Pancreatic Islet Vasculature Adapts to Insulin Resistance Through Dilation and Not Angiogenesis

Animal studies were performed in accordance with guidelines of the Vanderbilt University Institutional Animal Care and Use Committee. Adult male and female ob/wt mice on the C57BL/6J background were bred and genotyped according to protocol provided by The Jackson Laboratory (Bar Harbor, ME). The ob/ob and wt/wt controls were studied at 4, 8, and 16 weeks of age. GLUT4^+/− mice and their GLUT4^+/+ littermates were 48–50 weeks of age on mixed C57BL/6J/129 Sv/Black Swiss background, as previously described (12). For an insulin-resistant model of diet-induced obesity, male C57BL/6J mice (The Jackson Laboratory) were fed chow or a high-fat diet (HFD) starting at weaning until 20 weeks of age. The standard chow (LabDiet, cat. #5001) contained 48.7% carbohydrate, 23.9% protein, and 5.0% fat by weight with a total of 13.5% of dietary calories from fat. The HFD (Bio-Serv, Frenchtown, NJ, cat. #F3282) contained 36.3% carbohydrate, 20% protein, and 35.5% fat by weight with a total of 60% of dietary calories from fat. Food was changed twice weekly. Blood glucose and plasma insulin levels were measured in the fed and fasted state, as described (6).

Total islet RNA was extracted using an RNAqueous RNA isolation kit (Ambion, Austin, TX), and analysis was performed as described (13). TaqMan primers and probes were from Applied Biosystems (Supplementary Table 1).

Immunohistochemistry, morphometric analysis, and assessment of the functional islet vasculature were performed as described (6). Antigens were visualized with the primary antibodies listed in Supplementary Table 2. Secondary antibodies were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA).

Islet capillaries were visualized after a bolus of sulforhodamine-conjugated dextran, and islet blood flow was measured by tracking sulforhodamine-labeled erythrocytes (RBCs), as previously described (14). Internal islet capillary diameter was measured using Imaris 7.5 software (Bitplane, South Windsor, CT) on 20 capillaries in wt/wt and ob/ob mice (n = 3 islets/group).

Wild-type and ob/ob mice were perfused through cardiac puncture at a 1 mL/min flow rate with warm PBS for 5 min, followed by 10 mmol/L 4,5-diaminofluorescein diacetate (DAF-2DA; Alexis Biochemicals, Farmingdale, NY) containing 0.1 mmol/L l-arginine and 2 mmol/L CaCl₂ for 10 min, and a 10-min washout period with PBS (15). Production of nitric oxide (NO) converts the nonfluorescent dye, DAF-2 DA, to its fluorescent triazole derivative DAF-2T. Pancreata were removed and fixed, and cryosections were imaged, as described above. DAF-2T fluorescence intensity was measured using ImageJ software, and data were reported after background signal subtraction. Islets were deemed DAF-2T⁺ only if DAF-2T fluorescence intensity was twofold higher compared with background.

Ultrastructure of β-cells and islet vasculature were assessed by transmission electron microscopy (6). Basement membrane thickness was measured in calibrated images at original magnification ×8,800–15,000 using ImageScope software (Aperio, Vista, CA). Depending on the capillary size, the basement thickness was measured at 10–15 points along the capillary length and then averaged per each capillary.

The Student t test was used for comparisons of two groups. All data are presented as mean ± SEM, and a P value < 0.05 was considered significant.

