Loss of Stearoyl-CoA desaturase 1 leads to cardiac dysfunction and lipotoxicity

Over time, chronic consumption of excess dietary nutrients seems to exceed the maximum adipose storage capacity (reviewed in Virtue and Vidal-Puig, 2010). Beyond this threshold, it is hypothesized that excess lipids enter and become toxic in non-traditional storage tissues, including the heart (reviewed in Schaffer, 2016; Unger and Scherer, 2010). This toxic overflow of lipid metabolites from storage tissue resulting in ectopic deposition of lipids in peripheral organs is termed lipotoxicity, and is thought to have a causal role in the onset of co-morbidities associated with obesity and type 2 diabetes (Unger and Scherer, 2010). The nature of the lipids responsible for lipotoxicity is not well understood, but may comprise triglycerides (TG), diglycerides (DG), ceramides, free fatty acids and other lipid species. In addition to the diet, age also seems to exacerbate lipotoxicity, and the severity of pathophysiology associated with lipotoxicity is exacerbated in older humans (Drosatos, 2016), rats (Guo et al., 2007), mice (Koonen et al., 2007), Drosophila (Na et al., 2013; Sujkowski et al., 2012) and Caenorhabditis elegans (Copes et al., 2015).

This ‘lipotoxic’ process of abnormal lipid metabolism has been recapitulated in mammalian models ranging from adipocyte, hepatocyte and cardiomyocyte cell lines to the rodent heart, adipose and liver (Drosatos, 2016; Gaemers et al., 2011; Guo et al., 2007; Ogawa et al., 2018). Mammalian Chinese hamster ovary cells supplemented with the saturated fatty acid palmitate exhibit increased lipotoxicity and cell death, while the monounsaturated fatty acid (MUFA) oleate (also called oleic acid, OA) reduces lipotoxicity-associated apoptosis (Listenberger et al., 2003) consistent with a potential role for fatty acid saturation in this process. Culturing hepatocytes with palmitate also increases lipotoxicity-associated cell death and oxidative stress in a citric acid cycle- and calcium-dependent manner (Egnatchik et al., 2019). Incubating cardiomyocytes with palmitate increases the expression of stress genes, a response that could be alleviated by increasing fatty acid acylation or esterification as TG (Bosma et al., 2014). In 3T3-L1 differentiated adipocytes, inhibiting TG synthesis increases lipotoxicity-associated stress gene expression and fatty acid secretion (Chitraju et al., 2017), suggesting that TG can be protective against lipotoxicity. Mechanisms of lipotoxicity have also been explored in vivo in the mammalian heart. Mice overexpressing long chain acyl-CoA synthase (ACSL1) accumulate cardiac TG and exhibit cardiomyopathy, including increased heart failure (Chiu et al., 2001), whereas mice lacking ACSL1 show improved cardiac function, suggesting that decreasing TG or acyl-CoA pools can ameliorate cardiac lipotoxicity in these animals (Pascual et al., 2020 preprint). Cardiac lipotoxicity can also be improved by genetically decreasing DG, ceramide and free fatty acids in the heart, even when TG are increased (Liu et al., 2009). These and other studies lead to a model where lipotoxicity is a complex phenomenon that depends on fatty acid composition, esterification and location to determine the degree of pathophysiology that occurs in the heart (reviewed in Sletten et al., 2018).

To date, limited studies have been conducted using Drosophila as a model for lipotoxicity, with most focusing on metabolic syndrome-like phenotypes using diet or genetic manipulations (Birse et al., 2010; Diop et al., 2015; Guida et al., 2019; Hardy et al., 2015). In the fly, the expression of the lipogenic transcription factor (TF) stearoyl regulatory element binding protein (SREBP), the TF cofactor peroxisome proliferator-activated receptor-γ coactivator)/spargel (PGC-1) and downregulation of the insulin–target of rapamycin (TOR) pathway have been implicated in protection against cardiac lipotoxicity (Birse and Bodmer, 2011; Birse et al., 2010; Diop et al., 2015). As in mammals, sphingolipid metabolism plays a role in Drosophila cardiac lipotoxicity (Walls et al., 2018, 2020) and both mitochondrial and peroxisomal fatty acid metabolism are essential to protect against lipotoxicity (Bülow et al., 2018; Sellin et al., 2018). Loss of function and biochemical studies also indicate that SREBP, carbohydrate response element binding protein (ChREBP), TG and coenzyme-A are protective against lipotoxicity in the fly fat body (Lim et al., 2011; Musselman et al., 2013; Palanker Musselman et al., 2016). Taken together, many similarities in physiology and lipid biochemistry exist between flies and mammals, suggesting there are conserved mechanisms that regulate lipotoxicity.

The expression of the mammalian stearoyl-CoA desaturase (SCD) genes is repressed by the satiety hormone leptin (Cohen et al., 2002) and is activated by insulin (Waters and Ntambi, 1994), dietary cholesterol (Garg et al., 1988), saturated fat (Sampath et al., 2007) or sugar (Miyazaki et al., 2004), and TFs including SREBP (Tabor et al., 1998, 1999), ChREBP (Jeong et al., 2011), nuclear transcription factor-Y (NF-Y) (Yao et al., 2016), liver X receptor (LXR) (Zhang et al., 2014) and the cofactor PGC-1 (Lin et al., 2005). The stearoyl-CoA desaturase or ‘Desat’ class of lipid metabolic enzymes has not previously been shown to function in cardiac lipotoxicity but plays an important role in lipogenesis and is essential for the flexibility and utility of acyl chain substituents in phospholipids, TG and other complex lipids. The SCD1 gene encodes desaturase 1 or Desat1, which introduces a double bond and therefore desaturates palmitoyl-CoA and stearoyl-CoA to form palmitoleoyl-CoA and oleoyl-CoA, respectively (Paton and Ntambi, 2009). Global knockout of SCD1 produces reduced body size and adiposity (Ntambi et al., 2002); moreover, SCD1 is required for monounsaturated C18:1 fatty acid incorporation into TG (Attie et al., 2002). Interestingly, knockout of SCD1 in genetically obese ob/ob mice exacerbated the increased plasma glucose and insulin resistance typically seen in these mice (Flowers et al., 2007). These studies highlight the importance of SCD1 and its role in lipogenesis and overall metabolic homeostasis.

