Drosophila DNA/RNA methyltransferase contributes to robust host defense in aging animals by regulating sphingolipid metabolism

Innate immunity (Janeway and Medzhitov, 2002) is an evolutionarily conserved host defense mechanism present throughout the plant and animal kingdoms. It is the predominant form of defense against pathogens. Invertebrates lack an adaptive immune system and are thus an excellent model to study innate defense mechanisms in isolation. The fruit fly, Drosophila melanogaster, responds to microbial infections by mounting a defense (Akira et al., 2006; Anderson, 2000; Buchon et al., 2014; Ferrandon et al., 2007; Iwasaki and Medzhitov, 2010; Lemaitre and Hoffmann, 2007; Ligoxygakis, 2013; Uvell and Engström, 2007) against the invading organisms. The first line of defense comprises the external cuticle and epithelial barriers. Once the pathogen breaches these barriers and reaches the hemocoel, it encounters systemic defenses, both humoral and cellular. The humoral response encompasses the up-regulation of the defense genes and antimicrobial peptides (AMPs) from the fat body of Drosophila, melanization and the release of reactive oxygen species, while hemocytes (Agaisse et al., 2003; Williams, 2007) lead the cellular response, by efficiently phagocytosing and encapsulating microorganisms. Defense genes thus encode proteins/RNA that function to counteract the effect of the invader and repair the damage caused. The regulation and thereby expression of defense genes is controlled by a number of well-characterized signal transduction pathways like the Toll signaling pathway, immune-deficient (IMD) pathway, c-Jun N-terminal kinases (JNK) and the JAK-STAT pathway (Agaisse et al., 2003; Delaney et al., 2006; Govind and Nehm, 2004; Kounatidis and Ligoxygakis, 2012; Lemaitre and Hoffmann, 2007; Lemaitre et al., 1996; Matova and Anderson, 2010; Schneider, 2007; Silverman et al., 2003). Extracellular ligands and/or cell surface receptors sense signatures of systemic microorganisms and this signal is transduced via the aforementioned transduction pathways to activate the Drosophila NFκBs Dorsal, Dif and Relish (Brennan and Anderson, 2004; Govind, 1999; Hetru and Hoffmann, 2009; Ip et al., 1993; Kounatidis et al., 2017; Lemaitre et al., 1995; Tanji and Ip, 2005).

The immune response in Drosophila shows complex age-dependent phenotypes (Clark et al., 2014; Zerofsky et al., 2005). In terms of the cellular response, phagocytic activity declines by 30% in 1 month old flies and this correlates with a decline in the number of hemocytes (Horn et al., 2014; Mackenzie et al., 2011). Levels of expression of many defense genes vary greatly with age (Felix et al., 2012; Zerofsky et al., 2005), suggesting age­-dependent regulation of the immune response. The overall picture is complex and suggests compensatory mechanisms to deal with infection while aging. Longevity has also been linked to immune function, with many critical signaling networks that regulate longevity such as insulin-IGF like (IIL) and TOR pathways (Grewal, 2009; Johnson et al., 2013; Kapahi et al., 2017; Partridge et al., 2011) shown to communicate with the central immune pathways for robust regulation of host defense (DeVeale et al., 2004; Kounatidis et al., 2017; Unckless et al., 2015).

In this study, we characterized the immune response in flies via perturbation of the activity of Drosophila DnMt2 (henceforth Mt2), a cryptic DNA/RNA methyltransferase (MT). Vertebrates have multiple DNA MTs, classified as DnMt1, Mt2, DnMt3a and DnMT3b, based on their activity and structural features (Basu et al., 2016; Okano et al., 1998). In contrast, Mt2 is the only MT identified in Drosophila (Tang et al., 2003). Originally, Mt2 was characterized as a DNA-MT, but recent research suggests that Mt2 might function primarily as an RNA-MT (Goll et al., 2006; Schaefer et al., 2010), with methylation enhancing tRNA stability. Mt2-null flies (Mt2^−/−) do not show overt developmental abnormalities and their lifespan is near normal under non-stressed conditions (Lin et al., 2005). Under stress, Mt2^−/−flies live shorter lives (Lin et al., 2005), with a lifespan reduction of 5–20 days for heat shock and oxidative stress, respectively. Flies grown in overcrowded conditions develop melanotic spots (Durdevic et al., 2013), suggesting disturbances in immune function. Infection studies also suggest that Mt2 plays an important role in acute immune response to Drosophila C virus (DCV) by binding to and possibly methylating viral RNA (Durdevic et al., 2013).

