Peroxisome-associated Sgroppino links fat metabolism with survival after RNA virus infection in Drosophila

The fruit fly Drosophila melanogaster is a valuable model organism for the discovery and characterization of innate immune pathways, but host responses to virus infection remain incompletely understood. Here, we describe a novel player in host defense, Sgroppino (Sgp). Genetic depletion of Sgroppino causes hypersensitivity of adult flies to infections with the RNA viruses Drosophila C virus, cricket paralysis virus, and Flock House virus. Canonical antiviral immune pathways are functional in Sgroppino mutants, suggesting that Sgroppino exerts its activity via an as yet uncharacterized process. We demonstrate that Sgroppino localizes to peroxisomes, organelles involved in lipid metabolism. In accordance, Sgroppino-deficient flies show a defect in lipid metabolism, reflected by higher triglyceride levels, higher body mass, and thicker abdominal fat tissue. In addition, knock-down of Pex3, an essential peroxisome biogenesis factor, increases sensitivity to virus infection. Together, our results establish a genetic link between the peroxisomal protein Sgroppino, fat metabolism, and resistance to virus infection.

Weight measurement. Embryos were collected on apple juice-agar plates as previously described 28 . Fifty embryos were transferred to a single culture vial containing standard cornmeal-agar medium and cultured at 25 °C. Five to seven-day-old flies were collected and frozen in groups of ten individuals of a single sex. The weight of each group was determined on a precision scale and expressed as mass per individual fly.
Time to pupation assay. Fifty embryos were grown on standard cornmeal-agar medium at 25 °C, as described for weight measurement. The appearance of pupae was scored twice a day.
Quantification of triglycerides. Three pools of two flies were homogenized in 150 μL lysis buffer (1% NP-40 in PBS). Samples were heated for 5 minutes at 90 °C and allowed to cool down at room temperature; this step was repeated twice. Debris was pelleted by centrifugation at 16,000 g for 2 min, and supernatant was transferred to a new tube. Total protein concentration was quantified using the Pierce BCA protein assay kit (Thermo Scientific) on 25 μL of undiluted lysate, following the manufacturer's instructions. Triglycerides were measured with the Triglyceride Quantification kit (BioVision), following the manufacturer's instructions using a 1:40 dilution of the same lysate. Colorimetric measurements were performed at 570 nm using a Biotek Synergy 2 plate reader. All measurements were performed in triplicate, and triglyceride levels were normalized against protein levels.
Quantification of lipid peroxidation. Peroxidized lipids were quantified using the lipid peroxidation kit (K739, BioVision) following the manufacturer's instructions. Three pools of 20-40 young (2-4 days) and old (10-12 days) flies were lysed in 300 μL malondialdehyde (MDA) lysis buffer and homogenised on a QIAshredder column (QIAGEN) at 13,000 g for 10 min. Homogenates were diluted 1:4 before measurement. Total protein concentration was quantified using the Pierce BCA protein assay kit (Thermo Scientific) on 25 μL of undiluted lysate, following the manufacturer's instructions. Colorimetric measurements were performed at 532 nm using a Biotek Synergy 2 plate reader. All measurements were performed in duplicate, and lipid peroxidation levels were normalized against total protein content of the sample.
Virus and bacterial infection. Fly stocks were cleared of Wolbachia and persistent virus infections as previously described 28 . After anesthesia with CO 2 , three to five-day-old flies were inoculated with virus by intrathoraxical injection with a Nanoject II injector (Drummond), or pricked with a needle dipped in a freshly grown bacteria pellet (OD 600 = 100). Virus inocula contained 1,000 median tissue culture infectious doses (TCID 50 ) of DCV and CrPV; 14,000 TCID 50 of IIV-6; 3,000 TCID 50 of FHV; and 2,000 TCID 50 of DXV for all survival assays. An inoculum of 10,000 TCID 50 of DCV was used in experiments in which transcriptional responses were analyzed. Flies were transferred to fresh food every 3 days and survival was assessed daily. Lethality on the first day Scientific RepoRts | (2019) 9:2065 | https://doi.org/10.1038/s41598-019-38559-x was attributed to the injection procedure and excluded from the survival analysis. Unless noted otherwise, three pools of 10 to 15 flies were injected per condition with independent dilutions of virus stock.