Loss of desat1, the Drosophila homolog of mammalian SCD1, affects adiposity and fatty acid saturation in flies, similar to the effects shown in SCD1 knockout or conditional knockout mice. Reducing Desat1 in Drosophila leads to increased hyperglycemia, decreased organism size, delayed developmental progression and reduced lipid storage and insulin (Musselman et al., 2013; Parisi et al., 2013; Wang et al., 2016). Desat1 is also required for cuticular differentiation (Wang et al., 2016) as well as cuticular sex pheromone production (Marcillac et al., 2005) and mating behavior (Krupp et al., 2008). As in mammals, no role for Desat1 has been described in fly cardiac lipotoxicity.

In this study, we characterized the effects of desat1 loss of function in the fly fat body, analogous to both mammalian liver and adipose tissue in function and structure, on cardiac function, tissue-specific lipidomes, feeding and gut function. We found decreases in heart function and increased concentrations of potential lipotoxins in hearts from desat1 RNAi flies challenged with a high-sucrose diet (HSD). Moreover, we found desat1 RNAi flies aged on a HSD exhibited decreased feeding and increased excretion compared with the control genotype. Feeding the monounsaturated fatty acid OA partially recovered diabetes-like phenotypes. Taken together, our research presents a crucial role for Desat1 in cardiac function and lipid homeostasis.

All fly stocks were maintained on 5% dextrose–cornmeal–yeast–agar medium at 25°C with controlled humidity. White-eyed w¹¹¹⁸ control and UAS-desat1 RNAi (line # 33338) fly lines obtained from the Vienna Drosophila Resource Center (VDRC) were used for all experiments and maintained with 12 h/12 h day/night light cycling. A UAS-D2;r4-Gal4 line made from Lee and Park's r4-GAL4 driver (Lee et al., 2002) was used to drive expression of desat1 RNAi or as a trans-heterozygous control (r4,w¹¹¹⁸) for the genetic background of these flies. Females were used for all experiments because of high lethality in males. Adult female flies were segregated from males 24 h post-eclosion, giving the opportunity for copulation, and placed in vials of 20–30 flies. Subjects for all experiments were reared on 0.5 mol l⁻¹ sucrose media, then transferred to 0.15 mol l⁻¹ sucrose (5% sucrose, the control diet) or 1 mol l⁻¹ sucrose (34% sucrose; HSD) food within 24 h of eclosion. Aging flies were checked daily in order to preserve food quality. Control diet food consistency decreased at a quicker rate than HSD food so was this changed more frequently. Larvae used for supplementation studies were fed 0.5 mol l⁻¹ sucrose semi-defined media with or without OA (A16663, Alfa Aesar, Haverhill, MA, USA). Because the parental generation laying eggs were susceptible to being stuck in food supplemented with OA, eggs were collected and transferred to semi-defined media. Briefly, parental generation flies were allowed to mate for 24 h in a separate vial. Mated female flies were then transferred to bottles and allowed to lay eggs on 25% grape juice agar supplemented with yeast paste. Bottles each contained 10 females for egg laying. After 24 h, eggs were harvested, washed and deposited on fresh 0.5 mol l⁻¹ semi-defined sucrose media and allowed to develop. Larvae were allowed to develop to the wandering 3rd instar larval stage (wL3), then females were collected. Five larvae per replicate were used, with 8 biological replicates per dataset.

For adult TG assays, flies were aged for 3 weeks, weighed and frozen at −80°C prior to homogenization. Five flies per replicate with eight biological replicates were used for each assay. Flies were homogenized in PBS with 0.1% Tween-20 using electronic mortar and pestle and incubated at 65°C for 5 min to inactivate lipases. TG were analyzed using Infinity Triglyceride reagent (TR22421, Thermo Scientific, Waltham, MA, USA) and a VersaMax (Molecular Devices, San Jose, CA, USA) spectrophotometer at 540 nm optical density (OD) after a 5 min incubation at 37°C. Wandering L3 larvae for OA studies were collected from vials, washed and weighed prior to TG assay. Because OA is viscous, the parental generation died quickly; therefore, eggs were transferred from grape juice agar egg laying plates before hatching to 0.15 mol l⁻¹ or 0.5 mol l⁻¹ sucrose food with or without OA. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests or a one-way ANOVA was performed with Tukey's multiple comparison test to assess statistical significance between groups.

Larvae for hemolymph were collected at the wandering third instar stage. Larvae were washed, pierced with forceps and placed in a 0.5 ml centrifuge tube cut at the bottom with a razor and nested in a 1.5 ml centrifuge tube. Tubes were briefly spun, and hemolymph was extracted from 11 larvae per replicate. Infinity Glucose reagent (TR15421, Thermo Scientific) and a VersaMax (Molecular Devices) spectrophotometer at 340 nm OD after a 15 min incubation at 37°C. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests or one-way ANOVA were performed with Tukey's multiple comparison test to assess statistical significance between groups.

Fat bodies were dissected from 3 day old r4,w¹¹¹⁸ and r4>desat1 RNAi flies in PBS. Fifteen fat bodies were used to produce each RNA sample. Samples were homogenized and RNA extracted using Ribozol (97064-950, VWR, Radnor, PA, USA) reagent. Samples were treated with DNase (PIER89836, VWR) for 30 min prior to cleanup over a Qiagen RNeasy purification column (74104, Qiagen, Germantown, MD, USA). The total amount of RNA for each sample was calculated using a Qubit 2.0 fluorometer (Life Technologies/ThermoFisher Scientific, Waltham, MA, USA) and a Life Technologies Molecular Probes Qubit RNA BR Assay Kit (10210, Life Technologies/ThermoFisher Scientific). Reverse transcription was achieved using Bio-Rad iScript reagent (1708890, Bio-Rad, Hercules, CA, USA) and an Axygen Maxygene II bench top thermocycler (Axygen/Corning, Corning, NY, USA). Quantitative PCR (qPCR) was conducted using a Bio-Rad CFX Connect Real-Time System and SsoAdvanced SYBR-Green qPCR reagent (172-5272, Bio-Rad). A two-tailed Student's t-test was performed using Prism v.9.0 software to assess statistical significance between groups. Ribosomal Protein 49 coding mRNA (rp49) served as the internal control. Primers for rp49 were rp49-F: GCACTCTGTTGTCGATACCC; rp49-R: CAGCATACAGGCCCAAGAT. The primers that were generated for desat1 were:

desat1-7136F: 5′-TACTCGCTAAACTTCACTACCGC-3′;

desat1-7313R: 5′-CATCCTCAATCTCCTCCTTGGG-3′.

