Islet Microvasculature in Islet Hyperplasia and Failure in a Model of Type 2 Diabetes

All animal procedures were approved by the institutional animal care and use committee at the University of Michigan. Male and female fa/fa and ZDF rats (ZDF/Gmi-fa/fa; ZDF rats) or lean littermate controls were purchased from Charles Rivers (Indianapolis, IN). In these experiments, pancreata were harvested on 2 consecutive days between 4 and 5, 7 and 8, and 12 and 14 weeks of age and throughout this article will be referred to as 4-, 7-, and 12-week-old animals. Male rats were fed standard rat chow (Purina 5006) in a 12:12 light:dark cycle or a chow in which pioglitazone was added at 1.6 mg/g to give an average daily dose of pioglitazone of ∼40 mg/kg. To induce diabetes in female ZDF (Zucker fatty female [ZFF]) rats, they were fed a semipurified high-fat diet (48% fat, 16% protein; diet no. 13004; Research Diets, New Brunswick, NJ) from 8 to 13 weeks of age. Whole blood glucose was determined using a Hemocue (Angelholm, Sweden).

The use of custom rat islet Affymetrix GeneChips (Metabolex Rat Islet Oligonucleotide Arrays) used in these studies has been described previously (18).

Sections of fixed tissues (4% paraformaldehyde) were stained for nestin (1:100; Pharmingen, San Diego, CA) and insulin (1:2,000; Linco Research, St. Charles, MO) using a Vectastain ABC-Peroxidase kit (Vector Laboratories, Burlingame, CA). Biotinylated secondary antibodies and horseradish peroxidase–conjugated antibiotin were used to detect staining according to the manufacture’s protocols. For immunofluorescence staining, blocked slides were incubated with nestin (1:100), VEGF (1:200; Santa Cruz), or proliferating cell nuclear antigen (PCNA) (1:100; Sigma) antisera and were developed with appropriate fluorescent secondary antibodies (Molecular Probes).

Four to five random sections from each animal were stained for nestin and nuclei counterstained with DAPI (4′6-diamino-2-phenylindole). Two observers counted the total number of nuclei in all of the islets from each section and also quantified the number of nestin⁺ cells in each section that was associated with the nuclei. The scores from each observer were averaged (<5% variability in the counts between the observers) and the nestin⁺ cell-associated nuclei divided by the total number of nuclei.

Rat pancreas was excised and fixed by emersion in 2.5% glutaraldehyde in 0.1 mol/l Sorensen’s buffer, pH 7.4, and postfixed in 1% osmium tetroxide followed by en bloc stained with aqueous 3% uranyl acetate for 1 h, dehydrated, and embedded in Epon epoxy resin. Semithin sections were stained with toluidine blue for tissue orientation. Selected areas were ultrathin sectioned in 70-nm thickness and stained with uranyl acetate and lead citrate. They were examined using a Philips CM100 electron microscope at 60 kV. Images were recorded digitally using a Kodak 1.6 Megaplus camera system operated using AMT software (Advanced Microscopy Techniques, Danvers, MA).

Islets were isolated from lean control rats or male and female ZDF rats using collagenase digestion and ficoll purification as previously described (12). Total RNA was isolated from 100 to 150 handpicked islets or from cultures, purified on RNeasy Quick spin columns (Qiagen, Valencia, CA), and treated by DNA-free DNase (Ambion, Houston, TX). Maloney murine leukemia virus reverse transcriptase (Promega, Madison, WI) was used for the reverse-transcription reaction in accordance with the manufacturer’s instructions. cDNA was purified by Quick spin kit (Qiagen). Real-time PCR was performed on a DNA Engine Opticon PCR cycler (MJ Research, Waltham, MA). PCR protocol was one cycle of 15 min at 95°C, 45 cycles of 15 s at 95°C, 30 s at 56°C, 30 s at 72°C, and plate read, and 1 cycle of 10 min at 72°C, and 1 cycle of melting curve from 65 to 95°C. The following synthetic oligo-nucleotide primers were used in real-time PCR: rat nestin forward 5′-AGGTGGGTGCTCTAAGGTT-3′, reverse 5′-AGGATCTCACCTCCCTTGCT-3′; Rat Flk-1 (188 bp) forward 5′ctctacacttgtcgtgtgaag-3′, reverse 5′taaccatacgacttctggcga-3′; rat VEGF forward (182 bp) 5′- CCAGGCTGCACCCACGAC-3′, reverse 5′-AGCCCGCACACCGCATTAG-3′; and universal 18S forward (121 bp) 5′-ACTCAACACGGGAAACCTCACC-3′, reverse 5′-CCAGACAAATCGCTCCACCAAC-3′. Results of the real-time PCR data were represented as C_t values, where C_t was defined as the threshold cycle of PCR when the amplified product was first detected. The PCR and protocol for the 18S ribosome were the same as described above. ΔC_t was the difference in the C_t values derived from the specific gene being assayed and the 18S control, while ΔΔC_t represented the difference between the paired tissue samples, as calculated by the formula ΔΔC_t = ΔC_t of wild-type tissue − ΔC_t of other tissue. The N-fold differential expression in a specific gene of a sample compared with the wild-type counterpart was expressed as 2^ΔCt. Relative RNA equivalents for each sample were obtained by normalizing to 18S levels. Each of the three samples per group was run in duplicate to determine sample reproducibility, and the average relative RNA equivalents per sample pair was used for further analysis. Statistical analysis was performed using ANOVA.