Virus titration. Viral titers were determined by end-point dilution, as previously described 28 .
In vivo RNAi reporter assay. RNAi competency of adult flies was analyzed using a reporter assay, as described previously 21 . Briefly, three to five-day-old female flies were injected in the abdomen with a suspension containing lipofection reagent complexed with Firefly luciferase (Fluc) and Renilla luciferase (Rluc) reporter plasmids, along with Fluc specific or non-specific control (GFP) dsRNA. Fluc and Rluc activity was measured in fly homogenate, Fluc over Rluc ratios were calculated for each sample, and data are presented as fold silencing relative to the non-specific dsRNA control. qPCR analyses. DNA was isolated from flies with the QIAamp DNA blood mini kit (Qiagen) following the manufacturer's instructions and 25 ng of DNA was used as input in the qPCR to quantify IIV-6 levels. RNA was isolated from flies using Isol-RNA lysis Agent (5-Prime). cDNA synthesis was performed on 1 μg of DNase I (Ambion)-treated RNA using TaqMan  Fluorescence microscopy. For subcellular localization of Sgroppino and PMP34, 2 × 10 5 S2 cells (Invitrogen) per well were seeded in a 24-well plate. A day later, cells were transfected with 500 ng of pAc-RFP-Sgp and pAc-EGFP-PMP34, and, where applicable, 20 ng of dsRNA using Effectene transfection reagents (Qiagen) following the manufacturer's instructions. Two days post-transfection, cells were resuspended and seeded on coverslips coated with 50 μL Concavalin A (0.5 mg/μL) 32 . Two hours later, samples were fixed with 4% paraformaldehyde for 20 minutes. The cover slips were then washed in PBS and permeabilized in 0.1% Triton X-100 in PBS for 15 minutes. Nuclei were stained with Hoechst 33342 reagent (Sigma) for 5 minutes (1:15,000 dilution from a stock concentration of 10 mg/mL in PBS/0.1% Triton), washed in PBS, and mounted with Mowiol 40-88 (Omnilabo). Pictures were taken on an Olympus FV1000 confocal microscope and processed using FIJI 33 . Colocalization was analyzed using ICYsoftware (version 1.9.6.0 34 ). Briefly, the ICY spot detector 35 was used to automatically detect puncta for the GFP and RFP signal within regions of interest (ROI, hand-delimited cells). Spots were detected with wavelet scales 2 and 3 at a 100% sensitivity. ROI for WAT (Wavelet Adaptive Threshold) calculation was used to correct for variation in background signal between the regions of interest. Per region of interest the distances between the center of green and red puncta were calculated to define colocalization (distance below 4 pixels).
Colocalization was reported as percentage of RFP puncta that colocalized with GFP puncta.
Statistical analysis. Unpaired two-tailed Student's t-tests, as implemented in Graphpad Prism version 6, were used to compare differences in gene expression, viral RNA levels, and log-transformed viral titers. Survival assays were assessed using Kaplan-Meier analyses and log-rank tests, as implemented in SPSS Statistics (version 20, IBM). P-values below 0.05 were considered statistically significant.

Results
Sgroppino mutant flies are more sensitive to RNA virus infection. To identify novel genes induced upon virus infection, we previously analyzed the transcriptome of virus-infected flies, which were either mutant for the epigenetic regulator G9a or their wild-type controls, at 24 h post-infection (hpi) 21 . Components of the heat shock and Jak-Stat pathways were among the genes that were upregulated upon Drosophila C virus (DCV) infection; we analyzed their role in host defense previously 21,25 . Amongst the genes with the highest induction in G9a mutants was the uncharacterized gene CG13091, which we later named Sgroppino (Sgp) (Supplementary Table 1).
