Report

Targeting a ceramide double bond improves insulin resistance and hepatic steatosis

See allHide authors and affiliations

Science  26 Jul 2019:
Vol. 365, Issue 6451, pp. 386-392
DOI: 10.1126/science.aav3722

Ceramides in focus

Excess calorie intake can ultimately lead to a metabolic syndrome that interferes with fat or lipid metabolism. There are many different types of lipids, and it has been widely debated which are the true culprits underlying metabolic disorders. Chaurasia et al. report that ceramides are the major contributor to insulin resistance and fatty liver disease (see the Perspective by Kusminski and Scherer). This appears to be caused by the enzyme dihydroceramide desaturase 1 (DES1), which is normally involved in ceramide production by inserting a double bond into the backbone of the molecule. In mice fed a high-fat diet, deletion of DES1 improved glucose and lipid metabolism.

Science, this issue p. 386; see also p. 319

Abstract

Ceramides contribute to the lipotoxicity that underlies diabetes, hepatic steatosis, and heart disease. By genetically engineering mice, we deleted the enzyme dihydroceramide desaturase 1 (DES1), which normally inserts a conserved double bond into the backbone of ceramides and other predominant sphingolipids. Ablation of DES1 from whole animals or tissue-specific deletion in the liver and/or adipose tissue resolved hepatic steatosis and insulin resistance in mice caused by leptin deficiency or obesogenic diets. Mechanistic studies revealed ceramide actions that promoted lipid uptake and storage and impaired glucose utilization, none of which could be recapitulated by (dihydro)ceramides that lacked the critical double bond. These studies suggest that inhibition of DES1 may provide a means of treating hepatic steatosis and metabolic disorders.

Sphingolipids such as ceramides and dihydroceramides are products of fat and protein metabolism that have been shown to accumulate in humans and nonhuman primates with obesity and hyperlipidemia (14). These lipids have been implicated in a wide range of cellular processes related to metabolism, growth, and survival (5, 6). Unbiased lipidomic screens in large clinical cohorts have revealed particularly robust associations between serum and tissue levels of ceramides and/or dihydroceramides and comorbidities of obesity, including insulin resistance, type 2 diabetes, and major adverse cardiac events (715). Some clinics have begun using serum ceramide levels as a measure of cardiovascular disease risk (7).

Ceramides and dihydroceramides consist of a sphingoid base coupled to a variable fatty acid side chain. They can be distinguished from one another by the presence (ceramides) or absence (dihydroceramides) of a 4,5-trans double bond in their sphingoid backbone (Fig. 1A) (5). Studies in purified systems reveal that this double bond alters the biophysical properties of the molecules, modifying their elastic properties and packing behavior (16). In most tissues, this double bond is inserted by the Δ4-desaturase activity of the enzyme dihydroceramide desaturase 1 (DES1), which is resident in the endoplasmic reticulum of virtually all cells. A second dihydroceramide desaturase isoform (DES2), which also has C‐4 hydroxylase activity that enables it to produce phytoceramides that are important for barrier function, is present in tissues such as the skin, kidney, and intestines (5, 17). Both ceramides and dihydroceramides can serve as substrates for the enzymes that produce complex sphingolipids such as (dihydro)sphingomyelins and (dihydro)glucosylceramides (5, 18). The studies described herein investigated the role of these sphingolipids as causative agents in the development of insulin resistance and hepatic steatosis, which are major underlying causes of cardiometabolic disorders.

Fig. 1 Deletion of Degs1 from leptin-deficient ob/ob mice improves glucose homeostasis and resolves hepatic steatosis.

