Palmitoylated Cysteines in Chikungunya Virus nsP1 Are Critical for Targeting to Cholesterol-Rich Plasma Membrane Microdomains with Functional Consequences for Viral Genome Replication

Functional alphavirus replication complexes are anchored to the host cell membranes through the interaction of nsP1 with the lipid bilayers. In this work, we investigate the importance of cholesterol for such an association. We show that nsP1 has affinity for cholesterol-rich membrane microdomains formed at the plasma membrane and identify conserved palmitoylated cysteine(s) in nsP1 as the key determinant for cholesterol affinity. We demonstrate that drug-induced cholesterol sequestration in late endosomes not only redirects nsP1 to this compartment but also dramatically decreases genome replication, suggesting the functional importance of nsP1 targeting to cholesterol-rich plasma membrane microdomains. Finally, we show evidence that nsP1 from chikungunya and Sindbis viruses displays different sensitivity to cholesterol sequestering agents that parallel with their difference in the requirement for nsP1 palmitoylation for replication. This research, therefore, gives new insight into the functional role of palmitoylated cysteines in nsP1 for the assembly of functional alphavirus replication complexes in their mammalian host.

ABSTRACT

In mammalian cells, alphavirus replication complexes are anchored to the plasma membrane. This interaction with lipid bilayers is mediated through the viral methyl/guanylyltransferase nsP1 and reinforced by palmitoylation of cysteine residue(s) in the C-terminal region of this protein. Lipid content of membranes supporting nsP1 anchoring remains poorly studied. Here, we explore the membrane binding capacity of nsP1 with regard to cholesterol. Using the medically important chikungunya virus (CHIKV) as a model, we report that nsP1 cosegregates with cholesterol-rich detergent-resistant membrane microdomains (DRMs), also called lipid rafts. In search for the critical factor for cholesterol partitioning, we identify nsP1 palmitoylated cysteines as major players in this process. In cells infected with CHIKV or transfected with CHIKV trans-replicase plasmids, nsP1, together with the other nonstructural proteins, are detected in DRMs. While the functional importance of CHIKV nsP1 preference for cholesterol-rich membrane domains remains to be determined, we observed that U18666A- and imipramine-induced sequestration of cholesterol in late endosomes redirected nsP1 to these compartments and simultaneously dramatically decreased CHIKV genome replication. A parallel study of Sindbis virus (SINV) revealed that nsP1 from this divergent alphavirus displays a low affinity for cholesterol and only moderately segregates with DRMs. Behaviors of CHIKV and SINV with regard to cholesterol, therefore, match with the previously reported differences in the requirement for nsP1 palmitoylation, which is dispensable for SINV but strictly required for CHIKV replication. Altogether, this study highlights the functional importance of nsP1 segregation with DRMs and provides new insight into the functional role of nsP1 palmitoylated cysteines during alphavirus replication.

IMPORTANCE Functional alphavirus replication complexes are anchored to the host cell membranes through the interaction of nsP1 with the lipid bilayers. In this work, we investigate the importance of cholesterol for such an association. We show that nsP1 has affinity for cholesterol-rich membrane microdomains formed at the plasma membrane and identify conserved palmitoylated cysteine(s) in nsP1 as the key determinant for cholesterol affinity. We demonstrate that drug-induced cholesterol sequestration in late endosomes not only redirects nsP1 to this compartment but also dramatically decreases genome replication, suggesting the functional importance of nsP1 targeting to cholesterol-rich plasma membrane microdomains. Finally, we show evidence that nsP1 from chikungunya and Sindbis viruses displays different sensitivity to cholesterol sequestering agents that parallel with their difference in the requirement for nsP1 palmitoylation for replication. This research, therefore, gives new insight into the functional role of palmitoylated cysteines in nsP1 for the assembly of functional alphavirus replication complexes in their mammalian host.

INTRODUCTION

In the last decade, evidence has pointed toward the intricate relationship between host lipid metabolism and the replication of viral pathogens. Indeed, viruses can co-opt or reprogram lipid signaling, synthesis, and metabolism either to generate ATP, to extend cellular membranes, or to remodel membrane lipid content. These modifications will serve to create an environment that is optimal for viral replication. This need is dictated by the pivotal role played by membranes in almost all steps of the virus life cycle. Indeed, the importance of cellular lipids during the binding/entry process and assembly/budding of new infectious progeny into the extracellular space has long been appreciated (for review see reference 1). However, the discovery that viruses with a positive-stranded RNA genome [(+)RNA viruses] replicate in association with host cell membranes has expanded the regulatory function of lipids to genome replication. It is now well established that such viruses create membranous compartments, also called virus replication organelles, originating from the endoplasmic reticulum (ER), Golgi apparatus, mitochondria, peroxisomes, endosomes/lysosomes, or from the plasma membrane (PM) (2). In these compartments, viral replication proteins bind cell membranes with an affinity for determined lipid species (3, 4).

Cell membranes are composed of phospholipids, glycolipids, and cholesterol. Among other lipids, cholesterol constitutes a unique type of cellular membrane building block. It is responsible for regulating the fluidity and impermeability of lipid bilayers. In vertebrate cells, cholesterol homeostasis is maintained through de novo synthesis in the ER, with 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase being the rate-limiting enzyme, and through receptor-mediated intake from the extracellular medium in the form of low-density lipoproteins (LDLs) (5). In this scenario, LDLs are delivered to and hydrolyzed in late endosomes/lysosomes (LE/Ls) where free cholesterol is released. Then, cholesterol requires proper intracellular transport to exit the LE/Ls and reaches its final destination, mainly at the PM, in the Golgi apparatus, and the ER. This is ensured by Niemann-Pick C 1 and 2 (NPC1 and NPC2, respectively) proteins localized at the limiting membrane of lysosomes and in the lysosomal lumen, respectively (6). Once at the PM, cholesterol together with glycosphingolipids, glycophosphatidylinositol (GPI)-anchored proteins, and transmembrane proteins can cluster into discrete domains. These cholesterol-enriched detergent-resistant membrane microdomains (DRMs), also referred to as lipid rafts, have been identified as platforms for both endocytosis of penetrating viral particles and for progeny assembly and budding (7). More recently, cholesterol regulatory function was extended to the replication step of (+)RNA viruses. The local accumulation of cholesterol was proposed to contribute to the creation of a membrane microenvironment conducive to assembly and optimal functioning of replication complexes formed by members of the Flaviviridae family, including hepatitis C virus (HCV) (8–10) and West Nile virus (11), by members of Picornaviridae family, including coxsackievirus and poliovirus (12), and by plant viruses from the Bromoviridae family (brome mosaic virus [BMV]) (13). Consistent with the idea that cholesterol accumulation may be required for optimal activity of viral replication machinery, manipulation of cholesterol metabolism was found to impair the genome replication of taxonomically divergent (+)RNA viruses (8, 12, 14–18).

Alphaviruses are (+)RNA viruses, which are predominantly transmitted to vertebrates by mosquito vectors. Chikungunya virus (CHIKV) is an Old World alphavirus causing millions of infections in tropical and subtropical geographical areas, with a potential risk of spreading to regions with a temperate climate. It has recently received significant attention as a consequence of its reemergence in the Indian Ocean and Caribbean Islands before spreading worldwide (19). CHIKV, like other alphaviruses, replicates its genome in membranous niches derived from the host PM (20–22). The replication complex confined in these organelles, termed spherules, contains the four nonstructural proteins nsP1, nsP2, nsP3, and nsP4 encoded by the 5′ open reading frame (ORF) in the viral genome, which are expressed in the form of P123 and P1234 polyprotein precursors. In this complex, membrane binding is mediated by nsP1, the viral methyltransferase (MTase) and guanylyltransferase (GTase), which catalyzes the formation of the cap structure at the 5′ end of nascent positive-strand viral RNAs (23, 24). A putative α-helix central to nsP1 was proposed to play a crucial role for membrane attachment (25, 26). S-acylated cysteines located in the C-terminal region of this protein were then proposed to stabilize this interaction (27). The functional requirement for nsP1 interaction with membranes varies among alphaviruses. It is strictly required for Semliki Forest virus (SFV) nsP1 enzymatic activity (28), while it is dispensable for that of Sindbis virus (SINV) nsP1 (29). More specifically, alanine substitution of palmitoylated cysteines has only a marginal inhibitory effect on SINV replication (30, 31), while leading to the acquisition of compensatory mutations for SFV (31, 32). Recently, this mutation was reported to completely abolish the replication of CHIKV (33, 34).