To determine whether Sgroppino is important in host defense, we reduced Sgp expression by expression of an RNAi-inducing hairpin RNA (Sgp RNAi ) under control of the ubiquitous Actin-Gal4 driver and monitored survival rates upon viral challenge. Upon infection with DCV, a positive-sense RNA virus from the Dicistroviridae family, Sgp RNAi flies exhibited lower survival rates than control flies expressing the Actin-Gal4 driver only ( Fig. 1A; mean survival = 5.3 and 7.0 days, respectively; P < 0.001). We used reverse transcription followed by quantitative PCR (RT-qPCR) to confirm that Sgp knock-down was efficient, and found that Sgp mRNA levels were reduced by 80% in male and female flies ( Supplementary Fig. 1A). Next, we sought to confirm this observation using a mutant fly line (Sgp EY06744 ) containing a P{EPgy2} transposon insertion in the 3′-untranslated region of the Sgp gene 26 (Supplementary Fig. 1B). Sgp expression was reduced by 96% in these flies compared to wildtype control flies, and the insertion affected expression levels of both isoforms ( Supplementary Fig. 1C). For the remainder of our study we used this hypomorphic mutant, which we refer to as Sgp −/− . Sgroppino-deficient flies had no obvious defects in development and exhibited a similar lifespan as wild-type flies under standard laboratory conditions ( Supplementary Fig. 1D). Upon challenge with DCV, Sgp mutant flies had a reduced survival time compared to the wild-type control (y 1 w 1 ) flies ( Fig. 1B; mean survival = 4.4 and 7.1 days, respectively; P < 0.001), confirming the phenotype of Sgp RNAi flies (Fig. 1A). We verified that mock infection did not affect survival rates as a result of injury-associated stress and mortality (Supplementary Fig. 2A). Moreover, abiotic stresses, such as heat shock and starvation did not trigger premature mortality ( Supplementary Fig. 2B,C), excluding a broad sensitivity to various stressors.
We next challenged Sgp mutants with another dicistrovirus, cricket paralysis virus (CrPV). As for DCV, infected mutant flies succumbed earlier to CrPV challenge than wild-type flies ( To test whether Sgp mutants were also more sensitive to DNA virus infection, we challenged flies with invertebrate iridescent virus 6 (IIV-6). In contrast to our observations with RNA viruses, survival rates of Sgp mutants were slightly higher, with mean survival times of 16.5 days for wild-type flies and 19.1 days for Sgp mutants ( Fig. 1F; P = 0.02). Together, our results indicate that Sgroppino mutants are hypersensitive to infections with several RNA viruses, but not a DNA virus.
Higher DCV genomic RNA replication in Sgroppino mutant flies. To determine whether hypersensitivity to virus infection was accompanied by higher viral replication, we determined infectious titers using endpoint dilution assays and viral RNA levels by RT-qPCR over a 2-day time course after challenge with the panel of viruses of Fig. 1.
Upon DCV infection, we observed a ~6-fold increase in viral titers in Sgp mutants relative to wild-type controls at 24 hpi and a less pronounced, 3-fold difference at 48 hpi, but these differences did not reach statistical significance ( Fig. 2A). Next, we measured DCV RNA levels and observed an 18-fold increase in DCV levels at 24 hpi (P < 0.001) and a 5.5-fold increase at 48 hpi in Sgp mutants relative to wild-type flies (P < 0.01, Fig. 2B). In contrast, upon CrPV infection, no difference was found for viral titers and viral RNA levels, both at 24 and 48 hpi (Fig. 2C,D). Similarly, no significant increase in viral RNA levels could be detected in Sgp −/− flies upon FHV and DXV infection over the 2 days following infection (Fig. 2E,F). Finally, we measured viral DNA levels of IIV-6, and found no significant differences between Sgp mutant flies and wild-type controls at any of the time points analyzed

RNA interference and canonical immune pathways are intact in Sgroppino-deficient flies.