(A) Schematic depicting the DES1 reaction. Dihydroceramides are produced through a biosynthetic pathway fueled by palmitoyl-CoA and serine, which produces the sphingosine backbone that subsequently incorporates additional fatty acids. DES1 then inserts the highly conserved 4,5-trans double bond into the sphingoid backbone. Both ceramides and dihydroceramides serve as precursors for the more predominant, complex sphingolipids. (B to L) Adult ob/ob Degs1fl/fl and ob/ob Degs1Rosa26/ERT2-Cre mice (20 weeks old) were injected with tamoxifen (3 mg) once daily over a 5-day time period. Animals were euthanized and tissues collected at 31 weeks of age. (B) Degs1 mRNA expression and (C) ceramide/dihydroceramide ratios in tissues and serum; (D) body mass; (E) fat and lean mass when the animals were 30 weeks of age; (F) tissue mass; (G and H) glucose tolerance tests and area under the curve (AUC) when the mice were 23 weeks of age; (I and J) insulin tolerance tests and AUC when the mice were 24 weeks of age; (K) adipocyte size and liver lipid droplet area; and (L) plasma ALT and AST levels. For each experiment, n = 14 to 16 mice per group were used, except for the hematoxylin and eosin staining used to calculate adipocyte and liver droplet size (n = 3 to 8 mice) or the AST/ALT determinations (n = 7 to 13 mice). Values are expressed as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001 versus control. The “ns” denotes that the comparison was not significantly different. dhCer, dihydroceramide; Cer, ceramide.

Experiments using cultured cells suggest that ceramides and dihydroceramides have distinct, nonoverlapping functions, particularly with regard to insulin action and apoptosis (1, 5). To investigate the effects of manipulations targeting the 4,5 double bond in ceramides on insulin resistance and hepatic steatosis, we generated gene-targeted mice by excising exon 2 of the Degs1 gene (NP031879) encoding DES1. Specifically, we generated mice where exon 2 is flanked by LoxP sites, backcrossed them 10 times onto the C57BL/6 background, and bred them with mice expressing a tamoxifen-inducible Cre-recombinase inserted into the Rosa26 locus. The resultant conditional Degs1 knockout mice (Degs1Rosa26/ERT2-Cre) enable temporal control of gene depletion in a large number of tissues. Degs1 removal led to an increase in the proportion of sphingolipids lacking the 4,5 double bond in its sphingoid base (e.g., dihydroceramide and dihydrosphingomyelin) (fig. S1, A and B). Although we previously found that germline ablation of Degs1 was lethal (19), removing it from adult animals produced no major health abnormalities or discernable signs of stress or discomfort. Degs1 depletion reduced fat and liver mass, improved lipid and glucose handling (fig. S1, C to H), and elevated oxygen consumption (fig. S1K). No changes in respiratory exchange ratios, food intake, or ambulatory activity were observed (fig. S1, M to P).

We crossed these Degs1Rosa26/ERT2-Cre mice, as well as Degs1fl/fl littermates lacking Cre, with obese, leptin-deficient ob/ob mice that have impaired glucose tolerance and hepatic steatosis. At 20 weeks of age, we injected tamoxifen into the ob/ob Degs1fl/fl (control) and ob/ob Degs1Rosa26/ERT2-Cre offspring. The floxed allele from the Cre-containing strains was excised by administration of tamoxifen (Fig. 1B), which resulted in reduced ceramide/dihydroceramide ratios across a range of tissues (Fig. 1C and fig. S2, A to E). Other lipid classes, including glycerophospholipids (e.g., phosphatidylcholine, phosphatidylethanolamine, and phosphatidylserine), lysophospholipids, and sterols, were unaffected (fig. S2L). After Degs1 ablation, animals displayed notable improvements in glucose and insulin tolerance (Fig. 1, G to J). Degs1 depletion also halted weight gain in ob/ob mice by way of reduced fat mass (Fig. 1, D to F). Glucose and insulin tolerance tests were conducted before the onset of weight loss. Degs1 deletion did not affect oxygen consumption (VO2), carbon dioxide production (VCO2), ambulatory activity, or food intake but modestly reduced respiratory exchange ratios (fig. S2, F to I). Histological assessment revealed that Degs1 depletion resolved hepatic steatosis and decreased the size of adipocytes in the epididymal and subcutaneous adipose beds (Fig. 1K). No adipose browning was observed (fig. S2M). Circulating alanine transaminase (ALT) and aspartate transaminase (AST), which are measures of nonalcoholic fatty liver disease (steatohepatitis), were markedly reduced in the ob/ob Degs1 knockouts (Fig. 1L). Thus, global Degs1 depletion increased the proportion of sphingolipids lacking this conserved double bond and elicited a broad spectrum of metabolic benefits characterized by improvements in glucose and lipid handling.