To investigate if islet vascularization changes in the setting of insulin resistance, we studied three mouse models with different mechanisms of insulin resistance: 1) ob/ob mice with an inactivating mutation in the leptin gene (16,17), 2) mice heterozygous for inactivation of the GLUT4 glucose transporter gene (GLUT4^+/−) (12,18), and 3) mice fed an HFD. The ob/ob and HFD mice were both obese relative to their respective controls, but GLUT4^+/− mice had the same body weight as GLUT4^+/+ littermates (Supplementary Fig. 1A–C). When fasting plasma insulin was used as a measure of insulin resistance, ob/ob mice and HFD mice were more insulin-resistant than GLUT4^+/− mice (Supplementary Fig. 1E–G). Fasting and random blood glucose levels in ob/ob mice were moderately elevated at 4, 8, and 16 weeks of age, and they were noted to be glucose-intolerant by 4 weeks of age (Supplementary Fig. 2A and B). HFD and GLUT4^+/− mice had blood glucose levels similar to those of controls (Supplementary Fig. 2C and D). Islet size was significantly increased in all three insulin-resistant models but more profoundly in ob/ob and GLUT4^+/− mice compared with HFD mice (Supplementary Fig. 1H–P).

We reasoned that hyperplastic islets, like tumors, would need increased vascularization because of the increased β-cell number and increased metabolic demands associated with increased insulin biosynthesis and secretion. To assess the pancreatic islet vasculature, blood vessels were visualized by the intravital labeling with lectin-fluorescein isothiocyanate (FITC) (6) (Fig. 1A–F) or by using live imaging in vivo with a bolus of rhodamine-conjugated dextran (14) (Fig. 1G and H). These two methodologies take advantage of intravascular labeling that visualizes only functional vessels, and when combined with optical sectioning, allow for three-dimensional reconstruction of the islet angioarchitecture (Fig. 1A–H). This three-dimensional reconstruction revealed that capillaries in ob/ob islets were greatly enlarged as early as 4 weeks of age. The vessel enlargement was quite prominent within the islet core composed mainly of β-cells but was not observed in the islet periphery containing mainly non–β-cells (Fig. 1D–F and H and Supplementary Fig. 1H–M). Surprisingly, enlarged capillaries were detected not only in hyperplastic islets but also in medium and smaller size islets (Fig. 1D), suggesting that the capillary enlargement is independent of β-cell number per islet.

To better understand vascular phenotype in ob/ob islets, we used integrated morphometry to measure several vessel parameters using 10-μm-thick cryosections labeled with lectin-FITC. At 1 week of age, ob/ob mice were insulin-resistant (Supplementary Fig. 1D), but islet vascular parameters were similar to wt/wt mice (Fig. 1I–K). Although overall blood vessel area-to-islet area ratio increased in ob/ob compared with wt/wt islets by 4 weeks of age (Fig. 1I), islet vascular density progressively declined in ob/ob mice with age (Fig. 1J). At the same time, intraislet vessels became larger in ob/ob mice, as reflected by an 81% increase in the area/vessel parameter (Fig. 1K). This was further confirmed by measurement of the internal capillary diameter demonstrating that capillaries in ob/ob mice were more than twofold thicker (Fig. 1N). In contrast to islets, the vasculature of exocrine tissue was unchanged in insulin-resistant ob/ob mice (Fig. 1M and N). In addition, lectin-FITC labeling in the islets and exocrine tissue was colocalized with EC markers such as caveolin-1 and CD31, confirming that blood vessels in both pancreatic compartments were quiescent and not forming new sprouts (data not shown) (6,19). These data indicate that insulin resistance in the ob/ob model leads to increased vessel area per islet. However, this increase in islet vascular supply results not from increasing islet vessel density but, rather, from capillary size.

Because ob/ob mice have a number of metabolic changes and the leptin receptor is expressed in ECs (20), we examined islet vascularization in two additional models of insulin resistance: one associated with obesity (HFD mice) and the other one without obesity (GLUT4^+/− mice). Using EC labeling with caveolin-1, we found that islet vasculature in these two models (Fig. 2) was phenotypically similar to that of ob/ob islets (Fig. 1). The vessel area-to-islet area ratio was increased in HFD and GLUT4^+/− mice compared with their respective controls (Fig. 2C and H) and accompanied by reduced vessel density (Fig. 2D and I) and a 51–80% increase in the area-to-vessel ratio (Fig. 2E and J), demonstrating intraislet capillary enlargement. These results demonstrate that increasing islet vascular supply through islet capillary enlargement is a common adaptive response to insulin resistance.