Primary duct cultures from rats were grown in four-well plates for 5 days as previously described (24) and trypsinized for 5 min, after which the cells were resuspended in 500 μl vitrogen 100 (Collagen). Islets from 7-week-old lean control or 7- or 14-week-old ZDF rats were isolated on ficoll after collagenase digestion of the pancreas (12) and incubated for 24 h with or without matrix metalloproteinase (MMP)-9 to activate “angiogenesis” (25). Individual islets were placed with the trypsinized duct–derived cultures and after 3 days scored for starburst formation, by a blinded observer, as an indication of migration of the cells in culture.

ANOVA for single and repeated-measures ANOVA was used, where appropriate, to test for differences between groups, and Duncan’s multiple range test and Tukey’s Studentized range test were used for post hoc evaluations.

In performing Affymetrix-based gene expression profiles from male ZDF rats at 6 and 9 weeks of age (26), we noted an upregulation of a number of transcripts for genes known to be activated during hypoxia (Fig. 1) (27). In general, these genes were increased in the normoglycemic, 6-week-old ZDF rats and in some cases were further increased in the mildly hyperglycemic 9-week-old rats when compared with age-matched lean controls. This prompted experiments to assess the endocrine cell/vascular relationship in the islets of ZDF rats before, during, and after the induction of diabetes.

We (24) and others (28–30) have previously found that nestin, an intermediate filament protein, is a marker of endothelial cells in the islets of mice and humans. To confirm that we could use this easily detected antigen as an endothelial marker in rat islet, we performed confocal microscopy on pancreatic sections of control rats costained with monoclonal anti-nestin antibodies and the VEGF receptor Flk-1 (VEGF-R2). As shown in Fig. 2A–C, we found that there was 100% colocalization of Flk-1 with nestin within the islet. Flk-1⁺ cells in large vascular structures cells did not stain for nestin (Fig. 2A–C, asterisk), but some periductal and perivascular cells did show costaining (Fig. 2A–C, arrows). In a previous publication, we demonstrated that primary duct cultures could give rise to nestin⁺ cells. About 30% of primary duct–derived cells were nestin⁺ and also costained for Flk-1 (Fig. 2D). Thus, nestin staining can be used as a convenient marker for endothelial cells of the islet and in cultured cells.

To determine if alterations in the microvasculature of the islet occurs in a model of type 2 diabetes, we prospectively assessed potential changes in the microvasculature in the islets of male and female ZDF rats during islet expansion and islet failure and compared them with age-matched lean control and nondiabetic fa/fa rats. In these studies, blood glucose levels rose with time in the male ZDF rats compared with lean littermates and were frankly diabetic at 12 weeks of age (Table 1). Administration of a diet containing the thiazolidinedione pioglitazone prevented the onset of diabetes in the male rat. Lean control male rats and fa/fa rats remain euglycemic throughout the study. ZFF rats were significantly hyperglycemic only after administration of a proprietary high-fat diet for 4 weeks (31) (Table 1).

