FSTL3 deletion reveals roles for TGF-beta family ligands in glucose and fat homeostasis in adults

Abstract

Activin and myostatin are related members of the TGF-beta growth factor superfamily. FSTL3 (Follistatin-like 3) is an activin and myostatin antagonist whose physiological role in adults remains to be determined. We found that homozygous FSTL3 knockout adults developed a distinct group of metabolic phenotypes, including increased pancreatic islet number and size, beta cell hyperplasia, decreased visceral fat mass, improved glucose tolerance, and enhanced insulin sensitivity, changes that might benefit obese, insulin-resistant patients. The mice also developed hepatic steatosis and mild hypertension but exhibited no alteration of muscle or body weight. This combination of phenotypes appears to arise from increased activin and myostatin bioactivity in specific tissues resulting from the absence of the FSTL3 antagonist. Thus, the enlarged islets and beta cell number likely result from increased activin action. Reduced visceral fat is consistent with a role for increased myostatin action in regulating fat deposition, which, in turn, may be partly responsible for the enhanced glucose tolerance and insulin sensitivity. Our results demonstrate that FSTL3 regulation of activin and myostatin is critical for normal adult metabolic homeostasis, suggesting that pharmacological manipulation of FSTL3 activity might simultaneously reduce visceral adiposity, increase beta cell mass, and improve insulin sensitivity.

Fig. 1.

Growth rate and body weights of FSTL3 KO and WT mice. (A) Distribution of age versus body weight of WT and FSTL3 KO male mice is shown for 1- to 4-month-old animals. (B) Mean body weights of male (average age, 9 months; n = 20 WT and 34 KO) and female (average age, 11 months; n = 11 WT and 27 KO) WT and KO mice. (C) Weights of dissected quadriceps (Quad) and gastrocnemius (Gastroc) muscles expressed as a ratio to body weight (n = 25 WT and 58 KO).

Fig. 2.

Increased pancreatic islet size in FSTL3 KO mice. (A and B) Low-power photomicrographs of WT (A) and FSTL3 KO (B) pancreas immunocytochemically stained for insulin. (C and D) H&E staining of the same WT (C) and KO (D) tissues. Black arrowheads show numerous capillaries within the KO islet. (E and F) Immunofluorescence photographs showing insulin (red) and glucagon (green) localization in islets from WT (E) and KO (F) mice. In the KO islets, the majority of green staining is nonspecific staining of red blood cells within the islet, shown by open arrowheads. Glucagon producing α cells in both WT and KO islets are shown by solid arrowheads. (Magnifications: A and B, ×5; C–F, ×40.) (Scale bars, A and B, 200 μm; C–F, 50 μm.) (G–I) Histomorphometric analyses of pancreatic islets. Average islet size (G) (n = 36 WT and 99 KO) and β cell size (H) (n = 119 WT and 94 KO) in pancreas from WT and KO animals (n = 6 WT and 8 KO) (values shown in squared micrometers). (I) Size distribution of pancreatic islets in WT and KO animals; percent of total number of islets counted is plotted against islet size. (Inset) The mean number of islets seen per section of WT and KO animals. Both islets per section and average islet size is significantly increased (P < 0.05) in KO pancreas. All error bars are SEM.

Fig. 3.

Enhanced glucose metabolism in FSTL3 KO mice. (A) Glucose levels in tail blood of random-fed WT and KO mice are not different (n = 17 WT and 30 KO). (B) Serum insulin levels in these same mice are significantly greater (P < 0.05) in KO animals compared with WT littermates. (C) Glucose tolerance test showing mean glucose levels of five animals per group after i.p. injection of 2 g/kg glucose at time 0. ∗, P < 0.05. (D) Insulin tolerance test showing glucose levels as a ratio to untreated glucose concentration (at time −15 min). Insulin (1 unit/kg) was injected i.p. at time 0. Glucose levels were significantly lower in FSTL3 KO mice from 30–120 min. ∗∗, P < 0.01. A total of 24 WT and KO mice were examined in these studies. All error bars are SEM.

Fig. 4.

Liver phenotypes of FSTL3 KO mice. (A–D) H&E staining of livers from WT (A and C) and KO (B and D) animals. (E and F) Frozen sections of liver stained with Oil Red O from WT (E) and KO (F) animals at 9.5 months. (G and H) Liver sections stained with Periodic acid-Schiff's stain showing vastly reduced glycogen content in FSTL3 KO mice (H) compared with WT littermates (G). (Magnifications: A and B, ×10; C–H, ×40.) (Scale bars, A and B, 100 μm; C–H, 50 μm.) (I) Gluconeogenic genes PEPCK and G6Pase are significantly (P < 0.05) elevated in FSTL3 KO mice relative to WT mice by 6- and 4.5-fold, respectively, as analyzed by quantitative PCR. (J) Activin (5 ng/ml) significantly (P < 0.01) stimulated G6Pase gene expression relative to untreated HepG2 liver hepatoma cells, whereas myostatin (25 ng/ml) suppressed G6Pase under the same conditions (P < 0.05).

Fig. 5.

Fat depot mass is reduced in FSTL3 KO mice. (A and B) Reduced visceral abdominal fat pad mass in FSTL3 KO males (A) (n = 25 WT and 56 KO; P < 0.005) and females (B) (n = 19 WT and 42 KO; P < 0.05). (C) Percent body fat of WT and FSTL3 KO mice, as measured by dual emission x-ray absorptiometry (n = 6 WT and 7 KO). All error bars are SEM.