In vitro, SFV nsP1 has an affinity for anionic phospholipids, especially phosphatidylserine, phosphatidylglycerol, and cardiolipin; these lipid species significantly improve its capping activity (25, 28). In cells, however, nsP1 affinity for specific lipids remains mostly uninvestigated. In recent years, cholesterol metabolism was reported to be critical for the replication of the alphavirus genome. Indeed, SINV RNA replication and protein synthesis are significantly decreased in fibroblasts from patients with type A Niemann-Pick disease (NPD-A), which induces cholesterol and sphingolipid storage in LE/Ls (35). More recently, we reported that U18666A, a class II cationic amphipathic steroid 3-β-[2-(diethylamino)ethoxy]androst-5-en-17-one, and the anti-depressant drug imipramine, which both phenocopy NPD-A, are CHIKV inhibitors with potential activity against RNA replication steps (16). Here, we further question the importance of cholesterol metabolism in the alphavirus life cycle. We especially explore the outcome of cholesterol manipulation on nsP1 subcellular distribution and membrane anchoring. We show that nsP1 has affinity for cholesterol-rich membranes. Indeed, U18666A or imipramine redirect CHIKV nsP1 to Lamp2-positive endosomes where unesterified cholesterol accumulates. At the plasma membrane, nsP1 cosegregates with cholesterol-containing DRMs, as attested by membrane flotation assays. Investigating the molecular basis of nsP1 targeting to DRMs revealed that the cysteines, previously reported to be palmitoylated (34), are the main determinants for association with these domains. Interestingly, when expressed transiently in the context of a P1234 polyprotein precursor or by an infectious CHIKV, nsP2, nsP3, and nsP4 were found to cosegregate with nsP1 in cholesterol-rich membrane fractions, a property that was abolished when nsP1 cysteines were mutated. In a parallel study, we found that cysteine-mediated nsP1 association with cholesterol-rich membranes is conserved for SINV despite being less marked than for CHIKV. Moreover, SINV nsP1 was also less sensitive to U18666A-induced cholesterol sequestration. This phenotype with regard to nsP1 cholesterol partitioning parallels the reduced requirement for cysteine palmitoylation previously reported for SINV replication (30, 31). Altogether, this study provides clues on the proviral role of cholesterol in alphavirus replication, suggesting its regulatory function in nsPs association with the PM and presumably in the formation of functional replication complexes.

RESULTS

Cholesterol is pivotal for alphavirus genome replication.

First, we set out to study the involvement of cholesterol homeostasis in CHIKV replication. Cholesterogenesis was inhibited using lovastatin, an FDA-approved inhibitor of HMG-CoA reductase, that catalyzes the conversion of HMG-CoA to mevalonate in the cholesterol biosynthesis pathway (Fig. 1A) (36). Cholesterol availability at the PM was reduced using U18666A or imipramine. By targeting the NPC1 transporter, both drugs block the transfer of endocytosed cholesterol from late endosomes to different organelles, including the PM, without a significant effect on other lipid species (37). Additionally, U18666A inhibits enzymes of the cholesterol synthesis pathway (38). Each drug, used in a concentration range that was controlled to have limited toxicity (Fig. 1B, b, d to f), was added to HEK293T cells 30 min before infection with CHIKV-LR-5′GFP at a multiplicity of infection (MOI) of 0.5. Cells were maintained in culture with the drugs for 24 h, and then infection was monitored measuring levels of GFP reporter expressed by this recombinant virus. In each case, CHIKV infection was significantly reduced compared with the mock-treated condition (Fig. 1B, a, c to e).

FIG 1.

Cholesterol homeostasis regulates the CHIKV life cycle. (A) Cholesterol metabolism and biosynthesis inhibitors. (B) Drugs affecting cholesterol metabolism inhibit CHIKV infection. HEK293T cells treated for 30 min with the indicated concentrations of lovastatin (a), U18666A (c), and imipramine (e) were infected with CHIKV-LR-5′GFP at an MOI of 0.5. Infection was monitored after 24 h by quantification of the GFP reporter in the cell lysates. For each drug concentration, cell viability was determined after 24 h using the Cell Titer Glo assay (b, d, f). Values are expressed as a percentage of the control condition that was taken as 100%. A representative experiment is shown. Values are means of triplicates ± SEM. P values are calculated by comparing each treated condition with the mock.

We and others have demonstrated that the depletion of membrane cholesterol is deleterious for fusion of the virion and host membranes (39, 40). Because host membranes are also pivotal for replication of alphavirus genome through the creation of membranous replication organelles, we investigated whether cholesterol biosynthesis is also required for the postentry step of the CHIKV infection cycle. To this end, we performed experiments in which cells were treated with cholesterol metabolism and transport inhibitors 1 h after CHIKV infection. Under these conditions, lovastatin, U18666A, and imipramine decreased CHIKV genome replication (Fig. 2A). To definitively omit drug effects on viral entry, we finally took advantage of a CHIKV trans-replication system that recapitulates postentry events of the CHIKV life cycle (33). Plasmids CMV-P1234 and HSPolI-Fluc-Gluc, encoding the P1234 polyprotein and a replication-competent template RNA containing firefly and Gaussia luciferase reporter genes under the control of genomic and subgenomic viral promoters, respectively (Fig. 2B), were cotransfected in HEK293T cells. Consequences of cholesterol transport inhibition were assayed by adding increasing concentrations of lovastatin, U18666A, and imipramine to transfected cells and quantification of reporter activities in cell lysates. As depicted in Fig. 2C, reporter expression directed by both genomic and subgenomic promoters was decreased by the drugs compared with control conditions. Altogether, these results indicate that cholesterol homeostasis, including ongoing biosynthesis and transport of unesterified cholesterol to the host membranes, is pivotal for CHIKV genome replication.

FIG 2.

Cholesterol transport inhibitors regulate intracellular CHIKV replication. (A) Postinfection addition of lovastatin, U18666A, or imipramine reduces CHIKV replication. Increasing concentrations of drugs were added to HEK293T cells 1 h after CHIKV infection at an MOI of 1. After 24 h, CHIKV replication was monitored by quantification of GFP fluorescence in the cell lysates. Values, normalized to the protein content in the samples, are expressed as a percentage of the noninfected (NI) condition. (B) Schematics of CMV-P1234 and HSPolI-Fluc-Gluc constructs; plasmid backbones are shown. (C) HEK293T cells were transfected with plasmids depicted in (B) and treated with the indicated concentrations of lovastatin, U18666A, or imipramine at 2 h posttransfection. Firefly and Gaussia luciferase activities were determined after 24 h in culture. A representative experiment is shown. Values are expressed as a percentage of the untreated condition and are means of triplicates ± SEM. NT designates nontransfected control cells. P values are calculated by comparing each treated condition with the mock.

α-Helix and putative palmitoylated cysteines cooperate for CHIKV nsP1 binding to host membranes.