Here, we demonstrate that Mt2^−/− flies show an age­-dependent immune decline. The ability of adult flies to clear bacteria decreases dramatically by the fifteenth day post-eclosion. Adult hemocytes are sickle-shaped, with numbers in excess of that for a wild-type animal of the same age. The age­-dependent effects are correlated with perturbations in lipid homeostasis, suggesting that the decline may be a direct response to changes in critical lipid molecules involved in cellular homeostasis. We hypothesize that Mt2 regulates enzymes involved in lipid homeostasis and this function is essential for supporting a robust immune response as the animal ages.

Wild-type (Mt2^+/+; W1118), Mt2-null (Mt2^−/−; Dnmt2⁹⁹; Schaefer et al., 2010) and transgenic rescue Mt2-TG (w1118; pGeno>>Dnmt2-EGFP; Schaefer et al., 2008) flies were maintained on standard corn meal medium at 25°C. Mt2^−/− and Mt2-TG flies were provided by Dr Frank Lyko (DKFZ, Heidelberg, Germany) and Dr Matthias Schaefer (MFPL, Vienna, Austria) respectively. The lines were validated by measuring transcript levels in Mt2^−/− flies by genomic PCR to confirm the deletion as described by Schaefer et al. (2010) and by PCR followed by sequencing to confirm Mt2-TG flies (data not shown).

For survival assays, thirty 3 day old males from each genotype (Mt2^+/+, Mt2 ^−/− and Mt2-TG) were maintained on standard medium at 25°C. Another set of 30 flies, each pricked with 1× phosphate-buffered saline (pH 7, PBS; mock infected) or a 20 h old culture of ampicillin-resistant E. coli (DH5α), were also tested. Dead flies were removed and food vials changed every day. Surviving flies were scored for 2 weeks at both temperatures, i.e. 25°C and 29°C. Thirty flies were tested for each genotype for each condition in biological quadruplets. Kaplain–Meier and log rank (Mantel–Cox) tests were performed using GraphPad Prism 5.0 to analyze the data.

Male flies, 2, 15 and 30 days old, from each genotype (Mt2^+/+, Mt2^−/− and Mt2-TG) were pricked with E. coli and kept at 25°C for 6 h. Four live flies from each genotype were surface sterilized using 70% ethanol. Flies were air-dried and washed twice with autoclaved Milli-Q water under sterile conditions, crushed in 100 µl of LB medium and plated on ampicillin-containing agar plates. Colony count (colony-forming units, cfu) was taken and plotted in the form of a bar graph. The experiment was repeated thrice for each genotype. Results were analyzed using one-way ANOVA in GraphPad Prism 5.0.

Hemolymph was extracted as described previously (Neyen et al., 2014). In brief, 15 flies (1 and 15 day old males) from each genotype were placed on a 10 μm filter spin column (ThermoFisher, cat. no. 69705), covered with 4 mm glass beads (Zymoresearch, cat. no. S1001 Rattler™) and centrifuged for 20 min at 4°C, 10k rpm in a microcentrifuge. The extracted hemolymph was collected in 20 µl of 1× PBS solution containing 0.01% phenylthiourea, to prevent melanization of hemolymph, and counted using a Brightline hemocytometer as described elsewhere (Kacsoh and Schlenke, 2012). The experiment was repeated thrice for each genotype. The total number of hemocytes per fly was plotted and one-way ANOVA was performed in GraphPad Prism 5.0 to analyze the results.

Crystal cells were visualized by heating 30 third instar larva of the Mt2^+/+ and Mt2^−/− genotype at 60°C for 10 min. Photographs were taken using a Zeiss microscope (AxioVision) and crystal cells were counted using ImageJ software. The results were analyzed in GraphPad Prism 5.0 using Student's t-test.

