In insulin-resistant status such as obesity, failure of pancreatic islets to increase insulin secretion leads to diabetes. We sought to screen for the islet genes that facilitate islet adaptation to obesity by comparing gene expression profiles between two strains of obesity-prone inbred mice with different propensities for hyperglycemia. C57BL/6J and AKR/J were fed regular rodent chow or high-fat diet, after which islet morphology, secretory function, and gene expression were assessed. AKR/J had lower blood glucose and higher insulin levels compared with C57BL/6J mice on regular rodent chow or high-fat diet. Insulin secretion was 3.2-fold higher in AKR/J than C57BL/6J mice following intraperitoneal glucose injection. Likewise, glucose-stimulated insulin secretion from isolated islets was higher in AKR/J. Additionally, islet mass was 1.4-fold greater in AKR/J compared with C57BL/6J. To elucidate the factors associated with the differences in islet function, we analyzed the gene expression profiles in islets in AKR/J and C57BL/6J mice. Of 14,000 genes examined, 202 were upregulated and 270 were downregulated in islets from diet-induced obese AKR/J mice compared with C57BL/6J mice. Key genes involved in islet signaling and metabolism, e.g., glucagon-like peptide-1 receptor, sterol Co-A desaturase 1 and 2, and fatty acid desaturase 2 were upregulated in obese AKR/J mice. The expression of multiple extracellular matrix proteins was also increased in AKR/J mice, suggesting a role in modulation of islet mass. Functional analyses of differentially regulated genes hold promise for elucidating factors linking obesity to alterations in islet function.
- insulin secretion
- islet mass
the incidence of type 2 diabetes (DM) is increasing rapidly as a consequence of the obesity epidemic (18, 25). Normally, insulin resistance associated with obesity is compensated for by increasing pancreatic islet mass and insulin secretion (16, 30). However, in some patients β-cell adaptation is attenuated, leading to the onset of diabetes and related complications (30). Both genetic and environmental factors are important in determining the susceptibility to type 2 DM and are likely to play critical roles in the attenuation of β-cell adaptation in high-risk individuals (28). Although rare cases of a single gene mutation can cause type 2 DM, the genetic defects that account for the deterioration of islet function in the majority of type 2 DM are complex, polygenic, and still not fully understood (28, 30).
Inbred mice show wide variation in the propensity for obesity and diabetes and serve as valuable models to analyze genetic factors that confer susceptibility to the disease (35, 36, 39). Previous studies indicate that genetic factors have a strong influence on islet function and islet mass and thereby contribute to the disparity in glucose homeostasis between different lines of inbred mice (1, 3, 21). Perifused islets from C57BL/6J (Bl6) mice had significantly lower glucose-stimulated insulin secretion compared with A/J mice (21). The comparison of islet mass and total islet numbers in mice from seven different genetic backgrounds demonstrated that strains are the major determinant of these parameters (3). Therefore, the genes differentially regulated in pancreatic islets between inbred strains of mice may reflect factors that regulate islet size and its response to secretagogues.
In the present study we aimed to screen for the genes that promote islet compensation for insulin resistance by utilizing inbred mice. We used microarray analysis as it allows for the screening of a wide range of genes and has the potential to identify gene clusters that concomitantly regulate islet function. Moreover, the polygenic nature of diabetes is difficult to address with linkage analyses. Therefore, we performed gene expression profiles in pancreatic islets between AKR/J (AKR) and Bl6 mice, both of which are susceptible to diet-induced obesity but show different propensities for diet-induced hyperglycemia.
MATERIALS AND METHODS
Experiments were performed in accordance with the University of Pennsylvania Institutional Animal Care and Use Committee guidelines with its approvals. We housed 4 wk-old male Bl6 mice and AKR mice (Jackson Laboratories, Bar Harbor, ME) (n = 5/cage) in 12 h light/dark cycle, at ambient temperature of 22°C, and allowed them free access to food and water. Groups of mice were fed regular rodent chow (NC, 4 kcal% fat; no. 5001 from Lab Diet, Richmond, IN) or high-fat diet (HF, 45 kcal% fat; no. D124551i from Research Diets, New Brunswick, NJ) for up to 12 wk.
Bl6 mice and AKR mice NC or HF were anesthetized with pentobarbital sodium (50 mg/kg BW ip), and body composition was measured using dual-emission x-ray absorptiometry (PIXImus DEXA, General Electric, Madison, WI; Ref. 15).
Glucose tolerance test was performed after overnight fast by giving 2 gm/kg glucose ip. For insulin tolerance test, mice were fasted for 5 h and given Humalin 0.75 U/kg ip (Eli Lilly, Indianapolis, IN). Tail blood was drawn at various times for glucose measurement with OneTouch Ultra Glucometer (Lifescan; Johnson & Johnson, Milpitas, CA). To assess glucose-stimulated insulin secretion in vivo, mice were given 3 gm/kg glucose ip, and 20 μl of tail blood was obtained at the indicated time for insulin measurement by ELISA from Crystal Chem (Chicago, IL) using mouse insulin standard.
Islet isolation and ex vivo perifusion assay.