Alphavirus replication complexes are anchored to the host membranes thanks to the nsP1 membrane-binding capacity. While extensively reported for nsP1 encoded by SFV and SINV (25, 30), this feature has not yet been studied functionally for CHIKV. To investigate CHIKV nsP1 behavior with regard to host cell membranes, a plasmid encoding a GFP-fused CHIKV nsP1 protein (GFP-nsP1) (Fig. 3A) was generated and used to transfect HEK293T cells. The nsP1 membrane association was investigated by fractionation of transfected cells. Postnuclear extract prepared from GFP-nsP1-expressing cells was separated into membranous (P25) and cytosolic (S25) samples by differential centrifugation, as previously reported (41). Each fraction was resolved using SDS-PAGE and probed with anti-GFP antibodies and with antibodies against Na⁺/K⁺ ATPase or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) that, respectively, associate with membrane and cytosolic compartments. Under these conditions, GFP alone was detected in the cytosolic sample together with GAPDH (Fig. 3B and C). In contrast, GFP-nsP1 was detected in the membrane fraction also containing Na⁺/K⁺ ATPase. In parallel, HeLa cells expressing these proteins were analyzed using confocal microscopy. As expected, the fluorescence of individual GFP was diffuse in the cytoplasm and nucleus. However, GFP-nsP1 fluorescence overlapped with PM stained using wheat germ hemagglutinin (WGA) conjugated with Alexa Fluor 647 (Fig. 3D). As previously reported for related alphaviruses (42), expression of GFP-nsP1 generated huge membrane reshaping creating filopodia- and lamellipodia-like structures covering the entire cell surface, which stained positive for the green fluorescence. Of note, this profile was also observed for a C-terminal GFP-tagged nsP1-GFP protein as well as for an untagged nsP1 detected by the mean of anti-nsP1 serum (Fig. 3E), thereby supporting that GFP-nsP1 behaves as native nsP1 regarding localization and association with membranes. According to the focal plane chosen, a fraction of each of these proteins was also detected as small cytosolic aggregates, as illustrated in Fig. 3E.

FIG 3.

CHIKV nsP1 associates with the PM through conserved sequence motifs. (A) Organization of the CHIKV nsP1. Conserved amino acids involved in the central α-helix (amino acid [aa] 244 to 263) and those required for nsP1 acylation (aa 417 to 419) are indicated. Mutants used in this study are depicted. (B) Membrane affinity of GFP-nsP1, GFP-nsP1_3A, GFP-nsP1_W258A, and GFP-nsP1_DM was determined by cell fractionation assays. HEK293T cells transfected to express each GFP-fused protein were fractionated to produce cytosolic (S25) and membranous (P25) fractions. Equal amounts of fractions were analyzed by SDS-PAGE and Western blotting with anti-nsP1 or anti-GFP antibodies. Antibodies against Na⁺/K⁺ ATPase (ATPase) and GAPDH were used to control the separation of membranes and cytosolic compartments, respectively. For each experiment, a sample of the unfractionated cell lysate (CL) was run in parallel to determine the relative amount of each nsP1 mutant expressed. Cells expressing the GFP protein alone are shown as a control. (C) The intensity of bands in S25 and P25 samples from (B) was determined using Image J software and plotted as a histogram. (D) Subcellular localization of GFP-nsP1 and its mutant was determined by confocal imaging of transfected HeLa cells. Cell membranes are labeled with Alexa Fluor 647-conjugated wheat germ hemagglutinin (WGA). Scale bars are 5 μm. (E) Confocal microscopy imaging of HeLa cells transfected with expression plasmids for nsP1-GFP or for an untagged nsP1. The untagged nsP1 protein was detected with rabbit antiserum and Alexa 647-conjugated secondary reagents. Nuclei were stained with DAPI before analysis by confocal imaging. GFP-nsP1 at the plasma membrane and filopodia-like structures formed at the surface of transfected cells are indicated by white arrows. Cytosolic aggregates are indicated by blue arrows. Bars, 5 μm. (F) Lysates of cells expressing GFP-nsP1, GFP-nsP1_3A, GFP-nsP1_W258A, and GFP were subjected to a membrane flotation assay in an iodixanol gradient. Fractions collected from top to bottom were separated on an SDS-PAGE and probed with antibodies specific for nsP1. Lysates from GFP-expressing cells were probed with antibodies specific for GFP, flotillin-1 (FLOT1), or ATPase. Data are representative of three separate experiments.

Because alphavirus nsP1 was previously proposed to be trafficked to endosomes (43) and in light of our confocal microscopy analysis, we further questioned nsP1 subcellular distribution by performing fractionation assays that separated the PM from other cell membranes. The postnuclear extract was prepared from GFP-nsP1-expressing cells and separated by isopycnic centrifugation in a self-forming linear 10% to 20% to 30% iodixanol density gradient. Twenty-four samples were collected from top to bottom and assayed by Western blotting for GFP-nsP1 content. A roughly equal proportion of GFP-nsP1 was detected in the top fractions 1 and 2 and also in fractions 9 to 12 that all stained positive for the Na⁺/K⁺ ATPase membrane marker (Fig. 3F). These fractions also contained flotillin-1 (FLOT1) that is known to localize predominantly to the PM and endosomal compartments, i.e., late endosomes and recycling endosomes (44). In contrast, individual GFP segregated with fractions 17 to 24 at the bottom of the gradient that corresponded to the cytosolic compartment. Altogether, these results show that CHIKV nsP1 is a membrane-associated protein, which cofractionates equally with the PM and some internal membranes, probably endosomal in nature, suggesting that this protein may traffic between the two compartments.

We next questioned whether the functions of membrane-binding determinants previously identified in alphavirus nsP1 proteins are also conserved for CHIKV nsP1. Extensive analysis performed using SFV and SINV as models established that membrane association of nsP1 relies on a central conserved sequence that folds as an amphipathic α-helix when studied as a synthetic peptide in solution (25). In this sequence, a pivotal role in membrane anchoring was attributed to tryptophan at position 259 (W₂₅₉) that sinks into the phospholipid bilayer (25). This interaction was proposed to be reinforced by the presence of cysteine residue at position 420 (C420) in SINV or cysteine residues 418 to 420 in SFV (C418-420) which are covalently palmitoylated and render the protein highly hydrophobic (27, 30). These motifs are conserved in the CHIKV nsP1 sequence where W258 and C417-419 of residues correspond to W259 and C418-420 of SFV nsP1, respectively (Fig. 3A). Recently, C417-419 in CHIKV nsP1 were reported to be palmitoylated (34), therefore further attesting that nsP1 palmitoylation is a conserved feature among Old World alphaviruses. To study the contribution of each motif in CHIKV nsP1 membrane binding, we generated GFP-nsP1 mutants in which W258 (GFP-nsP1_W258A) or C417-419 (GFP-nsP1_3A) residues in the nsP1 sequence were replaced by alanine as well as a double nsP1 mutant bearing a combination of W₂₅₈A or C_417-419A mutations (GFP-nsP1_DM). Fluorescence microscopy of cells transfected with the corresponding plasmids evidenced a diffuse green fluorescence that predominated in the cytoplasm, which is in contrast with cells expressing GFP-nsP1 that was mainly detected at the plasma membrane (Fig. 3D). Fractionation experiments confirmed that GFP-nsP1_W258A and GFP-nsP1_3A were more abundant in the cytosolic fraction than GFP-nsP1, with a significant amount of each mutant protein remaining associated with cell membranes (Fig. 3B and C). The same cell extracts were then subjected to a membrane flotation assay in iodixanol gradient to appreciate their capacity to associate with the internal/plasma membrane compartments. GFP-nsP1_W258A and GFP-nsP1_3A proteins were both detected in fractions 18 to 24 of the gradient, thereby confirming that these proteins have a decreased membrane affinity (Fig. 3F). Each mutant protein was also present at the top of the gradient in fractions 1 to 2 corresponding to the PM and in fractions 9 to 12 corresponding to internal membranes. The distribution of mutant proteins in these fractions generally mimicked that of GFP-nsP1, except that GFP-nsP1_W258A was somewhat more abundant in internal membrane fractions. This was also reflected by the more diffuse localization of GFP fluorescence in cells expressing GFP-nsP1_W258A (compare panels b and d of Fig. 3D), indicating a possible role of W258 in CHIKV nsP1 affinity for the PM. Nevertheless, the observed differences were small, suggesting that W₂₅₈A and C_417-419A mutations, despite decreasing membrane affinity, do not prominently modify nsP1 distribution between intracellular and plasma membranes or nsP1 trafficking capacity. Interestingly, membrane association was further decreased when W₂₅₈A and C_417-419A mutations were combined in GFP-nsP1_DM (Fig. 3B and D). Altogether, these results establish that CHIKV nsP1 associates both with the plasma and internal membranes, an association dictated by the cooperation of motifs that contain C417-419 and W258 amino acids in this protein.

