Fibroblast growth factor 21 reflects liver fat accumulation and dysregulation of signalling pathways in the liver of C57BL/6J mice

Abstract

Fibroblast growth factor 21 (Fgf21) has emerged as a potential plasma marker to diagnose non-alcoholic fatty liver disease (NAFLD). To study the molecular processes underlying the association of plasma Fgf21 with NAFLD, we explored the liver transcriptome data of a mild NAFLD model of aging C57BL/6J mice at 12, 24, and 28 months of age. The plasma Fgf21 level significantly correlated with intrahepatic triglyceride content. At the molecular level, elevated plasma Fgf21 levels were associated with dysregulated metabolic and cancer-related pathways. The up-regulated Fgf21 levels in NAFLD were implied to be a protective response against the NAFLD-induced adverse effects, e.g. lipotoxicity, oxidative stress and endoplasmic reticulum stress. An in vivo PPARα challenge demonstrated the dysregulation of PPARα signalling in the presence of NAFLD, which resulted in a stochastically increasing hepatic expression of Fgf21. Notably, elevated plasma Fgf21 was associated with declining expression of Klb, Fgf21's crucial co-receptor, which suggests a resistance to Fgf21. Therefore, although liver fat accumulation is a benign stage of NAFLD, the elevated plasma Fgf21 likely indicated vulnerability to metabolic stressors that may contribute towards progression to end-stage NAFLD. In conclusion, plasma levels of Fgf21 reflect liver fat accumulation and dysregulation of metabolic pathways in the liver.

Figure 1. Physiological changes induced by the different dietary interventions at the age of 6, 12, 24, and 28 months.

Body weight (a) liver weight (b) and IHTG content (c) dramatically increased over time, except for the CR-fed animals which stayed lean. Significance (p-value) of age, diet and interaction were evaluated using two-way ANOVA. Error bars represent s.e.m. Prevalence of hepatic steatosis (e) and fibrosis (f) in the mice aging cohort at different age time points.

Figure 2. Plasma Fgf21 reflected the hepatic fat accumulation at middle and old age, but not at mature adult age.

(a) Different plasma Fgf21 levels induced by the dietary interventions at the age of 6, 12, 24, and 28 months. Correlation of IHTG and plasma Fgf21 for different dietary interventions (b) and ages (c). (d) Plasma Fgf21 was positively correlated with IHTG levels at older ages, while the young 6-month-old mice had elevated plasma Fgf21, despite their low IHTG levels (in red symbols). r values were calculated with Pearson’s correlations and their significance are indicated in the parentheses.

Figure 3. Plasma Fgf21 levels were strongly reflected by the expression of Fgf21 in the liver.

(a) Different expression levels of Fgf21 in the liver induced by the dietary interventions and age. (b) Significant positive correlation between plasma Fgf21 and Fgf21 expression in the liver. The association between plasma and expression levels of Fgf21 was also compared in other tissue types, (c) epidydimal white adipose tissue (eWAT), (d) muscle, and (e) colon tissue. r values were calculated with Pearson’s correlations and their significance are indicated in the parentheses.

Figure 4. Liver biological processes associated with plasma Fgf21 level.

(a) Heatmap depicting the correlation between gene modules (in rows) and phenotypes (in columns). The top values in each cell represents the correlation coefficient between the module and phenotype with the correlation p-value in parentheses. Red and green color represents positive and negative correlation, respectively. Correlation between plasma Fgf21 levels and hepatic expressions of Cd36 (b) and Klb (c). (d) Biological processes and regulators associated with MEturquoise, which is the most significant modules correlated with plasma Fgf21. Significant liver-specific functions and upstream regulators are reported in p-values and z-scores. Positive and negative z-score represent predicted activation and inhibition, respectively. (e) Correlation between plasma Fgf21 and expression levels of hepatocellular carcinoma-related genes within MEturqouise, which included genes related to extracellular matrix formation and angiogenesis (Col1a2 and Mmp14), inhibition of Wnt signalling (Frzb and Dkk3), and downstream target of Wnt signalling (Glul and Ccnd1). r values were calculated with Pearson’s correlations and their significance are indicated in the parentheses.

Figure 5. The differentially regulated pathways in NAFLD were reflected by plasma Fgf21 levels.

(a) Stress induced by NAFLD may activate NRF2, PPARα and MAPK, and the expression of their target genes. The expression levels of the NRF2 (b) PPARα (c) and MAPK (d) target genes were strongly correlated with IHTG and more modestly with plasma Fgf21 levels (in black and red color, respectively). r values were calculated with Pearson’s correlations and their significance are indicated in the parentheses.

Figure 6. The response to PPARα challenge test showed that the presence of NAFLD partially altered the response at the gene expression levels.

(a) Fgf21; (b) Mogat1; (c) G0s2; (d) Acot3; (e) Hmgcr; and (f) Pparα. The effect of PPARα agonist and NAFLD, as well as any interaction between them, were analysed using two-way ANOVA.