To probe tissue-specific mechanisms that account for this phenotype, we selectively deleted Degs1 from either the liver, adipose tissue, intestine, or myeloid cells. Although excising Degs1 from intestinal or myeloid cells had no effect on any metabolic parameters assayed (figs. S8 and S9), Degs1 removal from the liver and/or adipose tissue improved glucose and lipid handling (Fig. 2 and figs. S3 to S7). Specifically, we excised the gene from adipose tissue by crossing the Degs1fl/fl mice with mice expressing Cre recombinase under the control of the adiponectin promoter (20). Degs1 was deleted in the liver by injecting adeno-associated virus expressing Cre recombinase (under the control of the human thyroid hormone–binding globulin promoter) into Degs1fl/fl mice. In a subset of animals, we deleted the genes from both adipose tissue and liver simultaneously. Although changes in transcripts were restricted to the targeted tissue (i.e., liver or adipose), ceramide/dihydroceramide ratios were altered in multiple locales, supporting our previous observation that these sphingolipids transmigrate between tissues (Fig. 2, A and B, and figs. S3, A to C; S4, A and B; S5, A and B; and S6, A to D) (20). Using a metabolomic flux assay to measure the rate of incorporation of a stable serine isotope into ceramides, we confirmed that rates of ceramide production were substantially reduced in primary adipocytes isolated from the Degs1δAdipo mice (fig. S3C). All mouse strains were fed an obesogenic diet for 8 weeks before undergoing the metabolic phenotyping protocol. Gene depletion from either adipose tissue or the liver comparably improved glucose tolerance and decreased circulating levels of glucose (in fed and fasted mice), insulin, and fatty acids (Fig. 2, F to J). Depleting Degs1 from either compartment decreased liver fat, with the effect of the liver-targeted deletion being more robust (fig. S6D), and decreased levels of several inflammatory cytokines (fig. S6C). Knockout and control animals exhibited comparable body and tissue mass, VO2, VCO2, energy expenditure, respiratory exchange ratios, food intake, and ambulatory activity under both normal chow diet (NCD) and high-fat diet (HFD) feeding regimens (Fig. 2, C and E, and figs. S5, C to J, and S7, G to M). These findings suggest that the benefits caused by global Degs1 depletion could result from separate actions in adipose tissue or the liver, independent of effects on weight gain or locomotor activity.

Fig. 2 Tissue-specific deletion of Degs1 improves glucose homeostasis and resolves hepatic steatosis in mice.

(A to J) Degs1δΑdipo or Degs1fl/fl littermate (control) mice were placed on a HFD at 4 weeks of age. At 12 weeks of age, Degs1fl/fl mice were injected with AAV-Tbg-Cre to produce Degs1δLiver or Degs1δLiver/Adipo mice (i.e., where Degs1 is specifically deleted in the liver or liver plus adipose tissues, respectively). The other mice received control empty viruses (AAV-Tbg-Null). Euthanasia and tissue harvesting were performed when mice were 16 weeks of age. (A) Degs1 transcript levels from indicated tissues; (B) ceramide/dihydroceramide ratios in tissues and serum; (C) body mass; (D) fat versus lean mass; (E) tissue mass; (F) fed and fasted blood glucose; (G) glucose tolerance tests were performed in animals when they were 14 weeks of age; (H) insulin tolerance tests were performed when they were 15 weeks of age; (I) serum insulin; (J) serum FFAs (n = 3 to 7 mice per group; replicate cohorts of animals were phenotyped separately, with the data included in figs. S3 to S7). Values are expressed as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001 versus control. eWAT, epididymal white adipose tissue; sWAT, subcutaneous white adipose tissue; BAT, brown adipose tissue.