To characterize the mechanism underlying the hypersensitivity of Sgp mutant flies to virus infection, we asked whether canonical antiviral defenses were functional in those mutants. RNAi is one of the main antiviral immune pathways in Drosophila 12,13 . To test RNAi functionality, we used an in vivo sensor assay 21 that is based on an RNAi-inducing hairpin RNA that silences the inhibitor of apoptosis Death-associated inhibitor of apoptosis 1, also known as thread (Diap1 RNAi ) 27,36 . When driven specifically in the eye using the GMR-Gal4 driver, expression of Diap1 RNAi triggers severe apoptosis in the developing eye, characterized by loss of pigmentation and reduced size (Fig. 3A). Argonaute 2 (AGO2) mutant flies missing the central catalytic component of the RNAi pathway do not exhibit this phenotype, demonstrating its full dependence on the RNAi pathway 27,36 . We confirmed this here, as concomitant expression of Diap1 RNAi and Ago2 RNAi hairpins did not induce the eye phenotype (Fig. 3A). Simultaneous expression of Diap1 RNAi and Sgp RNAi hairpins, however, induced a similar eye phenotype as Diap1 RNAi in control flies (Fig. 3A), suggesting that the RNAi pathway is functional in Sgroppino-deficient flies.
To confirm this observation, we used a luciferase-based RNAi sensor assay to assess RNAi efficiency in Sgp −/− mutant flies, as described previously 25,36 . Silencing efficiency was measured in fly lysates, collected 3 days after in vivo transfection with Firefly (Fluc) and Renilla luciferase reporter plasmids together with either Fluc-specific dsRNA or control dsRNA. The strong reduction of silencing activity in Dicer-2 null mutants, compared to their wild-type controls (y 1 w 1 ), demonstrates that silencing of Fluc expression was RNAi-dependent (Fig. 3B). Silencing efficiencies in Sgp −/− mutant flies and controls were similar, confirming that RNAi is fully functional in Sgp −/− mutant flies (Fig. 3B). In addition to the RNAi pathway, viral infection has been shown to activate several immune pathways 15,37 . For example, the Jak-Stat pathway controls expression of genes encoding the stress-related proteins Turandot A and M (TotA and TotM) and the infection-induced gene virus induced RNA-1 (vir-1) 14,16 . The NF-κB-related Toll and Imd pathways, regulate expression of genes encoding antimicrobial peptides, such as Drosomycin, Metchnikowin, and Diptericin, which are secreted by the fat body upon bacterial challenge and in some cases upon viral infections [17][18][19] . To test whether these signaling cascades are functional in Sgp mutant flies, we monitored expression of these downstream genes by RT-qPCR, at 24 and 48 hours after DCV infection (Fig. 3C,D).
No significant induction of Jak-Stat or NF-κB-dependent genes was detected at 24 hpi with DCV in wild-type or Sgp mutant flies (Fig. 3C). Amongst the genes analyzed, we observed the highest induction (4-fold) for vir-1 in Sgp mutants. However, for none of the genes a significant difference was observed between mutant flies and controls at 24 hpi (Fig. 3C). At 48 hpi, expression of vir-1 was strongly increased in Sgp mutants (12.7-fold), whereas it was not induced in wild-type flies (0.8-fold, P < 0.05; Fig. 3D). This is most likely due to higher DCV levels in Sgp mutants ( Fig. 2A,B). We noted that induction of these canonical Jak-Stat dependent genes was consistently lower than previously reported with the same virus dose 21 , which is most likely due to the different genetic background. For the other Jak-Stat or NF-κB regulated genes, we observed only low induction upon infection, and, more importantly, no significant difference between wild-type and Sgp mutant flies. We also verified that constitutive expression levels of Jak-Stat dependent genes were similar in wild-type and Sgp mutant flies ( Supplementary Fig. 3A). Since only low expression of NF-κB-dependent genes was expected upon systemic viral challenge (reported previously in 14,17,21,38 ), we also measured the expression of Drosomycin, Metchnikowin, Drosocin, Diptericin B, Immune induced 1, and Cecropin A2 at 6 and 24 hpi with Gram positive (Micrococcus luteus) and Gram negative bacteria (Erwinia caratovora caratovora 15, Ecc 15) (Supplementary Fig. 3B-G). Overall, no significant differences in AMP induction levels were observed between Sgp mutants and control flies, at both time points and upon both challenges. Taken together, these date indicate that canonical antiviral defense mechanisms are intact in Sgroppino mutant flies.