Total RNA was extracted from all the samples 0 and 6 h post-infection (Direct-zol™ RNA MiniPrep, cat. no. R2050). cDNA was then synthesized from 1 μg total RNA using a high capacity cDNA synthesis kit (cat no. 4368814). Quantitative real-time PCR (qPCR; Applied Biosystems/Life Technologies) experiments were accomplished with a StepOnePlus machine (ABI) using SYBR Green (ABI, cat. no. 4368706). Relative gene expression was calculated after normalization to the control RpL32/rp49 mRNA. The primer sequences were as follows. Primers used for real-time PCR were Rp49 F: TAAGAAGCGCACAAAGCACT, R: GGGCATCAGATATTGTCCCT; eater F: TTAATTGTGGAAGTGGCTTCTGC, R: GGTTCCTCGACTACATCCCTTG; srp F: CCTTCCAGCGGATCGCATAC, R: GCTTAACGGATCGAACCAGGT; ush F: GATTGTTCCGATACCGCAGAA, R: GGGGTCTCTGGAATCGTTCAA; mt2 F: AGCCTGAGTGTAAAGGAAGTCA, R: ACAGATGAGTAAGTGCATCCGA; dipt F: GCGCGAAACGGTATAAAAAG, R: CTCGGCAAACATCACATCAT; AttD F: AAGGGAGTTTATGGAGCGGTC, R: GCTCTGGAAGAGATTGGCTTG; Drs F: CGTGAGAACCTTTTCCAATATGATG, R: TCCCAGGACCACCAGCATT. Primers used to amplify the Mt2 genomic locus were F: ATGGTATTTCGGGTCTTAGAAC, R: TTATTTTATCGTCAGCAATTTA.

Hemocytes from 1 and 15 day old adult males of the Mt2^+/+ and Mt2^−/− genotype were isolated as described above. A drop of hemolymph was allowed to settle on a silicon wafer for 30 min at room temperature. Hemocytes were then washed with 20 µl of 1× PBS; 20 µl of fixing solution (50% ethanol, 5% acetic acid and 1% paraformaldehyde) was added and the cells were kept overnight at 4°C in a clean chamber. The next day, cells were washed with 50%, 70%, 90% and 100% ethanol, air dried and imaged using Zeiss FE-SEM. Circularity index was calculated using ImageJ software (Circularity plugin). A perfect circle is indicated by a circularity value of 1.0 and as this value approaches 0, it indicates an elongated polygon.

Lipid isolation was done using a modified Folch extraction protocol (Kamat et al., 2015). Briefly, five whole adult males were crushed in 1 ml DPBS in a glass vial, 1 ml methanol was added and the mixture vortexed. Subsequently, 2 ml of chloroform (CHCl₃) was added to these samples and vortexed vigorously. The sample was then centrifuged at 2800 g for 5 min to separate the aqueous and organic phases. The organic phase (bottom) containing lipids was collected in a clean glass vial. To enrich for phospholipids, the aqueous layer was acidified using 2.5% (v/v) formic acid, and re-extracted using 2 ml CHCl₃, and the two phases were separated by centrifugation at 2800 g for 5 min. The two organic extracts were pooled and dried using N₂ gas. The sample was spotted on silica TLC plates using a glass capillary. The solvent system and plate development were as previously described (Rai et al., 2018). Briefly, the TLC was run using two different mobile phases sequentially. The first solvent was a mixture of n-hexane:diethyl ether:acetic acid (70:30:1). The first solvent was run halfway up to the top of the plate, after which the plate was air dried. The plate was then run in a solvent mixture of n-hexene:diethyl ether (59:1). The plate was dried and visualized by spraying with 10% (w/v) CuSO₄ in 8% (v/v) H₃PO₄ followed by baking in the oven above 150°C for 20 min. The plates were scanned and quantified using ImageJ software.

