Integrative proteomics and lipidomics reveals dual roles for lipid droplets in the host cell antiviral response

Ebony A Monson; Zahra Telikani; Jay L Laws; Abbey J Milligan; Irumi Amarasinghe; Sarah E Bamford; Ashley M Rozario; Monique L Smith; Vivian Tran; Quynh N Dinh; Nicholas Williamson; Adam Mechler; Boyin Liu; Chad J Johnson; Paul J Pigram; Michael Leeming; Shuai Nie; Jessica K Holien; Markus J Hofer; Donna R Whelan; Karla J Helbig

doi:10.1038/s41467-026-74016-w

. 2026 Jun 5;17:7189. doi: 10.1038/s41467-026-74016-w

Integrative proteomics and lipidomics reveals dual roles for lipid droplets in the host cell antiviral response

Ebony A Monson ^1,^2,^✉,^#, Zahra Telikani ^1,^2,^#, Jay L Laws ^1,^2,^#, Abbey J Milligan ^2,³, Irumi Amarasinghe ^1,², Sarah E Bamford ⁴, Ashley M Rozario ^2,³, Monique L Smith ^1,², Vivian Tran ^1,^2,⁵, Quynh N Dinh ^1,^2,^5,⁶, Nicholas Williamson ⁷, Adam Mechler ^2,⁸, Boyin Liu ⁹, Chad J Johnson ⁹, Paul J Pigram ⁴, Michael Leeming ⁷, Shuai Nie ⁷, Jessica K Holien ^10,^11,¹², Markus J Hofer ¹³, Donna R Whelan ^2,³, Karla J Helbig ^1,^2,^✉

PMCID: PMC13396240 PMID: 42248863

Abstract

Lipid droplets (LDs), once viewed as inert lipid storage sites, are now recognised as dynamic organelles central to cellular signalling and immunity. This study presents a dual-omics approach, integrating proteomics and lipidomics, to investigate LDs in the host antiviral response. In vivo and in vitro RNA viral infection models demonstrate that LDs rapidly remodel both their proteome and lipidome. The antiviral proteins RIG-I, MDA5, STAT1, STAT2, and viperin are specifically recruited to virus-induced LDs. Simultaneously, the LD lipidome shifts toward enrichment in long-chain polyunsaturated fatty acids and bioactive phospholipids, likely enhancing membrane dynamics and protein recruitment. Correlation analysis of matched proteomic and lipidomic datasets revealed extensive lipid–protein co-regulation on LDs and identified distinct correlation networks in which immune and antiviral proteins preferentially aligned with neutral lipid species. Enzymes involved in lipid metabolism and post-translational modifications are also upregulated, suggesting a mechanistic link between lipid remodeling and protein localisation. Functional assays utilizing artificial LDs revealed that arachidonic-acid and eicosapentaenoic-acid suppress viral replication and enhance type I and III interferon responses. These findings position LDs as key immunometabolic platforms in early antiviral defence.

Subject terms: Organelles, Virology, Innate immunity

Dual proteomics–lipidomics demonstrates that virus-induced lipid droplets rapidly recruit antiviral proteins and remodel lipids. These lipid–protein networks form signalling hubs, while specific long chain fatty acids enhance interferon responses and restrict viral replication.

Introduction

Lipid droplets (LDs) are present in the different cell types of all living organisms. Once thought to function solely as passive energy reservoirs, LDs are now recognised as highly dynamic cytoplasmic organelles that play critical roles in a wide range of signalling pathways^1–3. Unlike other organelles that are surrounded by a phospholipid bilayer, LDs are enclosed by a single phospholipid monolayer that encapsulates a neutral lipid core. Their surfaces are also decorated with proteins that are believed to drive their cellular functions^4,5, with recent advances in proteomic technologies revealing the high diversity of the LD proteome.

In mammals, LDs are upregulated by the host in multiple disease states, including cancer, cardiovascular disease, neurodegenerative disorders, and following infection of cells with bacteria, fungi, parasites and viruses (reviewed in refs. ^6–9). Historically, LDs were thought to serve primarily as nutrient-rich reservoirs exploited by pathogens for energy and lipid synthesis, however, emerging evidence reveals a more complex role for LDs in immunity with host-derived LDs actively contributing to cell survival following pathogen infection. They have been shown to recruit and concentrate antimicrobial proteins, and can act as platforms for innate immune signalling, including the induction of pro-inflammatory cytokines and interferon responses^10–12, highlighting LDs as not only metabolic hubs but also key players in the host defence machinery.

A successful innate immune response to virus infection requires LD upregulation¹⁰. In particular, during zika virus (ZIKV) and herpes simplex virus −1 (HSV-1) infection, LDs upregulated via non-homeostatic mechanism involving engagement of the EGFR, has been linked to the downstream production of type-I and III interferons (IFNs; IFN-β and IFN- λ)¹⁰. However, the mechanism underpinning this increase in IFN production remains elusive and may involve both changes in lipid and protein profiles. Successful antiviral responses rely on complex protein-protein interactions that require platforms for their assembly^1,13–15. LDs house viperin, a key antiviral protein that also regulates multiple antiviral signalling cascades^1,2,14, but beyond this, the role of LDs in facilitating other protein complex formations remains unknown. The LD proteome is both diverse and dynamic in nature, changing constantly to cellular cues, and we now know that changes in lipid species can drive alternate protein binding across cellular membranes^16–19. Specific lipids have also been shown to influence antiviral cytokine expression, in particular cellular cholesterol reduction can trigger antiviral signalling and there is a complex relationship between bioactive lipids and antiviral control²⁰.

Given the recent recognition of LDs as essential organelles in host driven antiviral immunity and the emerging appreciation of the intricate interplay between proteins and lipids, here we describe the development of technologies that, for the first time, enable simultaneous analysis of the LD proteome and lipidome in mammalian cells. Using these approaches, we demonstrate that LDs undergo rapid and coordinated changes in both protein and lipid composition in response to viral infection. These changes include the recruitment of key antiviral proteins and modifications in long-chain fatty acid content, which together support an effective antiviral response and contribute to broader host cell signalling processes.

Results

Brain derived lipid droplets recruit cellular defence proteins following LCMV infection in vivo

LDs are upregulated early following viral infection in mammalian cells and are essential for the production of a heightened antiviral IFN state¹⁰, however, the mechanisms that underpin their ability to drive IFN production and limit early viral replication remain unknown. To investigate the role upregulated LDs may play in cellular restriction of viral infection we selected a well-established neural model of lymphocytic choriomeningitis virus (LCMV) infection²¹. Mice were infected with LCMV intracranially, and their brains were harvested at 2- and 4-days post infection (dpi), modelling early-stage infection, as evidenced by increases in viral load and cytokine expression (Figs. 1A, 1B, S1A, S1B). LD abundance was significantly increased in the brains of virally infected mice at both 2 and 4 dpi (Fig. 1B), as we have also previously observed in vivo during influenza A and dengue viral infection in the lungs and brain respectively¹⁰. To further characterize the potential role of LDs in viral infection we developed techniques to cleanly extract LDs from whole murine brains (Figs. 1C, S1C-E) and performed comparative mass spectrometry profiling of proteins differentially associated with LCMV-LDs or SHAM-LDs isolated from the brain tissues of mice. In the virally infected brains, 4324 proteins were identified on LDs (Fig. 1D; Supplementary data 1; S1F). Of these, only 30 were found to be differentially upregulated compared to SHAM-LDs, with 24 proteins exclusively upregulated on LCMV-LDs, and only 2 downregulated (0.5%; Fig. 1D-F and S1F). Functional annotation enrichment analysis revealed that most of the upregulated LD proteins at 4 dpi belonged to ‘cellular defence response to viruses’ and ‘response to interferons’ (Fig. 1G), the predominant mammalian antiviral cytokines. Additionally, the enrichment of proteins within our LD subsets following viral infection did not follow the identical enrichment patterns within the whole cell lysate, further illustrating that LDs enrich specific proteins following viral infection (Fig. 1H).

Fig. 1 — A Experimental overview. Mice were intracranially infected with LCMV (500 PFU) or SHAM treated, and brains collected at 2 or 4 days post-infection (dpi) (n = 6 mice per condition). Created in BioRender. Helbig, K. (2026) https://BioRender.com/k6zo39lB Mice brain sections were immunostained for lipid droplets (LDs) using BODIPY 493/503 (green), anti-PLIN2 (red), and nuclei with DAPI (blue). LD number and total cell fluorescence (normalised to cell number) were quantified using ImageJ. Scale bar, 100 μm. Data are shown as mean ± SEM (n = 6 mice per condition; p < 0.0001). C LDs isolated from mouse brains were visualised by BODIPY staining and fluorescence microscopy or by transmission electron microscopy following plunge freezing. Scale bars, 50 μm (IF) and 200 nm (TEM). Image is representative of n = 3 biological replicates. D LC–MS/MS analysis comparing protein identity and abundance in whole brain lysates and isolated LD fractions at 2 and 4 dpi. Significantly up-and downregulated proteins indicated by red and blue circles, respectively. E Overlap analysis of significantly upregulated proteins following LCMV infection (vs SHAM), highlighting distinct proteomic signatures between whole brain lysates (red circle) and LD fractions at 2 (grey circle) and 4 dpi (blue circle). F Hierarchical clustering of the 18 significantly regulated LD-resident proteins at 4 dpi across all experimental groups. G Tree-plot summarising similarity clusters of the top 15 significantly enriched Gene Ontology terms across Biological Process, Molecular Function, and Cellular Component categories (adjusted p < 0.05). H Hierarchical clustering of proteins annotated as involved in “immune response to virus” (UniProt), comparing whole brain and LD fractions from SHAM and LCMV-infected mice at 4 dpi. I Quantification of LD localisation across brain cell types at 4 dpi in SHAM and LCMV-infected mice. Data represent mean ± SEM; statistical significance was determined by two-way ANOVA with Bonferroni post-hoc correction (n = 64 tissue sections across 6 mice). Source data are provided as a Source Data file.

The brain contains a very heterogenous mix of cell types, and LCMV has been demonstrated previously to infect both neurons and astrocytes²². To assess the cell type specificity of the upregulation of LDs in our LCMV model of murine brain infection we next performed immunofluorescence analysis on fresh frozen brain sections (Fig. S2). Both microglia and neurons were shown to express LD, although at lower levels, with no significant increase seen following viral infection. However, astrocytes demonstrated more than a two-fold increase in LDs following LCMV infection, showing them to be the largest contributor in the brain to the upregulated LD response to virus (Fig. 1I, S2). In vitro experiments also confirmed the ability of astrocytes to significantly upregulate LDs following activation of early innate signalling pathways with RNA viral mimic stimulation, whereas neuronal and microglial cell lines did not (Fig. S2).

Astrocyte lipid droplets recruit immune signalling proteins following viral infection

We next wanted to gain an astrocyte cell-type specific LD proteomic profile following response to viral infection. To do this in the absence of viral antagonism, we first described the proteomic shift of LDs following RNA viral mimic stimulation of early innate antiviral signalling pathways using dsRNA in primary immortalised astrocyte cells in vitro. As described previously by our team¹⁰, an upregulation of LDs was observed at both 8 and 24 hrs following dsRNA stimulation (Fig. S3A). LDs were isolated from the primary immortalised astrocyte cells and were morphologically characterised before profiling their proteome (Fig. 2A, B, S3B-D, Supplementary data 2). To assess if any bona fide LD-resident proteins were localising to the astrocyte LDs, we compared all 21-mammalian published LD-proteomes against ours (Supplementary data 3)^{11,12,23–39}. 18 LD-resident proteins were identified in our proteome and were characterised as likely “bona fide” LD proteins as they appeared in more than 75% of published proteomes (Fig. S4A). Additionally, 79.5% of our proteome has been described previously to be present on LDs (Fig. S4A).

Following stimulation with dsRNA, there was an upregulation of 10 and 56 proteins on LDs at 8 and 24 hrs respectively (Fig. 2C and Supplementary data 2). Functional categorisation of the entire LD proteome (Fig. 2Di) in comparison to the differentially regulated LD proteins following stimulation (24 hrs, Fig. 2Dii) revealed an altered distribution of LD proteins that saw an increase of proteins belonging to the ‘immune response’ and ‘lipid metabolism’ functional categories (Supplementary data 2).

Astrocytes have a robust immune response following viral infection^40,41. Of the ‘immune response’ category proteins shown to be upregulated in the whole cell following viral mimic stimulation, only a sub-set of these proteins were recruited to the LD (Fig. 2E), including IFIT3, DDX58 (RIG-I), ISG15, IFIT2, IFIT1, MX1, STAT1, RSAD2 (viperin). Using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) network analysis, these immune response proteins were seen to have a high level of predicted protein-protein interactions with each other, which also included interactions with CMPK2, a known interaction partner of RSAD2 required for its antiviral function (Fig. 2F)⁴². Other clusters observed with our upregulated LD proteins included interaction networks of Rab proteins and lipid metabolism proteins (LPCAT1, ACSL3, ACSL4, LSS, SQLE, HSD17B7, SDHL, LPCAT2, FAF2).

To further characterise the LD proteome response to viral infection, we next infected astrocytes with the model RNA virus, ZIKV (Zika virus), which we have previously demonstrated to productively infect primary immortal astrocytes (confirmed in this experiment by the high levels of viral proteins detectable in the whole cell lysate of our samples) (¹⁰ Figure S4B). Members of the ‘immune response’ category of proteins were also upregulated on the LD following ZIKV infection of the astrocyte cells at the same time point (24 h: Fig. 2G, Fig. S4C and Supplementary data 4). Immunoblotting of LD fractions confirmed the presence of selected novel LD localised proteins, both those that are enriched following dsRNA stimulation (RIG-I, STAT1, MX) or those present but not upregulated (STAT2, ZC3HAV1, TRIM25) (Fig. 2H, S4D). These results indicate that key antiviral proteins localise to the LD following activation of viral RNA sensors by either a viral mimic or active RNA viral infection.

STAT proteins dynamically localise to lipid droplets following virus infection in vitro

To date the only antiviral protein shown to localise to LDs in human cells is viperin (RSAD2), which orchestrates a heightened antiviral environment^2,19,20. Our proteomic analysis of LDs in primary immortalised astrocyte cells revealed that multiple members of the early antiviral innate immune signalling pathways thought previously to be cytoplasmic were present in the LD proteome, with expression of selected proteins being significantly upregulated following both pathway activation by dsRNA (in the absence of viral-mediated antagonism) or ZIKV infection (Fig. 2A,B, F; Supplementary data 2,S4). Many of these proteins are known to be involved in early antiviral signalling pathways or have known antiviral functions (Fig. 3A), including STAT1 and STAT2, which underpin the formation of the ISGF3 complex that initiates transcription of multiple antiviral proteins⁴³. To further confirm the ability of the STATs to localise to LDs we performed microscopy analysis. Confocal imaging confirmed the known LD localised protein, viperin^21,22, to be highly abundant on the outside of LDs, with novel localised proteins; STAT1 and STAT2 being less abundant on LDs, and colocalising to only a subset of cytoplasmic LDs following activation of innate immune pathways by dsRNA (Fig. 3B). To more confidently validate the localisation of these proteins to the LD and enumerate the percentage of LDs these antiviral proteins were localising to, we isolated LDs from cells overexpressing tagged mCherry-viperin, -STAT1, and -STAT2 (Fig. S5A). Viperin was present on 54.8% of LDs on average, in comparison to STAT1 localising to 44.9% and STAT2, 24.9%, which was also confirmed via immunoblotting (Fig. 3C-E).