Sgroppino partially localizes to peroxisomes.
To obtain more insights into the function of Sgroppino, we analyzed its intracellular localization in Drosophila S2 cells. In a previously published in silico analysis, Sgroppino was predicted as one of 17 Drosophila orthologs of the human fatty acyl-CoA reductase (FAR-1) gene (although not the closest ortholog). FAR-1 transforms fatty acyl CoA into fatty alcohol within the ether lipid synthesis pathway in peroxisomes 39 . To determine whether Sgroppino localizes to peroxisomes in Drosophila, we expressed Sgp fused to Red Fluorescent Protein (RFP) at its N-terminus from an Actin promoter-driven expression plasmid. As a marker for peroxisomes, we used an expression vector encoding N-terminally GFP-tagged Peroxisomal Membrane Protein 34 (PMP34), which contains a peroxisome membrane targeting signal and six transmembrane domains and localizes in the peroxisomal membrane 40 . To control for specificity and to determine whether Sgp knock-down affects peroxisome integrity, we cotransfected dsRNA targeting Sgp, PMP34, or, as a negative control, luciferase. Upon transfection with control dsRNA, we observed punctate cytoplasmic GFP staining demonstrating that PMP34-GFP localizes to peroxisomes, as expected (Fig. 4). Strikingly, Sgroppino-RFP repeatedly localized to the similar puncta, and the merged images reveal partial colocalization of Sgroppino and PMP34. We quantified colocalization in a total of 108 cells, defined as a distance below 4 pixels between the centers of RFP and GFP-positive puncta (Fig. 4A). We found that 50,4% of the Sgroppino-RFP colocalized with peroxisomal PMP34-GFP puncta, indicating that Sgroppino localizes, at least partly, to peroxisomes. As expected, RFP or GFP signals were strongly reduced upon knock-down of Sgp and PMP34, respectively, demonstrating the specificity of the fluorescent signal (Fig. 4B). Moreover, Sgp knock-down did not affect the distribution or density of PMP34-GFP puncta, suggesting that Sgp is not required for peroxisome biogenesis or integrity. These results demonstrate that Sgp partially localizes to peroxisomes in Drosophila S2 cells.

Sgroppino mutant flies have a defect in lipid metabolism.
Peroxisomes are intracellular organelles that are important for lipid metabolism, including ether lipid biosynthesis, α-oxidation of branched chain fatty acids, and β-oxidation of fatty acids. During β-oxidation, reactive oxygen species (ROS) are generated and peroxisomes contain enzymes (oxidases and catalases) that regulate oxidative stress 39 . Thus, we sought to determine whether Sgroppino mutants had major defects in metabolism or growth. Pupation is a highly regulated process in Drosophila that depends amongst others on hormonal signaling, circadian clock, and weight 41,42 . To test how the time to pupation of Sgroppino mutants compares to wild-type flies, we analyzed the formation of pupae in vials in which the same number of embryos had been placed. However, we found no significant difference in pupae formation between wild-type and Sgp mutant flies, suggesting that Sgroppino deficiency does not impact the timing of larval growth and the transition from the larval to the pupal stage (Fig. 5A).