Flies were aged for 3 weeks on a HSD and assessed for cardiac failure and recovery; food was changed routinely to prevent fly death and preserve food freshness. Flies were first subjected to a cardiac pacing assay (Paternostro et al., 2001; Wessells and Bodmer, 2004). In short, each replicate consisted of five flies that were anesthetized with FlyNap (trimethylamine; Carolina Biological Supply, Burlington, NC, USA), placed on a slide between two electrical contacts with conductive gel on either end of the fly's head and abdomen, with the head oriented towards the positive electrode. Then, 50 V was passed through the fly at 30 Hz for 30 s using a square wave stimulator (Phipps & Bird, Richmond, VA, USA). Flies were visually assessed for a rhythmic heartbeat at 30 and 120 s post-stimulation. No heartbeat after 30 s was considered a failure. Those hearts that failed were visually assessed for the ability to recover post-failure with a rhythmic heartbeat at 120 s. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). A contingency table analysis was performed using Fisher's exact test to assess statistical significance between groups.

Adult females were aged for 3 weeks on a HSD with flies being flipped routinely to preserve food freshness, anesthetized with FlyNap, and dissected in artificial hemolymph (Fink et al., 2009). A Lumenera LT225 camera (Teledyne-Lumenera, Nepean, ON, Canada) attached to a Zeiss inverted microscope was used to collect high-resolution videos. Five, 10 s long videos were taken per fly and analyzed; 12 replicates per genotype were used. The SOHA software (Fink et al., 2009) measured heart parameters associated with cardiac dysfunction: heartbeat, period, diastolic and systolic diameter, fractional shortening, arrhythmia, and both diastolic and systolic heart diameter. Each 10 s technical replicate per biological fly replicate were averaged and used for statistics. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests were performed to assess statistical significance between groups.

Adult females aged 3 weeks were filleted and the organs removed, leaving the heart and fat body. Abdomens were then removed and fixed in a 4% paraformaldehyde/PBS solution for 30 min and washed 3 times in PBS/0.1% Tween-20. Abdomens were then incubated with goat anti-pericardin (PC) (EC11, Developmental Studies Hybridoma Bank, Iowa City, Iowa, USA) rotating for 24 h at 4°C. Samples were then washed 3×15 min in PAXDG (PBS, 1% BSA,0.3% Triton X-100, 0.5% sodium deoxycholate and 5% normal goat serum, NGS). Following washing, samples were stained with Alexa Fluor 488 phalloidin actin stain (1:250) (A12379m Invitrogen) and F(ab)2 fragment, CF633 conjugated mouse anti-goat (1:500) (20130, Biotium, Fremont, CA, USA) with subsequent washing 3×15 min in PAXDG. All antibodies were diluted in PAXDG. Pelts were placed on a coverslip with Vecta Shield with DAPI (405 nm) (H-1200, Vector, Burlingame, CA, USA) and then placed on a slide for confocal microscopy. Hearts were imaged on a Zeiss LSM 510 META under 60× magnification.

Flies were aged for 3 weeks on a HSD. At 3 weeks, flies were anesthetized with CO₂ and weighed in groups of six. Groups were then frozen at −20°C for 30 min, returned to room temperature (23°C) and allowed to desiccate for 24 h. After desiccation, fly groups were reweighed. Mass was recorded and the difference calculated. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests were performed to assess statistical significance between groups.

Feeding and consumption assays were conducted similar to the ‘Con-Ex’ assay described in Shell et al. (2018). For feeding assays, flies were aged for 3 weeks on 1 mol l⁻¹ sucrose semi-defined media and then transferred to media supplemented with 2% FD&C Blue #1 dye for 2 h. Five flies per replicate were homogenized using an electronic mortar and pestle. Supernatants were read on a VersaMax (Molecular Devices) spectrophotometer at 630 nm. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests were performed to assess statistical significance between groups.

An excreta assay was used to determine the amount of food processed by the flies (Shell et al., 2018). Flies were aged on a 1 mol l⁻¹ sucrose diet for 3 weeks. At 3 weeks, 20 flies were transferred to fresh blue food (1 mol l⁻¹ sucrose food supplemented with 2% FD&C Blue #1) for 24 h. Flies produced excreta on white foam plugs pushed down to sit 1 cm from the media surface. Plugs were photographed using a Bio-Rad Chemi-Doc and the resulting photos were analyzed in ImageJ for number of excreta. Statistical analysis and graph preparation were conducted using Prism software (v.9.0). Two-tailed Student's t-tests were performed to assess statistical significance between groups.

Upon eclosion, 20 flies per vial were transferred to 0.15 or 1 mol l⁻¹ sucrose semi-defined food supplemented with 2% FD&C Blue #1. Flies were assessed daily for survival and ‘smurfing’ (Pereira et al., 2018; Rera et al., 2012). Briefly, flies exhibiting gut barrier dysfunction, i.e. blue dye leaching into hemolymph, were scored as ‘smurfed’. Dead flies were removed daily, and flies were flipped to new food every 2 days to ensure food freshness. Kaplan–Meier survival curves and log rank statistical analysis using Prism v.9.0 were used to determine significance. Photos depicting smurfed and non-smurfed flies were taken on an Olympus SZX12 stereo microscope at 25× magnification with a Jenoptik Progres Gryphax Subra digital camera and Progres Gryphax software (Jenoptik, Jena, Germany). Representative images were taken with r4,w¹¹¹⁸ flies because of low smurfing of r4>desat1 RNAi flies.