Cholesterol storage defect redirects CHIKV nsP1 to late endosomes.

Based on the observation that cholesterol synthesis and transport is required for CHIKV RNA replication and on the intimate relationship of CHIKV nsP1 with membranes, we next wondered whether cholesterol homeostasis has an impact on nsP1 behavior. We first investigated nsP1 subcellular localization with respect to cholesterol distribution. Cells transfected with plasmid-expressing GFP-nsP1 were incubated with U18666A or imipramine to generate a cholesterol entrapment in LE/Ls (45). This capacity was controlled using the antifungal antibiotic filipin III that forms a fluorescent complex upon association with unesterified cholesterol (46). In the presence of U18666A or imipramine, filipin III evidenced the formation of large fluorescent cytosolic aggregates that contrasted with the presence of cholesterol at the PM and more evenly distributed in the cytoplasm of mock-treated cells (Fig. 4A; data not shown). For U18666A, these clusters colocalized with Lamp2 (Fig. 4B), a LE/Ls marker, thereby confirming that this pharmacological agent stimulates the accumulation of unesterified cholesterol in late endosomes, as previously reported (47). Using these experimental conditions, we next assessed the impact of cholesterol storage defect on GFP-nsP1 subcellular localization and membrane affinity. Fractionation assays established that drug treatment did not increase the amount of cytosolic nsP1 (Fig. 4C, S25 fraction), indicating that it had no consequence on nsP1 membrane affinity. However, microscopy imaging revealed that, in the presence of U18666A, a significant part of GFP-nsP1 fluorescence was redirected from the PM to cytosolic aggregates. This signal overlapped with filipin III staining, as attested by a cross-sectional analysis of the fluorescent signals (Fig. 4D and E). In similar experimental conditions, no redistribution of individual GFP fluorescence was observed. Accordingly, upon cholesterol storage condition, GFP-nsP1 colocalized with unesterified cholesterol stored in late endosomes, as quantified by calculating Mander’s coefficient (Fig. 4F). This situation contrasted with that of mock-treated cells in which the GFP-nsP1 fluorescence colocalized with filipin III at the PM and did not overlap Lamp 2 staining. Altogether, our results argue that the inhibition of NPC1-mediated cholesterol transport by U18666A redirects GFP-nsP1 to LE/Ls where unesterified cholesterol accumulates. Retargeting GFP-nsP1 to these compartments has no significant impact on nsP1 membrane-binding capacity.

FIG 4.

Cholesterol trafficking inhibitors redirect nsP1 to endo/lysosomal compartments. (A) HeLa cells cultured in the presence of 12 μM U18666A were stained with filipin III to visualize cholesterol distribution. Cells cultured in medium containing an appropriate amount of vehicle are shown as control (mock). (B) Cells expressing the GFP-nsP1 protein were maintained for 24 h in the presence of 12 μM U18666A and then were colabeled with filipin III and anti-Lamp2 antibodies to visualize cholesterol and endo/lysosomes, respectively. Cells maintained with vehicle alone are shown as controls (mock). (C) Consequences of U18666A treatment on GFP-nsP1 membrane-binding capacity was determined by cell fractionation assays. Crude lysates (CLs) and cytosolic fractions (S25) prepared from mock-treated or U18666A-treated HEK293T cells were separated on SDS-PAGE and probed with antibodies against GFP or GAPDH. (D) HeLa cells transfected with a GFP or GFP-nsP1 expression plasmid and maintained with 12 μM U18666A or in medium alone (mock) for 24 h were stained with filipin III and DAPI and then processed for confocal imaging. (E) Histograms indicate the signal intensity profile of filipin III and green fluorescence along the white lines in (D). Bars, 5 μm. (F) Colocalization of filipin and GFP signals in U18666A-treated cells was determined by calculation of Mander’s overlap coefficients (n ≥ 20 cells).

C417-419, but not the W258 residue, determine CHIKV nsP1 sensitivity to cholesterol distribution.

Palmitoylation governs protein trafficking and association with membranes. A major focus of studies on protein palmitoylation has been the role of this modification in promoting an interaction with gangliosides and cholesterol, leading, at certain conditions, to protein translocation to raft/caveolae membrane domains (48). Given that C417-429 in CHIKV nsP1, required for optimal membrane association, have recently been reported to be palmitoylated (34), we investigated their contribution to GFP-nsP1 cholesterol affinity. The above-described experiments were repeated using cells transfected with the GFP-nsP1_3A plasmid. In the presence of U18666A, GFP-nsP1_3A fluorescence remained diffuse in the cytoplasm whether the cells were maintained in the presence of U18666A or with an appropriate concentration of vehicle (Fig. 5C). GFP-nsP1_W258A fluorescence was concentrated in cytosolic foci colocalized with filipin III fluorescence (Fig. 5A), as observed for GFP-nsP1. These phenotypes were confirmed by cross-sectional analysis of the fluorescent signals (Fig. 5B and D) and by calculation of Mander’s coefficients (Fig. 5E). Accordingly, W258, even if required for strengthening nsP1 membrane affinity, is not critical for cholesterol affinity. In contrast, a cysteine-to-alanine substitution in mutant GFP-nsP1_3A abolished sensitivity to U18666A, suggesting a critical role of palmitoylated cysteines in nsP1 cholesterol affinity.

FIG 5.

GFP-nsP1_W258A and GFP-nsP1_3A display different sensitivity to U18666A-induced cholesterol storage defect. HeLa cells expressing either GFP-nsP1_3A (A) or GFP-nsP1_W258A (C) were incubated with 12 μM U18666A for 24 h. Unesterified cholesterol was stained with filipin III, and cells were processed for confocal microscopy. (B and D) Histograms showing filipin III and GFP fluorescence intensity along the white lines indicated in (A) and (B), respectively. Bars, 5 μm. (E) Colocalization of filipin and GFP signals in U18666A-treated cells was determined by calculation of Mander’s overlap coefficients (n ≥ 20 cells).

