no. 48 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0911679106 Downloaded by guest on September 27, 2021 DIC Viperin BODIPY Overlay by immunofluoresence for colocalization with BODIPY and A ADRP. Consistent with the BMM⌽, viperin appeared to form a complete ring around the BODIPY-stained lipids within the lipid droplets and colocalized with ADRP located in the outer lipid droplet membrane (Fig. 1C).

N-Terminal Amphipathic ␣-Helices of Viperin and NS5A Are Necessary DIC Viperin ADRP Overlay and Sufficient for Lipid Droplet Association. Because previous stud- ies have shown that viperin localizes to the cytoplasmic face of the ER via its N-terminal amphipathic ␣-helix (12), we next investigated whether this ER-targeting domain was sufficient to localize viperin to lipid droplets. 293T cells transfected with either WT viperin or viperin lacking the N-terminal amphipathic ␣-helix (⌬1–42), as diagramed in Fig. 2A, were treated with oleic acid and then examined by immunofluorescence for localization B Fraction WC Mem LD with BODIPY and ADRP. Consistent with the BMM⌽ results, WT KO WT KO WT KO WT viperin surrounded the outer rim of the BODIPY-stained lipid droplets and colocalized with ADRP (Fig. 2B). However, GRp94 when the N-terminal amphipathic ␣-helix was removed, this ADRP mutant no longer surrounded the BODIPY-stained lipids nor colocalized with ADRP (Fig. 2B). Because HCV replicates on Viperin lipid droplets, and the NS5A protein of HCV also has an N-terminal amphipathic ␣-helix (7), we asked whether this helix was also a lipid droplet-targeting signal. A chimeric protein, in which the N-terminal amphipathic ␣-helix of viperin (1–42 C Viperin BODIPY Overlay amino acids) was replaced by that of NS5A (1–30 amino acids) (Fig. 2A), clearly localized to the outer edge of the BODIPY- stained lipids and colocalized with ADRP, indicating that the N-terminal amphipathic ␣-helix of NS5A was also sufficient to target viperin to lipid droplets (Fig. 2B). These results were confirmed by biochemically isolating lipid droplets with sucrose gradients and performing Western blot Viperin ADRP Overlay analyses as described above. 293T cells were transfected with either the vector control, WT viperin, viperin ⌬1–42, or the NS5A-viperin chimera, treated with oleic acid, and then exam- ined for viperin expression in the lipid droplet fraction. Consis- tent with the immunofluorescence results, WT viperin and the NS5A-viperin chimera were abundant in the lipid droplet frac- tion, while viperin ⌬1–42 was not expressed in the lipid droplet fraction (Fig. 2C). As shown previously, the presence of these Fig. 1. Viperin localizes to lipid droplets. (A and B) Primary mouse bone proteins in the lipid droplet fraction was not due to abundant ER marrow derived macrophages (BMM⌽) were treated with 200 U/mL IFN␣ and or lysosomal contamination, as evidenced by the low levels of 400 ␮M oleic acid overnight to induce viperin expression and lipid droplet GRp94 and TPP-1 in this fraction, respectively. MICROBIOLOGY formation, respectively. (A) BMM⌽ were fixed, permeabilized with saponin, These results suggested that the amphipathic ␣-helices of and then examined by immunofluorescence for viperin colocalization with viperin and NS5A are lipid-droplet-targeting domains. To de- the lipid droplet dye, BODIPY, and an anti-ADRP antibody. (B) BMM⌽ from termine if these amphipathic ␣-helices were both necessary and wild-type B6 (WT) or viperin knock-out (KO) mice were subjected to lipid sufficient to not only target viperin but also a soluble protein to droplet fractionation using a sucrose gradient. The whole cell lysate (WC), lipid droplets, we generated dsRed chimeric proteins in which membrane (Mem), and lipid droplet (LD) fractions were separated by SDS/ the N-terminal amphipathic ␣-helix of either viperin or NS5A PAGE and then analyzed by Western blot analysis for viperin expression using anti-GRp94 and anti-ADRP antibodies as markers for the ER and lipid droplets, was fused to dsRed. By fluoresence, dsRed alone did not respectively. (C) HepG2 cells that had been transiently transfected with viperin colocalize with BODIPY or ADRP in oleic acid-treated cells were fixed, permeabilized with saponin, and then analyzed by immunofluo- (Fig. 3A). However, both the NS5A-dsRed and viperin-dsRed rescence for localization to lipid droplets using BODIPY and an anti-ADRP chimeric proteins localized to lipid droplet membranes, as antibody as lipid droplet markers. evidenced by their colocalization with ADRP and their ring-like appearance around the BODIPY-stained droplets (Fig. 3A). Sucrose gradient and Western blot analyses of transfected and The whole cell, membrane, and lipid droplet fractions were oleic acid-treated cells confirmed these findings. Unlike dsRed, separated by SDS/PAGE, and viperin, GRp94, and ADRP were which was not detected in the lipid droplet fraction, the viperin- detected by Western blot analysis. Viperin was detected in the dsRed protein was abundantly present (Fig. 3B). lipid droplet fractions of WT but not viperin knockout BMM⌽. The ER resident protein GRp94 was not detected in the lipid Viperin Must Associate with the ER to Localize to Lipid Droplets. The droplet fraction, indicating that the presence of viperin in this mechanism by which lipid droplets are generated is still un- fraction was not due to ER contamination (Fig. 1B). known. Current models propose that neutral lipids accumulate between the ER membrane leaflets and then bud into the Viperin Localizes to Naturally Occurring Lipid Droplets in Hepatocytes. cytoplasm, allowing the cytosolic ER membrane to surround and To confirm that viperin also localizes to naturally occurring lipid enclose the lipids (15). Because viperin is located on the cytosolic droplets in addition to those induced by oleic acid, a hepatocyte face of the ER membrane, we decided to determine whether the cell line, HepG2, was transfected with viperin and then analyzed correlation between the ability of viperin to localize to the ER