The proportion of nestin⁺ endothelial cells in the islets of the experimental animals was determined by quantifying between 1,100 and 2,500 nuclei from histologically identified islets in four to five random, nonconsecutive sections from each of four different animals in each group and by determining the association of nuclei with nestin⁺ cells determined by immunostaining. As previously reported (7), the average number of cells within islets increased in the fa/fa and ZDF animals compared with lean controls, with a 70–90% increase in cells/islets in these populations compared with lean controls of the same age (Table 1). Serpentine nestin⁺ cells were present throughout the islet of all animals (Fig. 3A; fa/fa not shown). In general, the nuclei associated with the endothelial cells were smaller and more elongated than the nuclei in the endocrine cells. The relative nestin⁺ endothelial population was nearly doubled in the 4-week-old fa/fa and ZDF rat islets compared with age-matched lean controls. The increase in endothelial cell number was also seen in the 7-week-old ZDF rat (Fig. 3D) before the onset of overt hyperglycemia. At 12 weeks, the fa/fa islet-associated nestin⁺ cell number continued to be significantly greater than the number in lean controls, while the number of nestin⁺ cells in the 12-week-old ZDF rats fell and continued to fall at 18 weeks of age and paralleled the islet degeneration (Fig. 3A and E). In addition to hyperplasia, the endothelial cells in the 4- and 7-week-old ZDF rats appeared hypertrophic compared with lean control fa/fa rats (Fig. 3A; 7-week group). In male ZDF rats treated with pioglitazone, the relative increase in the proportion of nestin⁺ endothelial cells was maintained at levels seen in the 4- and 7-week-old ZDF animals (Fig. 3B and E), and the morphology of the cells was not different from that of control animals (Fig. 3B).

To independently confirm age-related changes in endothelial cell populations in the ZDF rat, islets were isolated from 4- and 12-week-old rats, and the levels of nestin and Flk-1 mRNA expression were determined. We also examined nestin and Flk-1 mRNA expression in 12-week-old low-fat–fed (nondiabetic) and high-fat–fed (diabetic) ZFF rats. Nestin and Flk-1 mRNA levels were significantly lower in 12-week-old male ZDF rats compared with 4-week-old animals and compared with age-matched lean animals (Fig. 3F). As with the nestin⁺ endothelial cell counts, pioglitazone prevented a significant fall in the nestin mRNA levels in parallel to preventing the onset of diabetes, although the mRNA levels of Flk-1 tended to be lower than those seen in the pre-diabetic ZDF rats (Fig. 3F). Similar decreases in nestin⁺ cell number were noted in the islets of 14-week-old female ZDF rats after 6 weeks of feeding a high-fat diet (not shown), which is in parallel to the development of hyperglycemia, and this was paralleled by a decrease in nestin and Flk-1 mRNA levels in islets isolated from the female rats (Fig. 3F). Interestingly, confocal examination of islets of ZFF rat pancreata costained with nestin and the cellular proliferation marker PCNA showed that a portion of the endothelial cells were replicating in the control-diet–fed ZFF rats as shown by the presence of PCNA in nestin⁺ cells. After the induction of diabetes, there was a paucity of nestin⁺ endothelial cells and few double-positive cells (Fig. 3C and D, arrows). There was evidence of endocrine cell replication, as detected by PCNA staining, of typical islet endocrine cells (Fig. 3C and D, asterisk). These results show that islets with active expansion in the nondiabetic state have proliferating nestin⁺ endothelial cells, while failing islets have a loss of these cells.

To better understand the loss of endothelial cells in this model of type 2 diabetes, we examined the ultrastructure of lean control and ZDF islets by transmission electron microscopy (TEM). The islets’ ultrastructure in 7-week-old lean rats showed cells with multiple electron-dense granules within vesicles (Fig. 4A, arrows), which is indicative of β-cells with intact epithelium. Higher-power images showed that the endothelium contained multiple fenestrations (Fig. 4B, arrows), which abut β-cells. In contrast, TEM images from 7-week-old normoglycemic male ZDF (Fig. 4C and D) rats showed several significant changes in the β-cells, including numerous vesicles without dense granules (Fig. 4C and D, arrow heads), as well as evidence of swollen mitochondria (Fig. 4C and D, asterisk) and dilated endoplasmic reticulum multiple (Fig. 4C and D, double white arrow). The ultrastructure of endothelium was variable in 7-week-old ZDF rats, with some areas showing normal-appearing fenestrations (Fig. 4C, arrows) and most areas showing significantly thickened endothelium (Fig. 4C and D, black double arrows), which likely corresponds to the hypertrophic endothelium seen in the pre-diabetic ZDF rat (Fig. 3A). In addition, there were structures within the thickened epithelium of the pre-diabetic animals that appeared as “target-like” structures. These vesicles contain electron-dense material, which may be endocytosed insulin (Fig. 4D, white arrowheads). In sections from 14-week-old ZDF rats, the endothelium was severely disrupted (Fig. 4E, arrows), with areas completely denuded of endothelium (Fig. 4E and F, asterisk), suggesting intra-islet hemorrhage. Autophagic vesicles containing degenerating mitochondria and other intracellular organelles were easily detected (Fig. 4F, arrows), suggesting a significant disruption in nutrient availability. There was no difference in the appearance of the acinar cells or endothelium in the exocrine portion of the pancreas (data not shown).