To further explore GFP-nsP1_3A behavior with regard to cholesterol, we took advantage of CD81, a heavily palmitoylated tetraspanin, segregating mainly with DRMs, as a lipid raft biomarker (49). Indeed, in our hands, this protein was mainly detected at the PM of untreated U2OS cells (Fig. 6A). It was redirected to cytosolic aggregates in cells cultured with U18666A or imipramine (Fig. 6B) where it colocalized with GFP-nsP1, as attested by Mander’s and Pearson’s coefficient calculation (Fig. 6D and E). CD81, therefore, appears as a sensitive cholesterol sensor. Using this property, we explored the importance of nsP1 palmitoylated cysteines for cholesterol dependency. Cells transfected to express GFP-nsP1_3A were incubated in the presence of U18666A. Despite inducing the clustering of CD81 into intracellular compartments, this treatment did not affect GFP-nsP1_3A distribution (Fig. 6C). Indeed, GFP-nsP1_3A remained detected as a diffuse cytoplasmic protein, as observed in mock-treated cells, thereby contrasting with the phenotype observed for GFP-nsP1. No colocalization of CD81 and GFP-nsP1_3A fluorescence was observed (Fig. 6D). Optical sectioning (z-stack) and three-dimensional (3D) volume reconstruction from image stacks confirmed that conversely to GFP-nsP1, GFP-nsP1_3A poorly colocalized with CD81 in drug-treated cells (Fig. 6E and F). These phenotypes were confirmed from cells cultured with imipramine. According to these results, nsP1 behaves similarly to CD81 with regard to cholesterol storage, a property that requires the presence of palmitoylated cysteines in nsP1 C terminus.

FIG 6.

C417-419 are key determinants for CHIKV nsP1 targeting to CD81-positive cholesterol-rich compartments. U2OS cells expressing the GFP (A), GFP-nsP1 (B), and GFP-nsP1_3A (C) were cultured under control conditions or in medium supplemented with 12 μM U18666A or 75 μM imipramine for 24 h. Then, the cells were probed with anti-CD81 monoclonal antibodies and Alexa Fluor 594-conjugated secondary antibodies. Nuclei were stained with DAPI, and the cells were processed for confocal microscopy. Scale bars are 5 μm. (D) Quantification of the degrees of colocalization of GFP fluorescence and CD81 staining in treated cells were determined by calculation of Mander’s and Pearson’s overlap coefficients (n ≥ 20 cells). (E and F) 3D image reconstruction was performed from z-stacks acquired from treated cells expressing either GFP-nsP1 or GFP-nsP1_3A.

CHIKV nsP1 partitions with DRMs.

Given that nsP1 subcellular localization is sensitive to cholesterol redistribution to the PM, we questioned its capacity to segregate with cholesterol-enriched membrane microdomains. Cholesterol is not uniformly distributed in membranes. In living cells, it concentrates in nanoscale assemblies, also enriched in sphingolipids and glycosylphosphatidylinositol (GPI)-anchored proteins, referred to as lipid rafts. These compartments are characterized biochemically by their insolubility in nonionic detergents, a property reflected in their name (detergent-resistant membrane microdomains [DRMs]) and by their light density on sucrose gradients. Therefore, they can be separated from nonraft membranes by centrifugation methods. Samples prepared from cells transfected with a GFP-nsP1 expression plasmid were treated with Triton X-100 at 4°C, separated on a 10% to 80% sucrose density gradient, and then analyzed by Western blot. Detergent-resistant fractions corresponding to DRMs were identified utilizing an antibody against FLOT1, a well-known raft-associated protein (50). In our experimental conditions, FLOT1 fractionated into light density fractions 1 and 2 at the top of the gradient, thereby identifying DRMs (Fig. 7A). In contrast, the nonraft marker Na⁺/K⁺ ATPase remained associated with fractions 7 to 9 of heavier density, corresponding to nonraft membranes and cytosolic compartment (detergent sensitive [DS]). Using this protocol, more than 85% of the GFP-nsP1 protein was detected in FLOT1-positive fractions at the top of the gradient, supporting its capacity to associate with cholesterol-enriched DRMs (Fig. 7A and B). To confirm GFP-nsP1 affinity for cholesterol-enriched microdomains, this experiment was repeated starting from cells cultured in the presence of the cholesterol-depleting agent methyl-β-cyclodextrin (β MCD). Due to its ability to sequester cholesterol in its hydrophobic pocket, β MCD extracts cholesterol from the lipid bilayer and disrupts DRMs (51). Under these cholesterol-depletion conditions, FLOT1 was redistributed from DRMs to DS fractions and sedimented at the bottom of the density gradient (Fig. 7C and D). Analysis of the same fractions with anti-nsP1 antibodies revealed that GFP-nsP1 was barely detectable in top fractions, while it accumulated in fractions of heavier density together with FLOT1. Therefore, GFP-nsP1 fractionation with DRMs is sensitive to cholesterol extraction from cell membranes. Finally, this experiment was repeated starting from cells expressing GFP-nsP1_W258A or GFP-nsP1_3A mutant proteins that displayed different sensitivity to U18666A. As observed for GFP-nsP1, GFP-nsP1_W258A mainly segregated with DRM fractions, attesting that raft association was not disrupted by mutation of the W258 residue in the putative α-helix of nsP1 (Fig. 7E and F). Analyzing the behavior of the GFP-nsP1_3A mutant protein revealed, in contrast, that most of this protein was absent from the top fraction and cosegregated with the DS fractions. These experiments demonstrate that the presence of GFP-nsP1 in DRMs is mainly dictated by acylated cysteines, while mutation in the putative α-helical peptide has only a marginal impact on this phenotype.

FIG 7.

CHIKV nsP1 partitioning with lipid rafts in human cells requires C417-419. (A) HEK293T cells were transfected with plasmids encoding GFP-nsP1 and were processed for raft isolation. Nine fractions of the density gradients, numbered from top to bottom, were collected and analyzed using SDS-PAGE. Proteins were probed with antibodies against nsP1, FLOT1, and Na+/K+ ATPase. Fractions containing detergent-resistant membranes (DRMs) and detergent soluble membranes (DSs) are indicated. (B) The amount of GFP-nsP1 and FLOT1 associated with each fraction was determined by densitometry scanning of immunoblots and are expressed as a percentage of total protein expression level. The diagram is representative of three experiments. (C and D) A similar experiment was performed starting from cells incubated for 30 minutes in the presence of 10 mM βMCD. (E) Cells expressing the GFP-nsP1_3A or GFP-nsP1_W258A mutants were processed as in (A). (F) Amounts of GFP-nsP1, GFP-nsP1_3A, or GFP-nsP1_W258A in DRM and DS fractions were determined by densitometry scanning of the immunoblots shown in (A) and (E) and expressed as a percentage of total protein signal.

nsPs associate with DRM fractions in cells with CHIKV RNA replication.

Unlike SFV, CHIKV replication is abolished by nsP1_3A mutation regardless of the type and growth temperature of cells (33, 34). In contrast, the nsP1_W258A mutation allows virus to grow efficiently in insect cells or in mammalian cells cultivated at 28°C (33, 52). Combined with our observations, this strongly indicates that the presence of nsP1 in DRMs is an absolute requirement for CHIKV genome replication. Therefore, we questioned whether nsP1 affinity for DRMs was conserved in cells containing functional CHIKV replicase complexes. To this end, HEK293T cells either transfected with a trans-replication system that reproduces CHIKV RNA replication (Fig. 2B) or infected with the CHIKV-LR-5′-GFP virus were used. DRM isolation followed by immunoblot analysis revealed that a significant fraction of nsP1 expressed in transfected or infected cells (~85% and ~40%, respectively) was detected in cholesterol-rich fractions (Fig. 8D and E). NsP1 is the only alphavirus nonstructural protein with membrane affinity. In the replication complex, nsP1 co-localizes with nsP2, nsP3, and nsP4 (53–56). Thus, it is suspected to play a critical role in the replication complex anchoring to the PM. Therefore, we investigated whether the capacity of nsP1 to segregate with DRMs has an impact on the association of other nsPs with specialized membrane microdomains. Probing gradient fractions prepared from transfected and infected cells with antibodies against nsPs established that a significant part of each of them was detected in DS fractions, thereby agreeing with the previously reported capacity of nsPs to be present in different cytosolic compartments (57). However, approximately 10% to 30% of total nsP3 and nsP4 levels were also present in raft fractions. In some, but not in all experiments, nsP2 was also detectable in DRMs, most probably reflecting the weaker interaction of nsP2 with other components of replicase complexes. In order to further assess the relevance for the presence of nsP2, nsP3, and nsP4 in DRMs fractions, these experiments were reproduced starting from cells transfected with a CHIKV trans-replicase system in which the nsP1 protein contained the C_418-420A mutation. As expected, this mutation prevented nsP1 association with DRMs (Fig. 8C). Concomitantly, nsP2, nsP3, and nsP4 were also excluded from these fractions. Altogether, these results confirm the capacity of nsP1, expressed in the context of trans-replicase or by infectious CHIKV, to associate with DRMs, an association that dictates targeting of other nsPs, albeit at low levels, to the cholesterol-enriched membrane microdomain. Furthermore, the C_418-420A mutation that inactivates the CHIKV trans-replicase (33) completely prevented the association of CHIKV nsPs with DRMs.