Fig. 3 — A Schematic of interferon signalling following viral stimulation highlighting proteins absent from the lipid droplet (LD) proteome (white), present on LDs (grey), or upregulated on LDs (red), based on RNA stimulation at 24 h. Created in BioRender. Helbig, K. (2026) https://BioRender.com/z9el4lmB Primary immortalised astrocytes were transfected with mCherry-tagged viperin, STAT1 or STAT2 (red) and stained with BODIPY 493/503 to visualise LDs (green) and DAPI to label nuclei (blue). Representative images and zoomed panels show association of mCherry-tagged proteins with LDs. Scale bar, 50 µm. Images represent n = 3 biological replicates. C Confocal microscopy of astrocytes transfected with mCherry, viperin-mCherry, STAT1-mCherry or STAT2-mCherry followed by LD isolation and purification. Purified LDs were imaged to assess protein localisation, with zoomed panels highlighting co-localisation. Scale bars: 400 µm (transfection panels), 50 µm (merged images) and 20 µm (zoom). D Quantification of mCherry co-localisation with isolated LDs. Each data point represents the average percentage of co-localisation per field of view containing ~500–3000 LDs (n = 25 fields) across two independent experiments. Data are shown as mean ± SEM; p < 0.0001, determined by unpaired two-tailed Student’s t-test with Holm–Šidák correction. E Immunoblot validation of mCherry-tagged proteins in whole-cell lysates and purified LD fractions from transfected astrocytes, confirming enrichment of LD-associated STAT proteins (n = 3). F Super-resolution imaging of LD–STAT1 interactions following dsRNA stimulation using single-molecule localisation microscopy (SMLM). Epifluorescence images of LDs (BODIPY 493/503, blue) were overlaid with SMLM images of STAT1 and phosphorylated STAT1 (Ser727 or Tyr701; Alexa Fluor 647, yellow) across a 72 h stimulation time course (n = 3 biological replicates). Scale bars, 5 µm; zoom inserts, 2 × 2 µm. G Quantification of LD–STAT1 co-localisation events per cell. Super-resolution data rendered using ThunderSTORM and merged with LD images prior to manual enumeration. Data represent mean ± SEM (n = 30 cells from 3 independent experiments). Statistical significance assessed by two-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.

Further confirmation of native STAT1 localisation to the LD, and its phosphorylated forms (STAT1-Tyrosine701; ph-STAT1(T) and STAT1-Serine 727; ph-STAT1(S)), was performed using super resolution microscopy (single molecule localisation microscopy, SMLM), where we tracked STAT1 co-localisation with LDs in the cell over a timeframe of 72 hrs post dsRNA stimulation (Fig. 3F, G and S5B). There was a significant increase in the number of co-localisation events of LDs with both STAT1 and ph-STAT1(T) at multiple time points (Fig. 3F, G), with no significant increase in association observed between ph-STAT1(S) and LDs. However, the interaction of ph-STAT1 (T) with the LD was the most significantly enhanced interaction from as early as 8 hrs post activation of early innate antiviral signalling pathways, when normalised to the increased cellular density of both LDs and STAT1 proteins (Fig. 3G; S5B, S5C, S5D). This was further supported by a doubling of non-random co-localisation events between the LD and ph-STAT1(T) by 48 hrs post activation in which the degree of colocalisation was equivalent to random levels of overlap as indicated by values over 1 (Figure S5B).

We have previously shown that virally driven LDs are upregulated via a non-homeostatic mechanism dependant on EGFR activation⁴⁴. To assess whether the localisation of either total STAT1 or ph-STAT(T) was altered during inhibition of virally driven LDs we performed the above experiments again in the presence of an EGFR inhibitor (Fig. 4). When normalising back to cellular LD number and STAT protein expression, we saw that loss of EGFR signalling significantly decreased the ability of both STAT1 and phSTAT-1(T) to localise to LDs. Indicating that these proteins preferentially bind to virally driven LDs.

Fig. 4 — Primary immortalised astrocyte cells were treated with 2 μM AG-1478 (EGFR inhibitor) 16 h prior to stimulation with dsRNA for 24 h. LD-STAT1 co-localisation events were visualised with single-molecule localisation microscopy (SMLM). The epifluorescence images of LDs were merged with the SMLM images of STAT1 and ph-STAT1 (T; tyrosine 701). Cells were immunolabelled and imaged for LDs (Bodipy (493/503), epi, blue) and STAT1 (AlexaFluor647, SMLM, yellow) (n = 3 independent biological replicates). Scale bars represent 5 µm, with zoom inserts representing 2 µm x 2 µm in size. Raw super-resolution data were analysed and rendered using the ThunderSTORM plugin in ImageJ before merging with epifluorescence LD images and enumeration of co-localisation events. Data represent mean ± SEM (n = 30 cells from 3 independent experiments). Statistical significance was assessed by two-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.

Lipid droplets interact with mitochondria to form signalosome complexes

Our proteomic analysis of virally driven LDs also demonstrated that both major viral RNA sensors form part of the LD proteome: MDA5 at baseline, as well as upregulated RIG-I (DDX58) following both ZIKV infection and dsRNA stimulation (Fig. 3A, Fig. 2, Supplementary data 2 and 4). Both of these RNA sensors require subsequent interaction with Mitochondrial Antiviral Signaling (MAVS) to drive antiviral cytokine expression, which is predominantly expressed on the mitochondria and is absent from the LD proteome (⁴⁵ Fig. 5A). We hypothesied that LDs may assist in signalosome formation of RIG-1 to facilitate activation of the RNA sensor adaptor protein, MAVS.

Fig. 5 — A Schematic depiction of potential signalosome interactions between LD localised RIG-I and mitochondrial localised MAVS. Created in BioRender. Monson, E. (2026) https://BioRender.com/6c1b687B, C Primary immortalised astrocyte cells were transfected with a RIG-I over-expression plasmid and stimulated with dsRNA for 24 hrs. Cells were live stained with MitoTracker Red prior to fixation and then all cells were stained with Autodot LD dye to visualise LDs (green), αRIG-I antibody (1:200) (purple) and αMAVS antibody (1:100) (blue). Scale bar, 50 μm. Data represent mean ± SEM. Statistical significance assessed by two-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Cells imaged over a 10 min timeframe. White arrows indicate interactions between the two organelles. Images analysed via Imaris image software for % interactions between LDs and mitochondria. n = 100 cells across 6 fields of view over 3 biological replicates. Scale bar, 50 μm. Images represent n = 3 biological replicates. D–G Cells were live stained, and infected with either ZIKV (MR766 strain) at an MOI1, stimulated with dsRNA viral mimic (Poly I:C) or treated with oleic acid (OA, 500 µM) for 8 hrs, and live imaged for LD/mitochondria movement in stimulated cells for 10 min (120 frames). Images analysed via Imaris image software for % interactions between LDs and mitochondria, and data further analysed for numbers of transient interactions between the two organelles. n = 100 cells across 6 fields of view over 3 biological replicates, scale bar, 50 μm. Data represent mean ± SEM. Statistical significance assessed by two-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). H Primary immortalised astrocyte cells were transfected with a RIG-I over expression plasmid and stimulated with dsRNA for 24 hpi. Cells live stained with MitoTracker Red prior to fixation and all cells stained with Autodot LD dye to visualise LDs (green), αRIG-I antibody (1:200) (purple) and αMAVS antibody (1:100) (blue). Scale bar, 50 μm. Images represent n = 3 biological replicates. Source data are provided as a Source Data file.

We first examined if LDs enhanced their interaction with mitochondria following viral infections. Live cell imaging analysis of astrocytes revealed that the frequency of LDs interacting with mitochondria increased significantly during both dsRNA and ZIKV infection, a phenomenon that did not occur when increasing LD numbers via oleic acid treatment of the cells (Fig. 5B-G, S6, Supp Movie 1-4). This increase in interactions was driven by transient interactions of LD and mitochondria (Fig. 5D, E), with some LDs interacting with mitochondria up to 15 times in a 10-minute timeframe, with an average of 6 interactions in virally infected cells compared to 1.8 in mock cells (Fig. 5F, G). Using confocal microscopy to image RIG-I and MAVS on the surface of both LDs and mitochondria we observed multiple instances of signalosome formation in individual cells, with LD localised RIG-I forming complexes with mitochondrial localised MAVS, following activation of early innate signalling pathways with an RNA viral mimic (Figs. 5H, S7A).

The structure of RIG-I displays no obvious structural features that would allow it to bind to a phospholipid membrane (Fig. 6A). To understand how RIG-1 is localising to the LD surface, we screened the protein structure using the DisoLipPred web server⁴⁶ for disordered lipid binding regions (DLBRs). We found that RIG-1 contains four main identifiable regions of DLBRs with 24 of the 925 (3%) amino acid residues consisting of amino acids that have a propensity to form DLBRs (Fig. 6B). We next used computational modelling to determine if a complex between RIG-I and the phospholipid monolayer of a LD was possible. The cryo-EM structure of a viral RNA mimetic in complex with RIG-1 (PDB code: 7TNY⁴⁷,) was docked onto a POPC lipid bilayer (to mimic an LD) and subjected to molecular dynamics simulations to determine if this interaction was stable. Across the 10nS simulation, there was little change in energy or RMSD, indicating a stable interaction between a lipid and the protein (Fig. 6 and S7B). This model suggests that LDs are able to interact with the Hel2 of RIG-I, shown to be essential as an autoinhibitory motif of RIG-I. This model would also leave the CARD domains of RIG-I free to interact with MAVS via the well-established mechanism⁴⁶. In summary, the modelling supported the formation of a signalosome between LDs, RIG-I, MAVs and the mitochondria⁴⁸.

The lipid droplet lipidome dynamically changes following activation of antiviral early innate signalling pathways

Viruses alter the cellular lipidome in a pro-viral manner to enhance their own replication cycles (as reviewed in ref.⁴⁸), however, these changes are usually examined at later infection time points, excluding the possibility that early cellular lipid changes may occur to facilitate a pro-host response. It is plausible that changes in lipid composition within the LD itself may assist in both altering the proteome of LDs as well as potentially enhancing the antiviral environment, however a lipidome analysis of LDs following viral infection has not yet been performed at early time points following viral infection prior to viral cellular remodeling. This manuscript has applied a dual-omics approach to better understand the regulation of lipids simultaneously with proteins on the virally driven LD following activation of the cellular early innate antiviral response, which may offer further insight into the role LDs may play during viral infection.

There were limited changes to lipids observed in the whole cell lysates following dsRNA stimulation of astrocyte cells despite the increase in the overall lipid quantity following innate immune activation (Fig. S8A), likely driven by the increase in cellular LDs. However, there were significant changes in relative abundance of the four lipid categories examined within the LD lipidome (Fig. 7A). Lipid changes within the LDs were seen as early as 8 hrs, with greater changes observed at 24 h (Fig. S8A, B), towards an increased abundance of glycerolipids and a decrease in sterols (Fig. 7A, Supplementary data 5). Further analysis of individual lipid classes was able to detect a total of 491 different lipid species, revealing that following activation of antiviral pathways, LDs generally altered their lipidome to increase the relative abundance of long-chain polyunsaturated triacylglycerols (chain lengths of 13-21 carbons, pink triangle, Figs. 7B, S8C, D), whilst decreasing saturated cholesterol esters (blue circles, Figs. 7B and S8C). A small but significant change was also observed in the phospholipids making up the LD membrane, with an increase in PE (phosphatidylethanolamine) and PI (phosphatidylinositol) lipids (Fig. 7C). Small changes in these membrane phospholipids are known to alter membrane curvature, stability, and their ability to incorporate proteins⁴⁹. There was also an upregulation of ether linkages in these structural lipids between 8 and 24 h (Fig. 7D), which are known to support membrane functional changes, including cellular signalling at lipid membranes^50–52.

Fig. 7 — A The relative abundance of major lipid categories (Glycerolipids, Glycerophospholipids, Sphingolipids and Sterols) in following groups; whole cell lysate post dsRNA stimulation at 8 and 24 h, LD fractions post dsRNA stimulation at 8 and 24 h. B Bubble plot of log2 fold changes in abundance of individual lipid species post dsRNA stimulation relative to mock at 24 h. Significance was determined by unpaired two-tailed students t test (n = 3). Individual lipid species are coloured by the class of lipid that they belong to. PC phosphatidylcholine; PE phosphatidylethanolamine; PI phosphatidylinositol; PS phosphatidylserine; LPC lysophosphatidylcholine; LPE lysophosphatidylethanolamine; DAG diacylglycerol; TAG triacylglycerol; Cer ceramide; SM sphingomyelin; CE cholesterol ester. Created in BioRender. Helbig, K. (2026) https://BioRender.com/118vs80C Changes in relative abundance of membrane phospholipids (PC phosphatidylcholine; PE phosphatidylethanolamine; PI phosphatidylinositol; PS phosphatidylserine; LPC lysophosphatidylcholine; LPE lysophosphatidylethanolamine) following dsRNA stimulation at 24 h in isolated LDs. D Bubble plot of log2 fold changes in relative abundance membrane phospholipids in isolated LDs following dsRNA stimulation from 8 h to 24 h. Individual phospholipid species (PC phosphatidylcholine; PE phosphatidylethanolamine; PI phosphatidylinositol; PS phosphatidylserine; LPC lysophosphatidylcholine; LPE lysophosphatidylethanolamine) characterized by abundance in LDs at 8 h relative to LDs at 24 h post dsRNA stimulation. Significance was determined by unpaired two-tailed students t test (n = 3).

Dual-omics correlation analysis reveals lipid–protein co-regulatory networks on lipid droplets

Our lipidomic and proteomic datasets were analysed from the same isolated LD fractions, allowing us to perform correlation analysis to map co-regulatory lipid:protein networks on LDs for the first time.

We investigated the correlation between the relative abundances of all protein-lipids pairs within LDs across our control and 24 hr post dsRNA treated cells to assess global trends in our dual-omics data sets (Supplementary data 6). Interestingly, many protein-lipid pairs showed very strong positive and negative correlations, with a small bias towards a negative correlation overall. However, we were able to detect greater than 100,000 protein:lipid pairs with an extremely strong positive linear correlation coefficient values (r > 0.9). Most lipid classes carried an overall averaged modest negative bias with the exception of cholesterol esters, sphingomyelins and ceramides, having individual species with very strong positive and negative correlations, and the lysophosphatidylcholines, being the only lipid class with a positive overall correlation coefficient mean. This may suggest that the lipid headgroup chemistry could strongly shape proteome co-regulation of the LD.

Hierarchical clustering of our correlation matrix revealed 4 main clusters (Fig. 8A, B), each with distinct groupings of protein:lipid interactions across each of the lipid classes (Fig. 8C). Cluster 1 represented proteins that appeared to have limited correlations with lipids, cluster 2, moderately lipid-negative aligned proteins, cluster 3, strongly lipid-aligned proteins and cluster 4 represented strongly lipid-anti-aligned proteins. Interestingly, the immune related proteins shown to be differentially regulated on LDs treated with our viral mimic for 24 hrs (Fig 2Dii) were all represented within either network cluster 4 (DLG1, FN1, DAK, LCP1, LGMIN, OTUD7B: all proteins that are downregulated on LDs at 24hrs post dsRNA) or in cluster 3, (CCNB2, MX1, IFIT1/2/3, SEC61A1, DDX58, STAT1, ISG15, RSAD2, RAB8B, RAB4A, RAB35, RAB10 and TMEM33: all proteins that are upregulated on LDs at 24 hrs post dsRNA). As can be seen in Fig. 8D, the proteins belonging to these individual clusters have opposed lipid correlations across most lipid classes. However, extracting the top 30 most positively and negatively correlated lipid species with each of these proteins revealed similarities in the fact that high positive correlations occurred predominantly with neutral lipids such as TGs and cholesterol esters, with negative correlations being enriched for membrane/sphingolipid species (Fig. 8E).

Additionally, using our dual-omics approach, we were able to show that the virally driven LD proteome supported the changes observed across phospholipid and glycerolipid classes, as well as the observed increases in long chain polyunsaturated fatty acids (PUFAs) (Fig. 9A, B, C). Lipid modification enzymes found within our proteome were heavily weighted towards involvement in glycerolipid and phospholipid metabolism (Fig. 9B), with high upregulation of the enzymes ACSL1, ACSL3 and ACSL4 (Fig. 8A), known to be involved in the synthesis of long-chain fatty acids being observed. Interestingly, LPCAT1 and 2 were also significantly upregulated on our LDs 24 hrs following activation of antiviral innate immune pathways (Fig. 2Dii). LPCAT1/2 together with the ACSL enzymes are known to be involved in the production pathways of bioactive lipid mediators, which are formed from PUFAs, with LDs being a known source of these antiviral bioactive mediators^8,50,51.