Tissue vascular development and adult blood vessel homeostasis are coordinated through intricate signaling networks of several angiogenic growth factor families (3,4,21,22). Islet endocrine cells express several angiogenic factors, but VEGF-A is a principal regulator of islet capillary density and vascular permeability (5,6). Because insulin resistance led to an increased ratio of blood vessel area to islet area, we asked if this increase in islet vascular supply was modulated by changes in islet angiogenic growth factor expression. To address this question, we investigated the expression profile of several angiogenic growth factor families, including VEGFs, angiopoietins, fibroblast growth factors, and ephrins, by quantitative RT-PCR in control and ob/ob islets. As expected, mRNA levels of Vegfa and its obligatory receptors Flt1 and Kdr were the most abundant among all of the angiogenic factors and receptors examined (data not shown). None of the angiogenic factors or their receptors were differentially expressed in ob/ob versus control islets (Fig. 3A and B), suggesting that islet vascular expansion associated with insulin resistance is not mediated by angiogenesis. A caveat of this analysis is that ob/ob islets may have fewer ECs per islet volume due to the increased β-cell population, and thus, angiogenic factor receptor mRNA could be unchanged due to the increased number of receptors on intraislet ECs. Consistent with the gene expression analysis in Fig. 3A, VEGF-A production was similar in wt/wt and ob/ob islets (Fig. 3C).

To further confirm that intraislet EC proliferation was not contributing to islet vascular changes, we assessed replication in ob/ob and wt/wt mice at 4 and 8 weeks of age. Although proliferating β-cells were abundant in ob/ob islets compared with controls (Fig. 3D, F, H, and J), we did not detect intraislet EC proliferation in either type of islets (Fig. 3E, G, I, and K). These data, together with those in Fig. 3A–C, indicate that islet angiogenesis during insulin resistance is very rare, with islet vasculature characterized by a relative stability and a very low mitotic activity of the intraislet EC population. Furthermore, these data suggest that even if the number of VEGF receptor 1 (encoded by Flt1) and VEGF receptor 2 (encoded by Kdr) receptors increased on intraislet ECs in ob/ob islets, this increase did not have an important biological effect.

In addition, transmission electron microscopy demonstrated that intraislet ECs in ob/ob islets had an abundance of fenestrations similar to that in control islets (Fig. 3L and M). Although the basement membrane thickness increased nearly twofold in ob/ob islets (Fig. 3N), this increase represents less than 1% of the islet vessel diameter (Fig. 1L). Taken together, our results show that islet vascular expansion associated with insulin resistance is not mediated by angiogenesis. Instead, adaptive changes in islet vasculature result from dilation of preexisting vessels.

To dissect the mechanisms of the islet blood vessel dilation, we first examined islet capillary support structures. In islets, this capillary support is provided mainly by endocrine cells and pericytes. Hellström at al. (23) showed previously that the lack of pericytes leads to EC hyperplasia, vessel dilation, and microaneurysm formation. Transmission electron microscopy revealed RBC extravasation indicative of intraislet hemorrhage formation, especially in large islets of ob/ob mice at 16 weeks of age (Supplementary Fig. 3). To assess the status of pericytes in ob/ob islets, pancreatic sections were labeled for two pericyte markers: NG2 and platelet-derived growth factor receptor β (PDGFRβ). We found that pericytes were present in ob/ob islets and as demonstrated by immunolabeling, and their area appeared to be increased compared with wt/wt islets (Fig. 4A–F). The ultrastructural analysis further demonstrated that pericytes around the intraislet capillaries were hypertrophied in ob/ob islets (Fig. 4G and H). This pericyte hypertrophy was associated with increased expression of the Pdgfb/Pdgfrb ligand-receptor system in ob/ob islets (Fig. 4I).