Guts were dissected from 1–2 day old adult females and 3 week aged females, depending on the experiment. Flies were anesthetized with FlyNap and dissected in PBS. Images of guts were taken with an Olympus SZX12 stereo microscope at 10× magnification with a Jenoptik Progres Gryphax Subra digital camera and Progres Gryphax software. Guts were measured using ImageJ with Mann–Whitney test, and statistical analysis was completed using Prism software (v.9.0). A Mann–Whitney test for non-parametric data was used to determine significance.

For ultra high performance liquid chromatography–tandem mass spectrometry (UHPLC-MS/MS; see below), 100 adult female hearts, 10 µl hemolymph and 20 fat bodies per replicate were collected from 3 week old flies anesthetized with FlyNap. Hearts included intact heart tubes with podocytes and alary muscles, with the fat body being carefully removed prior to collection. Adult hemolymph was collected using a 0.5 ml centrifuge tube pierced with a 20 gauge hypodermic needle and nested in a 1.5 ml centrifuge tube. Flies were pierced in the thorax and collected as previously described (Tuthill et al., 2020). Briefly, 10 µl of hemolymph was collected using a 25 gauge hypodermic needle from 200 flies per replicate, and centrifuged in groups of 40 flies per tube at 2348 RCF for 5 min. Adult heart and fat body were homogenized by hydraulic pressure and mechanical shearing using a 20 gauge needle in 200 µl of PBS (pH 7.4), with 190 µl extracted for UHPLC-MS/MS and 10 µl saved for protein. Volume was brought up to 200 μl with PBS. Statistical analysis and graph preparation were conducted using Prism software (v.9.0).

Drosophila adult hearts, hemolymph and fat bodies (see above) were analyzed by the South Eastern Center for Integrated Metabolomics (SECIM) at the University of Florida, Gainesville, FL, USA, as previously reported. Samples were micro-dissected and immediately frozen at −80°C in PBS before analysis. Lipids were extracted using a bacterial lipidomics extraction procedure (Folch et al., 1957) and normalized prior to injection to the lowest sample protein concentration and analyzed using techniques previously described (Tuthill et al., 2020). In short, lipids were analyzed using a Thermo Q-Exactive Orbitrap mass spectrometer/electrospray ionization and Dionex UHPLC. LipidMatch software was used to identify peaks (Koelmel et al., 2017). Standards for all lipid classes identified were spiked into samples prior to mass spectrometric analysis. All peak areas were normalized to the total ion current during UHPLC-MS/MS sample runs (Alfassi, 2004) with subsequent normalization by peak area sum during data examination. A principal component analysis powered by MetaboAnalyst 3.0 (Xia and Wishart, 2016) was used to estimate variation across the sample groups. The fragmentation libraries were edited to identify principal components or phosphatidyl choline species with the neutral loss of the phosphatidyl headgroup, trimethyl amine, and the loss of fatty acid tails. Graphs were made using Prism software (v.9.0). Two-tailed Student's t-tests were performed to assess statistical significance between groups.

Desat1 is essential for development, fatty acid desaturation and incorporation of fatty acids into triglycerides in Drosophila larvae (Musselman et al., 2013; Wang et al., 2016), but previous studies did not explore the role of Desat1 in adult physiology. Consistent with previous studies, 1 mol l⁻¹ sucrose produced larval lethality when desat1 was targeted by transgenic RNAi in the developing fat body. There was also a high degree of larval lethality on 0.7 mol l⁻¹ sucrose, so 0.5 mol l⁻¹ (17% sucrose) was the concentration used for larval rearing throughout this study in order to obtain healthy individuals in which to study the adult stage. Almost all of these eclosing flies were female, so we excluded males from our analyses. Young adult flies were transferred to HSD (1 mol l⁻¹ sucrose) after eclosion. First, we measured effects on adult mass and TG accumulation after 3 weeks of HSD. Loss of desat1 was sufficient to reduce body mass from 1 mg in genotype-matched r4,w¹¹¹⁸ controls to 0.76 mg in desat1 RNAi (r4>desat1 RNAi) animals (P<0.0001) (Fig. 1A). Similarly, TG content per fly was reduced from 65.28 µg mg⁻¹ in r4,w¹¹¹⁸ flies to 8.37 µg mg⁻¹ in desat1 RNAi flies (P<0.0001) for flies fed on HSD (Fig. 1B). Reverse transcription and qPCR performed in isolated fat bodies to measure the amount of desat1 mRNA showed an 89.5% reduction of desat1 mRNA (Fig. S1) in desat1 RNAi flies compared with r4,w¹¹¹⁸.

Because desat1 RNAi flies showed defects in TG storage after 3 weeks, and previous studies showed changes in fatty acid saturation in the heart lipid pools of HSD-fed flies after 5 weeks (Tuthill et al., 2020), we assessed cardiac function by measuring pacing-induced heart failure and recovery (Paternostro et al., 2001; Wessells and Bodmer, 2004). Adult desat1 RNAi female flies aged for 3 weeks on HSD showed a significant 24.6% increase in heart failure compared with HSD-fed r4,w¹¹¹⁸ flies (P<0.001), followed by a 28.8% decrease in recovery for desat1 RNAi compared r4,w¹¹¹⁸ flies fed the HSD (P<0.0001) (Fig. 2A,B). This role of fat body Desat1 in cardiac resilience was HSD specific, as we found that desat1 RNAi flies fed a control diet for 3 weeks had no increase in heart failure over r4,w¹¹¹⁸ flies (Fig. 2A) and all of these flies recovered a rhythmic heartbeat by 120 s (Fig. 2B). Moving forward, our studies focused on the effects of HSD on other aspects of cardiac physiology. Using SOHA, we found a 36% decrease in heart rate in desat1 RNAi flies (Fig. 2C) (P<0.01) with a decrease in both the systolic and diastolic means (Fig. 2E,F) (P<0.01 and P<0.01, respectively). In addition, desat1 RNAi fly hearts had longer systolic intervals compared with r4,w¹¹¹⁸ controls (Fig. 2G) (P<0.01) with a mean time of 0.22 s for desat1 RNAi and 0.09 s for r4,w¹¹¹⁸. There was no difference in time spent in diastole (Fig. 2H). Surprisingly, there was a decrease in arrhythmicity in desat1 RNAi flies compared with controls (P<0.05) (Fig. 2J), with no change in heart period or fractional shortening (Fig. 2D,I). When hearts were stained using anti-pericardin, a type-IV collagen and marker of fibrosis, there were no qualitative changes in organization, fibrosis or morphology between r4,w¹¹¹⁸ and desat1 RNAi flies (Fig. S2A–F) consistent with a specific role for Desat1 in heart function, rather than structure.