The progressive disruption of the endothelium in pre-diabetic ZDF rats may arise from decreases in production of endothelial survival factors or refractoriness of the endothelium to these factors either to intrinsic defects or due to the elaboration of antiangiogenic factors. VEGF-A, the most well-characterized proangiogenic factor, is expressed in the β-cells of islets (19,32,33), and disruption of islet VEGF-A results in abnormalities of islet endothelium and glucose intolerance (22). Immunohistochemistry of islets from different-aged lean control and ZDF rats showed that in all animals, VEGF-A was confined to the islets, and the number of cells that showed that VEGF staining in 14-week-old ZDF rats was decreased compared with those found in age-matched lean animals (Fig. 5A). However, the intensity of staining in the islet of the 15-week-old ZDF rat appeared greater than in the other groups of rats. Quantitative RT-PCR of VEGF-A mRNA from islets of pre-diabetic (7 weeks old) and diabetic (14 weeks old) ZDF rats showed that at both ages, VEFG-A mRNA levels were increased over that of 14-week-old lean control levels (Fig. 5B). At the same time, the insulin mRNA levels are decreased ∼60% in the islets of 12-week-old male ZDF rats compared with 7-week-old lean controls, similar to that previously reported (34) (Fig. 5B). This result suggests that VEGF-A levels are increased per β-cell in the islets of the 7-week-old mZDF rat, and there is a relative preservation of VEGF-A levels in the islets even as the islet fails. When compared with standard-diet–fed ZFF rats, ZFF fed the high-fat/high-carbohydrate diet showed a 30% increase in VEGF mRNA after 3 weeks of feeding (Fig. 5C). Pioglitazone treatment decreased the levels of VEGF-A mRNA to near control levels. We also measured the content of thrombospondin-1 (Tsp-1), a potent anti-VEGF peptide. Tsp-1 mRNA levels increased 2-fold in the 7-week-old pre-diabetic male ZDF rat and by 10-fold in the 12-week-old diabetic animal. Pioglitazone blunted the rise in Tsp-1, but the levels were still fourfold greater than control levels (Fig. 5B). The rise in Tsp-1 was most marked in the ZFF rat, where the diabetogenic diet increased Tsp-1 expression nearly 20-fold (Fig. 5C).

Following an 18-h incubation at 2 mmol/l glucose, the islet content of VEGF-A was elevated twofold in islets from 7-week-old ZDF rats compared with lean controls (Fig. 5D). Incubation in 20 mmol/l glucose increased the VEGF-A content in islets from 7-week-old lean animals by 4.1-fold and in the ZDF animals by 5.8-fold in the ZDF islets. Even accounting for the increase in islet size, these values show an increase in islet content of VEGF-A in the ZDF islets. Unlike insulin, there was no acute release of VEGF-A from islets following exposure to 20 mmol/l glucose for 60 min (data not shown). Over a 24-h incubation, secretion of VEGF-A into media was approximately twofold higher in islets from ZDF rats incubated at both low and high glucose concentrations (Fig. 5D). The increase in VEGF-A secretion was dependent on de novo translation, as cyclohexamide prevented the rise following incubation in 20 mmol/l glucose (data not shown). These data show that in the “failing” islets of the ZDF rat, there is preserved and even enhanced synthesis and secretion of VEGF-A.