Fig. 9 — A, B Distribution of identified LD resident proteins involved in lipid metabolism post dsRNA stimulation at 24 h based on the percentage abundance. Gene ontology was performed using the Uniprot database within Perseus. Proteins were grouped into 4 main categories (involved in regulating fatty acids metabolism). C Schematic of the lipidomic and related proteomic changes in LDs following infection. Overview of changes in the LD lipidome complemented by upregulation of LD proteins involved in lipid metabolism. Proteins involved in lipid metabolism (long/very long/ poly unsaturated fatty acids (PUFAs) metabolism) were identified as significantly upregulated post dsRNA stimulation. These proteins could potentially play roles in LDs accumulation and protein recruitment to LDs (SPG20 & AUP1), fatty acids elongation and poly unsaturation (e.g., ACSLs, CYBR5 & UBXD8), increase in Triglycerides (e.g., APOL1 & LDAH), decrease in cholesterol esters (e.g., NSDHL) & also phospholipid synthesis (e.g., LPCAT), which were the main alterations observed in LDs lipidomic profile post dsRNA stimulation in primary immortalised astrocytes at 24 h. Created in BioRender. Helbig, K. (2026) https://BioRender.com/0fd88i6D Heatmap displaying z-score of significantly changed LD fatty acids with highest and lowest log2 fold change in dsRNA stimulated primary immortalized astrocytes; Blue tiles refer to fatty acids with a z-score <0 with red tiles referring to fatty acids with a z-score > 0. E, F Primary immortalised astrocyte cells were infected with ZIKV (MOI 1) for 4 h prior to treatment with 100,000 aLDs/cell containing the top up/down regulated fatty acids for 16 h. RT-qPCR was performed to evaluate ZIKV, IFN-β, IFN-λ and viperin mRNA expression. All results are in comparison to RPLP0 expression. Data are presented as mean values ± SEM, P values were determined by unpaired two-tailed Student’s t test with a Holm-Sidak correction for multiple comparisons for 2 or more groups. Stimulated cells were statistically compared with their respective mock controls, ns = not significant. (n = 4 biological replicates). Source data are provided as a Source Data file.

Long and very-long chain fatty acids within lipid droplets promote an antiviral response

Virally driven LDs displayed enhanced levels of polyunsaturated triglycerides (Fig. 7B), with elongated chain length (Fig. S8D). To further understand the abundance of long chain fatty acids in LDs following activation of early innate immune signalling, we performed fatty acid chain length analysis across all lipid classes. As can be seen in Fig. 9D, there is a statistically significant increase in multiple fatty acid chain lengths from the long ( > 18) and very long ( > 22) categories present within LDs. These fatty acid changes belonged mainly to the phospholipid and glycerolipid classes, with a general decrease in these fatty acid chain lengths observed in sphingolipids (Fig. S9A).

The role of individual fatty acids within LDs has not previously been examined, although some of the identified individual fatty acids that are differentially regulated in our data have shown antiviral activity when added to cell cultures in vitro^52–54. We have recently optimised an artificial lipid droplet (aLD) system that is deliverable to multiple cells types at high levels⁵⁵. Using this system, we examined the potential of some of the most highly up and down regulated long chain fatty acids within our LDs following viral infection, for their ability to influence viral replication. Fatty acids were chosen from the top 10 regulated fatty acids based on their commercial availability and were incorporated into the neutral lipid component of our aLDs (addition of alternate fatty acids did not change aLD morphology; Fig. S9B, C). Notably, the delivery of aLDs themselves, composed of a neutral lipid TAG core and a complex phospholipid layer comprising PC, PE and PI, as we have delivered previously⁵⁵, was able to restrict ZIKV infection in vitro (Fig. 9E). This also coincided with increases in both type I and III IFN mRNA as well as the downstream ISG, viperin (Fig. 9F). Of the 8 long chain fatty acids delivered within aLDs, only 2 of the LD upregulated PUFAs were shown to directly restrict ZIKV replication, arachidonic acid (AA) and eicosapentaenoic acid (EPA), interestingly, delivery of both PUFAs within our aLDs also significantly enhanced the cellular transcription of both type I and III IFNs as well as viperin, indicating that these PUFAs may have a direct contribution to enhancing antiviral cytokine expression (Fig. 9E and F). Delivery of the four LD downregulated fatty acids in aLDs to cells, in all instances but one (NA: Nervonic acid) demonstrated a significant increase in ZIKV replication, indicating that these fatty acids may be detrimental to a pro-host environment following viral infection.

Discussion

The role of LDs in viral infection has been predominantly studied from a pro-viral point of view, where some viruses usurp the LD as a platform for assembly by manipulating its lipolysis and biogenesis to help enhance the viral life cycle⁸. However, recent research has demonstrated that LD numbers rapidly increase during early viral infection via an EGFR dependant mechanism^54–57. This pathway is distinct from the homeostatic turnover of LDs and is essential for the optimal production of antiviral cytokines^58–60, highlighting LDs as critical organelles in mounting an effective antiviral immune response. LDs are the main lipid storage organelle of the cell, and although the lipidome of the whole cell has been extensively studied during viral infection^56–59, lipidomic changes occurring within the LD is yet to be examined. Here, we use a powerful dual-omics approach to examine the proteomic and lipidomic changes of virally induced LDs in astrocyte cells, the main producers of LDs in the brain, to dissect the role they play in generating an antiviral host cell environment.

We found that the astrocyte LD proteome undergoes dynamic changes following viral infection, enhancing the presence of antiviral and signalling proteins. This included the recruitment of directly antiviral proteins such as IFIT1 and IFIT2, two ISGs known to bind viral RNA species and inhibit their translation^60–62. Additionally, RSAD2 (viperin), a protein that both directly restricts RNA viruses and enhances antiviral signalling through signalosome formation, was upregulated following infection^14,63,64. Signalosome formation is a requirement for efficient signalling intracellularly, to increase local concentrations of signalling components and promote weak interactions that may be required for enzyme activation. This process typically occurs at specific membrane platforms such as the ER for STING, the mitochondria and peroxisome for MAVS, and more recently, endosomes, which transiently accumulate RIG-I to facilitate poly-ubiquitylation^13,65,66. Alternatively, signalling molecules like cGAS can form self-organising centres to concentrate their reactions¹⁵. Here, we demonstrate that LDs serve as a platform for localising key signalling proteins required for the production of both type I interferons and ISGs: critical components of the antiviral response. While some of these proteins remain constant at the LD surface (MDA5, STAT2), others accumulated transiently following viral infection (RIG-I, STAT1). As previously described, LD biogenesis in this context is driven by a non-homeostatic, EGFR-dependent mechanism^19,65–72, and we demonstrate that STAT1 localisation to LDs was specific to virally driven LDs, not those formed under homeostatic conditions. Ultimately, transient localisation of proteins to the LD surface is not well understood; however, post-translational modifications (PTMs) are hypothesied to play a key role. Supporting this, ~8.4% of the LD proteome following innate immune activation comprised enzymes involved in PTMs (Fig. S10A, and as reviewed in refs. ^50–52).

In addition to proteomic changes, we observed alterations in the LD lipidome that may contribute to protein recruitment. Specifically, the external phospholipid membrane composition shifted towards a profile known to enhance membrane curvature, protein binding capacity and cellular signalling at lipid membranes^19,67–74. These structural changes were accompanied by large alterations in the neutral lipid content, with a shift towards long-chain fatty acids, including antiviral lipids previously shown to impair viral replication^52–54. Our dual-omics approach also enabled the identification of both global and specific protein–lipid correlation patterns within LDs. It revealed that enzymatic proteomic changes observed in the LD proteome likely drive much of the lipidomic alterations observed. Notable proteomic changes included increased expression of enzymes involved in fatty acid elongation and poly-unsaturation (e.g., ACSLs, CYBR5, UBXD8), triglyceride synthesis (e.g., APOL1, LDAH), and phospholipid modification (e.g., LPCAT), alongside reduced expression of cholesterol ester synthesis proteins (e.g., NSDHL). These findings highlight the power of a dual-omics strategy in elucidating the coordinated molecular changes that occur on LDs during viral infection. Additionally, correlation analyses revealed strong positive associations between neutral lipid enrichment and the upregulation of immune and antiviral proteins recruited to the LD surface (Fig. 8). Given our previous findings that LDs rapidly expand following activation of early innate immune signalling pathways¹⁰, this shift towards a higher neutral to membrane lipid ratio may facilitate enhanced access of neutral lipids for selective protein binding.

The antiviral roles of lipids remain relatively understudied, particularly given the vast diversity of lipid species. Notably, the specific functions of lipids within the LD itself have not previously been explored. Recently, interest has grown in the role of bioactive lipid mediators during viral infection^52,75,76, many of which are synthesised by LD-localised enzymes to generate a diverse array of signalling molecules. Eicosanoids represent a major class of these bioactive lipids, and many of the fatty acids upregulated in LDs following viral infection are known eicosanoid precursors. The delivery of two of these fatty acids (arachidonic acid and eicosapentaenoic acid) using an artificial lipid droplet delivery system was able to significantly decrease ZIKV replication in vitro. While the delivery of other enriched fatty acids such as docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA) within aLDs did not directly affect viral replication, it is plausible that they may have an in-direct overall pro-host antiviral response within the cell, and future work will need to be performed to assess this. Polyunsaturated fatty acids (PUFAs) such as arachidonic acid (AA) and eicosapentaenoic acid (EPA) have previously been shown to inhibit multiple enveloped viruses, including ZIKV infection^77,78, primarily through interference with early stages of the viral life cycle (e.g., attachment, fusion, or entry). Here we demonstrate a previously undescribed antiviral mechanism for these LD-localised PUFAs, where they are able to act from within the LD to drive significant upregulation of antiviral interferons and ISG mRNAs. Lastly, it cannot be overlooked that cellular lipids are in a constant flux, and fatty acids stored in LDs are frequently mobilised to generate new lipids for cellular functions. Thus, the rapid lipidomic changes observed in virally induced LDs may support broader metabolic reprogramming and signalling required during infection.

Although LD proteomes are understood to be dynamic in nature, the mechanisms underlying these changes remain poorly defined due to limited proteomic datasets^77,78. The LD proteome has previous been shown to be highly sensitive to bacterial LPS stimulation^{8,17,18,79,80}, however, we observed only a 17-protein overlap between LPS-induced and viral RNA-induced LD proteomic changes, suggesting a pathogen-specific reprogramming of the LD proteome (Fig. S10B S10C). Shared proteins were primarily involved in organelle trafficking and lipid metabolism, implying a core function of LDs in pathogen response: trafficking through the cell and modulating lipid species. (Fig. S10C). To our knowledge, this is the first study to apply a dual-omics approach to concurrently analyse the changing LD proteome and lipidome in mammalian cells. This powerful tool enabled us to associate changes in metabolic enzymes in the LD proteome with changes seen in both the increases of lipids having long-chain fatty acyl chains and the smaller alterations observed in the structural lipids of the LD membrane. Our approach also highlighted the striking presence of multiple enzymes involved in post-translational modification of proteins, and potentially lipids within the LD proteome. These findings support a model in which virally induced LDs not only coordinate protein localisation for antiviral signalling, but also drive the synthesis of bioactive lipid mediators, such as eicosanoids, that contribute to the cellular immune defence⁸⁰.

Until recently, the role of LDs in viral infections has been focused either on their use as a replication platform, with multiple viruses localising their proteins to its surface^81–84, or the ability of viruses to initially upregulate LDs prior to their depletion in cells to release various fatty acids and lipids to facilitate their respective viral replication lifestyles^85–88. In particular, multiple members of the Flaviviridae genus of viruses are reported to both utilise the LD as a scaffold for viral replication^81,82, but also to deplete LDs to release free fatty acids and cholesterols required to facilitate their viral lifecycle^85,88,89. This creates a tightly regulated and often paradoxical relationship, with the LD serving both as a structural hub for viral replication and as a metabolic reservoir that viruses strategically usurp to sustain their life cycle. However, members of the Flaviviridae viral family are renowned for their ability to ‘counteract’ innate immune responses^90–92 and the reported observations of viral manipulations of LDs are usually in the setting of advanced viral infection. Here, we show that triggering early innate antiviral signalling pathways induces LDs to undergo rapid, coordinated proteomic and lipidomic remodelling that promotes an antiviral cellular state, positioning them as active drivers of early innate immune defence rather than passive targets of viral exploitation.

LDs are rapidly upregulated following viral infection by non-homeostatic mechanisms and are critical in the early innate immune response facilitating a heightened antiviral environment⁹⁰. These studies highlight that viral driven LD upregulation coincides with dynamic changes to the LD lipidome and proteome to facilitate antiviral signalosome formation and the production of antiviral lipid species. Understanding these mechanisms may enable the development of next-generation antiviral strategies that enhance host cell immunity in a broad, pathogen-agnostic manner.

Methods

This research complies with all relevant ethical regulations of the institutions where the experiments were conducted, including the La Trobe University Ethics, Integrity and Biosafety committee, the Institutional Animal Care and Use Committee guidelines of the University of Sydney and the Institutional Biosafety Committee of University of Sydney.

Cells and culture conditions

Primary Immortalised Human Fetal Astrocytes were used throughout this study (referred to as primary immortalised astrocyte cells). These cells were maintained at 37 °C in a 5% CO₂ air atmosphere in DMEM (Gibco, Cat; 12430054) containing 10% fetal bovine serum (FBS) (Gibco, Cat; 10099141), 100 units/mL penicillin and 100 μg/mL streptomycin (Sigma-Aldrich, cat; P0781). Neurons (SH-Sy5Y) microglia (SIM-A9) were used to validate in vivo LD analysis and cells were cultured as above.

Lymphocytic Choriomeningitis Virus (LCMV) infection of mice

C57BL/6 mice were obtained from Australian BioResources (Moss Vale, NSW 2577) and housed under specific pathogen–free conditions in the animal facility at the University of Sydney, Sydney, Australia. Animal experiments were performed in accordance with the University of Sydney’s Animal Ethics Committee (1738/2020) and Institutional Biosafety Committee approval NLRD (22N004); mice were maintained under a 12 hr light/dark cycle at an ambient temperature of 20–23 °C and relative humidity of 40–60%, and with ample food and water. All experiments were done in accordance with the Institutional Animal Care and Use Committee guidelines of the University of Sydney. All mice were aged between 8 and 16 weeks at the time of infection. Mice were anesthetized with 100 μg ketamine and 1 μg xylazine per gram bodyweight and intracranial infection was performed by injecting 500 PFU of LCMV (strain LCMV Armstrong 53b) diluted in 20 μL of phosphate-buffered saline (PBS) with 1% fetal bovine serum (FBS). Sham-infected mice were used as controls and received the same volume of PBS with 1% FBS but without virus. Mice were weighed at the times indicated below, and percent weight change was calculated. Mice were euthanized at 2- or 4-days post infection, and the brains were removed, flash frozen in optimum cutting temperature (OCT) medium.

In vitro viral infection, viral mimics stimulation and plasmid transfection

Primary immortalised astrocyte cells were seeded at 5 × 10⁶ per T175cm² flask plates prior to infection with Zika virus (MR766 strain) at an MOI of 1. Cells were washed once with PBS, then infected with virus in serum free media for 4 hrs, followed by 20 h in DMEM supplemented with 10% fetal calf serum, at 37 °C, containing 5% CO₂. The viral mimic, poly I:C (dsRNA) (Invivogen) and all plasmid constructs used throughout the study were transfected into cells using PEI transfection reagent (Polyscience, Cat; 24765-1) at a concentration of 1 µg/ml. Viperin-mCherry and control mCherry plasmids were created as described previously⁹³. STAT1- and STAT2-mCherry tagged plasmids were kindly gifted to us from Associate Professor Greg Mosely (Monash University, Melbourne, VIC). RIG-I-wt was kindly gifted to us from Professor Stephen Polyak (University of Washington, Seattle, WA).