All lipid extractions were done as described above, with minor modifications (Kamat et al., 2015; Pathak et al., 2018). Briefly, the five whole adult males were washed with PBS (×3 times), and transferred to a glass vial using 1 ml PBS; 3 ml of 2:1 (v/v) CHCl₃:MeOH with the internal standard mix was added, and the mixture was vigorously vortexed. The two phases were separated by centrifugation at 2800 g for 5 min. The organic phase (bottom) was removed, 50 μl of formic acid was added to acidify the aqueous homogenate (to enhance extraction of phospholipids) and CHCl₃ was added to make the volume up to 4 ml. The mixture was vortexed, and separated using centrifugation as described above. The two organic extracts were pooled, and dried under a stream of N₂. The lipidome was re-solubilized in 200 μl of 2:1 (v/v) CHCl₃:MeOH, and 20 μl was used for targeted LC-MS/MS analysis. All the lipid species analyzed in this study were quantified using the multiple reaction monitoring high resolution (MRM-HR) method (Table S1) on a Sciex X500R QTOF LC-MS/MS fitted with an Exion-LC series quaternary pump. All data were acquired and analyzed using the SciexOS software. The LC separation was achieved using a Gemini 5U C18 column (Phenomenex, 5 μm, 50×4.6 mm) coupled to a Gemini guard column (Phenomenex, 4×3 mm). The LC solvents were: for positive mode, buffer A: 95:5 (v/v) H₂O:MeOH+0.1% (v/v) formic acid+10 mmol l⁻¹ ammonium formate; and buffer B: 60:35:5 (v/v) isopropanol (IPA):MeOH:H₂O+0.1% formic acid+10 mmol l⁻¹ ammonium formate; for negative mode: buffer A: 95:5 (v/v) H₂O:MeOH+0.1% ammonium hydroxide; and buffer B: 60:35:5 (v/v) IPA:MeOH:H₂O+0.1% ammonium hydroxide. All the MS/MS-based lipid estimations was performed using an electrospray ion source, with the following MS parameters: ion source: turbo spray, collision gas: medium, curtain gas: 20 l⁻¹ min⁻¹, ion spray voltage: 4500 V, temperature: 400°C. A typical LC run consisted of 55 min, with the following solvent run sequence post-injection: 0.3 ml min⁻¹ of 0% buffer B for 5 min, 0.5 ml min⁻¹ of 0% buffer B for 5 min, 0.5 ml min⁻¹ linear gradient of buffer B from 0 to 100% over 25 min, 0.5 ml min⁻¹ of 100% buffer B for 10 min, and re-equilibration with 0.5 ml min⁻¹ of 0% buffer B for 10 min. A detailed list of all the species targeted in this MRM-HR study, describing the precursor parent ion mass and adduct, and the product ion targeted can be found in Table S1B. All the endogenous lipid species were quantified by measuring the area under the curve in comparison to the respective internal standard and then normalized to the number of flies. Data are presented as means±s.e.m. of five biological replicates per group (Table S1).

Total protein was isolated from five flies per replicate per genotype. A 15 μg sample of proteome was incubated with 100 μmol l⁻¹ sphingosine-1-phosphate (S1P; S9666, Sigma) in a reaction volume of 100 μl in PBS at 37°C with constant shaking. After 30 min, the reaction was quenched with 350 μl of 2:1 (v/v) CHCl₃:MeOH, doped with 250 pmol internal standard, cis-10-heptadecenoic acid (C17:1 FFA). The mixture was vortexed, and centrifuged at 2800 g for 5 min to separate the aqueous (top) and organic (bottom) phase. The organic phase was collected and dried under a stream of N₂ gas, re-solubilized in 100 μl of 2:1 (v/v) CHCl₃:MeOH, and subjected to LC-MS analysis. A fraction of the organic extract (∼20 μl) was injected onto a Sciex X500R QTOF LC-MS/MS fitted with an Exion-LC series quaternary pump. LC separation was achieved using a Gemini 5U C18 column (Phenomenex, 5 μm, 50×4.6 mm) coupled to a Gemini guard column (Phenomenex, 4×3 mm). The LC solvents were: buffer A: 95:5 (v/v) H₂O:MeOH+0.1% ammonium hydroxide, and buffer B: 60:35:5 (v/v) IPA:MeOH: H₂O+0.1% ammonium hydroxide. A typical LC run consisted of 15 min post-injection: 0.1 ml min⁻¹ of 100% buffer A for 1.5 min, 0.5 ml min⁻¹ linear gradient to 100% buffer B over 5 min, 0.5 ml min⁻¹ of 100% buffer B for 5.5 min, and equilibration with 0.5 ml min⁻¹ of 100% buffer A for 3 min. All MS analysis was performed using an electrospray ionization source in an MS1 scan negative ion mode for product formation (free fatty acid from S1P). All MS parameters were the same as those described in the MS-based lipids profiling method above. Measuring the area under the peak and normalizing it to the internal standard quantified the product release for the lipid substrate hydrolysis assays. The substrate hydrolysis rate was corrected by subtracting the non-enzymatic rate of hydrolysis, which was obtained by using heat-denatured proteome (15 min at 95°C, followed by cooling at 4°C for 10 min, three times) as a control. Data are presented as means±s.e.m. of three biological replicates.

