Reduced circulating sphingolipids and CERS2 activity are linked to T2D risk and impaired insulin secretion

Abstract

Gestational diabetes mellitus (GDM), a transient form of diabetes that resolves postpartum, is a major risk factor for type 2 diabetes (T2D) in women. While the progression from GDM to T2D is not fully understood, it involves both genetic and environmental components. By integrating clinical, metabolomic, and genome-wide association study (GWAS) data, we identified associations between decreased sphingolipid biosynthesis and future T2D, in part through the rs267738 allele of the CERS2 gene in Hispanic women shortly after a GDM pregnancy. To understand the impact of the CERS2 gene and risk allele on glucose regulation, we examined whole-body Cers2 knockout and rs267738 knock-in mice. Both models exhibited glucose intolerance and impaired insulin secretion in vivo. Islets isolated from these models also demonstrated reduced β cell function, as shown by decreased insulin secretion ex vivo. Overall, reduced circulating sphingolipids may indicate a high risk of GDM-to-T2D progression and reflect deficits in CERS2 activity that negatively affect glucose homeostasis and β cell function.

Fig. 1.. The study design and chemometric analysis of metabolomics (i.e., metabolite profile + lipid profile).

(A) Overall study design. The SWIFT recruited 1035 pregnant women with GDM and included 1010 women who returned to a nondiabetic state after delivery. Within 8 years of follow-up, 219 women had progressed to T2D. The current nested case-control study examines 143 Hispanic women within the SWIFT cohort. Plasma profiling was performed on baseline samples taken at 6 to 9 weeks postpartum. Metabolomic and public GWAS data were integrated using metGWAS to explore potential genetic predispositions. Mechanistic studies on murine models and human samples offer a comprehensive investigation of the in silico analysis. (B) Principal component comparison. (C) A 2D-score plot of PLS-DA is to show the separation between case and control in the baseline (6 to 9 weeks postpartum) metabolomics data. (D) This graph describes the R2X, R2Y, and Q2 values of component 1. (E) A total of 1032 analytes were used for differential expression analysis. The pie chart shows the distribution of these analytes in different chemical groups. (F) Empirical Bayesian analysis plot, delta 0.1, FDR < 0.05. (G) The log₂(fold change) distribution of significantly altered analytes between the incident case and control group (FDR < 0.05).

Fig. 2.. Investigating underlying pathophysiology of metabolites associated with T2D development.

(A) Overall bioinformatic workflow using metGWAS. (B) KEGG pathway analysis identified sphingolipid metabolism as the only pathway with a large impact on T2D development. Pathway impact, as shown by the size of the point, indicates how “important” or relevant a given pathway is to a given disease or outcome, with 0 having no impact and 1 having maximal impact. (C) Roles of different ceramide synthase enzymes in synthesizing different sphingolipids (i.e., ceramides, sphingomyelins, etc.) with different chain lengths. (D) Simplified sphingolipid metabolism with significantly altered metabolites and inferred enzymes. (E) Genetic interaction analysis revealed 10 significantly associated gene loci for down-regulated sphingolipid metabolism. Arrows depict the possible causal influence of gene loci on T2D-related traits (e.g., BMI), as shown in the GWAS catalog database.

Fig. 3.. Female Cers2^−/− mice phenotype characterization.

(A) A flow diagram of in vivo mouse studies. (B and C) Body weight and fasting glucose measurements between Cers2^+/+, Cers2^+/−, and Cers2^−/− at 12 to 14 weeks of age (n = 10). (D) The GTT at 12 to 14 weeks of age (n = 10). The calculated area under the curve (AUC) of IpGTT has been inserted. (E) The insulin secretion during IpGTT (n = 8). (F) ITT (n = 11). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.

Fig. 4.. Blood profile and liver studies of female CerS2 mice.

(A) A flow diagram for blood chemistry profiling with the basic structure of different ceramides and sphingomyelins. The attached fatty acids (R) determine the chain length by the function of different ceramide synthase enzymes (i.e., Cers1 to Cers6). These enzymes have specificity for the chain length of the fatty acids. (B to G) Circulating sphingolipid profiling [sphingosines (So) and sphinganines (Sa)] (n = 3 mice per group). (H to M) Liver sphingolipid profiling (n = 3 mice per group). (N) Immunoblot for Cers2 expression in the liver. (O) The cholesterol, triglyceride, ALT, and AST concentrations in the blood of these mice (n = 3 mice per group). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.

Fig. 5.. Islet biology of female Cers2 phenotypes.

(A) A flow diagram of islets biology studies. (B) Immunoblot for Cers2 in islets. (C) The GSIS of isolated islets was measured by low glucose, high glucose, and KCl stimuli (n = 5). (D) Total insulin quantification (n = 6). (E) The representative pancreatic histology for the insulin-positive area. (F) Pancreatic histology analysis for insulin staining (n > 3). (G) The representative pancreatic histology for the glucagon-positive area. (H) Pancreatic histology analysis for glucagon-positive cells (n > 3). (I) The representative immunofluorescence images. (J and K) Immunofluorescence analyses for insulin and glucagon in dispersed islets using cytospin (n ≥ 4). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.

Fig. 6.. Islet membrane rigidity.

(A) The flow diagram of the membrane integrity (i.e., membrane rigidity) experiment using laurdan fluorescence in confocal laser scanning microscopy. The membrane rigidity is calculated using GP values [GP = (I2 − I1)/(I1 + I2)], where I2 represents blue, and I1 represents the red color. (B) A representative mouse islet from each group (i.e., the blue indicates rigidity). (C) A quantification of mouse islet membrane rigidity by calculating GP values [GP = (I2 − I1)/(I1 + I2)] (at least five islets were used from three mice of each group). (D) A flow diagram of the glucose uptake studies showed that islets were treated by 2-NBDG and measured the fluorescence using spinning disc confocal microscopy. (E) The cumulative fluorescence intensity was measured in glucose uptake studies (n = 3). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.

Fig. 7.. Cers2 polymorphic mouse characterization.

(A) A flow diagram to depict the Cers2 polymorphic mouse characterization. (B and C) The genotype and immunoblots show no difference in Cers2 transcript and protein levels; however, Sanger sequencing revealed the difference at the sequence level. (D) Comparison of temporal body weights between Cers2 polymorphic mice and their wild-type controls (n ≥ 6). WT, wild type. (E and F) CerS2 polymorphic mice exhibited no significant difference in fasting glucose, and fasting insulin under the chow diet (n = 10). (G) IpGTT on 12- to 14-week-old mice (n = 10). The calculated AUC from IpGTT has been inserted. (H) The secreted insulin measurement from the IpGTT with inserted AUC (n = 10). (I) IpITT analysis (n ≥ 6). (J and K) GSIS and total insulin analyses on Cers2 polymorphic mouse islets (n ≥ 6). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.

Fig. 8.. Human islet biology upon CERS inhibition using 300 μM FB1.

(A) A flow diagram of the human islet studies. DMSO, dimethyl sulfoxide. (B) The GSIS study (n = 6). (C) The total insulin contents (n = 6). (D and E) The membrane integrity loss experiments using laurdan fluorescence dye. The membrane rigidity is calculated using GP values [GP = (I2 − I1)/(I1 + I2)], where I2 represents blue and I1 represents red color. (F) The glucose uptake experiments using 2-NBDG fluorescence dye on healthy human islets (at least five islets from each donor, n = 3). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.005.