Immunofluorescence microscopy

For cultured cells, briefly, cells were grown in 24-well plates on 12 mm glass coverslips coated with gelatine (0.2% [v/v]) were washed with PBS, fixed with 4% paraformaldehyde in PBS for 15 min at room temperature and permeabilised with 0.1% Triton X-100 in PBS for 10 min. Cells were blocked with 5% BSA for 1 h, before antibody staining with αRIG-I (1:200; MA5-31715, Thermo Fisher Scientific), αMAVS (1:100; PA5-17256, Thermo Fisher Scientific). Cells were then incubated with Alexa Fluor 647 (1:200; A21236, Thermo Fisher) or Alexa Fluor 488 (1:200; A11008, Thermo Fisher Scientific) secondary antibody for 1 h. Mitochondria were stained by incubating live cells with MitoTracker® Red (Thermo Fisher Scientific) at 100 nM for 1 h. LDs were stained by incubating cells with Bodipy (493/503) at 1 ng/mL for 1 h and nuclei were stained with DAPI (Sigma-Aldrich, 1 µg/ml) for 5 min at room temperature. Samples were then washed with PBS and mounted with Vectashield Antifade Mounting Medium (Vector Laboratories). Preparation and staining of murine frozen brain sections were prepared following optimised protocols for tissue sections⁹⁴. Whole brains were sectioned sagittally. One section was snap-frozen and the other mounted immediately in OCT. Frozen sections were cut at 14 μM with a Leica CM 3050 S cryostat and mounted on microscope slides and stored at −80 °C. Sections were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. Sections were then washed with PBS, permeabilised with 0.1% Triton X-100 in PBS for 10 min, washed again and then blocked with 1% BSA for 30 min. Sections were stained for specific cell types: astrocytes (αGFAP polyclonal antibody, 1:1000; PA3-16727, Invitrogen), neurons (αNeuN polyclonal antibody, 1:2500; PA5-78639, Invitrogen) and microglia (αTMEM119 monoclonal antibody, 1:10,000; MA5-35043, Invitrogen). Sections were then washed and incubated with Alexa Fluor 555 secondary antibody at 1:200 for 1 h. Bodipy (493/503) was used to stain for LDs at 1 ng/mL for 1 h at room temperature, and nuclei were stained with DAPI for 5 min at room temperature. Images were then acquired using a Zeiss 800 confocal microscope. Unless otherwise indicated, images were processed using ImageJ analysis software. LD numbers were analysed using quantitative data from the single raw images in ImageJ (version 1.54p) using the particle analysis tool. LD/mitochondria interactions were analysed using Imaris software (version 10.2). Images captured on the Nikon TiE were processed in NIS elements (version 5.30) and ZEN Blue (version 3.3). For each condition, at least 9 fields of view were imaged at 60X magnification from different locations across each coverslip. LDs from at least 100 cells per biological replicate, with a minimum of n = 2 per experiment being analysed for both LD number and average LD size.

Transmission electron microscopy (TEM)

Lacey carbon grids (ProSciTech Pty Ltd) were glow-discharged using an Emitech K350 unit coupled to an Emitech K950X (Quorum Technologies Ltd). A 3 µL aliquot of isolated LDs was applied to each grid and plunge-frozen using an FEI Vitrobot (FEI Company). Each grid was then transferred to a Gatan 914 cryo-holder (Gatan, Inc.) and loaded into a JEOL JEM-2100 transmission electron microscope (JEOL Ltd). Images were acquired at 200 kV using SerialEM software (Version 4.2.16, University of Colorado, Boulder)⁹⁵ and recorded with an AMT NanoSprint15 MK-II camera (AMT Imaging).

Lipid droplet isolation

Isolation of LD from cells and brain tissues was performed using a Lipid Droplet Isolation Kit (Cell Biolabs; Cat; MET-5011). For cells, 5x T-175cm² flasks (5 × 10⁶ cells of primary immortalised astrocyte cells were trypsinised, pelleted at 1000 g for 5 mins, washed 2 times with 1 x PBS. Mouse brain tissues were thawed on ice and 200 mg of tissue surrounding the hippocampus was minced and put into sterile 1.5 mL microcentrifuge tube. Both cells and tissues were resuspended in 200 µl of reagent A (Cell Biolabs; Cat; MET-5011) and incubated on ice for 10 min with occasional vortexing. 800 µl of 1 x reagent B (Cell Biolabs; Cat; MET-5011) was added to the cells/ tissues and further incubated on ice for 10 min with occasional vortexing. Following incubation, cells/ tissues were carefully homogenised by being passed through a one inch 27-gauge needle attached to a 3 mL syringe five times. 600 µl of 1x reagent B was layered on top of the homogenates. Lysates were centrifuged for 3 h at 20,000 × g at 4 °C. 100 µL of the top layer containing the floating LDs was taken per condition and either stored at −80 °C for omic analysis of proteins and lipids or used for purity testing via western blot.

Dynamic light scattering (DLS) and nano particle tracking analysis (NTA) of isolated LDs

Following isolation, 10 µL of LDs was diluted into 1 mL of Buffer B(1x) for DLS measurements (Malvern Zetasizer Nano). All measurements were taken at a scattering angle of 90° For each sample, 5 independent rounds of 1 min measurements were performed with physical parameters as indicated: the refractive index as 1.33, viscosity as 0.89 cP, dielectric constant as 78.50 and measuring temperature as 25 °C. NTA was conducted on the LD samples using the ZetaView® system (Particle Metrix—ZetaView®, Inning am Ammersee, Germany) to directly quantify particles and assess size distribution. The total particle count was measured with the ZetaView QUATT, equipped with four lasers and utilizing ZetaView software version 8.05.14SP7. A 405 nm laser was used to monitor the sample chamber. Samples were diluted in Dulbecco’s PBS (DPBS) to achieve an average of n = 50–300 particles per field of view.

Visualisation and analysis of mCherry co-localisation to isolated lipid droplets

Primary immortalised astrocytes were seeded at 5 × 10⁶ per T175 cm² flask prior to transfection with mCherry, viperin-mCherry, STAT1-mCherry and STAT2-mCherry. Cells were trypsinised and LDs were isolated. Following LD isolation, LDs were stained with Bodipy (493/503) at a concentration of 1 ng/mL. 10 μL of purified LDs were spotted on a Nunc Lab-Tek II Chamber Slide System (Thermo Fisher Scientific) and were visualised via a Zeiss LSM 780 high-sensitivity laser scanning confocal microscope at 63x to determine both LDs stained with Bodipy (493/503) and mCherry expressing protein co-localisation. Image analysis was carried out using ImageJ, with LDs segmented using the Find Maxima function and a segmentation map was created. Segmentation maps were then used to separate interacting LDs and the Particle Analyser plugin was used to count LDs, create ROIs, and determine their sizes. To determine which LDs contained protein the LD area was isolated using an intensity threshold, and a binary image was created. The same method was used to find areas containing protein. The areas where both binaries overlapped were then determined via image calculator. This mask was then used in combination with the created ROIs to determine the presence or absence of protein in each LD.

SMLM image acquisition and processing

Switching buffer was applied to cells containing 80 µL 1 M mercaptoethylamine (MEA), 20 µL 1 M potassium hydroxide (KOH) and 0.8 µL 1 mg/mL Bodipy (493/503) in 1x PBS (pH 8.5) immediately prior to imaging. 8-well chamber slides were mounted on a custom SMLM setup based on ref.⁹⁶. Briefly, the setup is built around an Olympus IX-83 inverted fluorescence microscope equipped with a 100 × 1.49 oil immersion objective and a Photometrics Prime-95B sCMOS detector coupled to a pair of excitation lasers using appropriate dichroics and focal lenses (Semrock, Thorlabs). Diffraction-limited epifluorescence images of LDs were captured using 488 nm excitation at 8 mW (200 mW, Cobalt MLD), with 40 ms exposure. SMLM images were constructed by capturing 10,000 frames at 100 Hz with 200 mW 640 nm excitation and 10 ms exposure (iBeam Smart, Toptica).

Cytoplasmic lipid droplet co-localisation enumeration

Confocal image analysis

For each condition, at least 9 cells were imaged at 60X magnification from varying locations of the coverslip and were deconvoluted prior to analysis. Images were imported into Imaris (v9.5) with transfected cells isolated and masked to create a region of interest for co-localisation analysis. Co-localisation analysis was performed within Imaris using the auto threshold option comparing LDs and mCherry tagged proteins. Manders A Coefficient values were used to quantify co-localisation of mCherry tagged proteins to LDs.

Super resolution image analysis

For each time point, at least 10 fields of view were imaged at 100X magnification from varying locations across each well of the chamber slides. Images were imported to FIJI (ImageJ) and the LD channel converted to 16-bit before being scaled to 3000×3000, smoothed and binarized. The STAT channel was analysed using the ThunderSTORM plugin for FIJI to determine molecular coordinates from raw TIFF stacks and normalised Gaussian renderings. These images were converted to 16-bit, smoothed and binarized before being overlayed with the LD channel to form the final merged LD-STAT images. To determine the number of colocalisations per cell, images were further analysed using the interaction factor analysis plugin, which is specifically designed to assess dense SMLM data and normalizes for coincidental co-localisation by generating Monte Carlo-based random renderings. This analysis was used to generate a ratio of real co-localisation to the number expected if only random interactions were present such that a ratio of 1 indicates entirely random overlap, where 2 indicates twice as many interactions as randomly modelled. These ratios are referred to throughout as ‘co-localisation coefficient’. Number of co-localisations for each image was determined using this object co-localisation analysis plugin of ImageJ.

Western blotting

Lysates were subjected to SDS-PAGE. The proteins were transferred to 0.2 µm nitrocellulose membranes (Bio-Strategy, Campbellfield, VIC, Australia) and probed with primary antibodies. The primary antibodies used were: mouse monoclonal αCalnexin (1:1000; sc-23954, Santa Cruz Biotechnology), mouse monoclonal αmfn1 (1:1000; sc-166644, Santa Cruz Biotechnology), mouse monoclonal αACOX1 (1:1000; sc-517306, Santa Cruz Biotechnology), rabbit monoclonal αPerilipin-2 (1:2000; ab108323, Abcam), mouse monoclonal αRIG-I (1:1000; sc-376845 Santa Cruz Biotechnology), rabbit monoclonal αMX1 (1:2000; ab207414, Abcam), rabbit polyclonal αSTAT1 (1:1000; 9172, Cell Signaling Technology), rabbit monoclonal αSTAT2 (1:1000; A3588, ABclonal), rabbit polyclonal αZC3HAV1 (1:5000; ab154680, Abcam), rabbit monoclonal αPhospho-STAT1 (Tyr701) (1:1000; #7649, Cell Signaling Technology), rabbit polyclonal αmCherry (1:1000; 5993, BioVision) and rabbit polyclonal αTRIM25 (1:1000; ab86365, Abcam). To probe for ZIKV antigen, mouse α4G2 hybridoma fluid against flavivirus group antigen was used in a 1:1 ratio with 5% skim milk. Following 3 × 5 min washes with TBS wash buffer, the membrane was incubated with HRP conjugated secondary antibodies (Goat αMouse IgG (H + L) Secondary Antibody, HRP, 31430, Thermo Fisher Scientific) and (Goat αRabbit IgG (H + L) Secondary Antibody, HRP, 31460, Thermo Fisher Scientific) for 1 hr diluted 1:10,000. Following 5 × 10 min washes with TBS wash buffer, the membrane was incubated with GE (Amersham) or Femto (Thermo-scientific) Western Developer Reagent, dependent on the required sensitivity. The membranes were scanned using Amersham 600 chemiluminescence imager.

RNA extraction and real time PCR

In vitro RT-qPCR experiments to evaluate ZIKV replication and IFN expression were performed in 24-well plates with astrocyte cells seeded at 4.5 × 10⁴/well 24 h prior to infection with ZIKV (MOI 0.1) and treatment with artificial lipid droplets spiked with individual fatty acids. All experiments were performed at least in triplicate. In vivo RT-qPCR experiments to evaluate LCMV replication (targeting the nuclear protein) and activation of immune pathways were performed in triplicate from the brain tissue of culled mice used in the omic studies. For all RT-qPCR experiments, total RNA was extracted from cells using TriSure reagent (Bioline), with first strand cDNA being synthesised from total RNA and reverse transcribed using a Tetro cDNA synthesis kit (Bioline). Quantitative real-time PCR was performed in a CFX Connect Real-Time Detection System (BioRad) to quantitate target gene mRNA in comparison to the housekeeping gene RPLPO. Primers sequences can be found in Table 1 and their use has previously been published by us and others^10,97.

Table 1.

Primer sequences

Primer name	Sequence
RPLOP0-FP	5’-AGA TGC AGC AGA TCC GCA T-3’
RPLP0-RP	5’-GGA TGG CCT TGC GCA-3’
LCMV(NP)-FP	5’-CAGAAATGTTGATGCTGGACTGC-3’
LCMV(NP)-RP	5’-CAGACCTTGGCTTGCTTTACACAG-3’
ZIKV-FP	5’CAG CTG GCA TCA TGA AGA AGA AYC-3’
ZIKV-RP	5’CAC YTG TCC CAT CTT YTT CTC C-3’
IFN-β-FP	5’-AGA AAG GAC GAA CAT TGG GAA A-3’
IFN-β-RP	5’-TAG CAG AGC CCT TTT TGA TAA TGT AA-3’
IFN-λ -FP	5’-GAA GAG TCA CTC AAG CTG AAA AAC-3’
IFN-λ-RP	5’-AGA AGC CTC AGG TCC CAA TTC-3’
Viperin-FP	5’GTG AGC AAT GGA AGC CTG ATC-3’
Viperin-RP	5’-GCT GTC ACA GGA GAT AGC GAG AA-3’

Open in a new tab

Protein sample preparation for mass spectrometry

Proteins samples were precipitated from isolated LDs and whole cell via the S-trap micro protocol. Briefly 70 µl 2x lysis buffer (10% SDS in 50 mM TEAB) to 70 µl liquid sample [at a 1:1 ratio] so that final SDS is 5%. Samples were then sonicated for 5 min to recover absorbed protein. Samples were then centrifuged for 8 min at 13,000 g. 500 mM TCEP was added so that final concentration is 10 mM TCEP and incubate at 55 °C for 15 min to reduce thiol groups. 500 mM IAA was added to reach final 50 mM IAA and incubate at RT for 30 min to alkylate disulphides. 27.5% H₂PO₄ was then added so that the concentration is ~2.5% phosphoric acid. Samples were vortexed and had pH checked to ensure acidity. 6X binding/wash buffer (100 mM TEAB in 90% MeOH) was added to sample and mixed. Samples were centrifuged through the S-trap column at 4000 g for 30 s to trap proteins. Protein was then cleaned by adding 150 µl of binding/washing buffer, and centrifuged 3x at 4000 g for 30 s discarding flowthrough. S-Trap column was then centrifuged at 4000 g for 1 min to fully remove binding/wash buffer. Protein was digested by adding 20 µl of digestion buffer (50 mM TEAB + 1 µg Trypsin per 50 µg protein) and incubated at 37 °C overnight. Proteins were eluted by adding 40 µl of 50 mM TEAB in water to the S-Trap and incubated for 30 min. 40 µl of 0.2% formic acid in water was added to the S-Trap followed by centrifugation at 4000 g for 1 min. 40 µl of 50% acetonitrile (ACN) was added to the S-Trap and centrifuged at 4000 g for 1 min. Sample was placed in speedy vac to remove ACN and was followed by freeze drying the sample overnight.