Hinson and Cresswell PNAS ͉ December 1, 2009 ͉ vol. 106 ͉ no. 48 ͉ 20453 Downloaded by guest on September 27, 2021 1 42 362 A DIC dsRed BODIPY Overlay A WT Viperin Viperin Viperin 1 30 43 362 NS5A-Viperin NS5A Viperin DsRed 43 362 ∆1-42 Viperin Viperin 1-42 (Vip) DIC Viperin BODIPY Overlay DsRed B WT

1-30 (NS5A) DsRed

∆ 1-42 DIC dsRed ADRP Overlay

1-42 (Vip) DsRed NS5A-Vip

1-30 (NS5A) DIC Viperin ADRP Overlay DsRed

WT B Fraction WC Membrane LD Construct 1 2 3 1 2 3 1 2 3

∆ 1-42 GRp94

NS5A-Vip TPP-1

C Fraction WC Membrane LD ADRP Construct 1 2 3 4 1 2 3 4 1 2 3 4

GRp94

TPP-1 dsRed

ADRP Fig. 3. The N-terminal amphipathic ␣-helices of viperin and NS5A are Viperin necessary and sufficient to localize dsRed to lipid droplets. 293T cells were transiently transfected with the noted dsRed constructs and then treated with 400 ␮M oleic acid overnight to induce lipid droplet formation. (A) 293T cells were examined by immunofluoresence as described in Fig. 1A for localization Fig. 2. The N-terminal amphipathic ␣-helix of NS5A is sufficient to localize with the lipid droplet markers BODIPY and ADRP. (B) The cells were subjected viperin to lipid droplets. 293T cells were transiently transfected with the vector to lipid droplet fractionation using a sucrose gradient. The whole cell lysate control, wild-type viperin (WT), NS5A (1–30)Viperin (43–362), or viperin ⌬1–42 (WC), membrane (Membrane), and lipid droplet (LD) fractions were separated as diagramed in (A) and then treated with 400 ␮M oleic acid overnight to by SDS/PAGE and then analyzed by Western blot analysis for viperin expres- induce lipid droplet formation. (B) Transfected and oleic acid–treated 293T sion. GRp94, TPP-1, and ADRP served as markers for the ER, lysosomes, and cells were examined by immunofluoresence as described in Fig. 1A for local- lipid droplets, respectively. Gel Lanes: vector control (1), dsRed (2), and Vip ization with the lipid droplet markers BODIPY and ADRP. (C) The 293T cells (1–42) dsRed (3). were subjected to lipid droplet fractionation using a sucrose gradient. The whole cell lysate (WC), membrane (Membrane), and lipid droplet (LD) frac- tions were separated by SDS/PAGE and then analyzed by Western blot analysis for viperin expression. GRp94, TPP-1, and ADRP served as markers for the ER, to glutamic acid to disrupt ER association. Several triple mutants lysosomes, and lipid droplets, respectively. Gel Lanes: vector control (1), were obtained that localized to the ER, such as W26E/L8E/L1E, wild-type viperin (2), NS5A (1–30) Vip (43–362) (3), and viperin ⌬1–42 (4). in which the numbers represent the residues in the helix, with residue 9 in the protein sequence designated as residue 1 in the helix. Furthermore, other triple point mutants and lipid droplets could be used to further examine this model. were obtained, notably L20E/L27E/L31E, that did not associate Hydrophobic residues in the N-terminal amphipathic ␣-helix of with the ER (Fig. 4B). These mutants were then examined for viperin, represented as a helical wheel in Fig. 4A, were mutated lipid droplet association by immunofluorescence and colocal-