Earlier reports on Mt2^−/− flies indicated that they are sensitive to stress (Schaefer et al., 2010) and are susceptible to viral infection (Durdevic et al., 2013). In our study, we tested whether E. coli. infection had an effect on Mt2^−/− fly lifespan. Mt2^+/+, Mt2^−/− and Mt2-TG lines (see Materials and Methods) were either infected with E. coli or mock infected with sterile 1× PBS. Mock-infected flies, as a result of trauma caused by injury, showed a shorter lifespan than flies without injury. The mutants infected with E. coli had a significantly shorter lifespan than the mock-infected mutants. Infected Mt2^−/− flies, when compared with Mt2^+/+ flies, had an increased rate of lethality. In contrast, Mt2-TG animals showed a near-normal lifespan for both mock infection and infection experiments, suggesting a role for Mt2 in host defense against gram-negative bacteria. In order to get a more detailed picture of the role of Mt2 in the innate immune response, we tested the functionality of both the cellular and humoral arms of the immune response using bacterial clearance assays and by measuring changes in transcript levels of defense genes in Mt2^−/− flies before and after infection, as described in following sections.

We infected 2 day old adult Mt2^+/+, Mt2^−/− and Mt2-TG flies with a saturated, ampicillin (Amp)-resistant culture of E. coli. Six hours post-infection, the animals were crushed and processed, as described in Materials and Methods, to measure the decrease in E. coli numbers as a consequence of clearance by a robust immune response. The Mt2^−/− flies were an order of magnitude less efficient (Fig. 1B, compare 6 h Mt2^+/+with Mt2^−/−) in clearing the infection than the wild-type or the rescue line (Fig. 1B). The 2 day old Mt2^−/− flies are thus impaired in their ability to clear bacteria, suggesting that Mt2 activity supports host defense against bacteria.

Next, we performed age­-dependent analysis for the ability of adult flies to clear infection. Surprisingly, we found that Mt2^−/− flies showed a significant age-dependent loss in their ability to clear bacterial infection as compared with wild-type flies, an ability regained by replacing Mt2, as in the MT2-TG flies (Fig. 1C). Wild-type flies did not show significant loss in their ability to clear infections over 30 days. In stark contrast, 15 day Mt2^−/− flies cleared bacteria 8- to 10-fold less efficiently; 30 day old flies showed a similar deficiency, suggesting that there is a steep decline in the ability to clear infection from day 2 to day 15.