Nervous system input plays an important role in the regulation of islet blood flow to maintain homeostasis and react to normal fluctuations in metabolic demands (24,25). To determine if nervous system changes occur in chronic islet adaptation to insulin resistance, we examined the morphology of islet autonomic innervation. Global islet innervation was increased in hyperplastic ob/ob islets (Fig. 5A and B), as quantified by the density of nerve fibers labeled for neuronal class III β-tubulin, the neuron-specific tubulin marker (Fig. 5E). Islets in ob/ob mice also showed dense parasympathetic innervation, with numerous nerve fibers expressing the vesicular acetylcholine transporter (VAChT) (Fig. 5C and D). When quantified, the density of VAChT⁺ varicosities and the VAChT⁺ area were both increased in ob/ob islets (Fig. 5F and G). The change was not due to larger islets in ob/ob mice, because the similarly sized islets in wt/wt mice did not show the same increase in VAChT⁺ nerve fibers (Supplementary Fig. 4A and B). In contrast to parasympathetic innervation, islet sympathetic innervation was unchanged in ob/ob mice, as evaluated by staining for tyrosine hydroxylase (TH), the rate-limiting enzyme in catecholamine biosynthesis. Islets from ob/ob and wt/wt mice showed a comparable number of TH⁺ nerve fibers within the insulin⁺ islet core (Supplementary Fig. 4C and D). Similarly, ob/ob and wt/wt mice both showed a few TH⁺ β-cells per islet (Supplementary Fig 4C and D). These data demonstrate that ob/ob islets have enhanced parasympathetic innervation that may contribute to increased blood flow in ob/ob islets.

To measure islet blood flow, we used live imaging and tracked sulforhodamine-labeled RBCs in vivo (14). We used the RBC trajectory to calculate RBC speed and found that it was 30% higher in ob/ob islets (Fig. 5H). Carlsson et al. (26) found that ob/ob mice had a major increase in islet blood (fourfold) at 4 weeks of age. Our longitudinal analysis of islet capillary morphometry (Fig. 1K) complements the islet blood flow data by Carlsson et al. (26), indicating that the robust increase in islet blood flow at 4 weeks of age precedes adaptive changes in the islet capillary size. This suggests that islet capillary diameter responds to increased islet blood flow and that shear stress leads at least in part to islet capillary dilation.

Increased blood flow and shear stress induce transcription and translation of endothelial NO synthase (eNOS), which leads to increased synthesis and release of the potent vasodilator, NO (27). All three isoforms of NOS are present in wt/wt islets, but the mRNA level of eNOS (Nos3) was higher than neuronal NOS (Nos1) and inducible NOS (Nos2) (Fig. 6A). Moreover, ob/ob islets had higher expression of Nos3 but lower expression of Nos1 and Nos2 (Fig. 6B). We next measured NO production in vivo using the fluorescent NO indicator DAF-2DA. In ob/ob islets, DAF-2T fluorescence intensity increased by 30% (Fig. 6C–G), although some wt/wt and ob/ob islets had similar DAF-2T fluorescence intensity. These data indicate that islet vascular changes during insulin resistance are closely associated with increased islet parasympathetic innervation and islet blood flow and are at least partly mediated by increased NO production.

Pancreatic islets respond to the increased insulin demand associated with insulin resistance by increasing insulin output and expanding β-cell mass. These compensatory changes are critical for maintaining normoglycemia and avoiding diabetes. Because islets are highly vascularized and many of the adaptive changes in islets are accompanied by increased oxygen and nutrient requirements, we assessed islet vascularization in three independent mouse models of insulin resistance: leptin deficiency, HFD-induced obesity, and GLUT4 haploinsufficiency. Our results indicate that the islet vascular supply increases during insulin resistance. However, this vascular adaptation is mediated not by angiogenesis but, rather, by dilation of preexisting vessels.