We next determined the weight of adult flies, and noticed a significant increase in the weight of Sgp mutants compared to wild-type flies, both in males and females (Fig. 5B). The increased weight of adults was even noticeable visually, with Sgp mutants having larger abdomens than wild-type flies. In agreement, we observed that, after removal of the digestive track and reproductive system, a larger mass of fat body tissue remained loosely attached to the abdominal carcass (as illustrated in Supplementary Fig. 4). It formed large oleaginous droplets and appeared white (which reminded us of the Sgroppino cocktail). The Drosophila fat body is a multi-functional organ involved, for instance, in the storage of fat, and the secretion of humoral immune factors and endocrine mediators 43,44 . In this organ, adipocytes store energy in the form of glycogen and triglycerides, which may be recruited in response to energy demands of the insect. It was previously demonstrated that high calorie diet leads to the storage of excess triglycerides in large droplets in the fat body 45 . Our visual observation ( Supplementary Fig. 5), together with the increased weight of Sgp mutants (Fig. 5B) prompted us to quantify triglyceride levels. We found that Sgp mutant flies contained significantly higher amounts of triglycerides than wild-type flies at steady-state levels (Fig. 5C). Upon mock infection, triglyceride levels remained significantly different, but upon challenge with DCV, triglyceride levels were similar between Spg mutants and wild-type flies at 1, 2, and 4 dpi (Fig. 5D). Lipid peroxidation is defined as the degradation of lipids through oxidation of long chain fatty acids, which occurs partly in peroxisomes 46 . Quantification of malondialdehyde (MDA), a byproduct of lipid peroxidation, is quantifiable with a colorimetric assay and can be used as a proxy for lipid peroxidation levels, and thus, peroxisomal lipid degradation function. Using this assay, we measured levels of lipid peroxidation in young (2-4 day old) and aged (10-12 day old) flies, and found a significant increase in lipid peroxidation in young Spg −/− flies compared to control flies of the same age (Fig. 5E). Together, these observations suggest that Sgp mutants have a defect in lipid metabolism which is likely related to Sgroppino function in peroxisomes.
peroxisomes are required for antiviral host defense. As Sgroppino is important for antiviral defense and localizes to peroxisomes, we asked whether these organelles are necessary for host defense. To this end, we used RNAi to reduce expression of Pex3, an essential factor for de novo peroxisome biogenesis and function 47 . As previously shown, elimination or strong reduction of the number of peroxisomes is developmentally lethal 47 . To achieve a non-lethal reduction in the number of peroxisomes, we induced ubiquitous knock-down of Pex3 using the armadillo-Gal4 driver. We challenged Pex3-deficient flies (Arm > Pex3 RNAi ) and, as control, flies expressing a GFP RNAi hairpin (Arm > GFP RNAi ) with DCV and monitored survival rates. The mean survival time of Pex3 knock-down flies was 4 days, whereas it was 7 days for Arm > GFP RNAi control flies (P < 0.001) (Fig. 6A). We performed mock infections to verify that the early mortality observed in Pex3-deficient flies was indeed due to DCV infection ( Supplementary Fig. 5). We used RT-qPCR to confirm that Pex3 expression was modestly, but significantly reduced at 24 and 48 hpi (1.8 and 2.6 fold reduction, respectively, Fig. 6B). Under these conditions, viral RNA levels were increased 8.2 and 7.5-fold in Pex3-deficient flies relative to the controls at 24 and 48 hpi, respectively (Fig. 6C). We thus conclude that peroxisomes are necessary for effective host response to DCV infection, the direct involvement of Sgp in that function remains to be established.