The degeneration of islet microvasculature in the ZDF rat with intact VEGF-A secretion suggests an active inhibition of islet endothelial growth. To investigate this possibility, we utilized a modified protocol of Folkman et al. (35) who showed that islets from normal mice became “angiogenic” when treated with the protease MMP-9 via release of VEGF-A from the extracellular matrix. We isolated cells from primary duct cultures that contained ∼50% Flk-1⁺ cells (Fig. 2) and placed them in a vitrogen matrix (25). Islets from lean or ZDF rats were isolated and incubated for 24 h without or with MMP-9 to activate VEGF-A (35), embedded in the endothelial cell–containing vitrogen matrix, and scored for starburst formation, after 3 days, as an indication of migration of the cells in culture. No apparent growth of cells from the islets was noted when placed in vitrogen without duct-derived cells (Fig. 6A) over the 25-h period. Little migration of endothelial cells was seen in islets from lean control rats unless treated with MMP-9, as previously described (25). In contrast, there was enhanced migration of cells to the islets isolated from ZDF rats in the absence of MMP-9, with a smaller increase following MMP-9 treatment (Fig. 6A and B), suggesting endogenous activation of VEGF-A in these cultures. We noted that the cultures containing 7-week-old ZDF islets also contained small rounded cells and elongated cells. Islets of 14-week-old ZDF mice induced the duct-derived cells to round up in the vitrogen gel matrix (Fig. 6A), independent of whether MMP-9 was added to the cultures. The rounding up of cells was not seen in cocultures using 15-week-old islets from lean rats (data not show). These data suggest that a substance is secreted from the islet that has a detrimental effect on the duct-derived cells.

Expansion of the β-cell mass is a stereotypic response to insulin resistance. The increase in mass has been attributed to both β-cell replication and neogenesis (36). Since islets are imbedded in the larger pancreas, an expansion of the endocrine cell mass would require a vasculature that can grow in response to specific signals arising from the islet parenchyma or external signals. Studies in a number of systems, including the endocrine pancreas, have shown that cues for tissue growth and remodeling can arise from the vasculature (21). The vasculature is also critical to maintaining the health of mature tissue by providing nutrients and oxygen, as well as additional growth signals.

In the present study, we demonstrate morphological changes in the microvasculature of the pancreatic islet before the onset of hyperglycemia with a continued parallel degeneration of the microvasculature and β-cells following the onset of hyperglycemia. Morphologically, it appears that nestin is a useful marker for the endothelium of the adult rat islet microvasculature, which confirms recent studies in other systems (30,37). The increase in the relative number of nestin⁺ cells in the islets of nondiabetic fa/fa and ZDF rats suggests that the expansion of the vasculature is actively stimulated, perhaps via modulation of signals arising from β-cells in response to the insulin-resistant environment. An additional possibility is that the increase in islet mass via β-cell replication and/or neogenesis outstrips the vascular supply, causing a hypoxemic reaction.

In both normal development (38) and during tumor growth (39), hypoxia results in the development of additional vasculature through the elaboration of vascular growth factors, notably VEGF-A. VEGF-A seems to only be produced in the β-cell of the islet. We found no evidence of acute release of VEGF-A with glucose stimulation but did find that glucose increases the synthesis and release of VEGF-A in cultured islets in response to glucose. VEGF-A mRNA and protein levels were elevated in the islets in 7-week-old male ZDF rats and is intriguing in light of the observation that the endothelium of 7-week-old male ZDF rats show thickening and loss of endothelial fenestrations, similar to what has been described in the endothelium of mice following the inhibition of VEGF signaling by VEGF antibodies (40) or mice lacking VEGF-A in the pancreas (21). Interestingly, the latter mice demonstrate significant glucose intolerance, suggesting that changes in the endothelial morphology may contribute to abnormal insulin secretion. In the present study, the disruption of the vascular endothelium occurred before the elevation in glucose levels, suggesting that the degeneration may precede the failure of the islet.

Changes in capillary density in islets have been evaluated in previous studies. In db/db mice, there is capillary and pericyte hypertrophy (41). The apparent hypertrophy of endothelial cells is evident in both light microscopy and TEM of endothelial cells described in the present study. Similarly, in the Otsuka-Long-Evans-Tokushima fatty rat, islet degeneration leading to diabetes is associated with loss of capillary number within the islet (42). The phenomena of capillary hyperplasia followed by loss of capillary density in the male ZDF rat may not be limited to the islet. A recent study showed a similar phenomena occurring in the kidney of these rats. In this study, Gealekman et al. (43) observed hyperplasia of the endothelium of both the cortex and inner medulla in 8-week-old ZDF rats followed by a significant loss of capillary density at 22 weeks of age. Remarkably, they observed an identical increase in capillary growth from primary explants of the renal medulla from the younger rats and a marked diminution in growth from the 22-week-old ZDF rats, which is identical to the pattern we found in primary duct explants. The authors attributed part of these changes to oxidative stress and nitric oxide production changes, leading to reduced VEGF-A and Flk-1 receptor levels.