The earliest difference between Mt2^+/+ and Mt2^−/− we could find in the cellular response was in the third instar larvae. We found that crystal cells, which are platelet-like cells involved in melanization, were higher in number in Mt2^−/− animals than in Mt2^+/+ flies (Fig. S1A). This indicated that number of blood cells was not as well regulated in the mutant. These data led us to look closely at the number of hemocytes in adults as they age (Fig. 2A). The 15 day old wild-type and Mt2-TG animals had fewer hemocytes compared with the corresponding 1 day old flies. In contrast, the number of hemocytes significantly increased in Mt2^−/− flies with age. This would indicate that the increase in hemocytes with age is a Mt2^−/− fly-specific event. While counting the hemocytes using light microscopy, we also noticed that the hemocytes in Mt2^−/− animals had an ellipsoid, rice grain-like shape as compared with the circular shape of the wild-type hemocytes. To get a clearer picture, we employed SEM (see Materials and Methods). When compared with round wild-type hemocytes, the Mt2^−/− hemocytes appeared flat, folded and C-shaped (Fig. 2B), reminiscent of diseased sickle-shaped human red blood cells. Quantification of the roundness index of the SEM image data indicated a dramatic change in shape of the hemocytes in Mt2^−/− animals (Fig. 2B,C). This change in cell shape could account for the inefficiency of the Mt2^−/− hemocytes in clearing the bacterial load in the animal. Based on the above results, we tested transcript levels of serpent (srp), a gene involved in regulating hemocyte morphology and phagocytic function (Petersen et al., 1999; Rämet et al., 2002; Shlyakhover et al., 2018), in 15 day old flies. srp showed a reduction in transcript levels in Mt2^−/− flies (Fig. 2D). This suggests that Mt2 regulates srp/Srp expression directly or indirectly, affecting the cellular arm of immunity. We then measured transcript levels for eater (Kroeger et al., 2012) and u-shaped (Muratoglu et al., 2007), in 15 day old animals, genes known to be critical for hemocyte phagocytosis and hemocyte cell proliferation, respectively. We find that the levels of these transcripts were significantly lower in Mt2^−/− flies as opposed to wild-type and rescue flies (Fig. 2E), again indicating a decline in the ability of flies to mount an effective cellular transcriptional response to infection. The above data strongly suggests that Mt2 plays a key role in the maintenance of a healthy immune response in older flies via transcription of genes involved in the cellular arm of fly immunity. This Mt2 function appears to become more critical as the fly ages. We next tested the transcriptional levels of genes that code for the AMPs Diptericin (Dipt), Attacin D (AttD) and Drosomycin (Drs). These genes are activated by Toll/NFκB or IMD/NFκB signaling and serve as readout for these pathways.

qPCR data were used to measure whole-animal transcript levels of Dipt, AttD and Drs for day 2–3 and day 15 post-eclosion. For this experiment, males of the correct age were infected with E. coli and transcript levels were measured at 0 and 6 h post-infection. Wild-type flies 2 days post eclosion showed a 275-fold, 75-fold and 100-fold increase in transcript levels for Dipt, AttD and Drs, respectively, post-infection, in contrast to an 800-fold, 140-fold and 175-fold increase, respectively, for Mt2^−/− flies (Fig. 3A). For Mt2-TG flies, the transcript levels were similar to those for Mt2^+/+ flies. This suggests that, in younger (2 day old) Mt2^−/− animals, the humoral immune response is robust and may be stronger than that in wild-type flies. For 15 day old Mt2^−/− flies, transcript levels of all three genes were minimally responsive to infection (Fig. 3B), indicating a breakdown in signaling or lack of transcription by the NFκBs Dorsal, Dif and Relish.

The altered shape of hemocytes at day 15 led us to profile the lipid content of Mt2^−/− animals 2–15 days post-eclosion. A triacylglycerol (or triglyceride; TAG)-specific TLC analysis of the total adult fly lipidome from 2–15 day old flies showed a significant decrease in triglycerides in Mt2^−/− animals. There appeared to be a 30% decrease in TAG levels based on quantification of TLC bands from day 1 to day 15 (Fig. 4A). MS-based quantitative lipidomics was then used to measure changes in the total lipidome for 15 day old flies (Fig. 4B; Fig. S1B). We found that the immunomodulatory sphingolipids, S1P and ceramides accumulated >2-fold in Mt2^−/− flies as compared with their WT counterparts. Concomitantly, the downstream products of sphingolipid metabolism (Fig. 5A), TAGs and phosphoethanolamine (PE) showed a ∼25% decrease in Mt2^−/− flies (Fig. S1B). The levels of lipids in the Mt2-TG rescue line were comparable to wild-type. We found that levels of several other lipid classes including neutral lipids, phospholipids (except PE), sphingomyelins and sterols remained unchanged, indicating a specific role for Mt2 in the regulation of sphingolipid metabolism (Acharya and Acharya, 2005; Kraut, 2011; Saba and Hla, 2004), especially those important in immune signaling (Rivera et al., 2008). Next, we checked whether, as in case of immune regulation, Mt2 also regulates lipid homeostasis in an age-dependent manner. Indeed, Mt2^−/− flies showed comparable levels of S1P till day 3 post-eclosion but, by day 5, S1P had started to accumulate in these mutants as compared with controls (Fig. 4C). This accumulation was more profound as the fly aged (Fig. 4C). This accumulation of S1P led us to probe whether the enzyme Sply, which converts S1P to PE (Fig. 5A), is affected. We observed a direct correlation between S1P accumulation and the failure of Mt2^−/− flies to increase Sply activity with age as compared with controls (Fig. 4D).

