Fig 2: Effects of two 12-week energy restriction diets (HQ, high-nutrient-quality diet; LQ, low-nutrient-quality diet) or control group (CON) on fasting and postprandial plasma FGF21 levels. (A) Geometric mean (95% CI) change in fasting plasma FGF21 upon a 12-week diet, as tested by ANCOVA with adjustment for baseline fasting FGF21 levels. (B) Geometric mean (95% CI) change in postprandial FGF21 response (∆120–0 min) to a mixed meal upon a 12-week diet, as tested by univariate ANOVA.

Fig 3: Postprandial FGF21 response at baseline and correlations between plasma FGF21 and markers of metabolic health. (A) Box plot with individual data points of the change in plasma FGF21 levels from the fasting state to 120 min after the consumption of the mixed meal in the complete population at baseline (p = 0.055, as tested using a paired-samples T-test). The box plot represents the 5th percentile, first quartile, median, third quartile, and 95th percentile. (B) Individual postprandial FGF21 responses according to baseline HOMA-IR: ≤2.5 or >2.5 (p = 0.012 for the difference between groups, as tested with ANCOVA, adjusted for fasting FGF21, age, gender, and BMI. (C) Spearman correlations between fasting FGF21 and the postprandial FGF21 response with cardiometabolic parameters at baseline (left) and in response to a 12-week intervention (right). Abbreviations: BMI, body mass index; AT, adipose tissue; IHL, intra-hepatic lipid content; HbA1c, glycated hemoglobin A1c; iAUC, incremental area under the curve; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; ALAT, aminotransferase; ASAT, aspartate aminotransferase; γGT, gamma-glutamyl transferase.

Fig 4: Fasting plasma FGF21 levels (adjusted geometric means with 95% CI) according to tertiles (T1: lowest tertile, T3: highest tertile) of habitual (macro)nutrient intake as % of daily energy intake (en%) (A–C,E,F), habitual alcohol consumption (D), and sweet-taste preference (G). FGF21 levels were lower in the highest tertile of polysaccharide intake compared to the lower tertiles (overall p = 0.022; tested by ANCOVA with adjustment for gender and LSD post-hoc testing; * p = 0.035; ** p = 0.009) and did not differ between tertiles of other nutrient intakes.

Fig 5: RNA sequencing using LPS-induced ALI tissues with or without FGF21 gene deletion and bioinformatics analysis. a Diagrams showed the GO analysis for DEGs representing an association between impacted GO-BP terms and related DEGs. b Diagrams showed the KEGG analysis for DEGs to investigate the potential pathways. c–e Western blotting results and analysis of p-JAK2/JAK2, p-STAT3/STAT3, and p-STAT4/STAT4 in LPS-induced ALI model of WT mice and FGF21KO mice as well as their controls (n = 3). f–h Western blotting results and analysis of p-JAK2/JAK2, p-STAT3/STAT3, and p-STAT4/STAT4 in LPS-induced mice treated with or without FGF21 (n = 3). Multiple comparisons were analyzed by one-way ANOVA. *P < 0.05 and **P < 0.01, as indicated. Data are presented as mean ± SD.