Because we found changes in body fat composition and cardiac function when desat1 was knocked down in the fat body, we took a tissue-specific metabolomics approach to characterize lipid profiles of the fat body, hemolymph and heart to assess whether changes in cardiac function may be associated with the accumulation of toxic lipids. We first looked at changes in the tissue-specific metabolome by principal component analysis. All three sample types exhibited divergence of 95% confidence intervals, indicating variation in lipid profiles between r4,w¹¹¹⁸ and desat1 RNAi flies (Fig. S3A–C). We next examined the overall glycerolipid, phospholipid, lyso-phospholipid, ether lipid and sphingolipid lipid classes for each tissue and hemolymph by measuring absolute quantities detected. Glycerolipids were the most abundant class with 597.5 ng per fat body, on average. There was a near-significant decrease in fat body glycerolipids from 990 ng to 205 ng (P=0.07) (Fig. 3A) between r4,w¹¹¹⁸ and desat1 RNAi flies. Phospholipids, lyso-phospholipids and sphingolipids were similar in abundance with an average of 301.5, 330.1 and 301.9 ng, respectively. Ether lipids were the least abundant class in the fat body, with an average of 0.48 ng detected (Fig. 3A). Hemolymph contained significant decreases in both sphingolipids (P<0.05) and ether lipids (P<0.01) while all other classes remained unchanged (Fig. 3B). Sphingolipids decreased from 6.78 ng to 6.56 ng µl⁻¹ of hemolymph and ether lipids decreased from 0.08 ng to 0.02 ng µl⁻¹ of hemolymph (Fig. 3B). There were no significant differences in overall lipid classes in cardiac tissue (Fig. 3C). As expected, fewer unsaturated fatty acid substituents were observed in desat1 RNAi fly organs and hemolymph, although these sometimes failed to reach statistical significance (Fig. 3G–I,M–O).

Because we saw some variance in fat body glycerolipids, and loss of desat1 led to an inability of flies to store TG (Fig. 1), we looked at changes in TG and DG in all samples. Interestingly, we saw a decrease in TG content in fat body (P<0.05) (Fig. 3D) and a nearly significant increase in heart TG (P=0.06) (Fig. 3F), whereas TG in hemolymph remained unchanged (Fig. 3E) when comparing r4,w¹¹¹⁸ with desat1 RNAi flies. There were no significant changes in either heart or hemolymph DG content (Fig. 3K,L); however, there was a significant decrease in fat body DG content (Fig. 3J) (P<0.05). Because ether lipids have been shown to be differentially present in flies fed a HSD (Tuthill et al., 2020), we looked at relative levels of diacylglycerol ethers (DAGE) and plasmenyl- phosphatidylethanolamine (plasmenyl-PE). The fat body showed decreases in DAGE (P<0.001) from 85.3% to 11.0% (Fig. S3D) while the hemolymph showed an increase in DAGE from 5.9% to 13.0% (P<0.05) (Fig. S3E). DAGE showed a possible decrease in the heart; however, this was not significant (P=0.21) (Fig. S3F). We next assessed relative plasmenyl-PE levels in all samples and found a trend opposite to that seen for DAGE. Plasmenyl-PE was increased in the fat body (P<0.05) (Fig. S3G) with a decrease in hemolymph (P<0.05) (Fig. S3H). Cardiac tissue showed no change in plasmenyl-PE (P=0.15) (Fig. S3I).

Because desat1 RNAi flies are small and lean, and have cardiac defects and reduced lipid storage, we asked whether these changes might occur as a result of differences in feeding or gut function. We first assessed consumption in both desat1 RNAi and r4,w¹¹¹⁸ flies at 3 weeks. We found that desat1 RNAi flies had decreased feeding compared with r4,w¹¹¹⁸ flies (P<0.01) with desat1 RNAi flies consuming approximately 49.3% less food than r4,w¹¹¹⁸ flies over a 2 h period (Fig. 4A). We next asked whether the inability to store nutrient-derived energy as TG might be a result of impaired gut function. We measured excreta at 3 weeks and found desat1 RNAi flies excreted more than r4,w¹¹¹⁸ flies (P<0.0001) (Fig. 4B) with each desat1 RNAi fly excreting 4.4 more excreta particles than each r4,w¹¹¹⁸ fly over a 24 h period. We measured fly guts to see whether reduced transit time might be the result of a reduction in gut size. We measured gut length and width from 3 week desat1 RNAi and r4,w¹¹¹⁸ flies and found desat1 RNAi flies had significantly smaller guts than r4,w¹¹¹⁸ flies (P<0.01) (Fig. 4C,D), with an average length of 7.8 mm for desat1 RNAi flies and 8.9 mm for r4,w¹¹¹⁸ flies. This seemed to be a result of a developmental effect of desat1 RNAi in the fat body because gut size was also reduced by around 25% in newly eclosed adults, regardless of the diet (24.3% for control food, data not shown; 30.1% for 0.5 mol l⁻¹ sucrose, see Fig. 5F, both P<0.0001). To further determine whether gut function was being affected by loss of desat1 in the fat body, we conducted a smurf assay (Rera et al., 2012) to assess gut barrier dysfunction. Non-smurfed flies showed only blue dye in the gut, with no blue leakage into the surrounding tissue (Fig. S4A), while smurfed flies had blue dye in all tissues except ommatidia (Fig. S4B). We found that desat1 RNAi flies had significantly less smurfing over the course of their lifetime, compared with controls (P<0.001) (Fig. 4E). Lastly, we postulated that the reduced size and changes in metabolic homeostasis might be due in part to reduced water content in desat1 RNAi flies. We measured the average water content of desat1 RNAi and r4,w¹¹¹⁸ flies and found desat1 RNAi flies contained less water per fly (P<0.05) (Fig. 4F). Taken together, fat body desat1 RNAi may lead to a reduced ability to store nutrients as a result, in part, of changes in gut development.