To further explore whether inhibition of DES1 may improve metabolic disorders, we delivered short hairpin RNAs (shRNAs) targeting the Degs1 transcript to the liver of mice using adeno-associated viruses (AAVs). After screening 10 potential shRNA sequences (fig. S10A), we identified one (shRNA8) that produced strong Degs1 knockdown in vitro and in vivo, substantially reducing the levels of liver ceramides (fig. S10, B and C). Having confirmed target engagement, we studied the effect of sustained Degs1 knockdown on glucose homeostasis. In an intervention study, we generated diet-induced obese (DIO) mice by feeding 20-week-old C57BL/6 mice with an obesogenic HFD for 12 weeks. Mice were then injected with either low doses (1 × 1011 genome copies/mouse) or high doses (3 × 1011 genome copies/mouse) of AAVs expressing shRNA8 (AAV-Degs1) or a scrambled control (AAV-control). For comparison, we also conducted a prevention study by injecting the AAVs before starting the HFD feeding regimen. Twelve weeks after AAV administration, both low- and high-dose injections resulted in ~50% knockdown of Degs1 mRNA in liver (Fig. 3A), suggesting sustained target engagement. Lipidomic analysis revealed that the shRNA reduced ceramide/dihydroceramide ratios in the liver, adipose tissue, and plasma but not in muscle (Fig. 3, B and K, and fig. S10, D and E). In the intervention protocol, knockdown of liver Degs1 improved insulin sensitivity and glucose homeostasis (Fig. 3, D to H, and fig. S10F) without affecting body mass. Hyperinsulinemic-euglycemic clamp assessments revealed that the liver-restricted knockdown led to markedly improved whole-body insulin action, evidenced by the increased glucose infusion rate (GIR) needed to maintain euglycemia (Fig. 3I). The elevation in GIR caused by Degs1 knockdown was attributed to both increased glucose disposal (Fig. 3I) and enhanced suppression of hepatic glucose production (Fig. 3J), indicating that liver-tropic Degs1 knockdown improved both hepatic and nonhepatic insulin resistance. In the prevention protocol, AAV-Degs1 elicited the same improvement in glucose homeostasis but also reduced weight gain (Fig. 3, L to O). This latter effect on body habitus occurred independent of food intake (Fig. 3M), thus recapitulating the effect seen after ablation of the Degs1 gene in the Degs1δLiver animals.

Fig. 3 Hepatic Degs1 inhibition with AAV-Degs1-shRNA improves glucose homeostasis.

(A to J) Intervention study: 26-week-old C57BL/6 mice were placed on a HFD for 12 weeks. They were then injected with a single injection (1 × 1011 or 3 × 1011 genome copies/mouse, low dose or high dose, respectively) of AAV-Degs1 expressing shRNA8 (AAV-Degs1) or a scrambled control (AAV-control). Mice were analyzed 12 weeks later. (A) Quantification of Degs1 mRNA in the liver, eWAT, and muscle 12 weeks after administration of AAV-Degs1-shRNA or control; (B) liver and plasma ceramide/dihydroceramide ratios; (C) body mass; (D) fasting blood glucose; (E) glucose tolerance; (F) insulin tolerance; (G) fasting plasma insulin; (H) HOMA-IR; (I) glucose infusion rate (GIR), glucose disposal rate (GDR), insulin-stimulated GDR, and (J) hepatic glucose production. (K to O) Prevention study: 20-week old C57BL/6 mice were given a single injection (1 × 1011 or 3 × 1011 genome copies/mouse, low dose or high dose, respectively) of AAV-Degs1 expressing shRNA8 (AAV-Degs1) or a scrambled control (AAV-control). They were then placed on a HFD for 12 weeks. (K) Ceramide/dihydroceramide ratios in liver, adipose tissue, and plasma; (L) body mass; (M) food intake; (N) percent blood glucose change; and (O) insulin tolerance. n = 10 to 12 mice for both the intervention and prevention studies. Values are expressed as mean ± SEM, *P < 0.05, **P < 0.001, ***P < 0.0001, ****P < 0.00001 versus control.