Discussion
Although several mechanisms for antiviral host defense have been discovered in Drosophila over the past years, our knowledge remains incomplete. Here, we propose Sgroppino as a player in the antiviral host defense. Sgroppino-deficient flies are hypersensitive to infection with a panel of single-stranded RNA viruses infection, of which DCV replicates to higher levels. Sgroppino localizes to peroxisomes, and partial depletion of Pex3, a peroxisome biogenesis factor, increases sensitivity of adult flies to DCV infection, accompanied by an increase in virus replication. Overall, our data indicate that Sgroppino participates in the host response to viral infection, conceivably through its function in lipid metabolism within peroxisomes. The mechanism by which Sgroppino affects virus infection remains to be defined. Sgroppino is one of 17 Drosophila orthologs of fatty acyl-CoA reductases that are predicted to reduce fatty acids into fatty alcohols, an intermediate in the biosynthesis of waxes and ether lipids 39 . Ether lipids are thought to contribute to a decrease in membrane fluidity, and might act as scavengers for reactive oxygen species to avoid the oxidation of other exposed membrane lipids 48 . At the subcellular level, ether lipid deficiency alters cholesterol distribution, resulting in cholesterol accumulation in endosomal and lysosomal compartments, and causing structural changes in the ER and Golgi apparatus [49][50][51] . As both cholesterol and intracellular membranous networks are exploited by viruses for their replication 2,5,52-54 , it is possible that the observed dysregulation in lipid metabolism in Sgroppino-deficient flies generates an environment that is favorable for virus replication.
Strikingly, Sgp mutant flies harbor higher levels of virus when infected with DCV, but not other single-stranded RNA viruses (CrPV and FHV), even though higher mortality rates were observed for all three viruses. Sgroppino  56 , which may explain the more pronounced phenotype and higher DCV RNA load of Sgp mutants. We cannot exclude that hypersensitivity to CrPV and FHV infection is caused by tissue-specific differences in viral replication that are below the sensitivity threshold of our assays in entire flies. Another possibility is that Sgroppino also influences tolerance to infection, which is the ability of a host to endure an infection 28,57-59 . The fat body plays an essential role in the storage and release of energy. Fatty acids are stored in the form of triglycerides along with other neutral lipids in lipid droplets of adipocytes. Sgroppino is expected to consume fatty acids after conversion into fatty acyl-CoA for the production of ether-linked lipids, one of the major metabolic pathways in peroxisomes 39 . Intriguingly, studies in yeast revealed that peroxisomes form extensive physical contacts with lipid droplets, coupling metabolic pathways of both compartments. In the absence of Pex5, which leads to peroxisomal malfunction, lipids that fail to be oxidized accumulate in the cytoplasm 60 . It is possible that reduced consumption of lipids in Sgp mutant flies trigger the accumulation of unprocessed lipid inclusions in cells, explaining the increased mass of fat tissue in the fly abdomen.
Peroxisomes are mostly studied for their metabolic functions, but were recently found to play a role in immunity. In mammals, peroxisomes have been proposed as platforms for antiviral signal transduction, as had previously been reported for mitochondria 61,62 . RIG-I like receptors (RLR) can signal via MAVS on peroxisomes to drive expression of type III interferons, which have tissue-specific functions in antiviral immunity 63 . However, we did not find obvious defects in Jak-Stat or NF-κB signaling, suggesting that the Sgroppino phenotype is not caused by defects in canonical immune pathways. A recent study showed that peroxisomes play an essential role in phagocytosis of bacteria in Drosophila and mouse macrophages 64 . As phagocytosis also contributes to virus-specific immune responses in Drosophila 38 , it is possible that defects in phagocytic processes explains the hypersensitivity of Sgp mutants to viral infection.
Human diseases linked to peroxisome dysfunction, such as Zellweger Syndrome, are rare and difficult to treat. Recent development of fly models for peroxisomal defects 39,47,65,66 offer great promise to study peroxisomal functions in metabolism, and, as our results suggest, immunity.

Data Availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. . Data represent means and s.d. of three independent pools of (A,C) 5 or (B) 15 female flies for each genotype. Student's t-tests were used to compare the differences in Pex3 or DCV RNA levels (*P < 0.05, **P < 0.01).