Organisms have to manage energy in order to survive. Energy homeostasis is dependent on energy uptake, storage and expenditure. As feeding is a discontinuous process, energy is usually stored in the form of carbohydrates, proteins or lipids to maintain a continuous supply in times of need. The Drosophila fat body, oenocytes, gut, Malphigian tubules and special regions of the nervous system play key roles in metabolic regulation and energy homeostasis. Metabolic pathways are conserved between mammals and the fly, enabling the use of Drosophila as a powerful model system to gain a better understanding of the functioning of complex metabolic networks (Owusu-Ansah and Perrimon, 2014; Padmanabha and Baker, 2014; Rajan and Perrimon, 2013; Schlegel and Stainier, 2007), including those of lipids. A finely tuned network of regulators and inter-organ communication is necessary to balance the intake, storage and expenditure of energy, whereby a deregulation of such networks can cause malfunction and disease.

In addition to being storage molecules and playing structural roles in membranes, lipids have increasingly been shown to have roles in signaling. Lipids and the enzymes that modify and interconvert them constitute complex lipid signaling networks responsible for cellular and organismal homeostasis (Owusu-Ansah and Perrimon, 2014; Palm et al., 2012) (Fig. 5A summarizes Drosophila sphingolipid metabolic pathways). In sphingolipid metabolism, levels of storage metabolites such as S1P, ceramides and TAG have to be maintained in a dynamic manner for cellular homeostasis. Drosophila mutants have contributed to insights into critical roles for sphingolipids in biological function. For example, mutants for sphingosine kinases (Sphk), which generate the important intracellular and intercellular signaling molecule S1P, and S1P-lyase (Sply) (Lovric et al., 2017), which breaks S1P down, have interesting developmental defects. Sply mutants show severe flight muscle defects as well as activation of apoptosis in reproductive organs (Herr et al., 2003; Phan et al., 2007), presumably by accumulating S1P. Sphk mutants have reduced S1P and accumulate sphingosine; Sphk2 mutants show flight defects and reduced fecundity (Herr et al., 2003). Sply phenotypes can be rescued by mutations in lace, which codes for a serine palmitoyl transferase that is a critical rate-limiting step for ceramide synthesis. Ceramides act as regulators of apoptosis and have also been shown to directly affect phosphorylation of retinoblastoma (Rb) in response to TNFα signaling (Lee et al., 1996). S1P, in a mammalian context, functions via GPCRs and is suggested to regulate events such as cell shape change in PC12 cells (Edsall et al., 2001). The above studies on enzymes involved in different steps of sphingolipid metabolism confirm critical roles for the maintenance of balance between the different sphingolipid moieties and deleterious effects upon perturbation of this homeostatic balance.

We found that Mt2^−/− mutants were unable to deal with infections as they aged. As early as 15 days post-eclosion, mutant flies were severely compromised in terms of their ability to clear infection, with plasmatocytes being of a disproportionately high number but defective in shape. This finding parallels an imbalance in lipid homeostasis. Quantitative lipidomics confirmed that S1P levels were 4-fold higher than in controls, though sphingosine levels were normal. This would suggest, based on our current understanding of S1P regulation, that Sply activity is reduced. This was confirmed by enzyme activity assays in fly lysates that showed reduction of Sply activity (Fig. 4D). This reduction in activity does not appear to be the result of lower transcript levels as sply mRNA levels did not decrease significantly (data not shown). The phenotypes could be due to errors in translation resulting from tRNA methylation defects reported earlier in Mt2 mutants (Schaefer et al., 2010). Alternatively, sply could be regulated in a tissue/immune-specific manner in flies in a way similar to that in C. elegans, where expression of S1P lyase is regulated by GATAA-like transcription factors and is limited to the gut (Oskouian et al., 2005). In Drosophila, Srp is one of the GATAA-like transcription factors known to regulate aldehyde dehydrogenase (Abel et al., 1993) and immune-specific genes in a tissue-specific manner (Petersen et al., 1999; Senger et al., 2006). It would be interesting to see whether there is any regulatory link between Srp and Sply and whether Mt2 plays a key role in this communication.

