Whole transcriptome analysis and validation of metabolic pathways in subcutaneous adipose tissues during FGF21-induced weight loss in non-human primates

Figures

• 

  Figure 1. FGF21 treatment and weight loss associated transcriptional changes in AT. Gene expression in AT was measured by RNA sequencing. (a) Fold change and statistical significance of genes DE between treatment time points. Genes with fold change >0.26 or < −0.26 and FDR < 0.05 are highlighted in red. (b) Fold change of genes between pre- and post-treatment groups against the coefficient estimate for the association between change in gene expression and change in body weight, as determined by linear regression. Blue, significant DE genes in pre vs post comparison (as in A); red, genes with FDR < 0.25 in pre vs post that are also significantly associated with change in body weight, FDR < 0.25. Select genes of interest are labeled in black. n = 8–10 per time point in adipose tissue and n = 6–8 in skeletal muscle.

• 

  Figure 2. Correlation network analysis reveals several metabolism related gene modules in AT. WGCNA was used to group genes with coordinated expression into modules within AT. The relationship between select modules annotated with metabolic disease-related pathways is shown in a correlation network. Nodes are gene modules and edges represent the Spearman correlation between module eigengenes (blue, positive; red, negative). Only correlations stronger than r = 0.6 or −0.6 are shown. Node size depicts the number of genes in the module. Node color represents strength and direction of association of change in module expression with change in body weight. Bottom, short descriptions of modules included in the network, based on their overlap with known biological pathways (supplemental table S3). ECM, extracellular matrix.

• 

  Figure 3. FGF21-induced weight loss modulates adipogenesis, fat browning, lipid metabolism, and inflammatory pathways. (a) Expression levels of metabolism and immune-related gene module eigengenes by time point and (b) the association between change in module expression between pre- and post-treatment and decrease in body weight. White diamonds indicate mean expression at each time point. * FDR < 0.1, + FDR < 0.25 between pre- and post-treatment groups by Wilcoxon paired test (n = 8–10) (a), or by linear regression between change in module expression and change in body weight (n = 9) (b).

• 

  Figure 4. Divergent responses of genes involved in thermogenesis. (A) Gene expression heat map depicting select genes related to metabolic and inflammatory pathways. Each column depicts one sample from one treatment time point and animals appear in the same order within each time point. Each row represents one gene, and rows are hierarchically clustered to group genes with similar expression patterns. The module and the general ontological category each gene is a member of is depicted in columns between the heat map and gene names. n = 8–10 per group. +, genes with significantly different expression between pre and post time points (FDR < 0.05); o, genes whose changes in expression were significantly associated with decreased body weight (FDR < 0.25).

• 

  Figure 5. Overexpression of ANGPTL8 increased plasma triglyceride levels. ANGPTL8 was overexpressed in mice by hydrodynamic gene delivery. (a) Expression levels and (b) association with change in body weight in FGF21-treated NHP. (c-f) Mice were given empty plasmid, or plasmid encoding Exendin-4, N-terminal or C-terminal mouse albumin-fused human ANGPTL8 sequence. (c) Change in body weight and (d) cumulative food intake over time. (e) Blood glucose after glucose tolerance test and (f) plasma triglycerides at day 14. NS non-significant; (a-b) + FDR < 0.25 by linear model; (c-f) Data are displayed as mean plus standard error, n = 7–8 per group, *p < 0.05, **p < 0.01, ***p < 0.001 by one-way ANOVA with Dunnett’s post-hoc test.

• 

  Figure 6. Overexpression of NRG4 reduced weight gain in DIO mice. NRG4 was overexpressed in mice by hydrodynamic gene delivery. (a) Expression levels and (b) association with change in body weight in FGF21-treated NHP. (c–f) Mice were given empty plasmid, or plasmid encoding mouse FGF21, native mouse Nrg4, or mouse Nrg4 fused to human IgG1Fc. (c) Change in body weight over time. (d) Change in fat mass from day 0 to day 24. (e) Blood glucose after glucose tolerance test and (f) plasma triglycerides at day 14. NS non-significant; (a,b) ** FDR < 0.01, ***FDR < 0.001 by linear model; (c–f) Data are displayed as mean plus standard error, n = 7–8 per group, *p < 0.05, **p < 0.01, ***p < 0.001 by one-way ANOVA with Dunnett’s post-hoc test.

• 

  Figure 7. Summary of transcriptomic changes in AT in response to FGF21-induced weight loss. FGF21 treatment and resulting weight loss in NHP induced transcriptional changes in several metabolic pathways in AT, as measured by RNA sequencing of tissue biopsy punches. As a biopsy punch of adipose tissue contains a mixture of cell types, some of the transcriptional changes observed may originate from innervating neurons (green), immune cells (blue), or adipocytes (yellow). Blue arrows signify pathways or genes that were increased post-treatment compared to pre-treatment, and whose increase was associated with greater weight loss. Red arrows signify pathways or genes that were decreased at the post-treatment time point and whose increase was associated with greater weight loss.