The reason for altered capillary growth may be due to the paradoxical production of antiangiogenic factors in the islets of the ZDF rat. In parallel to the disruption of endothelial integrity in vivo, islets cultured in vivo show abnormal effects in cocultures, with endothelial cells providing evidence for the production of vascular growth inhibitory factors. Tsp-1 has potent antiangiogenesis activity (44,45), and mRNA for Tsp-1 is increased ∼2-fold in pre-diabetic ZDF rats and ∼5- to 10-fold in the islets of diabetic animals. The antiangiogenic effect of Tsp-1 is initiated by binding to the scavenger receptor CD36, which induces apoptosis via p38 and Jun NH₂-terminal kinase. Tsp-1–null mice have enlarged hypervascular islets (46), making them an attractive candidate for the antiangiogenic effect. However, crossing Tsp-1^−/− mice with db/db^lepr−/− mice does not prevent diabetes (M.K.T., C.F.B., unpublished observations), suggesting that Tsp-1 does not play a major role in the development of diabetes in db/db^lepr−/− mice, but we cannot rule out a role in the ZDF model.

We recently reported that inhibition of MMPs can delay the development of islet failure in the ZDF rat model (47). MMPs are a family of zinc-dependent enzymes responsible for matrix remodeling in several disease states (48). MMPs have been shown to play a role in the degradation of basal lamina and disruption of the blood-brain barrier in animal models of intracerebral hemorrhage. MMP-9, an important activator of VEGF in islets (25), has been found to be associated with intracerebral hemorrhage (49), and inhibition of MMPs has been shown to be helpful in reducing the edema associated with intracerebral hemorrhage (50,51). The finding that islets isolated from 7-week-old rats induced endothelial cell migration without the need for MMP likely reflects the increased expression of MMPs, which we have previously documented in the ZDF rat (47). In addition to our proposed decrease in responsiveness of islet endothelial cells to VEGF, an increased MMP activity could contribute to the intra-islet hemorrhages observed in the present study.

Recently, Johansson et al. (52) presented data suggesting that VEGF from islets could stimulate secretion of hepatocyte growth factor from vascular endothelial cells, which in turn stimulates proliferation of β-cells. Similar to our findings in the expansion phase of the ZDF rat, they found an increase in proliferation of islet endothelial cells in pregnancy, which corresponded to an active phase of active β-cell expansion. We did not identify a change in hepatocyte growth factor expression in islet extracts by gene profiling in the ZDF population either before or after the development of diabetes (Fig. 1), although the expression levels were quite low.

Finally, the ability of the thiazolidinediones to prevent diabetes in the male ZDF rat is thought to be via its insulin-sensitizing actions, although there may be direct effects on the islet. Pioglitazone did not prevent the expansion of the islet but prevented both the decrease in endothelial numbers and endothelial hypertrophy and preserved β-cell mass. While pioglitazone caused a slight, but significant, reduction in islet VEGF-A mRNA levels, it was even more effective in lowering Tsp-1 mRNA. It may be that, either directly or indirectly, thiazolidinedione treatment reduces the production of substances, which results in the disruption of islet endothelial cells.

While this manuscript was under review, Homo-Delarche et al. (53) reported that gene expression analysis of the islets of spontaneously diabetic Goto-Kakizaki rats showed an upregulation of mRNA for extracellular matrix components and inflammatory markers, similar to what we have previously reported (47). In addition, they found increases in the expression of a number of genes described in this report (collagen, vimentin, thrombospondin 4, and lactate dehydrogenase A). The authors also found a variable decrease in the vascular density of the islet in diabetic rats, as well as evidence of endothelial hypertrophy. These latter findings are identical to the findings in the present study and suggest that disruption of the endothelium may be part of the pathology underlying islet failure.

In summary, we found that both in vivo and in an in vitro system, islet endothelia go through a biphasic pattern with early hyperplasia followed by eventual endothelial cell loss. Islet hyperplasia in response to insulin resistance is associated with an increase in the number of endothelial cells. In those animals prone to islet failure, the number of endothelial cells falls in parallel to islet failure. Disruption of the endothelial architecture preceded the development of hyperglycemia, suggesting that a key pathology in the development of diabetes in this model is failure to maintain an intact vasculature. The loss of the vasculature may be due to either an intrinsic alteration in the endothelial cells or the elaboration of a factor from the islets, which causes death of the endothelial cells.

This work was supported by grants from the American Diabetes Association, the Michigan Diabetes Research and Training Center (DK-20572), and an Investigator Initiated Award from Takeda Pharmaceutical North America.