20454 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0911679106 Hinson and Cresswell Downloaded by guest on September 27, 2021 A

D Fraction WC Membrane LD Construct 1 2 3 4 1 2 3 4 1 2 3 4 GRp94

TPP-1

Wild-type: LAARLLSLFQQQLGSLWSGLAILFCWLRIALGWL ADRP W26E/L8E/L1E: EAARLLSEFQQQLGSLWSGLAILFCELRIALGWL L20E/L27E/L31E: LAARLLSLFQQQLGSLWSGEAILFCWERIAEGWL Viperin

B DIC Viperin Calnexin Overlay Fig. 4. ER association is necessary for viperin to DIC BODIPY Viperin Overlay localize to lipid droplets. (A) Helical wheel drawing E of the N-terminal amphipathic ␣-helix of viperin. T1 Amino acid residues 9–42 were diagramed in a he- lical wheel using a helical wheel projection pro- gram (http://rzlab.ucr.edu/scripts/wheel/wheel. cgi?sequence ϭ ABCDEFGHIJLKMNOP&submit ϭ Submit). The circles, diamonds, and pentagons rep- T1-2 resent hydrophilic, hydrophobic, and potentially positively charged residues, respectively. The hydro- phobic and hydrophilic residues are colored in green T1-3 and red, respectively, with a greater hydrophobicity or hydrophilicity corresponding to a darker, purer color. Neutral residues are represented in yellow and potentially charged residues are in blue. Residues L20E/L27E/L31E W26E/L8E/L1E WT L20E/L27E/L31E W26E/L8E/L1E WT T1-4 that were mutated to glutamic acid have been out- lined in black and the sequences of the amphipathic C DIC Viperin BODIPY Overlay ␣-helix for wild-type viperin (WT), W26E/L8E/L1E, and L20E/L27E/L31E are shown below with the mu- T1-5 tated residues highlighted in bold. (B) 293T cells were transiently transfected with the noted triple helical point mutants in which the noted hydropho- bic residues were mutated to glutamic acid and then T1-6 examined for ER localization using anti-calnexin as an ER marker. (C and D) 293T cells were transiently transfected with the noted triple helical point mu- tants and treated with 400 ␮M oleic acid overnight. T1-7 The cells were examined for lipid droplet localiza- L20E/L27E/L31E W26E/L8E/L1E DIC Viperin ADRP Overlay tion using BODIPY and an anti-ADRP antibody as lipid droplet markers as described in Fig. 1A (C)orby MICROBIOLOGY T1-8 Western blot analysis of sucrose gradients as de- scribed in Fig. 2C (D). Gel Lanes: vector control (1), wild-type viperin (2), W26E/L8E/L1E (3), and L20E/ L27E/L31E (4). (E) 293T cells were transfected with the indicated sequential helical turn deletion mu- tants in which one (T1), two (T1–2), three (T1–3), etc. helical turns were deleted, and then treated with 400 ␮M oleic acid overnight. These cells were exam-

L20E/L27E/L31E W26E/L8E/L1E ined by immunofluorescence as described in Fig. 1A.