These findings from Degs1 knockout mice suggest that ceramides induce selective insulin resistance (21), a term previously reported to describe a prediabetic condition characterized by defective insulin action toward glucose metabolism (i.e., increased gluconeogenesis in liver and impaired glucose uptake in muscle and adipose tissue) coupled with an apparent lack of insulin resistance toward lipids (i.e., enhanced triglyceride production). The disruptive effects of ceramide on glucose homeostasis are attributable to its inhibitory actions against the anabolic enzyme Akt or protein kinase B (Akt/PKB) (18, 22), an insulin-stimulated kinase that represses gluconeogenesis in the liver, increases glucose uptake in adipose tissue and muscle, and inhibits production of glucagon in pancreatic alpha cells. We found that exogenous ceramides, but not dihydroceramides, impaired insulin stimulation of the Akt/PKB enzyme in cultured hepatocytes (Fig. 4A) and adipocytes (fig. S11A) and that liver-specific ablation of Degs1 enhanced activation of Akt/PKB in vivo (Fig. 4, B and C). By comparison, the ceramide effects on lipid metabolism result from several separable mechanisms that lead to increased uptake and storage of fatty acids in the liver and decreased mobilization and oxidation of fatty acids in adipose tissue. RNA sequencing (RNA-seq) experiments revealed that Degs1 ablation dramatically lowered hepatic expression of sterol regulatory element binding transcription factor 1 (Srebf1), a master regulator of triglyceride and sterol production. Srebf1 was the most highly down-regulated gene in Degs1δLiver knockout animals fed a NCD (Fig. 4D) and sixth most down-regulated gene in Degs1δLiver knockout mice maintained on a HFD (fig. S11B). Quantitative polymerase chain reaction (PCR) experiments confirmed that Degs1 depletion decreased expression of Srebf1 and several of its transcriptional targets (e.g., Dgat1, Dgat2, Cidea, and Elovl1) that encode proteins that enhance triglyceride storage (Fig. 4E). These data are consistent with prior studies showing that exogenous ceramides induce these targets in the liver (23). Degs1 depletion inhibited fatty acid uptake into hepatocytes (Fig. 4F) owing to the ability of ceramides, but not dihydroceramides, to alter the subcellular distribution of the fatty acid translocase CD36 (Fig. 4G). Degs1 depletion enhanced mitochondrial complex activity in adipose tissue (Fig. 4H), a finding that is consistent with our published data indicating that ceramides impair complex activity in adipocytes (20). We also found that ceramides, but not dihydroceramides, inhibit hormone-sensitive lipase (HSL) (Fig. 4, I to L), blocking the release of fatty acids from the triglyceride droplet. Prior published studies, together with our additional data, indicate that the ceramide effectors protein kinase Cζ or protein phosphatase 2A likely mediate the sphingolipid’s effects on AKT (22, 2427), Srebf1 (28), CD36 (29), and HSL (Fig. 4, K and L).

Fig. 4 Ceramides, but not dihydroceramides, induce selective insulin resistance.