Lipid Extraction

Lipids were purified according to a modified protocol⁹⁸. Briefly, 10 µl of SPLASH Lipidomix (Avanti Polar Lipids) was spiked in each sample as internal standards. Lipids from the whole cell lysates and LD fractions were extracted by diluting lysates with methanol (with 0.01% BHT) so that the final concentration of the sample was 60% v/v MeOH containing 0.01% BHT. Lysates were further diluted with MeOH and ChCl₃ so ratio of total H2O:CHCl₃: MeOH was 0.74:1:2 and lysates were centrifuged at 14,000 x g for 15 min to separate phases. Supernatants were collected and dried via speedvac centrifugation prior to analysis via LC-MS/MS.

Quantitative proteomics and functional annotation analyses

Proteins were identified by mass spectrometry and relatively quantified by a liquid chromatography approach. Peptide samples were analysed by LC-MS/MS using an Ultimate 3000 UHPLC coupled to an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, San Jose, CA). Solvent A is 0.1% formic acid (FA) / 5% dimethyl sulfoxide (DMSO) in water and solvent B is 0.1% FA / 5% DMSO in acetonitrile (ACN). Each sample was injected onto a PepMap C18 trap column (75 μM X 2 cm, 3 μM, 100 Å, Thermo Fisher Scientific, San Jose, CA) at 5 μL/min for 6 min using 0.05% trifluoroacetic acid (TFA) / 3% ACN in water and then separated through a PepMap C18 analytical column (75 μM × 50 cm, 2 μM, 100 Å, Thermo Fisher Scientific, San Jose, CA) at a flow rate of 300 nL/min. The temperature of both columns was maintained at 50 °C. During separation, the percentage of solvent B in mobile phase was increased from 3% to 23% in 89 min, from 23% to 40% in 10 min and from 40% to 80% in 5 min. Then the columns were cleaned at 80% solvent B for 5 min before decreasing the % B to 3% in 1 min and re-equilibrating for 8 min. The spray voltage, temperature of ion transfer tube and S-lens of the Orbitrap Elite mass spectrometer were set at 1.9 kV, 275 °C and 60%, respectively. The full MS scans were acquired at m/z 300 – 1650, a resolving power of 120,000 at m/z 200, an auto gain control (AGC) target value of 1.0 × 106 and a maximum injection time of 200 ms. The top 20 most abundant ions in the MS spectra were subjected to linear ion trap rapid collision induced dissociation (CID) at q value of 0.25, AGC target value of 5 × 10³, maximum injection time of 25 ms, isolation window of m/z 2 and NCE of 30%. Dynamic exclusion of 30 s was enabled. Data was searched by MaxQuant 1.4 against the UniProt homo sapien protein database. Trypsin was selected as enzyme. LFQ quantification and match between run were enabled and all other settings were default. The data sets (dsRNA, ZIKV, LCMV and respective controls) were imported into Perseus software (version 1.6.12.0). Label-free quantification (LFQ)/ or MS2 values were log2 transformed. Reverse database hits, potential contaminants, proteins only identified by site (a peptide carrying a modified residue), and proteins with 1 or less unique/razor peptides were removed from the matrices prior to t-test statistical analysis. The permutation FDR corrected p- value less than 0.05 alongside a minimum of 3-5 measurements per group required for significance allowed identification of significantly upregulated or downregulated proteins following stimulation/infection of dsRNA, ZIKV or LCMV, and their respective controls. The transformed and filtered data was exported into the web-based software VolcaNoseR (v1.03), to visualise significantly changed proteins with a criterion of having a log2 fold change < -2 or > 2 and an FDR corrected P-value < 0.05. Hierarchical clustering was performed across all replicates and was visualised via the “ComplexHeatmap” package. Functional enrichment analysis was performed on significantly enriched proteins via the R package (v4.5.0) “Clusterprofiler” and visualised using “enrichplot” in the form of a tree-plot. The t-test significant protein lists from each condition were visualised via their interactions using the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) database. The interaction data from STRING can be enhanced and adapted in Cytoscape (3 series) to produce enhanced network visualisation. Clustering of the network was performed using ClusterONE (Clustering with Overlapping Neighbourhood Expansion, 1.0) with a p-value < 0.05 cut-off.

Quantitative lipidomics

Samples were analysed by ultrahigh performance liquid chromatography (UHPLC) coupled to tandem mass spectrometry (MS/MS) employing a Vanquish UHPLC coupled to an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA), with separate runs in positive and negative ion polarities. Solvent A was 6/4 (v/v) acetonitrile/water with 5 mM medronic acid and solvent B was 9/1 (v/v) isopropanol/acetonitrile. Both solvents A and B contained 10 mM ammonium acetate. 10 µl of each sample was injected into an Acquity UPLC HSS T3 C18 column (1 ×150 mm, 1.8 µm: Waters, Milford, MA, USA) at 50 °C at a flow rate of 60 μL/min for 3 min using 3% solvent B. During separation, the percentage of solvent B was increased from 3% to 70% in 5 min and from 70% to 99% in 16 min. Subsequently, the percentage of solvent B was maintained at 99% for 3 min. Finally, the percentage of solvent B was decreased to 3% in 0.1 min and maintained for 3.9 min. All MS experiments were performed using an electrospray ionization source. The spray voltages were 3.5 kV in positive ionisation-mode and 3.0 kV in negative ionisation-mode. In both polarities, the flow rates of sheath, auxiliary and sweep gases were 25 and 5 and 0 arbitrary unit(s), respectively. The ion transfer tube and vaporizer temperatures were maintained at 300 °C and 150 °C, respectively, and the ion funnel RF level was set at 50%. In the positive ionisation-mode from 3 to 24 min, top speed data-dependent scan with a cycle time of 1 s was used. Within each cycle, a full-scan MS-spectra were acquired firstly in the Orbitrap at a mass resolving power of 120,000 (at m/z 200) across an m/z range of 300–2000 using quadrupole isolation, an automatic gain control (AGC) target of 4e5 and a maximum injection time of 50 milliseconds, followed by higher-energy collisional dissociation (HCD)-MS/MS at a mass resolving power of 15,000 (at m/z 200), a normalised collision energy (NCE) of 27% at positive mode and 30% at negative mode, an m/z isolation window of 1, a maximum injection time of 35 milliseconds and an AGC target of 5e4. For the improved structural characterisation of glycerophosphocholine (PC) lipid cations, a data-dependent product ion (m/z 184.0733)-triggered collision-induced dissociation (CID)-MS/MS scan was performed in the cycle using a q-value of 0.25 and a NCE of 30%, with other settings being the same as that for HCD-MS/MS. For the improved structural characterisation of triacylglycerol (TG) lipid cations, the fatty acid + NH3 neutral loss product ions observed by HCD-MS/MS were used to trigger the acquisition of the top-3 data-dependent ion trap CID-MS3 scans in the cycle using a q-value of 0.25 and a NCE of 30%, with other settings being the same as that for HCD-MS/MS. Dynamic exclusion of 15 s was enabled and only ions with charge state of 1-3 were selected for fragmentation.

Lipid Identification and functional annotation analyses

LC-MS/MS data was searched through MS Dial 4.90. The mass accuracy settings are 0.005 Da and 0.025 Da for MS1 and MS2. The minimum peak height is 50000 and mass slice width is 0.05 Da. The identification score cut off is 80%. Post identification was done with a text file containing name and m/z of each standard in SPLASH® LIPIDOMIX® Mass Spec Standard (Cat. 330707, Avanti Polar Lipids, Birmingham, AL, USA). In positive mode, [M + H]+, [M + NH4]+ and [M + H-H2O]+ were selected as ion forms. In negative mode, [M-H]- and [M + CH3COO]- were selected as ion forms. All lipid classes available were selected for the search. PC, LPC, DG, TG, CE, SM were identified and quantified at positive mode while PE, LPE, PS, LPS, PG, LPG, PI, LPI, PA, LPA, Cer, CL were identified and quantified at negative mode. The retention time tolerance for alignment is 0.1 min. Lipids with maximum intensity less than 5-fold of average intensity in blank was removed. All other settings were default. All lipid LC-MS features were manually inspected and re-integrated when needed. These four types of lipids, 1) lipids with only sum composition except SM, 2) lipid identification due to peak tailing, 3) retention time outliner within each lipid class, 4) LPA and PA artifacts generated by in-source fragmentation of LPS and PS were also removed. The shorthand notation used for lipid classification and structural representation follows the nomenclature proposed previously⁹⁹. Quantification of lipid species in the unit of pmol from each sample was achieved by comparison of the LC peak areas of identified lipids against those of the corresponding internal lipid standards in the same lipid class and the quantity of each lipid standard at pmol. Since the lipid class of analyte and internal standard are identical but no co-ionization of analyte and IS are achieved, it’s categorized as level 3 quantification by Lipidomics Standards Initiative (lipidomicstandards.org/). Finally, the lipid species at the class, subclass or molecular species levels were normalized to either the total lipid concentration (i.e., mol% total lipid), or total lipid-class concentration (i.e., mol% total lipid class). For the lipid classes without correspondent stable isotope-labelled lipid standards, the LC peak areas of individual molecular species within these classes were normalised as follows: the MG species against the DG (18:1D7_15:0); the LPG against the PG (18:1D7_15:0), the LPA against the PA (18:1D7_15:0) and the LPS against the PS (18:1D7_15:0). All lipidomic data was imported to Excel for normalization and further processing. Significant differences between relative abundance of individual lipid species in two sample groups was acquired using unpaired two-tailed students t test (n = 3; biological replicates). Stack bar charts comparing relative abundance of major lipid categories and classes were made using Prism Version 8.4.3 software. Bubble plots of log2 fold changes in relative abundance of individual lipid species were made using R packages “ggplot2”. Relative quantification of fatty acid chain length among all 491 identified lipid species was analysed and plotted using R packages “Heatmap” and relative intensities were compared based on z-score. Principal component analysis (PCA) was performed using built in R packages (“stats”, “prcomp()”, “t”, “tibble”).

Preparation of artificial lipid droplets (aLDs)

Artificial lipid droplets (aLDs) were prepared based on previously published protocols⁵⁵. A total of 1.3 mg of phospholipids including DOPC (1,2-dioleoyl-sn-glycero-3-phosphocholine), DOPE (1,2-dioleoyl-sn-glycero-3 phosphoethanolamine) and l-α-phosphatidylinositol (Soy PI) (sodium salt), dissolved in chloroform at 32 mM, were transferred to a 1.5 mL microcentrifuge and mixed in a molar ratio of 9:3:1, respectively (MERK-Avanti Polar Lipids). ATTO ATTO 647 N PPE (ATTO-TEC GmbH), a fluorescence membrane probe, was added to the lipid mixture at a final concentration of 1 mM. The mixture was then dried under a stream of N2 followed by hydrating the dried phospholipid film by addition of 100 μL of prewarmed Buffer B (20 mM HEPES, 100 mM KCl, 2 mM MgCl2, pH 7.4) incubated at 37 °C for 1 min. This was followed by the addition of 5 mg TAG (triglycerides; Triolein- MERK-Avanti Polar Lipids) and vortexed for 24 cycles of 10 s on and 10 s off to reach a crude emulsion. The emulsion was centrifuged at 1000 × g for 5 min at RT, and the infranatant was collected. Individual fatty acids were added to the phospholipid mixture at 2% of the total amount of lipids (0.4 mol). The fatty acids used in this study include Arachidonic Acid (5 mg/ml, CAT#181198, Sigma Aldrich), cis-5,8,11,14,17-Eicosapentaenoic acid (50 mg/ml, CAT# E2011, Sigma Aldrich), cis-4,7,10,13,16,19-Docosahexaenoic acid (50 mg/ml, CAT#D2534, Sigma Aldrich), all-cis-7,10,13,16,19-Docosapentaenoic acid (50 mM, CAT#D1797, Sigma Aldrich), Lignoceric acid (30 mg/ml, CAT# L6641, Sigma Aldrich), Nervonic acid (30 mg/ml, CAT#N1514, Sigma Aldrich), Heptacosanoic Acid (50 mM, AMBH97BA00FA, Ambeed, Inc.), Tricosanoic acid (590 mM, CAT# T6543, Sigma Aldrich). The size distribution and particle concentration of aLDs was quantified using the ZetaView® system (Particle Metrix—ZetaView®, Inning am Ammersee, Germany). Morphology and general characteristics were assessed using fluorescence microscopy on a Nikon TiE microscope at 60× magnification.

Computational modelling

The cryo-EM structure of a viral RNA mimetic in complex with RIG-1 (PDB code: 7TNY,) was manually docked onto a POPC lipid bilayer (to mimic a LD) created using the simulations setup module in Maestro (Schrodinger L.L.C). The interface of RIG-I was chosen based upon bioinformatics predictions (DisoLipPred, 8.0) which highlighted amino acids 783-790 (IQTHEKFI) within the HEL2 domain as hypothetical binders of lipids. The protein-lipid-RNA system was then prepared using the Protein Preparation Wizard (Schrodinger L.L.C) to assign bond orders, add hydrogens, optimise hydrogen bonding networks, and minimise the structure with the OPLS4 force field. The complex was then solvated in an orthorhombic box using the TIP3P water model, with a buffer distance of 10 Å from the solute. Na⁺ and Cl⁻ ions were added to neutralise the system and mimic a physiological salt concentration (0.15 M). A 10-nanosecond molecular dynamics (MD) simulation was then performed using the Desmond simulation package (Schrödinger, LLC) with trajectory snapshots were saved every 100 ps for analysis. Figures were created using The PyMOL Molecular Graphics System, Version 3.0 Schrödinger, LLC.

Bona fide LD protein identification

Bona Fide LD proteins were identified by comparing all proteins identified across all 21 mammalian published proteomes to our dataset using excel LOOKUP function.

Correlation analysis between lipids and proteins

The pairwise linear correlation coefficient (rho) was generated using the MATLABA 2025a function ‘corr’ (rho = corr(X,Y)). Here we examined the 8 and 24 h simulation timepoints individually, to generate a pair of correlation matrices between Mock, and dsRNA stimulations, comparing every combination of lipid (489 in total) and protein (3870 in total) identified in our LD proteomic and lipidomic analysis.

Statistical analysis and reproducibility

Student’s t tests were used for statistical analysis between 2 groups, with experiments with 2 or more experimental groups statistically analysed using either an ordinary one, or two-way multiple comparison ANOVA or multiple t tests using the Holm-Sidak method for corrections for multiple comparisons with P < 0.05 considered to be significant. Omics data was analysed and plotted using R packages Version 4.3.0 as stated in relevant sections. All statistical analysis (unless otherwise indicated) was performed using Prism Version 8.4.3 software (GraphPad, La Jolla, United States). All experiments were performed in biological triplicate (unless otherwise stated), and technical duplicates were also performed for RT-PCRs. Error bars represent mean ± SEM, with a P value less than 0.05 considered to be significant. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information^{(3.5MB, pdf)}

41467_2026_74016_MOESM2_ESM.pdf^{(152.6KB, pdf)}

Description Of Additional Supplementary File

Supplementary Data 1^{(3MB, pdf)}

Supplementary Data 2^{(2.8MB, pdf)}

Supplementary Data 3^{(5MB, xlsx)}

Supplementary Data 4^{(3.5MB, pdf)}

Supplementary Data 5^{(3.1MB, pdf)}

Supplementary Data 6^{(24.2MB, xlsx)}

Supplementary Movie 1^{(31MB, mp4)}

Supplementary Movie 2^{(8.9MB, mp4)}

Supplementary Movie 3^{(33.6MB, mp4)}

Supplementary Movie 4^{(11MB, mp4)}

Reporting summary^{(7.8MB, pdf)}

Transparent Peer Review file^{(840.6KB, pdf)}

Source data

Source data^{(2MB, xlsx)}

Acknowledgements

We acknowledge that the bulk of this project was conducted on the traditional homelands of the Wurundjeri Woi wurrung people. We would like to thank the La Trobe University Bioimaging Facility for their assistance with image acquisition, data collection and image analysis essential to this work. We also acknowledge the use of the Bio21 Mass Spectrometry and Proteomics Facility at the University of Melbourne and thank them for their continued support of our projects. We would like to thank Associate Professor Greg Mosely from Monash University (Melbourne, Australia) for gifting us the STAT1 and 2 mCherry constructs, Professor Stephen Polyak from the University of Washington (Seattle, WA) for gifting us the RIG-I-wt construct and Prof Paul Herzog from the Hudson Institute (Melbourne, Australia) for gifting us the MAVS antibody. SEB, and PJP gratefully acknowledge the support of the Australian Office of National Intelligence. This work was performed in part at the Australian National Fabrication Facility (ANFF), a company established under the National Collaborative Research Infrastructure Strategy, through the La Trobe University Centre for Materials and Surface Science. All schematic diagrams throughout this manuscript were created using BioRender.