ization with BODIPY and ADRP. Although W26E/L8E/L1E and L20E/L27E/L31E to examine their degradation rate. These localized to lipid droplets, L20E/L27E/L31E did not costain with analyses showed that, although W26E/L8E/L1E, which retained either BODIPY or ADRP (Fig. 4C). lipid droplet association, had a significantly shorter half-life, L20E/ Lipid droplet fractionation and Western blot analyses confirmed L27E/L31E had comparable stability to WT viperin and the ⌬1–42 these findings and showed that while W26E/L8E/L1E was abundant mutant (Fig. S1). The reason for the accelerated degradation of in lipid droplets, L20E/L27E/L31E was not present in the lipid W26E/L8E/L1E is unknown but may relate to instability induced by droplet fraction (Fig. 4D). Interestingly, these analyses also revealed imperfect ␣-helix formation. Despite this, however, this mutant that the lack of ER association subjected L20E/L27E/L31E to associated normally with ER membranes and lipid droplets. cleavage by an unknown cellular protease, as evidenced by its To further examine the relationship between ER association smaller size. This was not only true for this mutant but also for and lipid droplet localization, we used a series of viperin mutants several other helical wheel point mutants that did not localize to the in which the eight helical turns of the N-terminal amphipathic ER. To determine whether detection of L20E/L27E/L31E on lipid ␣-helix were sequentially deleted. These mutants were previ- droplets was because of decreased protein stability, we performed ously examined for their ability to associate with the ER, and it radiolabeling pulse/chase analyses of WT, ⌬1–42, W26E/L8E/L1E, was found that only four helical turns were necessary for ER