The role of Desat1 is to produce unsaturated lipid species; therefore, we sought to rescue desat1 RNAi phenotypes by feeding with C18:1 MUFA OA. We first assessed growth and TG accumulation in larvae because larvae eat vigorously and survived well on the supplementation diet. Wandering third instar r4,w¹¹¹⁸ and desat1 RNAi female larvae were reared from embryos on 0.5 mol l⁻¹ sucrose food with or without 1% OA. The mass of r4,w¹¹¹⁸ larvae did not vary when OA was added, while desat1 RNAi larvae supplemented with OA had a 77.2% increase in mass compared with the same genotype fed unsupplemented food (P<0.0001) (Fig. 5A). To test the effect of OA on lipogenesis, we measured whole larval TG. r4,w¹¹¹⁸ larvae reared on 0.5 mol l⁻¹ sucrose food with OA showed a nearly significant 37.8% increase in TG (P=0.07) (Fig. 5B) consistent with studies in other systems showing that OA induces lipogenesis (Lounis et al., 2017). desat1 RNAi larvae fed 0.5 mol l⁻¹ food with and without OA supplementation had significantly less TG than the r4,w¹¹¹⁸ control (P<0.01) (Fig. 5B). When desat1 RNAi larvae were fed OA-supplemented food, they had a 57.4% increase in TG over unsupplemented desat1 RNAi larvae (P=0.27, not statistically significant by ANOVA, P<0.01 by two-tailed t-test when compared directly) (Fig. 5B). When measured as TG content per animal, there was a statistically significant increase upon OA supplementation in desat1 RNAi larvae (P<0.08 by ANOVA, P<0.0001 by two-tailed t-test), increasing from 19.7 mg per animal to 42.5 mg per animal, whereas in the control genotype, TG per animal did not increase significantly (P=0.88 by ANOVA, P=0.59 by two-tailed t-test) (data not shown). Because desat1 RNAi in larvae produces hyperglycemia (Musselman et al., 2013), we measured hemolymph glucose levels in wandering third instar larvae fed OA (Fig. 5C). Larval hemolymph glucose was increased in desat1 RNAi larvae reared on both supplemented and unsupplemented diets, compared with r4,w¹¹¹⁸ controls (Fig. 5C). Interestingly, hemolymph glucose concentrations decreased when both r4,w¹¹¹⁸ and desat1 RNAi larvae were reared on OA-supplemented food, compared with unsupplemented food, with a 12.6% decrease for r4,w¹¹¹⁸ (not significant by ANOVA, P<0.05 by t-test), and a 19.2% decrease for desat1 RNAi larvae (P<0.05 by both tests).

Because we saw phenotypic rescue from larval OA supplementation but were unable to assess cardiac failure in larvae, we developed a 0.5% OA supplementation diet on which adults could survive for 3 weeks. A dramatic 26.1% increase in mass was seen in desat1 RNAi adults fed 1 mol l⁻¹ sucrose HSD supplemented with 0.5% OA, compared with those on HSD alone (P<0.0001), whereas there was no increase in the mass of the r4,w¹¹¹⁸ flies on HSD when supplemented with OA (Fig. 5D). Similarly, there was a significant 470% increase in TG content per fly mass in desat1 RNAi flies when supplemented with OA (P<0.01) and no increase in the r4,w¹¹¹⁸ flies (Fig. 5E). To further explore the ability of OA to rescue desat1 RNAi phenotypes, we assessed gut size in adult flies. Young, 1–2 day old desat1 RNAi female flies had reduced gut size compared with r4,w¹¹¹⁸ flies (Fig. 5F) (P<0.0001). Interestingly, supplementation with OA was able to recover gut size in desat1 RNAi flies (P<0.001), whereas OA reduced gut size in the r4,w¹¹¹⁸ flies (P<0.05) (Fig. 5F) so that no difference between the genotypes was observed on supplemented food. Lastly, to assess whether OA supplementation is sufficient to recover the detrimental cardiac effects of fat body desat1 RNAi, we measured failure and recovery of 3 week old adults after cardiac pacing. OA supplementation drastically reduced (46.6%) the rate of pacing-induced heart failure in desat1 RNAi flies compared with those on an unsupplemented diet (Fig. 5G) (P<0.01) whereas there was no change in the failure rate of the r4,w¹¹¹⁸ flies upon OA supplementation. Additionally, recovery of failed hearts at 120 s post-stimulation was increased 134.1% in OA-supplemented desat1 RNAi flies compared with unsupplemented desat1 RNAi flies (P<0.0001), with no improvement seen after OA supplementation in the r4,w¹¹¹⁸ flies (Fig. 5H).

In this study, we used loss-of-function genetics as a means to study the role of fatty acid saturation in a Drosophila model for metabolic syndrome. The loss of desat1 in the fat body had both tissue-autonomous and systemic effects, as it led to reduced fat body lipogenesis and reduced cardiac resilience accompanied by changes in the cardiac metabolome. We further found that the loss of desat1 in the fat body had profound effects on feeding and nutrient processing accompanied by changes in gut morphology. An OA-supplemented diet rich in MUFAs recovered several detrimental phenotypes.