FIG 8.

nsPs expressed by a CHIKV trans-replicase or an infectious CHIKV cosegregate with DRMs depending of nsP1 C417-419. DRMs were isolated from HEK293T cells transfected with CMV-P1234 and HSPolI-Fluc-Gluc plasmids (depicted in Fig. 2B) (A), cells infected with CHIKV-LR-5′GFP (MOI, 0.1) (B), and cells transfected with CMV-P1_3A234 and HSPolI-Fluc-Gluc plasmids (C). After 24 h, fractions were prepared and analyzed as described for Fig. 7A; immunoblots were probed with antibodies against each CHIKV nonstructural protein as indicated. (D and E) Amounts of nsPs in DRM and DS fractions were determined by densitometry scanning of the immunoblots shown in (A) and (B), respectively, and expressed as a percentage of total protein signal.

Conservation of nsP1-directed cholesterol affinity among divergent Old World alphaviruses.

Despite the presence of palmitoylated cysteine residues in the C-terminal region of nsP1 is a conserved feature of distantly related Old World alphaviruses (30, 34), functional studies have highlighted differences with regard to a cysteine requirement for genome replication. Cysteine-to-alanine mutations are lethal for CHIKV (33, 34) but are well tolerated by SINV (30). In the light of such differences and of the herein evidenced role of the CHIKV nsP1 cysteines in lipid raft association, we next questioned whether SINV nsP1 behaves similarly with respect to cholesterol. To achieve this, a plasmid encoding a GFP-fused SINV nsP1 protein was generated (Fig. 9A) and used for the transfection of HEK293T cells. Then, we tested SINV nsP1 partitioning to DRMs by membrane flotation assays. As previously observed for CHIKV nsP1, SINV nsP1 was also detected in the top fractions of the density gradient, supporting its ability to partition with cholesterol-rich membrane microdomains. However, contrasting with CHIKV, for which almost 85% of nsP1 associated with DRM fractions, a roughly equal proportion of the SINV nsP1 was associated with DRMs and DS (Fig. 9B). We further investigated SINV GFP-nsP1 affinity for cholesterol by testing its sensitivity to U18666A in confocal microscopy experiments. In control conditions, SINV GFP-nsP1 was detected at the PM, including in filopodia-like membrane protrusions that were abundantly observed as reported before for the untagged protein (42) and for CHIKV GFP-nsP1 (Fig. 9C). In the presence of U18666A, SINV GFP-nsP1 was still mainly detected at the PM, while cholesterol stained with filipin was concentrated in intracellular storage compartments as expected. Under these conditions, colocalization of SINV GFP-nsP1 with cholesterol-enriched endosomes was unfrequently detected. This is in contrast with the results obtained for CHIKV nsP1 (Fig. 4D) (Mander’s coefficient of 0.394 ± 0.032 and 0.217 ± 0.001 for CHIKV and SINV nsP1, respectively). Next, we investigated the role of palmitoylated cysteine by repeating these experiments, starting from cells expressing a cysteine-to-alanine SINV nsP1 palmitoylation-negative mutant protein (GFP-nsP1_C420A). As shown in Fig. 9B, the DRM association of GFP-nsP1_C420A was significantly reduced compared with that of SINV GFP-nsP1 (Fig. 8A). Analyzing SINV GFP-nsP1_C420A subcellular localization in U18666A-treated cells confirmed that this mutant did not colocalize with filipin-labeled cholesterol (Fig. 9D). Altogether, these results suggest that SINV nsP1 is targeted to DRMs, a property that depends on palmitoylated cysteine, as previously observed for CHIKV nsP1. However, compared with the CHIKV counterpart, SINV nsP1 is equally abundant in DS compartments and displays only modest sensitivity to U18666A, suggesting a reduced affinity for cholesterol.

FIG 9.

SINV nsP1 affinity for cholesterol. (A) Organization of the SINV nsP1 and mutant studied. Conserved amino acids involved in the central α-helix (aa 245 to 264) and palmitoylated cysteine (C420) are indicated. The mutant used in this study is depicted. (B) HEK293T cells transfected with either SINV GFP-nsP1 or GFP-nsP1_C420A plasmids were processed for DRM isolation as described for Fig. 7. Fractions (numbered from top to bottom) were probed in immunoblot with antibodies against GFP. (C) HeLa cells were transfected with a plasmid encoding a SINV GFP-nsP1 protein for 4 h and cultured for an additional 24 h in the presence of 12 μM U18666A. Cells were labeled with filipin III to visualize cholesterol. Controls cells (mock) were maintained with vehicle. (D) HeLa cells transfected to express the GFP-nsP1_C420A mutant were processed as in (C). Bars, 5 μm.

DISCUSSION

The present study identifies CHIKV nsP1 as a lipid-raft cosegregating protein with an affinity for cholesterol. We defined cysteine residues that can be palmitoylated as the molecular determinant important for this targeting. In the context of cells with ongoing CHIKV RNA replication, nsP1, together with a fraction of other nsPs, partitions with cholesterol-rich DRMs. Together with evidence that drugs reducing cholesterol availability at the PM impair CHIKV RNA replication, our results support that nsP1 targeting to cholesterol-rich PM microdomains may have a functional importance for viral genome replication.

Cholesterol is a main component of membranes. Together with sphingolipids, it segregates into discrete microdomains, referred to as lipid rafts or DRMs, present both on the inner and the outer leaflet of the PM (58, 59). These membrane domains with a size on the nanometer scale are highly dynamic. They accumulate a subset of membrane proteins, mainly GPI-anchored proteins, transmembrane proteins, and acylated cell components (60, 61). Based on these properties, rafts were seen as platforms that compartmentalize cellular processes with an important function in receptor-ligand interaction, signal transduction, and endocytosis (60). Herein, we establish that CHIKV nsP1 associates with the PM. In this compartment, a pool of nsP1 is targeted to cholesterol-containing DRMs (Fig. 10a), an association that was reversed by βMCD cholesterol-removal agent. Affinity for cholesterol was further supported by investigating nsP1 behavior with regard to U18666A- or imipramine-mediated cholesterol sequestration in LE/Ls. We found that intracellular cholesterol storage resulted in nsP1 accumulation in endosomes without consequence on overall nsP1 membrane binding ability (Fig. 10b). Altogether, these results indicate that the availability of cholesterol at the PM is required for appropriate targeting of CHIKV nsP1 to this compartment.

FIG 10.