Hinson and Cresswell PNAS ͉ December 1, 2009 ͉ vol. 106 ͉ no. 48 ͉ 20455 Downloaded by guest on September 27, 2021 association (12). Consistent with the helical wheel point mutants, viperin overexpression nor the absence of viperin in IFN-treated we found that when a sufficient number of turns were deleted to cells altered lipid droplet size, shape, or formation suggests that disrupt ER association, viperin also no longer localized to lipid viperin does not inhibit HCV by altering the ability of this virus droplets. Specifically, mutants with deletions of one (T1), two to generate and induce lipid droplets for replication. However, (T1–2), three (T1–3) or four (T1–4) sequential helical turns, we cannot rule out the possibility that in the context of HCV which were previously shown to localize to the ER, also localized infection, viperin is able to alter either lipid droplet formation or to lipid droplets, as determine by immunofluorescence with the ability of the HCV proteins to localize to this organelle. This BODIPY staining. Conversely, when five (T1–5), six (T1–6), is a particularly interesting possibility given that previous studies seven (T1–7), or eight (T1–8) helical turns were deleted, which suggested that viperin can alter ER membranes and apparently eliminates ER localization (12), association with lipid droplets induce ER membrane curvature (12). This ability to distort and was not observed (Fig. 4E). curve ER membranes could also affect lipid droplet formation or morphology. Therefore, until viperin localization and func- Viperin Does Not Alter the Number, Size, or Location of Lipid Droplets. tion are analyzed in the context of HCV infection, these possi- Finally, we examined whether viperin expression affected lipid bilities cannot be completely excluded. Furthermore, given that droplet formation. BMM⌽ isolated from WT and viperin knock- recent studies have shown that HCV brings ER membranes in out mice were examined using BODIPY staining and flow close apposition to lipid droplets, such analyses will have to be cytometry to quantitate the lipids present in droplets. WT performed carefully to distinguish ER localization from lipid BMM⌽, which were derived from a CD45.2-positive C57/B6 droplet localization (4). mouse, and viperin knockout BMM⌽, which were derived from An alternative possibility is that viperin affects the lipid a CD45.1-positive C57/B6 mouse, were cocultured for 5 days and content of lipid droplets. Previous studies with influenza virus then treated with 400 ␮M oleic acid overnight. The cells were indicated that viperin may alter the lipid content of cells by harvested and stained for CD45.1, CD45.2, viperin and binding to farnesyl diphosphate synthase (FPPS), an enzyme BODIPY. After gating on the individual CD45.1-positive and involved in the generation of cholesterol (14). Given that lipid CD45.2-positive populations, the cells were examined for both droplets are lipid storage organelles that are derived from the viperin expression and BODIPY staining. There was no differ- ER where cholesterol synthesis occurs, it is possible that the ence in the intensity of the BODIPY staining in the two ability of viperin to localize to lipid droplets provides another populations (Fig. S2A). Immunofluorescence analyses of WT mechanism to alter cellular lipids. Viperin could affect the type and viperin knockout BMM⌽ showed that there were no or quantity of lipids that accumulate in the ER membrane leaflet differences in the average number of lipid droplets per cell (Fig. and thus affect the lipid content of lipid droplets. Although it is S2C). In addition, microscopic analysis also showed that there still unclear if and how HCV uses the lipids within lipid droplets was no difference in the size or location of the lipid droplets in during its life cycle, HCV is linked with altered lipid metabolism WT and viperin knockout BMM⌽ (Fig. S2B). and steatosis (3, 17), a pathological effect that may be linked to lipid droplets. Discussion Regardless of the means by which viperin limits HCV repli- Viperin is an anti-microbial protein that clearly localizes to lipid cation, this study shows that viperin, like NS5A, localizes to lipid droplets. Given that several pathogens, notably HCV, are droplets through its N-terminal amphipathic ␣-helix. Similar to thought to induce and replicate on this intracellular organelle NS5A, this domain is also responsible for the ER localization of (1), these findings raise possibilities about how the cell responds viperin (7, 12). Therefore, these ␣-helices are both ER-targeting and fights these infections. Specifically, because viperin has been and lipid droplet-targeting domains. Previous studies indicated shown to be induced by HCV infection and to inhibit the that NS5A only minimally localizes to lipid droplets when replication of this virus (10, 11), the localization of viperin to expressed alone and that interactions with the HCV core are lipid droplets may reflect the mechanism that viperin uses to necessary to recruit NS5A to lipid droplets (4). However, our inhibit HCV. In addition, because the N-terminal amphipathic study examined only the N-terminal amphipathic ␣-helix of ␣-helices of both viperin and NS5A are sufficient to target both NS5A and did not examine the localization of the full-length viperin and dsRed to lipid droplets, it is clear that these helices protein. Therefore, the amphipathic ␣-helix may be sufficient to are lipid droplet-targeting domains and that viperin and NS5A localize NS5A to lipid droplets but other domains or an altered may use the same mechanism of localization. Finally, because conformation of the full-length protein in the absence of other viperin was already shown to localize to the cytoplasmic face of viral components may mask or regulate this domain so that it is the ER by its N-terminal amphipathic ␣-helix, the localization of only exposed at certain times during infection. viperin to lipid droplets provides further information on how Similar to our findings, other studies have identified hydro- lipid droplets may be generated. phobic domains that target proteins to lipid droplets. Some of the Viperin has been previously shown to be highly induced upon best characterized are the hydrophobic domains of caveolin-1 HCV infection and to limit HCV replication by an unknown and caveolin-2, which have been shown to be necessary to target mechanism (10, 11). In fact, when viperin was overexpressed in these proteins to lipid droplets (18, 19). Although caveolins HCV replicon-containing cell lines, these cells showed a signif- differ from viperin in that they encode two hydrophobic domains icant reduction in HCV RNA levels, which were comparable to that insert into membranes and generate a cytosolic hairpin, the HCV RNA levels when the cells were treated with 100 U/mL altering hydrophobic residues within these domains prevented IFN␣ (11), suggesting that viperin may be one of the most caveolin-1 or caveolin-2 from associating with lipid droplets (18, important IFN-induced that limits HCV replication. Fur- 19). Furthermore, these analyses indicated that the packing of ther analyses showed that deleting the N-terminal 50 amino acids the hydrophobic residues in these domains is specifically impor- of viperin significantly reduced the ability of viperin to limit tant for localizing to lipid droplets, as altering some of these HCV replication (11), indicating that the N-terminal amphi- residues did not affect association with other membranes, such pathic ␣-helix that localizes viperin to both the ER and lipid as the plasma membrane and Golgi (19). Because lipid droplets droplets is necessary to limit HCV replication. are derived from ER membranes, the packing and hydropho- Viperin could limit HCV replication through several mecha- bicity of the N-terminal amphipathic ␣-helices of viperin and nisms. Previous studies have shown that disrupting the ability of NS5A may be suited for binding to both the ER and lipid droplets the HCV core to localize to lipid droplets reduces the ability of and may facilitate their localization to lipid droplets in the HCV to produce progeny virions (16). The fact that neither context of infection.