Despite the dramatic changes in β-cell mass and function during insulin resistance, islets, unlike tumors, surprisingly maintain normal expression of the major angiogenic factor secreted by islet endocrine cells, Vegfa, and its signaling receptors Kdr and Flt1. Several recent reports indicate that the precise regulation of VEGF-A signaling is critical for the maintenance of vascular homeostasis in developing and mature islets (28–30), and our data suggest that this tight regulation of VEGF-A signaling is conserved even in insulin resistance. Increased VEGF-A expression by endocrine progenitors or β-cells stimulates proliferative angiogenesis that leads to β-cell loss (28,29). Although islets in insulin-resistant animals showed increased β-cell proliferation, there was no proliferation of intraislet ECs, indicating that islet vasculature is quiescent. Prior studies using prediabetic Zucker diabetic fatty rats and HFD-fed mice suggested a causal relationship between increased islet VEGF-A secretion and islet hypervascularization in insulin resistance, but these studies did not normalize islet VEGF-A production for the difference in the total cell number between hyperplastic and control islets (30,31). Our data show that these studies overestimated VEGF-A production per islet and that this is the reason for the difference with our results. Furthermore, because the number of islet capillaries is strictly regulated by VEGF-A (5,6), the fact that islets compensate by increasing the vessel size and not by changing the vessel number, EC proliferation, or EC fenestration is biologic evidence of VEGF-A–independent vascular adaptation. Formation of larger vessels was noted when Ang-1 was overexpressed in the skin (32), but our analysis suggests that the angiopoietin signaling system is not involved in the increased islet capillary size associated with insulin resistance.

Islet capillaries not only increased in size but also underwent several changes at the ultrastructural level. Although EC fenestrations were maintained in the insulin-resistant state, the thickness of the vascular basement membrane nearly doubled, the pericytes associated with islet capillaries became hypertrophied, and we noted RBC extravasation. Similar changes in basement membrane and pericytes were reported by Nakamura et al. (33) in islets of diabetic db/db mice. We showed that the thickness of the islet capillary basement membrane increased proportionally with the capillary diameter but did not observe hallmarks of fibrosis, such as presence of fibroblasts as reported by Nakamura et al. (33) in db/db mice with long-standing diabetes (34). Because the capillary basement membrane in islets is exclusively synthesized by intraislet ECs (35), its remodeling during insulin resistance could be caused by changes in ECs triggered by increased islet blood flow (36). Thus, the increased basement membrane thickness in the absence of diabetes could be viewed as an initial adaptive response of intraislet ECs rather than fibrosis, as proposed by Agudo et al. (30). Furthermore, Agudo et al. (30) suggested that the increased area of islet vascular collagen IV labeling during insulin resistance was indicative of fibrosis. Our data do not support this conclusion and instead indicate that the increased area of collagen IV, which marks blood vessel surface, is mainly due to the increased islet capillary size and not fibrosis.

The role of pericytes in normal islet function is not completely understood, but in other tissues, pericytes are very important in supporting blood vessel stability (23,37,38). ECs recruit pericytes to the vessel wall mainly by secreting PDGFB homodimer that binds to PDGFRβ receptor on the pericyte surface. Inactivation of PDGFB or PDGFRβ results in pericyte loss that leads to EC hyperplasia, vessel dilation, and vascular leakage (23,37,38). In the ob/ob model, pericytes were not lost, and actually, Pdgfb/Pdgfrb expression and pericyte area were slightly increased. We thus postulate that this compensation is insufficient in the presence of significantly increased islet vascular perfusion and leads to vessel instability and hemorrhaging. Our gene expression and vascular morphometry data strongly argue against the notion that the primary cause of vascular leakage in the insulin resistance setting is increased islet VEGF-A expression (30,31).

Our data indicate that insulin resistance leads to significantly increased expression of eNOS (Nos3), a rate-limiting enzyme in NO synthesis. NO plays a key role in regulating endothelium-dependent vasodilation (39). Several complementary mechanisms may lead to increased Nos3 expression and NO production resulting in NO-mediated islet capillary dilation during insulin resistance. The NO-mediated vasodilatory response of intraislet ECs may be stimulated by hemodynamic shear stresses imposed on endothelium by higher islet blood flow. ECs respond to changes in blood flow through the process known as mechanotransduction (27,36). We propose that this response increases expression of eNOS and thus results in greater availability of NO.