Model for nsP1/cholesterol interplay in transfected and CHIKV-infected cells. (a) Palmitoylated cysteines in wild-type nsP1 are critical for targeting to cholesterol-rich membranes microdomains. In infected cells, the preferential association of nsP1 with DRMs increases its local concentration which may facilitate its oligomerization and dictates the localization of a fraction of other nsPs to lipid rafts. In this scenario CHIKV efficiently replicates its genome. (b) Imipramine and U18666A, by inhibiting cholesterol trafficking to the PM, generate nsP1 accumulation in Lamp2-positive late endosomes/lysosomes. Association of nsP2, nsP3, and nsP4 to cholesterol-rich microdomains is no longer observed. Under these conditions, CHIKV genome replication is abolished by a mechanism that remains to be fully elucidated. (c) nsP1_3A that weakly interacts with the plasma membrane is not targeted to DRMs. In this situation, CHIKV genome replication is impaired. For simplicity, only mature nsPs are shown in this model. In a natural infection, the membrane association begins by modification and membrane targeting of polyprotein precursors.

In the last decade, the biochemical or biophysical underpinnings that govern nsP1 association with membranes have been the focus of research. For SFV and SINV, the central α-helical motif in nsP1 spanning amino acid residues 245 to 264 was proposed as the main determinant for membrane anchoring to lipid bilayers, with W259 residue being critical for hydrophobic interactions with the phospholipid acyl chains (25). Conserved acylated cysteine(s) in nsP1 were proposed to tighten this membrane interaction (26, 27, 30, 34). Using cell fractionation assays, we show that both W₂₅₈A substitution in the putative α-helix of CHIKV nsP1 and C_417-419A mutations indeed decrease nsP1 affinity for cell membranes. Interestingly, combining W₂₅₈A and C_417-419A mutations further reduced nsP1 membrane association, thereby suggesting that in the CHIKV nsP1 the two domains may synergize for membrane interaction, a situation that was not described for other alphaviruses. Analyzing the contribution of these membrane interaction determinants in nsP1 targeting to lipid rafts revealed that the W₂₅₈A mutation had only a slight effect on nsP1 cofractionation with cholesterol-rich domains. This mutation also slightly reduced PM association of nsP1 and facilitated its association with internal membranes. This may indicate that the W258 residue is important for membrane/plasma membrane targeting of nsP1 but not for its palmitoylation, DRM association, and enzymatic activity. How these properties correlate with the proposed role of the W258 residue as one of membrane anchors of nsP1 is currently unclear. For proper understanding of the somewhat controversial data regarding the importance of the W258 residue, the structure of the membrane-bound α-helical peptide of nsP1 (25) should be compared with the structure of membrane-bound enzymatically active nsP1 which, to this date, is not available. In contrast to the W₂₅₈A mutation, the effects of the C_417-419A mutation on CHIKV nsP1 were unambiguous. This mutation, previously reported to prevent CHIKV nsP1 palmitoylation and replication (33, 34), dramatically reduced DRM cofractionation (Fig. 10c). Moreover, in contrast with wild-type nsP1, nsP1_C417-419A sequestration with Lamp2 in endosomes could not be observed under U18666A or imipramine treatment. According to these experiments, acylation appears as critical to direct nsP1 to cholesterol-enriched membrane microdomains. This result parallels previous evidence regarding the role of palmitoylation in cellular (p59^fyn and p60^src) (62) or viral (influenza hemagglutinin) (63) protein association with rafts. This observation raised the question of the functional outcome of nsP1 association with cholesterol-rich DRMs.

Because other alphavirus nonstructural proteins cannot directly associate with membranes, nsP1 plays a decisive role in proper targeting and membrane binding of the replication complex (24). Starting from cells infected with CHIKV, we found that a fraction of each of the four nsPs was associated with DRMs. These results were confirmed in cells transfected with plasmids encoding a CHIKV trans-replication system, in which other nonstructural proteins cosedimented with DRMs depending on the integrity of nsP1 C417-419 residues. Currently, the functional importance of nsPs targeting to rafts is unknown. Nevertheless, we observed that, in addition to nsP1 sequestration into late endosomes, U18666A and imipramine also generated a significant drop in CHIKV genome replication. This raises the question of the functional consequences of nsP1 mistargeting and of its impact on replication complex assembly.

In the last two decades, alphaviruses encoding nsP1 mutants with a reduced membrane-binding ability have been the focus of different studies. They established that nsP1 palmitoylation had only a mild impact on SINV or SFV infectivity, highlighting the essential role of the W259 residue in nsP1’s central α-helix for both membrane anchoring and genome replication (30, 31). For CHIKV, the W₂₅₈A mutant is viable, albeit with a temperature-sensitive phenotype (33). In contrast, the C_417-419A mutation results in complete inactivation of the CHIKV replicase, leading to nonfunctional enzymes unable to synthesize any viral RNAs both in mammalian and insect cells (33, 34, 52). Here, the direct comparison of SINV and CHIKV revealed that SINV nsP1 partitions with DRMs but to a lesser extent than observed for CHIKV nsP1. This phenotype was equally dependent upon conserved cysteine in nsP1. Moreover, SINV nsP1 targeting to the PM was less sensitive to cholesterol manipulation by U18666A. These discrepancies, therefore, parallel the differences in the nsP1 palmitoylation requirement previously reported for CHIKV and SINV replication (33). However, counterintuitively, SINV genome replication was also sensitive to U18666A and cholesterol sequestration (this study and reference 35), albeit less than observed for CHIKV (data not shown). Conversely to CHIKV, this might not reflect dependency on PM rafts but instead cholesterol-induced alteration of endosomes where SINV preferentially replicates, as proposed by others (35, 64).

In the last decade, cholesterol-rich membrane microdomains, beyond their role in virus entry and exit, have also been identified as platforms for the assembly and anchoring of replication complexes produced by a broad range of RNA viruses, including HCV, picornaviruses, or flaviviruses (9, 10, 12, 18, 65). A dramatic reduction of viral RNA replication was observed upon membrane cholesterol extraction or under conditions reducing cholesterol availability (12, 14, 66, 67), establishing functional importance for this association. Besides changing membrane composition and fluidity, which may affect different interactions between virus-encoded replicase subunits, cholesterol partitioning in membranes also attracts cellular proteins, with functions in cell signaling and intracellular trafficking. We can assume that targeting CHIKV nsP1 and possibly other nsPs at these sites may favor interaction with raft-associated cellular factors indispensable for viral replication. The exact necessity for nsP1 partitioning to DRMs, therefore, will deserve further investigations.

MATERIALS AND METHODS

Antibodies and reagents.

The following antibodies/reagents and respective dilutions were used in this study: rabbit polyclonal antisera against CHIKV nsP1, nsP2, nsP3, and nsP4 (all in-house; 1:1,000); monoclonal antibodies against GAPDH (1:1,000) (Santa Cruz Biotechnologies Inc.); Na⁺/K⁺-ATPase (1:50,000) (Ab76020; Abcam), GFP (1:1,000) (Chromotek); Lamp2 (1:1,000); CD81 (1:500) (clone JS-81; BD Biosciences); and flotillin-1 (1:1,000) (BD Biosciences). Secondary antibodies conjugated to horseradish peroxidase or Alexa Fluor were purchased from Jackson Immunoresearch and Thermo Fisher Scientific, respectively. Filipin III, U18666A, lovastatin, and methyl-β-cyclodextrin were purchased from Sigma-Aldrich. Imipramine was obtained from Abcam and WGA-647 from Thermo Fisher Scientific.

Cells.

HEK293T cells (ATCC number ACS-4500), HeLa cells (ATCC number CRM-CCL2), BHK-21 cells (ATCC number CCL-10), and U2OS cells (ATCC number HTB-96) used for propagation; and Vero cells (ATCC number CCL-81) used for titration of the CHIKV were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Thermo Fischer Scientific) supplemented with penicillin and 10% fetal calf serum (FCS; Lonza) and grown at 37°C in a 5% CO₂ atmosphere. The viability of cells incubated with drugs for 24 h was measured using the CellTiter 96 AQueous one solution cell proliferation assay (Promega) according to the manufacturer’s protocol.