Author contributions

Conceptualization: E.A.M., K.J.H. Methodology: E.A.M., S.N., N.W., J.L.L., M.L.S., Z.T., C.J., A.J.M., A.M.R., S.E.B., A.M., P.J.P., J.K.H., D.R.W., K.J.H. Investigation: E.A.M., J.L.L., Z.T., A.J.M., A.M.R., I.A., M.L.S., V.T., Q.D., C.J.J., M.J.H., S.E.B., J.K.H., S.N., B.L., M.L., K.J.H. Visualization: E.A.M., J.L.L., Z.T., A.J.M., B.L., A.M.R., I.A., C.J.J., S.N., D.R.W., K.J.H. Funding acquisition: E.A.M., A.J.M., A.M.R., D.R.W., A.M., K.J.H. Project administration: E.A.M., K.J.H., D.R.W. Supervision: E.A.M., D.R.W., K.J.H. Writing—original draft: E.A.M., K.J.H. Writing—review & editing: E.A.M., Z.T., J.L.L., I.A., A.J.M., D.R.W., S.N., K.J.H. All authors participated in reviewing and editing the manuscript.

Peer review

Peer review information

Nature Communications thanks Filipe Pereira-Dutra and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

K.J.H., A.M., and D.R.W. disclose support for the research of this work from the National Health and Medical Research Council New Ideas Grant (grant numbers APP1181434 and APP2038548). EAM discloses support for publication of this work from the National Health and Medical Research Council Investigator Grant (grant number APP2033787), the CASS Foundation Medicine/Science grant and Jack Brockhoff Foundation Medical grant. D.R.W. is funded by the Australian Research Council Discovery Early Career Research Award (grant number DE200100584) and along with A.J.M., A.M.R. are supported by the Holsworth Biomedical Research Initiative under the auspices of the Bendigo Tertiary Education Anniversary Foundation. S.E.B., P.J.P. discloses support for publication of this work from the Australian Office of National Intelligence – National Intelligence and Security Discovery Research Grant (grant number NI210100127). Lastly, E.A.M., KJH disclose support from La Trobe University Internal funding schemes. Z.T., J.L.L., I.A., M.L.S., V.T., Q.N.D., N.W., B.L., C.J.J., M.L., S.N., J.K.H. and M.H. declare no relevant funding for this manuscript.

Data availability

All mass spectrometry proteomics data sets have been deposited to the ProteomeXchange consortium via the PRIDE partner repository, with the dataset identifier PXD066369 and PXD066446 (https://www.ebi.ac.uk/pride/archive). Lipidomic data is deposited in the Metabolight comprehensive data repository [https://www.ebi.ac.uk/metabolights/] under accession number MTBLS14458. All other data generated in this study is provided in the main text, or in the Supplementary Information/Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Ebony A. Monson, Zahra Telikani, Jay L. Laws.

Contributor Information

Ebony A. Monson, Email: e.monson@latrobe.edu.au

Karla J. Helbig, Email: k.helbig@latrobe.edu.au

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74016-w.

References

1.K. M. Crosse, E. A. Monson, A. B. Dumbrepatil, Viperin binds STING and enhances the type-I interferon response following dsDNA detection. Immunol. Cell Biol. 99, 373-391 (2021). [DOI] [PubMed]
2.Dumbrepatil, A. B. et al. Viperin interacts with the kinase IRAK1 and the E3 ubiquitin ligase TRAF6, coupling innate immune signaling to antiviral ribonucleotide synthesis. J. Biol. Chem.294, 6888–6898 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Arrese, E. L., Saudale, F. Z. & Soulages, J. L. Lipid droplets as signaling platforms linking metabolic and cellular functions. Lipid Insights7, 7–16 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.R. Dhiman, S. Caesar, A. R. Thiam, B. Schrul, Mechanisms of protein targeting to lipid droplets: A unified cell biological and biophysical perspective. Semin. Cell Dev. Biol., 10.1016/j.semcdb.2020.03.004 (2020). [DOI] [PubMed]
5.Thiam, A. R. & Beller, M. The why, when and how of lipid droplet diversity. J. Cell Sci.130, 315–324 (2017). [DOI] [PubMed] [Google Scholar]
6.Olzmann, J. A. & Carvalho, P. Dynamics and functions of lipid droplets. Nat. Rev. Mol. Cell Biol.20, 137–155 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Onal, G., Kutlu, O., Gozuacik, D. & Dokmeci Emre, S. Lipid droplets in health and disease. Lipids Health Dis.16, 128 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.E. A. Monson, A. M. Trenerry, J. L. Laws, J. M. Mackenzie, K. J. Helbig, Lipid droplets and lipid mediators in viral infection and immunity. FEMS Microbiol. Rev., 10.1093/femsre/fuaa066 (2021). [DOI] [PMC free article] [PubMed]
9.Zadoorian, A., Du, X. & Yang, H. Lipid droplet biogenesis and functions in health and disease. Nat. Rev. Endocrinol.19, 443–459 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Monson, E. A. et al. Intracellular lipid droplet accumulation occurs early following viral infection and is required for an efficient interferon response. Nat. Commun.12, 4303 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.M. Bosch et al, Mammalian lipid droplets are innate immune hubs integrating cell metabolism and host defense. Science370, (2020). [DOI] [PubMed]
12.Rösch, K. et al. Quantitative lipid droplet proteome analysis identifies annexin A3 as a cofactor for HCV particle production. Cell Rep.16, 3219–3231 (2016). [DOI] [PubMed] [Google Scholar]
13.Dixit, E. et al. Peroxisomes are signaling platforms for antiviral innate immunity. Cell141, 668–681 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Khantisitthiporn, O. et al. Viperin interacts with PEX19 to mediate peroxisomal augmentation of the innate antiviral response. Life Sci. Alliance4, e202000915 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Du, M. & Chen, Z. J. DNA-induced liquid phase condensation of cGAS activates innate immune signaling. Science361, 704–709 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Álvarez, R. et al. Escribá, G protein-membrane interactions I: Gαi1 myristoyl and palmitoyl modifications in protein-lipid interactions and its implications in membrane microdomain localization. Biochim. Biophys. Acta1851, 1511–1520 (2015). [DOI] [PubMed] [Google Scholar]
17.Yamamoto, E., Akimoto, T., Kalli, A. C., Yasuoka, K. & Sansom, M. S. P. Dynamic interactions between a membrane binding protein and lipids induce fluctuating diffusivity. Sci. Adv.3, e1601871 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Cho, W. & Stahelin, R. V. Membrane-protein interactions in cell signaling and membrane trafficking. Annu. Rev. Biophys. Biomol. Struct.34, 119–151 (2005). [DOI] [PubMed] [Google Scholar]
19.D. Casares, P. V. Escribá, C. A. Rosselló, Membrane lipid composition: effect on membrane and organelle structure, function and compartmentalization and therapeutic avenues. Int. J. Mol. Sci. 20, (2019). [DOI] [PMC free article] [PubMed]
20.York, A. G. et al. Limiting cholesterol biosynthetic flux spontaneously engages type I IFN Signaling. Cell163, 1716–1729 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Hofer, M. J. et al. Mice deficient in STAT1 but not STAT2 or IRF9 develop a lethal CD4+ T-cell-mediated disease following infection with lymphocytic choriomeningitis virus. J. Virol.86, 6932–6946 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Bonthius, D. J., Nichols, B., Harb, H., Mahoney, J. & Karacay, B. Lymphocytic choriomeningitis virus infection of the developing brain: critical role of host age. Ann. Neurol.62, 356–374 (2007). [DOI] [PubMed] [Google Scholar]
23.Bersuker, K. et al. A proximity labeling strategy provides insights into the composition and dynamics of lipid droplet proteomes. Dev. Cell44, 97–112.e7 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Beilstein, F., Bouchoux, J., Rousset, M. & Demignot, S. Proteomic analysis of lipid droplets from Caco-2/TC7 enterocytes identifies novel modulators of lipid secretion. PLoS One8, e53017 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Saka, H. A. et al. Chlamydia trachomatis infection leads to defined alterations to the lipid droplet proteome in epithelial cells. PLoS One10, e0124630 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Kramer, D. A., Quiroga, A. D., Lian, J., Fahlman, R. P. & Lehner, R. Fasting and refeeding induces changes in the mouse hepatic lipid droplet proteome. J. Proteom.181, 213–224 (2018). [DOI] [PubMed] [Google Scholar]
27.Menon, D. et al. Quantitative lipid droplet proteomics reveals Mycobacterium tuberculosis induced alterations in macrophage response to infection. ACS Infect. Dis.5, 559–569 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Dahlhoff, M. et al. Characterization of the sebocyte lipid droplet proteome reveals novel potential regulators of sebaceous lipogenesis. Exp. Cell Res.332, 146–155 (2015). [DOI] [PubMed] [Google Scholar]
29.Khan, S. A., Wollaston-Hayden, E. E., Markowski, T. W., Higgins, L. & Mashek, D. G. Quantitative analysis of the murine lipid droplet-associated proteome during diet-induced hepatic steatosis. J. Lipid Res.56, 2260–2272 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Wang, W. et al. Proteomic analysis of murine testes lipid droplets. Sci. Rep.5, 12070 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Zhang, H. et al. Proteome of skeletal muscle lipid droplet reveals association with mitochondria and apolipoprotein a-I. J. Proteome Res.10, 4757–4768 (2011). [DOI] [PubMed] [Google Scholar]
32.Liu, P. et al. Chinese hamster ovary K2 cell lipid droplets appear to be metabolic organelles involved in membrane traffic. J. Biol. Chem.279, 3787–3792 (2004). [DOI] [PubMed] [Google Scholar]
33.Brasaemle, D. L., Dolios, G., Shapiro, L. & Wang, R. Proteomic analysis of proteins associated with lipid droplets of basal and lipolytically stimulated 3T3-L1 adipocytes. J. Biol. Chem.280, 4004 (2005). Vol. 279 (2004) 46835–46842. [DOI] [PubMed] [Google Scholar]
34.Fujimoto, Y. et al. Identification of major proteins in the lipid droplet-enriched fraction isolated from the human hepatocyte cell line HuH7. Biochim. Biophys. Acta Mol. Cell Res.1644, 47–59 (2004). [DOI] [PubMed] [Google Scholar]
35.Sato, S. et al. Proteomic profiling of lipid droplet proteins in hepatoma cell lines expressing hepatitis C virus core protein. J. Biochem.139, 921–930 (2006). [DOI] [PubMed] [Google Scholar]
36.Bartz, R. et al. Dynamic activity of lipid droplets: protein phosphorylation and GTP-mediated protein translocation. J. Proteome Res.6, 3256–3265 (2007). [DOI] [PubMed] [Google Scholar]
37.Brasaemle, D. L., Dolios, G., Shapiro, L. & Wang, R. Proteomic analysis of proteins associated with lipid droplets of basal and lipolytically stimulated 3T3-L1 adipocytes. J. Biol. Chem.279, 46835–46842 (2004). [DOI] [PubMed] [Google Scholar]
38.Hou, J., Liu, Y., Lu, Z., Liu, X. & Liu, J. Biochemical characterization of RNase HII from Aeropyrum pernix. Acta Biochim. Biophys. Sin.44, 339–346 (2012). [DOI] [PubMed] [Google Scholar]
39.Wu, C. C., Howell, K. E., Neville, M. C., Yates III, J. R. & McManaman, J. L. Proteomics reveal a link between the endoplasmic reticulum and lipid secretory mechanisms in mammary epithelial cells. Electrophoresis21, 3470–3482 (2000). [DOI] [PubMed] [Google Scholar]
40.Lindqvist, R. et al. Fast type I interferon response protects astrocytes from flavivirus infection and virus-induced cytopathic effects. J. Neuroinflammation13, 277 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Pfefferkorn, C. et al. Abortively infected astrocytes appear to represent the main source of interferon beta in the virus-infected brain. J. Virol.90, 2031–2038 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Kambara, H. et al. Negative regulation of the interferon response by an interferon-induced long non-coding RNA. Nucleic Acids Res. 42, 10668–10680 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Au-Yeung, N., Mandhana, R. & Horvath, C. M. Transcriptional regulation by STAT1 and STAT2 in the interferon JAK-STAT pathway. JAKSTAT2, e23931 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Jacobs, J. L. & Coyne, C. B. Mechanisms of MAVS regulation at the mitochondrial membrane. J. Mol. Biol.425, 5009–5019 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Katuwawala, A., Zhao, B. & Kurgan, L. DisoLipPred: accurate prediction of disordered lipid-binding residues in protein sequences with deep recurrent networks and transfer learning. Bioinformatics38, 115–124 (2021). [DOI] [PubMed] [Google Scholar]
46.Wang, W. & Pyle, A. M. The RIG-I receptor adopts two different conformations for distinguishing host from viral RNA ligands. Mol. Cell82, 4131–4144.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Onomoto, K., Onoguchi, K. & Yoneyama, M. Regulation of RIG-I-like receptor-mediated signaling: interaction between host and viral factors. Cell. Mol. Immunol.18, 539–555 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Omasta, B. & Tomaskova, J. Cellular lipids-hijacked victims of viruses. Viruses14, 1896 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Papin, M., Bouchet, A. M., Chantôme, A. & Vandier, C. Ether-lipids and cellular signaling: a differential role of alkyl- and alkenyl-ether-lipids? Biochimie215, 50–59 (2023). [DOI] [PubMed] [Google Scholar]
50.Kuwata, H. & Hara, S. Role of acyl-CoA synthetase ACSL4 in arachidonic acid metabolism. Prostaglandins Other Lipid Mediat144, 106363 (2019). [DOI] [PubMed] [Google Scholar]
51.Saliakoura, M. et al. The ACSL3-LPIAT1 signaling drives prostaglandin synthesis in non-small cell lung cancer. Oncogene39, 2948–2960 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Feng, Y. et al. Antiviral activity of eicosapentaenoic acid against Zika virus and other enveloped viruses. Front. Pharmacol.16, 1564504 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.X. Suo, et al. DHA and EPA inhibit porcine coronavirus replication by alleviating ER stress. J. Virol., e0120923 (2023). [DOI] [PMC free article] [PubMed]
54.Chang, C.-Y. et al. DHA attenuated Japanese Encephalitis virus infection-induced neuroinflammation and neuronal cell death in cultured rat Neuron/glia. Brain Behav. Immun.93, 194–205 (2021). [DOI] [PubMed] [Google Scholar]
55.Z. Telikani et al, The phospholipid composition of artificial lipid droplets enhances their deliverability and facilitates a broad Biodistribution in vivo and in vitro. Acta Biomater., 10.1016/j.actbio.2025.05.002 (2025). [DOI] [PubMed]
56.Farley, S. E. et al. A global lipid map reveals host dependency factors conserved across SARS-CoV-2 variants. Nat. Commun.13, 3487 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Chen, Q. et al. Metabolic reprogramming by Zika virus provokes inflammation in human placenta. Nat. Commun.11, 2967 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Leier, H. C. et al. A global lipid map defines a network essential for Zika virus replication. Nat. Commun.11, 3652 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Hehner, J. et al. Glycerophospholipid remodeling is critical for orthoflavivirus infection. Nat. Commun.15, 8683 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Poddar, D. et al. The interferon-induced protein, IFIT2, requires RNA-binding activity and neuronal expression to protect mice from intranasal vesicular stomatitis virus infection. MBio15, e0056824 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Vladimer, G. I., Górna, M. W. & Superti-Furga, G. IFITs: Emerging roles as key anti-viral proteins. Front. Immunol.5, 94 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Pichlmair, A. et al. Superti-Furga, IFIT1 is an antiviral protein that recognizes 5’-triphosphate RNA. Nat. Immunol.12, 624–630 (2011). [DOI] [PubMed] [Google Scholar]
63.K. M. Crosse et al, Viperin binds STING and enhances the type-I interferon response following dsDNA detection. Immunol. Cell Biol. 10.1111/imcb.12420 (2020). [DOI] [PubMed]
64.Saitoh, T. et al. Antiviral protein Viperin promotes Toll-like receptor 7- and Toll-like receptor 9-mediated type I interferon production in plasmacytoid dendritic cells. Immunity34, 352–363 (2011). [DOI] [PubMed] [Google Scholar]
65.Chen, K.-R. et al. Endosomes serve as signaling platforms for RIG-I ubiquitination and activation. Sci. Adv.10, eadq0660 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Ishikawa, H. & Barber, G. N. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature455, 674–678 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Pöhnl, M., Trollmann, M. F. W. & Böckmann, R. A. Nonuniversal impact of cholesterol on membranes mobility, curvature sensing and elasticity. Nat. Commun.14, 8038 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Sunshine, H. & Iruela-Arispe, M. L. Membrane lipids and cell signaling. Curr. Opin. Lipidol.28, 408–413 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Brown, M. F. Curvature forces in membrane lipid-protein interactions. Biochemistry51, 9782–9795 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Israelachvili, J. N., Mitchell, D. J. & Ninham, B. W. Theory of self-assembly of hydrocarbon amphiphiles into micelles and bilayers. J. Chem. Soc.72, 1525 (1976). [Google Scholar]
71.Grabon, A., Bankaitis, V. A. & McDermott, M. I. The interface between phosphatidylinositol transfer protein function and phosphoinositide signaling in higher eukaryotes. J. Lipid Res.60, 242–268 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Ma, X. et al. Validating an artificial organelle: Studies of lipid droplet-specific proteins on adiposome platform. iScience24, 102834 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Bozelli, J. C. et al. Membrane curvature allosterically regulates the phosphatidylinositol cycle, controlling its rate and acyl-chain composition of its lipid intermediates. J. Biol. Chem.293, 17780–17791 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Dean, J. M. & Lodhi, I. J. Structural and functional roles of ether lipids. Protein Cell9, 196–206 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Kohn, A., Gitelman, J. & Inbar, M. Interaction of polyunsaturated fatty acids with animal cells and enveloped viruses. Antimicrob. Agents Chemother.18, 962–968 (1980). [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Goc, A., Niedzwiecki, A. & Rath, M. Polyunsaturated ω-3 fatty acids inhibit ACE2-controlled SARS-CoV-2 binding and cellular entry. Sci. Rep.11, 5207 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Olarte, M.-J., Swanson, J. M. J., Walther, T. C. & Farese, R. V. Jr, The CYTOLD and ERTOLD pathways for lipid droplet-protein targeting. Trends Biochem. Sci.47, 39–51 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
78.I. Amarasinghe, et al., Cellular communication through extracellular vesicles and lipid droplets. J. Extracell. Biol. 2, (2023). [DOI] [PMC free article] [PubMed]
79.Vitrac, H., MacLean, D. M., Jayaraman, V., Bogdanov, M. & Dowhan, W. Dynamic membrane protein topological switching upon changes in phospholipid environment. Proc. Natl. Acad. Sci. USA. 112, 13874–13879 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
80.Sheppe, A. E. F. & Edelmann, M. J. Roles of eicosanoids in regulating inflammation and neutrophil migration as an innate host response to bacterial infections. Infect. Immun.89, e0009521 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
81.Miyanari, Y. et al. The lipid droplet is an important organelle for hepatitis C virus production. Nat. Cell Biol.9, 1089–1097 (2007). [DOI] [PubMed] [Google Scholar]
82.Samsa, M. M. et al. Dengue virus capsid protein usurps lipid droplets for viral particle formation. PLoS Pathog.5, e1000632 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
83.Yue, M. et al. Coronaviral ORF6 protein mediates inter-organelle contacts and modulates host cell lipid flux for virus production. EMBO J.42, e112542 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
84.Criglar, J. M., Estes, M. K. & Crawford, S. E. Rotavirus-induced lipid droplet biogenesis is critical for virus replication. Front. Physiol.13, 836870 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
85.Stoyanova, G. et al. Zika virus triggers autophagy to exploit host lipid metabolism and drive viral replication. Cell Commun. Signal.21, 114 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
86.Dias, S. S. G. et al. Metabolic reprogramming and lipid droplets are involved in Zika virus replication in neural cells. J. Neuroinflammation20, 61 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
87.Heaton, N. S. & Randall, G. Dengue virus-induced autophagy regulates lipid metabolism. Cell Host Microbe8, 422–432 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
88.Zhang, J. et al. Flaviviruses exploit the lipid droplet protein AUP1 to trigger lipophagy and drive virus production. Cell Host Microbe23, 819–831.e5 (2018). [DOI] [PubMed] [Google Scholar]
89.T. X. Jordan, G. Randall, Dengue virus activates the AMP kinase-mTOR axis to stimulate a proviral lipophagy. J. Virol. 91, (2017). [DOI] [PMC free article] [PubMed]
90.Miorin, L., Maestre, A. M., Fernandez-Sesma, A. & García-Sastre, A. Antagonism of type I interferon by flaviviruses. Biochem. Biophys. Res. Commun.492, 587–596 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
91.Diamond, M. S. Mechanisms of evasion of the type I interferon antiviral response by flaviviruses. J. Interferon Cytokine Res.29, 521–530 (2009). [DOI] [PubMed] [Google Scholar]
92.Gale, M. Jr & Foy, E. M. Evasion of intracellular host defence by hepatitis C virus. Nature436, 939–945 (2005). [DOI] [PubMed] [Google Scholar]
93.Helbig, K. J. et al. Viperin is induced following dengue virus type-2 (DENV-2) infection and has anti-viral actions requiring the C-terminal end of viperin. PLoS Negl. Trop. Dis.7, e2178 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
94.X. Zhou, B. B. Moore, Lung section staining and microscopy. Bio. Protoc.7, (2017). [DOI] [PMC free article] [PubMed]
95.Mastronarde, D. N. Automated electron microscope tomography using robust prediction of specimen movements. J. Struct. Biol.152, 36–51 (2005). [DOI] [PubMed] [Google Scholar]
96.Whelan, D. R., Holm, T., Sauer, M. & Bell, T. D. M. Focus on super-resolution imaging with direct Stochastic Optical Reconstruction Microscopy (dSTORM). Aust. J. Chem.67, 179 (2014). [Google Scholar]
97.Rebejac, J. et al. Meningeal macrophages protect against viral neuroinfection. Immunity55, 2103–2117.e10 (2022). [DOI] [PubMed] [Google Scholar]
98.Folch, J., Lees, M. & Sloane, G. H. Stanley, A simple method for the isolation and purification of total lipids from animal tissues. J. Biol. Chem.226, 497–509 (1957). [PubMed] [Google Scholar]
99.Liebisch, G. et al. Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. J. Lipid Res.61, 1539–1555 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information^{(3.5MB, pdf)}