We also observed that islets during insulin resistance have increased parasympathetic innervation, while maintaining normal sympathetic innervation. Islet parasympathetic nerves release neurotransmitters such as ACh, vasoactive intestinal peptide, gastrin-releasing peptide, and pituitary adenylate cyclase-activating polypeptide and play critical roles in regulating hormone release and in determining blood vessel tone (40–43). When glucose is infused specifically in the brain, signals from the central nervous system are transmitted through the vagus nerves to enhance insulin secretion and islet blood flow (25). In addition, glucose infusion directly into the duodenum triggers a vagal relay to induce similar effect (24). The increased density of cholinergic nerves that we observed in ob/ob islets supports the importance of the parasympathetic nervous system in mediating the increased insulin secretion and islet blood flow in obese mice and rats (44–46). For example, ACh activates muscarinic receptors on ECs, which results in the release of NO leading to relaxation of vascular smooth muscle cells and vasodilation (47). Further studies are required to understand the mechanisms regulating islet parasympathetic innervation during insulin resistance and the relationship between parasympathetic innervation and adaptive changes in islet vascularization.

Our three-dimensional reconstruction of islet capillaries demonstrated that the dilation was present only in the islet core containing mostly β-cells and not around the periphery composed of other islet hormone-producing cell types. This unique arrangement of islet angioarchitecture suggests that β-cells may be involved in islet capillary dilation. Because insulin secretion is markedly increased during insulin resistance, it is possible that the local increase in insulin levels activates the insulin receptor signaling cascade in intraislet ECs leading to phosphatidyl inositol 3-kinase-Akt-mediated eNOS activation, increased NO production, and vasorelaxation (48).

Here we show that the pancreatic islet adaptation to insulin resistance is not limited to changes within β-cells but also involves islet-specific, neurovascular remodeling. To accommodate the increased demand for insulin delivery into the peripheral circulation, islet capillaries expand by dilation and not by angiogenesis. Based on our data, we propose a model where eNOS-dependent islet capillary dilation is modulated by complementary signals derived from β-cells, parasympathetic nerves, and islet blood flow (Fig. 7). In addition, islet capillary dilation is accompanied by basement membrane remodeling and pericyte hypertrophy. These changes in the structural support of islet capillaries may be insufficient to maintain capillary function, resulting in microaneurysm and β-cell damage. The adaptive changes in islet vascularization may influence the ability of islets to respond to insulin resistance and could be vital for preventing β-cell failure and development of type 2 diabetes.

This work was supported by a Merit Review Award from the Veterans Affairs Research Service (BX000666); by National Institutes of Health (NIH) grants DK-69603, DK-68764, DK-089572, DK-66636, DK-63439, and DK-072473; and by grants from JDRF, the Vanderbilt Mouse Metabolic Phenotyping Center (NIH Grant DK-59637), and the Vanderbilt Diabetes Research and Training Center (Islet Procurement and Analysis Core, NIH Grant DK-20593). Imaging was performed with support from the Vanderbilt University Medical Center Cell Imaging Shared Resource (NIH grants CA-68485, DK-20593, DK-58404, HD-15052, DK-59637, and EY-08126).

No potential conflicts of interest relevant to this article were reported.

C.D., M.B., R.B.R., L.N., M.S., T.T., and A.C.P. researched data and designed experiments. C.D., M.B., R.B.R., L.N., C.T., and A.S. performed experiments and reviewed data. E.H.L. reviewed data. C.D., M.B., R.B.R., E.H.L., and A.C.P. wrote the manuscript. C.D., M.B., R.B.R., L.N., E.H.L., C.T., A.S., M.S., T.T., and A.C.P. reviewed the manuscript. A.C.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented in abstract form at the 69th Scientific Sessions of the American Diabetes Association, New Orleans, Louisiana, 5–9 June 2009.