(A) Hepa cells were treated with vehicle, C2-ceramide (Cer), or C2-dihydroceramide (dhCer) for 2 hours before stimulation with insulin (1 μM, 10 min). Lysates were resolved by SDS-polyacrylamide gel electrophoresis, and Western blots were performed using antibodies recognizing phosphorylated Akt (p-Akt) or total Akt. (B) HFD-control and HFD-Degs1δLiver mice (as described in Fig. 1) were fasted for 5 hours before receiving a single dose of insulin (1 U/kg). The animals were euthanized 30 min later and tissues were flash frozen. p-Akt and Akt levels were determined by Western blotting as in (A). (C) C57BL/6 mice underwent the dietary and phenotyping regimen described for the intervention study (Fig. 3). Insulin stimulation of Akt/PKB phosphorylation was determined by multiplex analysis (as described in the supplementary materials). (D) RNA-seq experiments were performed on livers from control and Degs1δLiver animals under chow-fed conditions. Similar RNA-seq data are presented in the supplementary materials (fig. S11B) for animals maintained on the HFD. (E) Quantitative PCR determination of transcripts for SREBP-target genes in livers isolated from HFD-fed control, HFD-Degs1δLiver, HFD-Degs1δAdipo, and HFD-Degs1δLiver/Adipo mice (n = 3 to 5 per group, treatment regimen described in Fig. 2). (F) FFA uptake in primary hepatocytes isolated from control and Degs1Rosa26/ERT2-Cre mouse. (G) CD36 immunofluorescence in Hepa cells after treatment with either vehicle, Cer, or dhCer, as described in the supplementary materials. Scale bars, 10 μm. (H) Mitochondrial complex activity in WAT obtained from the control and Degs1Rosa26/ERT2-Cre mice maintained on a chow diet (2 weeks after receiving the five doses of tamoxifen). (I) Western blots depicting phosphorylated [pHSL(S563)] or total HSL in primary adipocytes treated with vehicle, Cer, C6-dihydroceramide (dhCer), and/or isoproterenol (Iso), as described in the supplementary materials. (J) Quantification of pHSL(S563) from (I) (n = 3 independent experiments). (K) Western blots depicting Iso-induced phosphorylation of pHSL (S563 and S660 phosphorylation sites) after treatment with Cer for 4 hours and/or the PP2A inhibitor microcystin-LR (MLR) for 1 hour in primary adipocytes. (L) Quantification of pHSL (S563) and pHSL (S660) from (K) (n = 4 independent experiments). (M) Sphingolipid flux was determined by measuring the incorporation of l-serine-13C3,15N into ceramide in primary adipocytes treated with Iso (n = 4). Values are expressed as mean ± SEM, *P < 0.05, **P < 0.001, ***P < 0.0001 versus control.

The evolutionary basis for the ceramide actions described here is of interest. Why would cells respond to ceramides by shutting down insulin signaling while simultaneously activating a parallel pathway to enhance triglyceride accumulation? We surmise that ceramides serve as gauges of free fatty acid (FFA) excess when the energetic needs of the tissue have been met and the storage capacity (i.e., triglyceride synthesis pathway) is saturated. When this occurs, the excess fatty acids spill over into the sphingolipid synthesis pathway, producing ceramides and other sphingolipids that initiate cellular responses that prevent aberrant membrane dissolution by detergent-like FFAs. The following activities are consistent with this hypothesis: (i) Ceramides (and their product sphingomyelins) are key components of detergent-resistant membrane (DRM) domains that serve as scaffolds to support FFA diffusion (3032). (ii) Ceramide induction of Srebf1 increases expression of the enzymes that convert the incoming FFAs into triglycerides (e.g., DGATs) (Fig. 4E) and synthesize cholesterol, the key binding partner of ceramide and sphingomyelin in DRMs (33). (iii) Although FFAs can diffuse through cellular membranes (3438), CD36 facilitates their uptake and enhances their esterification (39, 40). By recruiting CD36 to DRMs (29), ceramides stimulate the conversion of FFAs into acyl–coenzyme A (CoA) to mitigate their detergent-like properties. (iv) Ceramide blockage of Akt/PKB inhibits the uptake and oxidation of glucose and amino acids (1, 6), leading to the preferential utilization of FFAs for energy. An ancillary benefit is that this inhibits the synthesis of new fatty acids. (v) Ceramide inhibition of lipolysis prevents liberation of additional fatty acids from the lipid droplet. (vi) Ceramide impairment of mitochondrial complexes decreases mitochondrial efficiency, enabling the cell to burn more fatty acids before maximally altering mitochondrial membrane potential. (vii) Ceramide synthases (CERS enzymes) respond to FFA by translocating to the nucleus, where they repress transcription of lipases (41, 42), thus providing an additional means by which the ceramide pathway decreases cellular FFA levels. Although this last action would be unaffected by Degs1 depletion, the finding supports the supposition that the sphingolipid pathway’s primary purpose is to protect the cell in times of FFA excess. We conclude that chronic elevation of tissue ceramides leads to persistent impairment in metabolic homeostasis, driving insulin resistance and hepatic steatosis.