41467_2026_74016_MOESM2_ESM.pdf^{(152.6KB, pdf)}

Description Of Additional Supplementary File

Supplementary Data 1^{(3MB, pdf)}

Supplementary Data 2^{(2.8MB, pdf)}

Supplementary Data 3^{(5MB, xlsx)}

Supplementary Data 4^{(3.5MB, pdf)}

Supplementary Data 5^{(3.1MB, pdf)}

Supplementary Data 6^{(24.2MB, xlsx)}

Supplementary Movie 1^{(31MB, mp4)}

Supplementary Movie 2^{(8.9MB, mp4)}

Supplementary Movie 3^{(33.6MB, mp4)}

Supplementary Movie 4^{(11MB, mp4)}

Reporting summary^{(7.8MB, pdf)}

Transparent Peer Review file^{(840.6KB, pdf)}

Source data^{(2MB, xlsx)}

Data Availability Statement

[CR1] 1.K. M. Crosse, E. A. Monson, A. B. Dumbrepatil, Viperin binds STING and enhances the type-I interferon response following dsDNA detection. Immunol. Cell Biol. 99, 373-391 (2021). [DOI] [PubMed]

[CR2] 2.Dumbrepatil, A. B. et al. Viperin interacts with the kinase IRAK1 and the E3 ubiquitin ligase TRAF6, coupling innate immune signaling to antiviral ribonucleotide synthesis. J. Biol. Chem.294, 6888–6898 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR3] 3.Arrese, E. L., Saudale, F. Z. & Soulages, J. L. Lipid droplets as signaling platforms linking metabolic and cellular functions. Lipid Insights7, 7–16 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.R. Dhiman, S. Caesar, A. R. Thiam, B. Schrul, Mechanisms of protein targeting to lipid droplets: A unified cell biological and biophysical perspective. Semin. Cell Dev. Biol., 10.1016/j.semcdb.2020.03.004 (2020). [DOI] [PubMed]

[CR5] 5.Thiam, A. R. & Beller, M. The why, when and how of lipid droplet diversity. J. Cell Sci.130, 315–324 (2017). [DOI] [PubMed] [Google Scholar]

[CR6] 6.Olzmann, J. A. & Carvalho, P. Dynamics and functions of lipid droplets. Nat. Rev. Mol. Cell Biol.20, 137–155 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Onal, G., Kutlu, O., Gozuacik, D. & Dokmeci Emre, S. Lipid droplets in health and disease. Lipids Health Dis.16, 128 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.E. A. Monson, A. M. Trenerry, J. L. Laws, J. M. Mackenzie, K. J. Helbig, Lipid droplets and lipid mediators in viral infection and immunity. FEMS Microbiol. Rev., 10.1093/femsre/fuaa066 (2021). [DOI] [PMC free article] [PubMed]

[CR9] 9.Zadoorian, A., Du, X. & Yang, H. Lipid droplet biogenesis and functions in health and disease. Nat. Rev. Endocrinol.19, 443–459 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Monson, E. A. et al. Intracellular lipid droplet accumulation occurs early following viral infection and is required for an efficient interferon response. Nat. Commun.12, 4303 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.M. Bosch et al, Mammalian lipid droplets are innate immune hubs integrating cell metabolism and host defense. Science370, (2020). [DOI] [PubMed]

[CR12] 12.Rösch, K. et al. Quantitative lipid droplet proteome analysis identifies annexin A3 as a cofactor for HCV particle production. Cell Rep.16, 3219–3231 (2016). [DOI] [PubMed] [Google Scholar]

[CR13] 13.Dixit, E. et al. Peroxisomes are signaling platforms for antiviral innate immunity. Cell141, 668–681 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Khantisitthiporn, O. et al. Viperin interacts with PEX19 to mediate peroxisomal augmentation of the innate antiviral response. Life Sci. Alliance4, e202000915 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.Du, M. & Chen, Z. J. DNA-induced liquid phase condensation of cGAS activates innate immune signaling. Science361, 704–709 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Álvarez, R. et al. Escribá, G protein-membrane interactions I: Gαi1 myristoyl and palmitoyl modifications in protein-lipid interactions and its implications in membrane microdomain localization. Biochim. Biophys. Acta1851, 1511–1520 (2015). [DOI] [PubMed] [Google Scholar]

[CR17] 17.Yamamoto, E., Akimoto, T., Kalli, A. C., Yasuoka, K. & Sansom, M. S. P. Dynamic interactions between a membrane binding protein and lipids induce fluctuating diffusivity. Sci. Adv.3, e1601871 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Cho, W. & Stahelin, R. V. Membrane-protein interactions in cell signaling and membrane trafficking. Annu. Rev. Biophys. Biomol. Struct.34, 119–151 (2005). [DOI] [PubMed] [Google Scholar]

[CR19] 19.D. Casares, P. V. Escribá, C. A. Rosselló, Membrane lipid composition: effect on membrane and organelle structure, function and compartmentalization and therapeutic avenues. Int. J. Mol. Sci. 20, (2019). [DOI] [PMC free article] [PubMed]

[CR20] 20.York, A. G. et al. Limiting cholesterol biosynthetic flux spontaneously engages type I IFN Signaling. Cell163, 1716–1729 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Hofer, M. J. et al. Mice deficient in STAT1 but not STAT2 or IRF9 develop a lethal CD4+ T-cell-mediated disease following infection with lymphocytic choriomeningitis virus. J. Virol.86, 6932–6946 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Bonthius, D. J., Nichols, B., Harb, H., Mahoney, J. & Karacay, B. Lymphocytic choriomeningitis virus infection of the developing brain: critical role of host age. Ann. Neurol.62, 356–374 (2007). [DOI] [PubMed] [Google Scholar]

[CR23] 23.Bersuker, K. et al. A proximity labeling strategy provides insights into the composition and dynamics of lipid droplet proteomes. Dev. Cell44, 97–112.e7 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Beilstein, F., Bouchoux, J., Rousset, M. & Demignot, S. Proteomic analysis of lipid droplets from Caco-2/TC7 enterocytes identifies novel modulators of lipid secretion. PLoS One8, e53017 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR25] 25.Saka, H. A. et al. Chlamydia trachomatis infection leads to defined alterations to the lipid droplet proteome in epithelial cells. PLoS One10, e0124630 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Kramer, D. A., Quiroga, A. D., Lian, J., Fahlman, R. P. & Lehner, R. Fasting and refeeding induces changes in the mouse hepatic lipid droplet proteome. J. Proteom.181, 213–224 (2018). [DOI] [PubMed] [Google Scholar]

[CR27] 27.Menon, D. et al. Quantitative lipid droplet proteomics reveals Mycobacterium tuberculosis induced alterations in macrophage response to infection. ACS Infect. Dis.5, 559–569 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Dahlhoff, M. et al. Characterization of the sebocyte lipid droplet proteome reveals novel potential regulators of sebaceous lipogenesis. Exp. Cell Res.332, 146–155 (2015). [DOI] [PubMed] [Google Scholar]

[CR29] 29.Khan, S. A., Wollaston-Hayden, E. E., Markowski, T. W., Higgins, L. & Mashek, D. G. Quantitative analysis of the murine lipid droplet-associated proteome during diet-induced hepatic steatosis. J. Lipid Res.56, 2260–2272 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Wang, W. et al. Proteomic analysis of murine testes lipid droplets. Sci. Rep.5, 12070 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Zhang, H. et al. Proteome of skeletal muscle lipid droplet reveals association with mitochondria and apolipoprotein a-I. J. Proteome Res.10, 4757–4768 (2011). [DOI] [PubMed] [Google Scholar]

[CR32] 32.Liu, P. et al. Chinese hamster ovary K2 cell lipid droplets appear to be metabolic organelles involved in membrane traffic. J. Biol. Chem.279, 3787–3792 (2004). [DOI] [PubMed] [Google Scholar]

[CR33] 33.Brasaemle, D. L., Dolios, G., Shapiro, L. & Wang, R. Proteomic analysis of proteins associated with lipid droplets of basal and lipolytically stimulated 3T3-L1 adipocytes. J. Biol. Chem.280, 4004 (2005). Vol. 279 (2004) 46835–46842. [DOI] [PubMed] [Google Scholar]

[CR34] 34.Fujimoto, Y. et al. Identification of major proteins in the lipid droplet-enriched fraction isolated from the human hepatocyte cell line HuH7. Biochim. Biophys. Acta Mol. Cell Res.1644, 47–59 (2004). [DOI] [PubMed] [Google Scholar]

[CR35] 35.Sato, S. et al. Proteomic profiling of lipid droplet proteins in hepatoma cell lines expressing hepatitis C virus core protein. J. Biochem.139, 921–930 (2006). [DOI] [PubMed] [Google Scholar]

