Activin B improves glucose metabolism via induction of Fgf21 and hepatic glucagon resistance

Abstract

Orchestrated hormonal interactions in response to feeding and fasting play a pivotal role in regulating glucose homeostasis. Here, we show that in obesity, the production of follistatin-like 3 (FSTL3), an endogenous inhibitor of Activin B, in adipose tissue is increased in both mice and humans. The knockdown of FSTL3 improves insulin sensitivity and glucose tolerance in diabetic obese db/db mice. Notably, the overexpression of Activin B, a member of the TGFβ superfamily that is induced in liver sinusoidal endothelial cells by fasting, exerts multiple metabolically beneficial effects, including improvement of insulin sensitivity, suppression of hepatic glucose production, and enhancement of glucose-stimulated insulin secretion, all of which are attenuated by the overexpression of FSTL3. Activin B increases insulin sensitivity and reduces fat by inducing fibroblast growth factor 21 (FGF21) while suppressing glucagon action in the liver by increasing phosphodiesterase 4 B (PDE4B), leading to hepatic glucagon resistance and resultant hyperglucagonemia. Activin B-induced hyperglucagonemia enhances glucose-stimulated insulin secretion by stimulating glucagon-like peptide-1 (GLP-1) receptor in pancreatic β-cells. Thus, enhancing the action of Activin B which improves multiple components of the pathogenesis of diabetes may be a promising strategy for diabetes treatment.

Fig. 1. FSTL3 was increased in obesity and negatively modulated glucose homeostasis.

a, b Relative expression levels of FSTL3 in visceral (a) or subcutaneous fat (b) of subjects. n = 188 (male n = 86; female n = 102). c, d Negative correlation of FSTL3 mRNA expression in visceral (c) or subcutaneous fat (d) with whole-body insulin sensitivity. The glucose infusion rate (GIR) was determined by hyperinsulinemic-euglycemic clamp studies. e, f Blood glucose levels of DIO mice received Ad-LacZ or Ad-Fstl3 during GTT (e, n = 7 mice per group) or ITT (f, Ad-LacZ; n = 6 mice, Ad-Fstl3; n = 7 mice). g Blood glucose levels of male db/db mice treated with a control antisense oligonucleotide (ASO) or Fstl3-targeting ASO (described in “Materials and method”) during GTT. h Relative expression levels of Fstl3 in the adipose tissues of mice used in (g). n = 8 mice per group. Data were shown as mean ± SEM. Statistical significance was determined by Spearman’s rank correlation test (a–d); unpaired two-tailed Student’s t-test (e, g, h); two-way ANOVA with Šídák’s multiple comparison test (e, f) as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data File.

Fig. 2. Overproduction of Activin B improved glucose metabolism in obese mice.

a, b Blood glucose (a) and plasma insulin levels (b) of DIO mice treated with Ad-LacZ or Ad-INHBB during GTT. c Blood glucose levels during ITT. n = 8 per group. d–f Blood glucose (d) and plasma insulin levels (e) of DIO mice treated with Ad-INHBB with or without Ad-FSTL3 during GTT. f Blood glucose levels during ITT. Ad-LacZ; n = 8, Ad-INHBB; n = 7, Ad-INHBB + Ad-FSTL3; n = 8. g Relative expression levels of Inhibin β subunits (Inhba, Inhbb, Inhbc, and Inhbe) in liver. n = 5 per group. h–k Relative expression levels of Inhbb gene in the liver of ad libitum fed, fasted, or refed mice (h, n = 5), fractionated liver of wild-type mice (i, n = 5), isolated CD11b+ cells and CD146+ cells (LSECs) from the liver of ad libitum fed (white bars) or fasted (black bars) mice (j, n = 3), or in liver tissues of conditional Inhbb knockout mice (k, the sample size (n) is indicated within the graph.). l–s Metabolic phenotyping in DIO mice treated with AAV-TBG-EGFP or AAV-TBG-INHBB mice. l, m Body weight changes (l) and ambient blood glucose levels (m). EGFP; n = 14, INHBB; n = 18. n–p Blood glucose levels during GTT (n, n = 8) or ITT (o, EGFP; n = 8, INHBB; n = 6) and plasma insulin levels during GTT in (n). q Whole-body energy expenditure measured in a metabolic cage. n = 8 mice per group. r Relative expression levels of genes in BAT. GFP; n = 6, INHBB; n = 8. s Representative images of H&E staining of BAT. Data were shown as mean ± SEM. Statistical significance was determined by unpaired two-tailed Student’s t-test (a, g, n, r); two-way ANOVA with Šídák’s (a–c, l–n, o–q), or Dunnett’s (d–f) multiple comparison test; one-way ANOVA Sidak’s multiple comparison test (h) as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. #p < 0.05, ##p < 0.01, ###p < 0.001 versus Ad-INHBB treated mice in (d–f). Source data are provided as a Source Data File.

Fig. 3. Activin B improved insulin sensitivity via the production of FGF21.

a, b Relative expression levels of Fgf21 in liver (a, n = 8) and plasma FGF21 concentration (b, n = 6) of mice treated with Ad-LacZ or Ad-INHBB. c, d Relative expression levels of Fgf21 (c) and Inhbb (d) in the liver. n = 5 mice per group. e Dose-dependent effects of recombinant Activin B protein on Fgf21 expression. n = 3. f–k Relative expression levels of Fgf21 in the liver (f), body weight changes (g), body fat (h), blood glucose levels during ITT (i), blood glucose levels (j), and plasma insulin levels (k) during GTT of wild-type or Fgf21 deficient mice treated with indicated AAV. (WT, EGFP; n = 7, WT, INHBB; n = 7, Fgf21 KO, EGFP; n = 6, Fgf21 KO, INHBB; n = 7). Data were shown as mean ± SEM. Statistical significance was determined by unpaired two-tailed Student’s t-test (a, b); two-way ANOVA with Tukey’s (c, d, g, i–k) or Šídák’s (f, h, j) multiple comparisons test; one-way ANOVA with Dunnett’s multiple comparisons test (e) as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data File.