The lipidomics data also indicate that ceramide levels are higher while neutral lipids are reduced, suggesting more than one link in lipid metabolism affected in Mt2^−/− mutants. The 3-fold increase in ceramide levels suggests either a backflow from sphingosine, which is maintained at normal levels, or increased activity of enzymes that metabolize ceramide. Curiously, TAG levels are low, which may suggest that the conversion of ceramide to TAG via diacylglycerol (DAG) is overactive in order to compensate for the low TAG levels. The decreased TAG levels suggest either a need for energy in the animal or a malfunction of enzymes (Fig. 5A) maintaining homeostatic levels of TAG.

The defective sickle-shaped hemocyte morphology (Fig. 2B,C) suggests architectural problems in maintaining the shape of the cell, with lipid homeostasis being a prime candidate. As sphingholipids are critical for membrane architecture (Adada et al., 2015; Kraft, 2016), the aberrant morphology and subsequent inability to function as macrophages may be a consequence of a reduction of sphingolipids. Mutations in S1P lyase have been implicated in the regulation of cell shape (Adada et al., 2015), with our data suggesting that its malfunction is a specific cause of the sickle morphology. The correlation between an imbalance in lipid homeostasis and host defense is a less explored area of research. It is understood that with environmental or nutrient stress, accumulation of lipids or signaling intermediates can interfere with immune regulation (Ertunc and Hotamisligil, 2016). Sphingolipid imbalance has been specifically linked to a number of studies (Bandhuvula and Saba, 2007; Bektas et al., 2010; Park et al., 2013; Rivera et al., 2008; Vijayan et al., 2017; Weber et al., 2009), but universal mechanisms are lacking.

Our study puts the spotlight on age-dependent regulation of lipid homeostasis and immune function. Mt2 activity, either through regulation of the transcription of critical genes or by regulation of the translation of protein products, is important for a robust immune response in the aging animal (Fig. 5B). The absence of Mt2 function triggers an age-dependent decline in both the cellular and humoral arms of the immune response. The mechanism that Mt2 utilizes for such a systemic regulation is unclear because of the uncertainties related to Mt2 function in Drosophila. Mt2 function has a history of dispute (Krauss and Reuter, 2011; Schaefer and Lyko, 2010; Yoder and Bestor, 1996) over its importance in the growth and development of the organism and also its molecular function. Low levels (0.1–0.6%) of genomic 5-methylcytosine (5mC) have been detected in Drosophila (Capuano et al., 2014; Panikar et al., 2015; Takayama et al., 2014), with dynamic, developmental stage-specific alteration in methylation patterns in Mt2-null animals (Panikar et al., 2018; Takayama et al., 2014). Under normal conditions, complete knockdown of Mt2 has no visible survival defects, not only in flies but also in rat and plant models (Goll et al., 2006). This led to a belief that Mt2 is not a vital gene for the organism. We, along with others, show that Mt2 is required for increased lifespan under stress conditions. Here, in addition, we propose a novel function for Mt2 in regulating a steady increase in Sply activity, a phenomenon essential to keep S1P levels in check as the fly ages. In the absence of Mt2 function, this regulatory mechanism is lost, and S1P starts to accumulate with age, leading to adverse effects on the ability of the fly to deal with infection. Our study thus uncovers a novel and unexpected relationship between Mt2-mediated activity, age-associated lipid homeostasis and the robust nature of the immune response.

Dr Ajeet Singh is thanked for access to the LC-MS facility at CAMS, NCL Venture Center, India. Vijay Vittal is thanked for technical assistance with SEM data collection. D.D. is ISRO Chair Professor at Savitribai Phule Pune University.