[CR36] 36.Bartz, R. et al. Dynamic activity of lipid droplets: protein phosphorylation and GTP-mediated protein translocation. J. Proteome Res.6, 3256–3265 (2007). [DOI] [PubMed] [Google Scholar]

[CR37] 37.Brasaemle, D. L., Dolios, G., Shapiro, L. & Wang, R. Proteomic analysis of proteins associated with lipid droplets of basal and lipolytically stimulated 3T3-L1 adipocytes. J. Biol. Chem.279, 46835–46842 (2004). [DOI] [PubMed] [Google Scholar]

[CR38] 38.Hou, J., Liu, Y., Lu, Z., Liu, X. & Liu, J. Biochemical characterization of RNase HII from Aeropyrum pernix. Acta Biochim. Biophys. Sin.44, 339–346 (2012). [DOI] [PubMed] [Google Scholar]

[CR39] 39.Wu, C. C., Howell, K. E., Neville, M. C., Yates III, J. R. & McManaman, J. L. Proteomics reveal a link between the endoplasmic reticulum and lipid secretory mechanisms in mammary epithelial cells. Electrophoresis21, 3470–3482 (2000). [DOI] [PubMed] [Google Scholar]

[CR40] 40.Lindqvist, R. et al. Fast type I interferon response protects astrocytes from flavivirus infection and virus-induced cytopathic effects. J. Neuroinflammation13, 277 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Pfefferkorn, C. et al. Abortively infected astrocytes appear to represent the main source of interferon beta in the virus-infected brain. J. Virol.90, 2031–2038 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR42] 42.Kambara, H. et al. Negative regulation of the interferon response by an interferon-induced long non-coding RNA. Nucleic Acids Res. 42, 10668–10680 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Au-Yeung, N., Mandhana, R. & Horvath, C. M. Transcriptional regulation by STAT1 and STAT2 in the interferon JAK-STAT pathway. JAKSTAT2, e23931 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Jacobs, J. L. & Coyne, C. B. Mechanisms of MAVS regulation at the mitochondrial membrane. J. Mol. Biol.425, 5009–5019 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Katuwawala, A., Zhao, B. & Kurgan, L. DisoLipPred: accurate prediction of disordered lipid-binding residues in protein sequences with deep recurrent networks and transfer learning. Bioinformatics38, 115–124 (2021). [DOI] [PubMed] [Google Scholar]

[CR46] 46.Wang, W. & Pyle, A. M. The RIG-I receptor adopts two different conformations for distinguishing host from viral RNA ligands. Mol. Cell82, 4131–4144.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Onomoto, K., Onoguchi, K. & Yoneyama, M. Regulation of RIG-I-like receptor-mediated signaling: interaction between host and viral factors. Cell. Mol. Immunol.18, 539–555 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Omasta, B. & Tomaskova, J. Cellular lipids-hijacked victims of viruses. Viruses14, 1896 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR49] 49.Papin, M., Bouchet, A. M., Chantôme, A. & Vandier, C. Ether-lipids and cellular signaling: a differential role of alkyl- and alkenyl-ether-lipids? Biochimie215, 50–59 (2023). [DOI] [PubMed] [Google Scholar]

[CR50] 50.Kuwata, H. & Hara, S. Role of acyl-CoA synthetase ACSL4 in arachidonic acid metabolism. Prostaglandins Other Lipid Mediat144, 106363 (2019). [DOI] [PubMed] [Google Scholar]

[CR51] 51.Saliakoura, M. et al. The ACSL3-LPIAT1 signaling drives prostaglandin synthesis in non-small cell lung cancer. Oncogene39, 2948–2960 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Feng, Y. et al. Antiviral activity of eicosapentaenoic acid against Zika virus and other enveloped viruses. Front. Pharmacol.16, 1564504 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.X. Suo, et al. DHA and EPA inhibit porcine coronavirus replication by alleviating ER stress. J. Virol., e0120923 (2023). [DOI] [PMC free article] [PubMed]

[CR54] 54.Chang, C.-Y. et al. DHA attenuated Japanese Encephalitis virus infection-induced neuroinflammation and neuronal cell death in cultured rat Neuron/glia. Brain Behav. Immun.93, 194–205 (2021). [DOI] [PubMed] [Google Scholar]

[CR55] 55.Z. Telikani et al, The phospholipid composition of artificial lipid droplets enhances their deliverability and facilitates a broad Biodistribution in vivo and in vitro. Acta Biomater., 10.1016/j.actbio.2025.05.002 (2025). [DOI] [PubMed]

[CR56] 56.Farley, S. E. et al. A global lipid map reveals host dependency factors conserved across SARS-CoV-2 variants. Nat. Commun.13, 3487 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR57] 57.Chen, Q. et al. Metabolic reprogramming by Zika virus provokes inflammation in human placenta. Nat. Commun.11, 2967 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR58] 58.Leier, H. C. et al. A global lipid map defines a network essential for Zika virus replication. Nat. Commun.11, 3652 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR59] 59.Hehner, J. et al. Glycerophospholipid remodeling is critical for orthoflavivirus infection. Nat. Commun.15, 8683 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR60] 60.Poddar, D. et al. The interferon-induced protein, IFIT2, requires RNA-binding activity and neuronal expression to protect mice from intranasal vesicular stomatitis virus infection. MBio15, e0056824 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Vladimer, G. I., Górna, M. W. & Superti-Furga, G. IFITs: Emerging roles as key anti-viral proteins. Front. Immunol.5, 94 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Pichlmair, A. et al. Superti-Furga, IFIT1 is an antiviral protein that recognizes 5’-triphosphate RNA. Nat. Immunol.12, 624–630 (2011). [DOI] [PubMed] [Google Scholar]

[CR63] 63.K. M. Crosse et al, Viperin binds STING and enhances the type-I interferon response following dsDNA detection. Immunol. Cell Biol. 10.1111/imcb.12420 (2020). [DOI] [PubMed]

[CR64] 64.Saitoh, T. et al. Antiviral protein Viperin promotes Toll-like receptor 7- and Toll-like receptor 9-mediated type I interferon production in plasmacytoid dendritic cells. Immunity34, 352–363 (2011). [DOI] [PubMed] [Google Scholar]

[CR65] 65.Chen, K.-R. et al. Endosomes serve as signaling platforms for RIG-I ubiquitination and activation. Sci. Adv.10, eadq0660 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Ishikawa, H. & Barber, G. N. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature455, 674–678 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Pöhnl, M., Trollmann, M. F. W. & Böckmann, R. A. Nonuniversal impact of cholesterol on membranes mobility, curvature sensing and elasticity. Nat. Commun.14, 8038 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Sunshine, H. & Iruela-Arispe, M. L. Membrane lipids and cell signaling. Curr. Opin. Lipidol.28, 408–413 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR69] 69.Brown, M. F. Curvature forces in membrane lipid-protein interactions. Biochemistry51, 9782–9795 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR70] 70.Israelachvili, J. N., Mitchell, D. J. & Ninham, B. W. Theory of self-assembly of hydrocarbon amphiphiles into micelles and bilayers. J. Chem. Soc.72, 1525 (1976). [Google Scholar]

[CR71] 71.Grabon, A., Bankaitis, V. A. & McDermott, M. I. The interface between phosphatidylinositol transfer protein function and phosphoinositide signaling in higher eukaryotes. J. Lipid Res.60, 242–268 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR72] 72.Ma, X. et al. Validating an artificial organelle: Studies of lipid droplet-specific proteins on adiposome platform. iScience24, 102834 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Bozelli, J. C. et al. Membrane curvature allosterically regulates the phosphatidylinositol cycle, controlling its rate and acyl-chain composition of its lipid intermediates. J. Biol. Chem.293, 17780–17791 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR74] 74.Dean, J. M. & Lodhi, I. J. Structural and functional roles of ether lipids. Protein Cell9, 196–206 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR75] 75.Kohn, A., Gitelman, J. & Inbar, M. Interaction of polyunsaturated fatty acids with animal cells and enveloped viruses. Antimicrob. Agents Chemother.18, 962–968 (1980). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR76] 76.Goc, A., Niedzwiecki, A. & Rath, M. Polyunsaturated ω-3 fatty acids inhibit ACE2-controlled SARS-CoV-2 binding and cellular entry. Sci. Rep.11, 5207 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR77] 77.Olarte, M.-J., Swanson, J. M. J., Walther, T. C. & Farese, R. V. Jr, The CYTOLD and ERTOLD pathways for lipid droplet-protein targeting. Trends Biochem. Sci.47, 39–51 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR78] 78.I. Amarasinghe, et al., Cellular communication through extracellular vesicles and lipid droplets. J. Extracell. Biol. 2, (2023). [DOI] [PMC free article] [PubMed]

[CR79] 79.Vitrac, H., MacLean, D. M., Jayaraman, V., Bogdanov, M. & Dowhan, W. Dynamic membrane protein topological switching upon changes in phospholipid environment. Proc. Natl. Acad. Sci. USA. 112, 13874–13879 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR80] 80.Sheppe, A. E. F. & Edelmann, M. J. Roles of eicosanoids in regulating inflammation and neutrophil migration as an innate host response to bacterial infections. Infect. Immun.89, e0009521 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR81] 81.Miyanari, Y. et al. The lipid droplet is an important organelle for hepatitis C virus production. Nat. Cell Biol.9, 1089–1097 (2007). [DOI] [PubMed] [Google Scholar]

[CR82] 82.Samsa, M. M. et al. Dengue virus capsid protein usurps lipid droplets for viral particle formation. PLoS Pathog.5, e1000632 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR83] 83.Yue, M. et al. Coronaviral ORF6 protein mediates inter-organelle contacts and modulates host cell lipid flux for virus production. EMBO J.42, e112542 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR84] 84.Criglar, J. M., Estes, M. K. & Crawford, S. E. Rotavirus-induced lipid droplet biogenesis is critical for virus replication. Front. Physiol.13, 836870 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR85] 85.Stoyanova, G. et al. Zika virus triggers autophagy to exploit host lipid metabolism and drive viral replication. Cell Commun. Signal.21, 114 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR86] 86.Dias, S. S. G. et al. Metabolic reprogramming and lipid droplets are involved in Zika virus replication in neural cells. J. Neuroinflammation20, 61 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR87] 87.Heaton, N. S. & Randall, G. Dengue virus-induced autophagy regulates lipid metabolism. Cell Host Microbe8, 422–432 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR88] 88.Zhang, J. et al. Flaviviruses exploit the lipid droplet protein AUP1 to trigger lipophagy and drive virus production. Cell Host Microbe23, 819–831.e5 (2018). [DOI] [PubMed] [Google Scholar]

[CR89] 89.T. X. Jordan, G. Randall, Dengue virus activates the AMP kinase-mTOR axis to stimulate a proviral lipophagy. J. Virol. 91, (2017). [DOI] [PMC free article] [PubMed]

[CR90] 90.Miorin, L., Maestre, A. M., Fernandez-Sesma, A. & García-Sastre, A. Antagonism of type I interferon by flaviviruses. Biochem. Biophys. Res. Commun.492, 587–596 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR91] 91.Diamond, M. S. Mechanisms of evasion of the type I interferon antiviral response by flaviviruses. J. Interferon Cytokine Res.29, 521–530 (2009). [DOI] [PubMed] [Google Scholar]

[CR92] 92.Gale, M. Jr & Foy, E. M. Evasion of intracellular host defence by hepatitis C virus. Nature436, 939–945 (2005). [DOI] [PubMed] [Google Scholar]

[CR93] 93.Helbig, K. J. et al. Viperin is induced following dengue virus type-2 (DENV-2) infection and has anti-viral actions requiring the C-terminal end of viperin. PLoS Negl. Trop. Dis.7, e2178 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR94] 94.X. Zhou, B. B. Moore, Lung section staining and microscopy. Bio. Protoc.7, (2017). [DOI] [PMC free article] [PubMed]

[CR95] 95.Mastronarde, D. N. Automated electron microscope tomography using robust prediction of specimen movements. J. Struct. Biol.152, 36–51 (2005). [DOI] [PubMed] [Google Scholar]

[CR96] 96.Whelan, D. R., Holm, T., Sauer, M. & Bell, T. D. M. Focus on super-resolution imaging with direct Stochastic Optical Reconstruction Microscopy (dSTORM). Aust. J. Chem.67, 179 (2014). [Google Scholar]

[CR97] 97.Rebejac, J. et al. Meningeal macrophages protect against viral neuroinfection. Immunity55, 2103–2117.e10 (2022). [DOI] [PubMed] [Google Scholar]

[CR98] 98.Folch, J., Lees, M. & Sloane, G. H. Stanley, A simple method for the isolation and purification of total lipids from animal tissues. J. Biol. Chem.226, 497–509 (1957). [PubMed] [Google Scholar]

[CR99] 99.Liebisch, G. et al. Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. J. Lipid Res.61, 1539–1555 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Integrative proteomics and lipidomics reveals dual roles for lipid droplets in the host cell antiviral response

Ebony A Monson

Zahra Telikani

Jay L Laws

Abbey J Milligan

Irumi Amarasinghe

Sarah E Bamford

Ashley M Rozario

Monique L Smith

Vivian Tran

Quynh N Dinh

Nicholas Williamson

Adam Mechler

Boyin Liu

Chad J Johnson

Paul J Pigram

Michael Leeming

Shuai Nie

Jessica K Holien

Markus J Hofer

Donna R Whelan

Karla J Helbig

Abstract

Introduction

Results

Brain derived lipid droplets recruit cellular defence proteins following LCMV infection in vivo

Fig. 1. LCMV infection upregulates LDs and remodels the LD proteome in vivo.

Astrocyte lipid droplets recruit immune signalling proteins following viral infection

Fig. 2. Proteomic profiling of astrocyte RNA-LDs reveals recruitment of immune signalling proteins.

STAT proteins dynamically localise to lipid droplets following virus infection in vitro

Fig. 3. STAT proteins localise to lipid droplets.

Fig. 4. STAT protein localisation to lipid droplets is EGFR driven.

Lipid droplets interact with mitochondria to form signalosome complexes

Fig. 5. Lipid droplets interact with mitochondria to form a signalosome with RIG-1 and MAVS.

Fig. 6. Structural Modelling reveals that RIG-1 localises to lipid droplet membrane.

The lipid droplet lipidome dynamically changes following activation of antiviral early innate signalling pathways

Fig. 7. The lipidome of lipid droplets changes significantly following dsRNA stimulation in primary immortalised astrocytes.

Dual-omics correlation analysis reveals lipid–protein co-regulatory networks on lipid droplets

Fig. 8. Correlation analysis demonstrates co-regulated protein and lipid networks on lipid droplets.

Fig. 9. Long-chain fatty acids upregulated in lipid droplets during infection are antiviral.

Long and very-long chain fatty acids within lipid droplets promote an antiviral response

Discussion

Methods

Cells and culture conditions

Lymphocytic Choriomeningitis Virus (LCMV) infection of mice

In vitro viral infection, viral mimics stimulation and plasmid transfection

Immunofluorescence microscopy

Transmission electron microscopy (TEM)

Lipid droplet isolation

Dynamic light scattering (DLS) and nano particle tracking analysis (NTA) of isolated LDs

Visualisation and analysis of mCherry co-localisation to isolated lipid droplets

SMLM image acquisition and processing

Cytoplasmic lipid droplet co-localisation enumeration

Confocal image analysis

Super resolution image analysis

Western blotting

RNA extraction and real time PCR

Table 1.

Protein sample preparation for mass spectrometry

Lipid Extraction

Quantitative proteomics and functional annotation analyses

Quantitative lipidomics

Lipid Identification and functional annotation analyses

Preparation of artificial lipid droplets (aLDs)

Computational modelling

Bona fide LD protein identification

Correlation analysis between lipids and proteins

Statistical analysis and reproducibility

Reporting summary

Supplementary information

Source data

Acknowledgements

Author contributions

Peer review

Peer review information

Funding

Data availability

Competing interests

Footnotes