Viruses.

The pCHIKV-LR-5′GFP, full-length molecular clone of CHIKV (LR2006OPY1 strain) with a GFP reporter (68) was linearized and transcribed in vitro using the mMessage mMachine kit (Ambion-Life Technologies). A total of 1 μg of RNA was then transfected with Lipofectamine 2000 (Thermo Fisher Scientific) into 10⁵ HEK293T cells, and the cells were incubated at 37°C for 24 h. Culture medium was collected, and virus stock was amplified on BHK-21 cells. After 48 h at 37°C, the supernatant was collected, filtered through a 0.45-μm membrane, aliquoted, and stored at –80°C. The titers of the viral stocks were determined using plaque assay, as previously reported (39).

Infection with CHIKV-reporter viruses.

The cells (70% to 80% confluence) were rinsed once with phosphate-buffered saline (PBS) before infection with CHIKV-LR-5′GFP diluted to achieve the desired MOI. For preinfection experiments, the cells were preincubated with the virus for 1 h or 2 h before drug addition. For postinfection experiments, the cells were incubated with drugs for 30 min before infection. After 24 h in culture, the cells were lysed with radioimmunoprecipitation assay (RIPA) buffer. GFP reporter fluorescence was measured directly from the cell lysate using an Infinite F200PRO fluorometer (Tecan). Values were normalized to the protein content in the sample determined using the bicinchoninic acid (BCA) assay (Pierce).

Plasmids and transfection.

The sequence encoding CHIKV nsP1 was amplified by PCR using pCHIKV-LR-5′GFP as a template; the obtained fragment was cloned into the pEGFP-C1 plasmid as previously described (69). Sequences encoding nsP1_W258A, nsP1_C417-419A, and nsP1_DM were generated using a Quikchange site-directed mutagenesis kit by Agilent. SINV nsP1 and nsP1_420A were obtained using PCR and pTOTO1101 (70) or its derivative, pSINV-C₄₂₀A1 (30), as templates. These inserts were cloned in frame with eGFP in the peGFP-C1 plasmid. Cells were transfected with obtained plasmids using the JetPei reagent (Polyplus Transfection) according to manufacturer recommendations.

Trans-replication assay.

For trans-replication assays, CMV-P1234 (71) or CMV-SINV-P1234 (72) plasmids encoding the nonstructural polyprotein from CHIKV or SINV, respectively, were cotransfected together with the HSPolI-Fluc-Gluc (33) or HSPolI-SINV-tFluc-Gluc (73) plasmids encoding replication-competent template RNA of CHIKV or SINV containing firefly and Gaussia luciferase reporter sequences placed under the control of genomic and subgenomic promoters, respectively. Equal amounts of plasmids were transfected into HEK293T cells using JetPei transfection reagent. After 24 h in culture, cells were washed in PBS and lysed using passive lysis buffer (Promega). The expression of firefly and Gaussia luciferase was determined using the Dual-Glo luciferase assay system (Promega) and a Spark luminometer (Tecan). Reporter activities were normalized to the protein content in the sample determined using the BCA assay (Pierce).

Immunoblotting.

Samples were separated by SDS-PAGE and then transferred to a polyvinylidene difluoride (PVDF) membrane (Hybond, Amersham, England). Membranes were blocked using 5% nonfat dry milk in PBS and were probed with appropriate primary antibodies. After wash steps with PBS and 0.1% Tween 20, the membranes were probed with horseradish peroxidase (HRP)-conjugated secondary antibodies. The revelation was performed by incubating the membranes with either Luminata Forte (Merck) or Clarity Max (Bio-Rad) solutions, and then image acquisition was done using a Chemidoc instrument (Bio-Rad).

Cell fractionation and membrane flotation assays.

Cells were incubated in hypotonic buffer (10 mM Tris/HCl [pH 7.4] and 10 mM NaCl supplemented with protease inhibitors) for 10 min on ice and then lysed with a Dounce homogenizer (30 to 40 strokes). The lysates were clarified by low-speed centrifugation at 1,000 × g for 10 min. Obtained postnuclear supernatants (PNSs) were then adjusted to a final concentration of 500 mM NaCl and were incubated for 30 min on ice. After ultracentrifugation at 25,000 × g for 20 min, the cytosolic (supernatant, S25) and membrane fraction (pellet, P25) were collected. P25 samples were solubilized in lysis buffer composed of 1% Brij 96 in 20 mM Tris/HCl [pH 7.5] before analysis. For membrane flotation experiments, cells were resuspended in 250 mM sucrose in PBS supplemented with protease inhibitors and then lysed with a Dounce homogenizer (30 to 40 strokes). Cell lysates were spun at 1,000 × g for 10 min to pellet the nuclei. The supernatant referred to as crude lysate (CL) was then adjusted to 30% iodixanol concentration by mixing Optiprep (Axis-Shield). CL (4 ml) was loaded at the bottom of a centrifuge tube and then overlaid with 4 ml 20% iodixanol and then 4 ml 10% iodixanol. The gradient was then spun 200,000 × g at 4°C for 16 h in a Beckman SW41 rotor. Finally, 24 fractions were collected from top to bottom.

Detergent-resistant membrane isolation.

Cells were lysed on ice in TNE buffer (10 mM Tris-HCl [pH 7.5], 100 mM NaCl, and 10 mM EDTA) containing 0.5% Triton X-100 for 30 min. Lysates were then further treated with the Dounce homogenizer and then clarified by low-speed centrifugation at 1,000 × g for 10 min to obtain the PNS. PNS (0.5 ml) was adjusted to 60% sucrose by adding 1.5 ml of 80% sucrose TNE (wt/vol). The lysates were layered over 500 μl of 80% sucrose TNE and then covered with 2 ml of 50% sucrose TNE, 6 ml of 38% sucrose TNE, and 1.5 ml of 10% sucrose TNE. The sucrose gradients were centrifuged at 100,000 × g at 4°C for 18 h in an SW41 rotor (Beckman Coulter). Nine fractions were then collected and analyzed by immunoblotting.

Immunofluorescence microscopy and image analysis.

Cells grown on glass coverslips were washed with PBS and then fixed with 4% paraformaldehyde/PBS (Sigma-Aldrich) for 10 min. For intracellular labeling, the cells were permeabilized with 0.1% Triton X-100 in PBS and blocked for 30 min with PBS containing 0.2% bovine serum albumin. Incubation with primary antibody was performed for 1 h. After washes with PBS, secondary reagents were added for 30 min. 4′,6-Diamidino-2-phenylindole (DAPI) (Sigma-Aldrich) was used to stain the nuclei. Filipin or WGA-647 staining was performed by incubation at room temperature for either 1 h or 10 min, respectively. After final washes, coverslips were mounted with Prolong Gold antifade mounting medium (Thermo Fisher Scientific). Images were acquired using a Leica SP5-SMD scanning confocal microscope equipped with a 63× 1.4 numerical aperture Leica Apochromat oil lens at the Montpellier Resources Imaging platform. Image analysis was performed utilizing Fiji ImageJ and the JACoP plugin. For Mander’s analysis, the threshold and region of interest were consistent between the different conditions. 3D reconstruction was performed by Imaris software.

Statistical analysis.

All of the analyses (unpaired Student’s t test) were performed using GraphPad Prism version 6 (GraphPad Software Inc.). A P value of <0.05 was considered statistically significant. The following designations are used on figures: * P < 0.05, ** P < 0.001, ***P < 0.0001, and **** P < 0.0001.