This unifying theory provides an important framework for understanding how ceramides can elicit the entire spectrum of defects that underlie cardiometabolic disease. The idea also enables the development and testing of additional hypotheses regarding the regulatory events that control fuel utilization. For example, we predicted that β-adrenergic agonists, which liberate FFAs and enhance glucose utilization during times of stress, would decrease rates of ceramide production in order to release all of the energy stores during times of acute fuel need (e.g., by releasing fatty acids, activating Akt/PKB, and limiting triglyceride storage). Using the aforementioned flux method to quantify the incorporation of stable serine isotopes into the sphingoid backbone, we determined that the β-adrenergic agonist isoproterenol completely blocked ceramide biosynthesis in adipocytes (Fig. 4M).

Our data provide important resolutions regarding the relative contribution of ceramides versus dihydroceramides in the pathogenesis of metabolic diseases. Although in vitro studies have suggested that ceramides, rather than dihydroceramides, likely contributed to the inhibition of insulin signaling, several lipid profiling studies have identified dihydroceramides as particularly strong markers of metabolic dysfunction (7, 9). In particular, it has been reported that circulating dihydroceramides were strong prognostic indicators of future development of type 2 diabetes, appearing as early as 9 years before disease onset (8). We instead propose that dihydroceramides are not causative but likely serve as markers of increased flux of fatty acids through the biosynthetic pathway that produces the toxic ceramides. Our findings presented here indicate that ceramides contribute to metabolic disease and may offer an opportunity for clinical intervention.

Supplementary Materials

science.sciencemag.org/content/365/6451/386/suppl/DC1

Materials and Methods

Figs. S1 to S11

References (4346)

References and Notes

Acknowledgments: We are grateful for support from the Metabolomics, Histology, Genomics, and Metabolic Phenotyping Cores at the Health Sciences Center of the University of Utah. Funding: The authors receive research support from the National Institutes of Health (DK112826 and DK108833 to W.L.H. and DK115824 and DK116450 to S.A.S.), the Juvenile Diabetes Research Foundation (JDRF 3-SRA-2019-768-A-B to W.L.H.), the American Diabetes Association (S.A.S.), the American Heart Association (S.A.S.), the Margolis Foundation (S.A.S.), and the USDA (2019-67018-29250 to B.C.). B.C. received a pilot grant from the Diabetes Research Center at Washington University in St. Louis of the NIH under award number P30DK020579. R.F.P. received a doctoral fellowship from FAPESP (process number 2014/17619-6). T.S.T. and J.L.W. received support from the NIH through the Ruth L. Kirschstein National Research Service Award 5T32DK091317. Studies conducted at Merck Research Laboratories in Kenilworth, NJ, were supported by Merck & Co., Inc. Author contributions: S.A.S., J.L., D.E.K., and B.C. conceived of the project, designed the experiments, and wrote the manuscript. B.C., T.S.T., R.M.M., J.L., Y.L., L.W., J.L.W., A.Po., C.R.S., R.F.P., V.K., G.I.L., M.M.S., S.S., H.Z., D.G.M., S.F.P., Y.C., Y.Q., A.Pe., M.W., X.S., J.Y., C.N.N., A.D.H., L.W., M.D.E., D.H.S., J.R., and W.L.H. performed experiments and analyzed data. J.A.M., J.E.C., M.P. aided in lipid quantification. Competing interests: S.A.S. is a cofounder, J.R. is on the board of directors, and D.E.K. is a consultant of Centaurus Therapeutics. R.M.M., J.L., S.S., H.Z., D.G.M., S.F.P., Y.C., Y.Q., A.Pe., M.W., X.S., J.Y., C.N.N., A.D.H., L.W., M.D.E., and D.E.K. are or were employees of Merck Research Laboratories. Data and materials availability: Degs1fl/fl mice can be provided upon completion of a material transfer agreement to S.A.S. All data are available in the manuscript or the supplementary materials. RNA-seq data have been deposited in Gene Expression Omnibus under the accession number GSE132056.
View Abstract

Stay Connected to Science

Navigate This Article