20456 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0911679106 Hinson and Cresswell Downloaded by guest on September 27, 2021 The fact that both viperin and the dsRed chimeric proteins Transfection. Cells were transfected with the noted constructs using Lipo- appeared to completely surround the BODIPY-stained lipid fectamine 2000 (Invitrogen) following the manufacturer’s instructions droplets provides further support for the canonical model of ⌽ ⌽ lipid droplet generation, in which neutral lipids bud from the BMM Isolation. BMM were isolated from the hind limbs of C57/B6 wild-type and viperin knockout mice (21) and cultured in RPMI medium 1640 containing cytosolic leaflet of the ER membrane, which then becomes a 20% FBS, L-glutamine, non-essential amino acids, sodium pyruvate, and 20 hemi-membrane surrounding the lipid droplet (1, 2, 15). Alter- ng/mL M-CSF (R&D Systems). BMM⌽ were harvested and used after 5–7 days native models, such as the bicelle model, postulate that the lipids of culture. that accumulate in the ER leaflet are excised, along with both the cytosolic and luminal leaflets (1, 20). However, with this Sucrose Gradients and Western Blot Analysis. Cells were treated with 400 ␮M model, viperin might be expected to localize to only part of the oleic acid overnight. On the following day, the cells were harvested, washed lipid droplet membranes, depending on both the fluidity of the once with PBS, and then resuspended in cold hypotonic lysis (HLM) buffer (20 mM Tris⅐HCl and 1 mM EDTA, pH 7.4) with a protease inhibitor mixture (Roche membranes and whether viperin is relocalized on the lipid Applied Science). Cells were Dounce homogenized on ice and the cellular droplet membranes after they are formed. debris was pelleted at 1,000 ϫ g for 10 min at 4 °C. The supernatant was Regardless of the precise model by which lipid droplets are collected and brought to 20% sucrose using a 60% sucrose stock solution in generated, the results presented here provide further evidence HLM buffer. This solution was layered in the bottom of an ultracentrifuge that ER membranes are centrally involved in the process and tube, followed by a 5% sucrose solution in HLM buffer and HLM buffer alone. indicate that the cell may have evolved strategies to localize The gradients were centrifuged for 30 min at 13,000 rpm with no brake at 4 °C. The top, lipid droplet layer and lower membrane pellet were harvested. The anti-viral proteins to lipid droplets to inhibit specific pathogens. lipid droplet fraction was centrifuged at 14,000 ϫ g for 10 min at 4 °C and the Future studies are needed to investigate how viperin inhibits infranatant was removed with a syringe. Protein concentrations were deter- these pathogens and if viperin affects lipid droplet formation or mined using a Bradford assay and equal amounts of whole cell lysates and composition in the context of these infections. membrane fractions were loaded onto SDS/PAGE gels. Due to the exceedingly low protein concentration in the lipid droplet fractions, the same amount of Materials and Methods protein in the whole cell and membrane fractions could not be used for the Cell Lines, Antibodies, Constructs, and Reagents. 293T and HepG2 cells were lipid droplet fractions; therefore, an equal amount of protein for the lipid droplet fraction was loaded for each sample. The SDS/PAGE gels were trans- maintained in Dulbecco’s modified Eagle’s medium with 5% bovine calf ferred to PVDF membranes (Millipore) and then probed with the indicated serum. Mouse anti-viperin (MaP.VIP) and rabbit anti-calnexin were described antibodies. in ref. 12. The following antibodies were purchased commercially: anti-GRp94 (Assay Designs; SPA-850), anti-TPP-1 (Abcam; ab54685), anti-ADRP (Abcam; Immunofluorescence. Transfected 293T cells or primary BMM⌽ were plated ab52355 and Progen; 651102), and anti-dsRed (Clontech; 632496). BODIPY onto glass coverslips and treated with 400 ␮M oleic acid overnight. The cells 493/503 and fluorescent conjugated goat anti-rabbit and goat anti-mouse were fixed in 3.7% formaldehyde, washed, and then permeabilized with secondary antibodies were purchased from Molecular Probes, and HRP- 0.05% saponin. The cells were stained with the indicated antibodies or conjugated secondary antibodies were purchased from Jackson Laboratories. BODIPY 493/503 fluorescent dye. After washing, the cells were mounted onto All viperin and dsRed constructs were generated as described in ref. 12. The slides using ProLong Gold antifade reagent (Invitrogen) and then examined NS5A construct was a kind gift from Dr. Darius Moradpour (University of under a 63ϫ objective using a Leica confocal microscope. Freiburg, Germany). The N-terminal amphipathic ␣-helix of NS5A was PCR amplified and then fused to the C terminus of viperin. Oleic acid was pur- ACKNOWLEDGMENTS. We thank Dr. Crina Paduraru for technical advice and chased from Sigma and complexed with BSA in PBS. Mouse IFN-␣ (IFN␣) was Mrs. Nancy Dometios for manuscript preparation. This work was supported by purchased from R&D Systems. the Ellison Foundation and the Howard Hughes Medical Institute.