Previous studies have shown that flies aged on a HSD had dramatic changes in fat content, including increased TG and a reduced saturation index (Tuthill et al., 2020), accompanied by impaired cardiac function (Na et al., 2013). We therefore characterized cardiac phenotypes and lipid profiles in aged HSD-fed r4,w¹¹¹⁸ and desat1 RNAi adult flies. In our initial studies, we found desat1 RNAi male survival to be extremely impaired on the 0.5 mol l⁻¹ sucrose diet, compared with females, leading us to focus our studies solely on female flies. Reduced male fitness has also been seen in other genotypes undergoing dietary challenges. In addition to unpublished observations from genetic screening from our own lab, one published study found that male flies challenged on a high-glucose, high-yeast diet had decreased survival relative to females on the same diet (Stefana et al., 2017). A recent study using starvation found that males had greater fluctuation in lipid levels and reduced survival during starvation, compared with females (Wat et al., 2020), indicating that males are less resilient in the face of impaired lipid homeostasis. High lethality in desat1 RNAi males could suggest that Desat1 is even more important in males than it is in females. Both male larvae and adult flies express significantly higher levels of desat1 in their fat bodies compared with females (Leader et al., 2018; Musselman et al., 2018). Male flies are smaller and have less adipose compared with females; therefore, the maximum adipose expandability is lower in males than in females. There are also some data to support a role for biological sex as a variable controlling lipid homeostasis in humans. Women undertake more subcutaneous fat synthesis than men (Edens et al., 1993), indicating women are more adept at accommodating fluctuations in energy homeostasis, which may contribute to a decrease in the incidence of metabolic syndrome in women, compared with men (Ford, 2005).

We found impairment in overall lipogenesis in desat1 RNAi flies. Like the Drosophila fat body, cultured hepatocytes deficient in SCD1 activity also showed a decrease in total lipid, TG and DG (Lounis et al., 2017). Inhibition of SCD1 in 3T3-L1 preadipocytes led to decreased unsaturated fatty acid synthesis and decreased transcripts for genes involved in TG synthesis (Ralston et al., 2014). Surprisingly, mouse adipose-specific loss of SCD1 did not affect TG content but did increase glucose catabolism (Hyun et al., 2010). Likewise, knockout of SCD1 in both liver and adipose reduced unsaturated fatty acid substituents but did not affect overall obesity (Flowers et al., 2012). Thus, the tissue-specific contributions of SCD1 to whole animal lipid homeostasis in this model remain unclear. In our study, fat body desat1 RNAi flies fed an obesogenic HSD showed decreased adiposity and body mass, consistent with a central role for the fat body in systemic metabolism. An insufficient pool of unsaturated fatty acids is likely to be one effect of desat1 RNAi, because we were able to rescue some desat1 RNAi phenotypes by OA supplementation. Because MUFA are essential for TG synthesis and storage, and SCD1 mutants lack the ability to produce MUFA, increasing endogenous MUFA in liver SCD1 knockout mice had beneficial effects similar to those observed in flies (Aljohani et al., 2019). Liver-specific knockout of SCD1 in mice using Cre/Lox led to decreases in body mass, hepatic TG and plasma TG under a high-carbohydrate feeding paradigm while supplementation with OA was able to rescue these phenotypes (Miyazaki et al., 2007). Interestingly, supplementation with OA improves physiology in a range of diabetes and obesity models, suggesting that fatty acid saturation is particularly important in the face of metabolic overload. Here, we show supplementation of HSD with OA partially restored TG storage and hemolymph glucose. Liver-specific knockout of SCD1 in mice fed a HSD led to a decrease in SREBP mRNA, indicating SCD1 is an important regulator of lipogenic gene expression (Miyazaki et al., 2007). Commensurate with our findings, supplementing SCD1-deficient hepatocytes with OA restores SREBP signaling as well as improving lipid droplet abundance, and DG and TG content (Lounis et al., 2017), and OA feeding also recovers obesity and TG turnover in liver-specific SCD1 knockout mice (Miyazaki et al., 2007). Two potential links between loss of SCD1 and pathophysiology may be endoplasmic reticulum stress and inflammation. Mice fed a HSD with liver-specific knockout of SCD1 showed increases in inflammatory cytokines including TNF-α as well as markers for endoplasmic reticulum stress, which were improved upon dietary supplementation with OA (Liu et al., 2016). Future studies will explore the cellular mechanisms by which pathophysiology arises in desat1 RNAi flies.

Because flies fed on a HSD have been shown to suffer from cardiac dysfunction (Na et al., 2013), we assessed heart function and found that Desat1 has a previously undescribed role in cardiac physiology. A protective role for MUFA is consistent with lipid profiles in mice and humans, where saturation is inversely correlated with heart function (Ander et al., 2003; Li et al., 2018). As MUFA promote lipogenesis, these findings fit with a model where reducing maximum adipose expandability via desat1 RNAi leads to lipotoxicity in the heart. We first employed a cardiac pacing paradigm to stress fly hearts after chronic HSD feeding in r4,w¹¹¹⁸ and desat1 RNAi flies. desat1 RNAi flies subjected to cardiac pacing stress had an increased failure rate coupled with a diminished ability to recover from stress. Consistent with impaired heart function, fat body desat1 RNAi flies exhibited evidence of restrictive cardiomyopathy, with reduced diastolic diameter, systolic diameter and heart rate, similar to defects seen in cardiac myosin and troponin-T fly and murine mutants (Cammarato et al., 2008; Viswanathan et al., 2014). Flies of our HSD-fed control (r4,w¹¹¹⁸) genotype had heart rates similar to those of other studies in aged flies; however, HSD-fed, desat1 RNAi flies had a decreased heart rate at 3 weeks, similar to older flies, which display reduced heart rate and increased pacing-induced heart failure, compared with younger flies (Wessells et al., 2004). Leptin deficient ob/ob mice lacking SCD1 have improved cardiac performance as well as decreased cardiac lipid accumulation, indicating SCD1 is deleterious in genetically obese mice (Dobrzyn et al., 2010); however, nothing has been reported of the role of SCD1 role in cardiac performance in HSD-fed mice in a non-genetically obese background.

