Differential metabolic and multi-tissue transcriptomic responses to fructose consumption among genetically diverse mice

Abstract

Understanding how individuals react differently to the same treatment is a major concern in precision medicine. Metabolic challenges such as the one posed by high fructose intake are important determinants of disease mechanisms. We embarked on studies to determine how fructose affects differential metabolic dysfunctions across genetically dissimilar mice, namely, C57BL/6 J (B6), DBA/2 J (DBA) and FVB/NJ (FVB), by integrating physiological and gene regulatory mechanisms. We report that fructose has strain-specific effects, involving tissue-specific gene regulatory cascades in hypothalamus, liver, and white adipose tissues. DBA mice showed the largest numbers of genes associated with adiposity, congruent with their highest susceptibility to adiposity gain and glucose intolerance across the three tissues. In contrast, B6 and FVB mainly exhibited cholesterol phenotypes, accompanying the largest number of adipose genes correlating with total cholesterol in B6, and liver genes correlating with LDL in FVB mice. Tissue-specific network modeling predicted strain-and tissue-specific regulators such as Fgf21 (DBA) and Lss (B6), which were subsequently validated in primary hepatocytes. Strain-specific fructose-responsive genes revealed susceptibility for human diseases such that genes in liver and adipose tissue in DBA showed strong enrichment for human type 2 diabetes and obesity traits. Liver and adipose genes in FVB were mostly related to lipid traits, and liver and adipose genes in B6 showed relevance to most cardiometabolic traits tested. Our results show that fructose induces gene regulatory pathways that are tissue specific and dependent on the genetic make-up, which may underlie interindividual variability in cardiometabolic responses to high fructose consumption.

Keywords: Fructose; Insulin resistance; Metabolic syndrome; Obesity; Personalized nutrition; Transcriptome.

Fig. 1.. Body weight and body composition changes and caloric intake in three strains of mice in response to fructose consumption.

A–C: Cumulative change in body weight in B6 (A), DBA (B) and FVB (C) mice fed normal Chow diet with water or 8% fructose over 12 weeks. Two-way ANNOVA was used to test the differences between treatment groups and time points, followed by Sidak post-hoc analysis to examine treatment effects at individual time points. * denotes P < 0.05 from the post-hoc analysis. D: body weight gain at the end of experiment. E-H: Body composition change in three strains of mice, with fat mass (E), percent fat mas (F), lean mass (G), percent lean mass (H) measured using NMR. I: Individual fat masses of DBA mice. BAT: brown adipose tissue; gWAT: gonadal white adipose tissue; rWAT: retroperitoneal white adipose tissue; mWAT: mesenteric white adipose tissue; scWAT: subcutaneous white adipose tissue. J: Caloric intake of three stains of mice. Error bars in the graph are standard errors. E-K: * denotes P < 0.05 and ** denotes P < 0.01 by two-sided Student’s t-test. Sample size n=8–12/group/strain.

Fig. 2.. Glucose metabolism and lipid changes in three strains of mice in response to fructose consumption.

A–C: Glucose tolerance analysis using IPGTT were conducted at baseline (0 day), 1^st, 4^th,9^th and 12^th week shown as area under the curve (AUC). D: Plasma glucose. E: Plasma insulin. F: Plasma triglyceride. G: Plasma total cholesterol. H: HDL. I: LDL. J: Unesterified cholesterol. K: Free fatty acid levels. A–K: sample size n=8–10 per group. *P < 0.05 and **P < 0.01 by two-sided Student’s t-test.

Fig. 3.. Differentially expressed genes (DEGs) and representative over-represented pathways in adipose, liver and hypothalamus of three mouse strains.

A–C: Venn diagrams of adipose (A), liver (B), and hypothalamus (C) DEGs and pathways across three strains. D: Fold change and significance of gene expression differences between fructose and control groups for known target genes of fructose, ChREBP (encoded by Mlxipl), SREBP-1c (encoded by Srebf1) in the three mouse strains. Fold change for each gene is the normalized average expression level in fructose mice over that of the control mice. * P < 0.05, ** P < 0.01, ^# FDR<5%, ^## FDR<1% in RNAseq analysis using a linear model. Sample size n=6/group/tissue/strain for liver; n=3–6/group/tissue/strain for adipose tissues; n=4/group/strain for hypothalamus tissue.

Fig. 4.. Gene subnetworks and top network key drivers (KDs) of DEGs in three strains of mice and validation of select strain-specific liver KDs in primary hepatocytes.

A: Top adipose KDs and subnetworks. B: Top liver KDs and subnetworks. C: Top hypothalamus KDs and subnetworks. D: Direct response to fructose for DBA-specific KD Fgf21 is stronger in DBA hepatocytes than in B6 hepatocytes. E: Direct response to fructose for B6-specific KD Lss is stronger in B6 hepatocytes than in DBA hepatocytes. D–E: P values for strain, fructose concentration, and time points are from 3-way ANOVA. At each concentration and time point, p values between mouse strains are from Tukey post-hoc analysis. Sample size n=3 (2 replicates/mice) per strain per concentration per time point. F: Expression changes of genes in Fgf21-driven network after siRNA-mediated knockdown of Fgf21 in DBA primary hepatocytes. Genes surrounding Fgf21 were selected from Fig. 4B. G: Expression changes of genes in Lss-driven network after siRNA-mediated knockdown of Lss in B6 primary hepatocytes. Genes surrounding Lss were selected from Fig. 4B. H: Gene expression of DEGs not shown in Fgf21/Lss gene networks (negative controls). F-H: NC represents scrambled siRNAs as negative controls. *P < 0.05 and **P < 0.01 by two-sided Student’s t-test. Sample size n=3 (2 replicates/siRNA).

Fig. 5.. Relationship between DEGs and metabolic traits in mice.

A: Numbers of strain-specific DEGs that are correlated with metabolic traits in our fructose study (P < 0.01 by Pearson correlation). B–D: Select examples of DEGs in liver correlating with adiposity in DBA mice. E–G: Select examples of DEGs in liver correlating with LDL in FVB mice.