1. Guo Y, Cordes KR, Farese RV, Jr, Walther TC (2009) Lipid droplets at a glance. J Cell Sci 12. Hinson ER, Cresswell P (2009) The N-terminal amphipathic ␣-helix of viperin mediates 122:749–752. localization to the cytosolic face of the endoplasmic reticulum and inhibits protein 2. Fujimoto T, Ohsaki Y, Cheng J, Suzuki M, Shinohara Y (2008) Lipid droplets: A classic secretion. J Biol Chem 284:4705–4712. organelle with new outfits. Histochem Cell Biol 130:263–279. 13. Mottola G, et al. (2002) Hepatitis C virus nonstructural proteins are localized in a 3. Barba G, et al. (1997) Hepatitis C virus core protein shows a cytoplasmic localization and modified endoplasmic reticulum of cells expressing viral subgenomic replicons. Virol- MICROBIOLOGY associates to cellular lipid storage droplets. Proc Natl Acad Sci USA 94:1200–1205. ogy 293:31–43. 4. Miyanari Y, et al. (2007) The lipid droplet is an important organelle for hepatitis C virus 14. Wang X, Hinson ER, Cresswell P (2007) The -inducible protein viperin inhibits production. Nat Cell Biol 9:1089–1097. influenza virus release by perturbing lipid rafts. Cell Host Microbe 2:96–105. 5. Boulant S, et al. (2008) Hepatitis C virus core protein induces lipid droplet redistribution 15. Martin S, Parton RG (2006) Lipid droplets: A unified view of a dynamic organelle. Nat in a microtubule- and dynein-dependent manner. Traffic 9:1268–1282. Rev Mol Cell Biol 7:373–378. 6. Boulant S, et al. (2006) Structural determinants that target the hepatitis C virus core 16. Boulant S, Targett-Adams P, McLauchlan J (2007) Disrupting the association of hepa- protein to lipid droplets. J Biol Chem 281:22236–22247. titis C virus core protein with lipid droplets correlates with a loss in production of 7. Brass V, et al. (2002) An amino-terminal amphipathic ␣-helix mediates membrane associ- infectious virus. J Gen Virol 88:2204–2213. ation of the hepatitis C virus nonstructural protein 5A. J Biol Chem 277:8130–8139. 17. Moriya K, et al. (1997) Hepatitis C virus core protein induces hepatic steatosis in 8. Elazar M, et al. (2003) Amphipathic helix-dependent localization of NS5A mediates transgenic mice. J Gen Virol 78:1527–1531. hepatitis C virus RNA replication. J Virol 77:6055–6061. 9. Elazar M, Liu P, Rice CM, Glenn JS (2004) An N-terminal amphipathic helix in hepatitis 18. Fujimoto T, Kogo H, Ishiguro K, Tauchi K, Nomura R (2001) Caveolin-2 is targeted to C virus (HCV) NS4B mediates membrane association, correct localization of replication lipid droplets, a new ‘‘membrane domain’’ in the cell. J Cell Biol 152:1079–1085. complex proteins, and HCV RNA replication. J Virol 78:11393–11400. 19. Ostermeyer AG, Ramcharan LT, Zeng Y, Lublin DM, Brown DA (2004) Role of the 10. Helbig KJ, Lau DT, Semendric L, Harley HA, Beard MR (2005) Analysis of ISG expression hydrophobic domain in targeting caveolin-1 to lipid droplets. J Cell Biol 164:69–78. in chronic hepatitis C identifies viperin as a potential antiviral effector. Hepatology 20. Ploegh HL (2007) A lipid-based model for the creation of an escape hatch from the 42:702–710. endoplasmic reticulum. Nature 448:435–438. 11. Jiang D, et al. (2008) Identification of three interferon-inducible cellular enzymes that 21. Qiu LQ, Cresswell P, Chin KC (2009) Viperin is required for optimal Th2 responses and inhibit the replication of hepatitis C virus. J Virol 82:1665–1678. T-cell receptor-mediated activation of NF-kappaB and AP-1. Blood 113:3520–3529.

Hinson and Cresswell PNAS ͉ December 1, 2009 ͉ vol. 106 ͉ no. 48 ͉ 20457 Downloaded by guest on September 27, 2021