Fig. 4. Activin B suppressed hepatic gluconeogenesis in an insulin-independent manner.

a, b Blood glucose (a) and plasma insulin levels (b) of STZ mice treated with Ad-LacZ or Ad-INHBB during GTT. Ad-LacZ; n = 8, Ad-INHBB; n = 7. c Relative expression levels of genes regulating hepatic gluconeogenesis of STZ mice treated with Ad-LacZ or Ad-INHBB. Ad-LacZ; n = 6, Ad-INHBB; n = 4. d PEPCK protein levels in the liver from 4 independent animals. e Blood glucose levels of STZ mice during PTT. n = 5 mice per group. f, g Fasting blood glucose levels (f) or blood glucose levels during PTT (g) in STZ-treated Inhbb-Lyve1 KO or control mice. Inhbb-Flox; n = 6, Inhbb-Lyve1 KO; n = 9. h Glucose production in primary hepatocytes treated with recombinant Activin B. n = 4. i–k Relative expression levels of Pck1 treated with recombinant Activin B with or without Forskolin and Dexamethasone (Frk/Dex) (i, n = 3), subtype-specific ALK inhibitors (j, n = 3), or treated with adenovirus encoding constitutive active mutant of ALK2 (Ad-ALK2(QD)) (k, n = 3). Data were presented as means ± SEM. Statistical significance was determined by unpaired two-tailed Student’s t-test (a, f, g); two-way ANOVA (g) or with Šídák’s (a, c, j), Dunnett’s (e), or Tukey’s (i) multiple comparison test, one-way ANOVA with Dunnett’s (e) or Šídák’s (h, k) multiple comparison test as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. #p < 0.05 versus normal mice in (e). Source data are provided as a Source Data File.

Fig. 5. Hepatic glucagon signaling was suppressed by Activin B.

a–c Blood glucose (a, n = 6 mice per group), hepatic glucagon signaling (b), and cAMP contents (c, the sample size (n) is indicated within the graph) in response to glucagon in mice treated with Ad-LacZ or Ad-INHBB. d–f Blood glucose (d, n = 6 mice per group), hepatic glucagon signaling (e), and cAMP contents (f, the sample size (n) is indicated within the graph) in response to glucagon in Inhbb-Lyve1 KO mice. g, h Plasma glucagon levels in fasted adenovirus treated wild-type mice (g, n = 6 mice per group) or Inhbb-Lyve1 KO mice (h, n = 6 mice per group). i, j Relative expression levels of genes encoding phosphdiesterase four subtypes in liver of mice treated with Ad-LacZ or Ad-INHBB (i, n = 3 mice per group) or Inhbb-Lyve1 KO mice (j, n = 3 mice per group). k–m Changes in hepatic cAMP contents (k, the sample size (n) is indicated within the graph) or blood glucose (l, the sample size (n) is indicated within the graph) in response to glucagon challenge and plasma glucagon levels (m, n = 5 mice per group) in mice treated with Ad-LacZ or Ad-INHBB with or without PDE4 inhibitor (PDE4i). Data were shown as mean ± SEM. Statistical significance was determined by unpaired two-tailed Student’s t-test (a, d, g–j); two-way ANOVA (a, d) or with Šídák’s (c, f), or Tukey’s (k–m) multiple comparison test as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data File.

Fig. 6. Activin B enhanced glucose-stimulated insulin secretion.

a, b Plasma levels of insulin (a) and c-peptide (b) in response to bolus glucose challenge (in vivo GSIS test). n = 6 mice per group. c, d Plasma levels of insulin (c) and c-peptide (d) of Inhbb-Lyve1 KO mice during in vivo GSIS test (Inhbb-Flox; n = 7, Inhbb-Lyve1 KO; n = 9). e Effects of Activin B on glucose-stimulated insulin secretion (GSIS). Isolated islets were cultured with or without Activin B and stimulated with high glucose (22.2 mM l-glucose). Data were presented as secreted insulin concentration normalized to insulin content. n = 6. f Pancreatic insulin contents. n = 5. g–i Plasma levels of glucagon (g), active GLP-1 (h), and active GIP (i) in mice received Ad-INHBB mice during in vivo GSIS test. n = 6 mice per group. j, k Plasma levels of glucagon in fasting (j) or plasma insulin during in vivo GSIS test (k) in Ad-INHBB received mice pretreated with or without Pramlintide. The sample size (n) is indicated within the graph. l, m Plasma insulin levels in Ad-INHBB received mice pretreated with or without PDE4 inhibitor (l, n = 5 mice per group) or GLP-1R blocker Ex9–39 (m, the sample size (n) is indicated within the graph). Data were shown as mean ± SEM. Statistical significance was determined by unpaired two-tailed Student’s t-test (a–d, f); two-way ANOVA (g–i, m) or with Šídák’s (a–d) or Tukey’s (e, j–n) multiple comparison test as compared to the respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data File.