No reports describe cardiac function in adipose or liver-specific SCD1 knockout mice, although links between lipid composition and cardiac physiology during metabolic syndrome have been studied in both mice and humans, where MUFA are able to improve heart function in some studies (Clifton and Keogh, 2017) but not in others (Jakobsen et al., 2009). Specifically, a number of unsaturated lipids are associated with improved cardiac health in diabetic mice, patients, and cultured cardiomyocytes (Gillman, 1997; Rasmussen et al., 1993; Yang et al., 2016; Zadeh Hashem et al., 2016). Heart lipid storage has been shown to be increased in both diabetic patients and Zucker diabetic rats with increased cardiac failure (Sharma et al., 2004), suggesting that cardiac TG is deleterious. However, increasing TG via cardiac overexpression of the TG biosynthetic enzyme DGAT in mice leads to an increase in cardiac performance and survival (Liu et al., 2009), implying that not all cardiac TG accumulation is bad. Perhaps the substituent composition of the TG is also important. In this present study, we saw increases in the relative abundance of saturated cardiac TG species with a commensurate decrease in unsaturated species. It could also be that DG act as toxic lipid species. While we did not see significant increases in cardiac DG content or saturation, there was an average increase that may have been obscured by the low signal to noise ratio in these rather small samples. Coupled with the detrimental cardiac phenotypes that were rescued by OA supplementation, it is possible that saturated lipid species as DG and/or TG are particularly harmful to the fly heart. Interestingly, supplementation of mouse diets with long chain MUFA attenuates atherosclerosis by activating PPAR-γ (Yang et al., 2016). Cardiomyocytes supplemented with palmitate showed increased endoplasmic reticulum stress (Bosma et al., 2014). While palmitate has been shown to be lipotoxic in cell culture (Egnatchik et al., 2019; Listenberger et al., 2003), OA and palmitoleic acid supplementation are able to reverse the effects of palmitate-induced lipotoxicity in cell culture (Hetherington et al., 2016; Zadeh Hashem et al., 2016). In the present study, flies fed a HSD supplemented with OA showed increases in protective TG storage, body mass and recovery. Elevated hemolymph glucose was observed, suggesting that MUFA may be therapeutic. We further demonstrated that OA supplementation is sufficient to recover lipid storage, and developmental and cardiac phenotypes associated with saturated fatty acid accumulation, indicating the importance of MUFA in lipid homeostasis and physiology.

Surprisingly, we found changes in fly feeding behavior, excretion and gut structure and function in Desat1-deficient flies. Our results showed a decrease in desat1 RNAi fly feeding with an increase in excretion, suggesting that desat1 RNAi flies have reduced hunger and/or have difficulty in absorption of nutrients. It is possible that the loss of fat body desat1 may have endocrine effects that reduce neuronal taste sensing or increase the sense of satiety. Recent studies using Drosophila showed that HSD led to increased feeding behavior in wild-type flies as a result of a blunted neuronal sweet tasting and synaptic response (May et al., 2019). In this study, we saw decreased feeding in desat1 RNAi flies, suggesting taste or other aspects of feeding behavior may be affected. Despite reduced feeding, we observed an increased number of excreta in desat1 RNAi flies, consistent with an effect of fatty acid composition on digestive function. Mice deficient in SCD1 have been shown to have increased gut inflammation and instances of tumorigenesis, while supplementation of their diet with OA ameliorates these phenotypes (Ducheix et al., 2018). In mice, feeding with a high-glucose or high-fructose diet led to increased gut permeability, with a decrease in gut microbiota (Do et al., 2018). A high-sugar diet had similar effects on wild-type flies, with increased gut permeability (Pereira et al., 2018) while desat1 RNAi flies showed evidence of reduced gut permeability using the smurf assay. Many flies in the current study died without smurfing although it has been suggested that smurfing precedes death in various species of flies, nematodes and fish (Dambroise et al., 2016). Other studies indicate that not all flies smurf prior to dying (Rera et al., 2012; Pereira et al., 2018). Decreased dye leakage is a complex phenotype and may represent a number of aspects of animal physiology. The increase in excretion coupled with the reduced feeding and reduced gut and organism size in fat body desat1 RNAi flies could mean that transit time for food is quicker, decreasing nutrient and dye absorption and contributing to the apparent decreases in gut barrier leakage and fat storage. Further studies are needed to better understand the mechanisms underlying this surprising role of Desat1 in feeding behavior and gut function.

There is controversy whether desaturase activity is helpful or deleterious to type 2 diabetic patients. In humans, one study showed reduced SCD1 activity in subcutaneous adipose tissue from diabetic patients but no effect on insulin levels (Bódis et al., 2018) while another study showed increased SCD1 activity in the subcutaneous adipose tissue of obese, type-2 diabetic patients (García-Serrano et al., 2011). In both human diabetes patients and their cultured myotubes, increases in muscle SCD1 expression were correlated with abnormal lipid metabolism (Hulver et al., 2005), highlighting the complex potential effects of perturbing lipid saturation. Extensive research has been conducted into the effect of MUFA and polyunsaturated fatty acids (PUFA) on cardiac health (Ander et al., 2003). One study in human subjects has shown that saturated fatty acids are more prevalent in patients with metabolic syndrome compared with PUFA and MUFA (Hosseinpour-Niazi et al., 2015); moreover, unsaturated lipids were less metabolically detrimental than saturated species (Luukkonen et al., 2018). Interestingly, another study indicated plant-derived MUFA have a positive effect on patients with cardiovascular disease (Zong et al., 2018). Obese women who consumed a diet with increased MUFA had improved weight loss (Kaippert et al., 2015). Overall, it seems to be clear that there is a benefit to a diet with increased MUFA and/or PUFA and a decrease in saturated fatty acid intake. Our data are consistent with a model where saturated fatty acids are detrimental and MUFA are better for individuals who overeat.

We would like to thank the Vienna Drosophila Resource Center for fly stocks and the University of Florida's Southeast Center for Integrated Metabolomics core facility, especially Dr Richard Yost and Dr Timothy Garrett.