Glucose intolerance and impaired insulin secretion in pancreas-specific signal transducer and activator of transcription-3 knockout mice are associated with microvascular alterations in the pancreas

Abstract

Maintenance of glucose homeostasis depends on adequate amount and precise pattern of insulin secretion, which is determined by both beta-cell secretory processes and well-developed microvascular network within endocrine pancreas. The development of highly organized microvasculature and high degrees of capillary fenestrations in endocrine pancreas is greatly dependent on vascular endothelial growth factor-A (VEGF-A) from islet cells. However, it is unclear how VEGF-A production is regulated in endocrine pancreas. To understand whether signal transducer and activator of transcription (STAT)-3 is involved in VEGF-A regulation and subsequent islet and microvascular network development, we generated a mouse line carrying pancreas-specific deletion of STAT3 (p-KO) and performed physiological analyses both in vivo and using isolated islets, including glucose and insulin tolerance tests, and insulin secretion measurements. We also studied microvascular network and islet development by using immunohistochemical methods. The p-KO mice exhibited glucose intolerance and impaired insulin secretion in vivo but normal insulin secretion in isolated islets. Microvascular density in the pancreas was reduced in p-KO mice, along with decreased expression of VEGF-A, but not other vasotropic factors in islets in the absence of pancreatic STAT3 signaling. Together, our study suggests that pancreatic STAT3 signaling is required for the normal development and maintenance of endocrine pancreas and islet microvascular network, possibly through its regulation of VEGF-A.

Figure 1

Generation of p-KO mice and assessment of STAT3 deletion in pancreatic tissue. A, Exon 21 and part of exon 22 of STAT3 gene were flanked by loxP sites (open triangle) in STAT3^fl/fl mice. In p-KO mice, Cre-mediated recombination removes sequences between the loxP sites (KO) and results in specific inactivation of STAT3 gene in the pancreas without affecting STAT3 expression in other tissues. B, STAT3 protein levels were assessed by Western blot analysis using polyclonal STAT3 antibody. Equal amounts of brain and pancreatic tissue extracts from KO and control mice were resolved by SDS-PAGE. STAT3 was barely detectable in the pancreas of STAT3 KO mice, whereas it was expressed at a similar level in the brain as control. Tubulin was used as loading control. C, Immunohistochemical analysis of STAT3 and insulin expression was performed on 5-μm pancreatic cryosections of 12-wk-old p-KO and control mice. STAT3 (red) was localized throughout the islet area in control but was undetectable in p-KO mice. Scale bar, 50 (a and c) and 5 μm (b and d).

Figure 2

Impaired glucose tolerance and normal insulin sensitivity in STAT3 KO mice. A, Glucose levels of p-KO mice and control littermates were measured before and at 15, 30, 60, 90, and 120 min after ip glucose injection. p-KO mice (filled circle, n = 14) showed higher glucose levels than control (open circle, n = 12) after glucose challenge. *, P < 0.05; **, P < 0.01. B, Insulin tolerance tests were performed on fed p-KO (filled circle, n = 11) and control (open circle, n = 11) mice. Blood glucose levels were similarly affected by insulin at all time points in p-KO and control mice. Data are presented as means ± sem.

Figure 3

STAT3 KO mice exhibited impaired glucose-induced insulin release in vivo. A, Plasma insulin levels were measured before and at 8, 15, and 30 min after ip glucose injection in p-KO (filled circle, n = 12) and control mice (open circle, n = 11) after overnight fasting. Insulin levels were lower in p-KO mice after glucose challenge. B, Total glucose-stimulated insulin secretion, calculated by integrating area under curve in A, was reduced in p-KO mice (gray bars, n = 12) during the first 15 min and the entire 30 min of stimulation compared with control mice (white bars, n = 11). Data are presented as means ± sem. *, P < 0.05.

Figure 4

Normal insulin secretion from perifused isolated islets of STAT3 KO mice. A, Glucose-stimulated insulin secretion was measured in isolated islets from p-KO and control mice. Groups of 20 islets were incubated for 1 h in KRH buffer containing 3 mm glucose (basal) before switching (arrow) to 20 mm glucose (stimulatory). There was no difference in basal and glucose-stimulated insulin secretion in islets from p-KO (filled circle, n = 4) and control (open circle, n = 4) mice. B, Net glucose-stimulated insulin secretion, calculated as the sum of insulin amount per islet in all fractions during the first 15 min (first phase) or the entire stimulation period (total) after baseline subtraction. No difference was observed in the first phase or total insulin secretion between p-KO (gray bars, n = 4) and control (white bars, n = 4) mice. Data are presented as means ± sem. C, Representative traces of NAD(P)H fluorescence from p-KO (blue) and control (red) mouse islets (n = 12 for each group). The islets were perifused in 3 mm glucose before the perfusion buffer was switched to 20 mm glucose (dotted line). Both p-KO and control groups displayed similar time course. D, Representative traces of calcium recordings from p-KO (blue) and control (red) mouse islets (n = 18 and 13 for p-KO and control, respectively). The islets were initially perifused in 3 mm glucose before the perfusion buffer was switched to 20 mm glucose (dotted line). Cytosolic calcium levels were measured using fura-2AM. Both p-KO and control groups exhibited similar calcium responses.

Figure 5

Reduced islet microvascular density and lower VEGF-A expression level in STAT3 KO mice. A, Five-micrometer cryosections of p-KO and control mouse pancreas were immunolabeled with STAT3 (red) and endothelial marker CD31 (green). B, Vascular density, measured as CD31-stained area relative to the whole pancreas was reduced in p-KO (gray bar, n = 3 mice) compared with control mice (white bar, n = 3 mice). Images taken from 10 pancreatic sections of each animal were used for analysis. C, Vascular density in the islets (e.g. inside yellow boundary in A) and exocrine pancreas (e.g. outside yellow boundary in A) was similarly assessed as in B, Vascular density was significantly reduced in the islets (**, P < 0.01) and marginally reduced in the exocrine tissue (#, P = 0.06). Data are presented as means ± sem. Images taken from 10 pancreatic sections of each mouse, three mice per group, were used for analysis. D, VEGF-A mRNA levels were analyzed by quantitative real-time PCR from total RNA extracted from isolated islets. P-KO mice (gray bar, n = 4) have lower VEGF-A expression than control mice (white bar, n = 4). *, P < 0.05. E, VEGF-A content in isolated islets from p-KO mice (gray bar, n = 3) was significantly lower than that from control mice (white bar, n = 3). VEGF-A level was determined by ELISA. Data are presented as means ± sem. *, P < 0.05. F, Five-micrometer cryosections of p-KO and control mouse pancreas were immunolabeled with VEGF (red) and 4′,6′-diamino-2-phenylindole (blue). Green boundary denotes islets. G, VEGF-A levels in the islets and exocrine pancreas were measured by integrating the gray levels inside and outside the islets. VEGF-A levels in the islets and exocrine pancreas decreased significantly in p-KO mice. **, P < 0.01. Multiple images were taken from pancreatic sections of each mouse, and three mice per group were used for analysis.