Mice were anesthetized with pentobarbital sodium (50 mg/kg ip), and pancreatic islets were isolated using collagenase digestion followed by Ficoll density gradient centrifugation as was described before (10, 20). Then islets were hand picked under a dissecting microscope (SMZ 800; Nikon, Melville, NY). The purity of islet preparation judged by dithizone staining was ∼90%. Around 100 freshly isolated islets were loaded to a perifusion apparatus and perifused for 35 min with the Krebs buffer (pH 7.4) containing 2.2 mM Ca2+, 0.25% bovine serum albumin, and 10 mM HEPES under 5% CO2 atmosphere at 37 °C without glucose. Then perifusion was continued with the Krebs buffer containing increasing concentrations of glucose, ramped from 0 to 30 mM at 0.8 mmol/min. At the end of each experiment, islets were tested for the maximum insulin secretion by adding 30 mM KCl in the perifusate (14). Samples were collected at 1 ml/min for insulin measurement by radioimmunoassay (Linco Research, St. Charles, MO). At the end of experiments, islets in a perifusion apparatus were recovered on the filter and stored at −80°C for DNA and insulin measurement as described before (14). The threshold of glucose-stimulated insulin secretion was obtained as the glucose concentration that resulted in significantly higher level of insulin secretion compared with the baseline in the perifusion assay.
Paraffin-embedded secretions were prepared from pancreas fixed with 10% buffered formalin overnight and visualized with 1:1,000 guinea pig anti-insulin antibody (Linco Research) as described before (14). An image of a pancreatic section with the maximum footprint from each mouse was captured by a color video camera attached to a Nikon light microscope. Pancreas area, total tissue area, and β-cell area were measured using IP lab (Scanalytics, Fairfax, VA). Islet area (%) was calculated as (β-cell area/pancreas area). Islet weight was calculated as (weight of prefixed pancreas) × (β-cell area/total tissue area) and divided by body weight to adjust for the difference in body weight between Bl6 and AKR.
RNA extraction and gene expression analyses.
RNA was extracted from freshly isolated islets using RNeasy kit (Qiagen, Valencia, CA), and cDNA was generated by SprintPowerScript for cDNA synthesis (Clontech, Mountain View, CA) using 500 ng of islet RNA as a template. Gene expression was analyzed by ABI Prism 7900HT sequence detection system (Applied Biosystems, Foster City, CA) with commercial primers for the system. The results were expressed using 36B4 gene expression as an internal standard.
Microarray analyses were performed on quadruplicate RNA samples of pancreatic islets from AKR and Bl6 mice placed on HF for 3 mo. Pancreases from two mice were combined to yield one sample of islet RNA. All protocols were conducted as described in the Affymetrix GeneChips Expression Analysis Technical Manual (Affymetrix, Santa Clara, CA) using 5 μg total RNA and GeneChip Mouse Expression Arrays MOE 430v2 (Affymetrix). For data analyses, probe intensity data (cel files) were input into ArrayAssist Lite version 3.4 (Stratagene, La Jolla, CA), and expression values for the probe sets were calculated using GCRMA. Affymetrix “present,” “absent,” or “marginal” flags were also calculated. Subsequently intensity and flag data were imported into GeneSpring GX version 7.3.1 (Agilent Technologies, Palo Alto, CA) and filtered to retain probe sets flagged as present in at least three out of eight samples. Finally, the statistical test Significance Analysis of Microarrays (v2.23b; Stanford University, Palo Alto, CA) was applied using a two-class unpaired analysis, and differentially expressed genes were identified using a fold change cutoff of ≥1.8 (up or down) and a false discovery rate of 0.08%. The gene lists thus obtained (472 genes) were uploaded to DAVID (http://david.abcc.ncifcrf.gov), and cell compartment annotation and functional annotation chart were obtained. The genes present in at least three samples out of eight (22,245 probe sets) were used as the background list. Microarray data were submitted to Gene Expression Omnibus (www.ncbi.nlm.nih.gov/projects/geo) under the accession number GSE10639.
The data obtained from in vivo analyses and islet morphometry are presented as means ± SE. Differences between two groups were assessed with unpaired Student's t-test. Regression lines were obtained using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego CA). Multiple parameters in Figs. 1 and 3 were analyzed by ANOVA test. P < 0.05 was considered significant.
Both AKR and Bl6 are prone to obesity, but AKR maintain lower glucose levels with higher insulin levels on HF.
Inbred strains of mice that effectively increase insulin secretion on HF serve as an ideal model to identify genes that promote islet adaptation to insulin resistance. Since AKR are known to develop diet-induced obesity (DIO) like Bl6 but are more glucose tolerant on HF (31, 39), we evaluated their glucose homeostasis and islet function on NC and HF. On NC, AKR were slightly heavier than Bl6 (Fig. 1A, P < 0.05). However, serum glucose levels were significantly lower in AKR (Fig. 1B, P = 0.0001) while insulin levels were higher in AKR (Fig. 1C, P < 0.05). Insulin tolerance testing revealed that AKR are resistant to hypoglycemic effect of insulin compared with Bl6 (Fig. 1D; ANOVA test, P < 0.0001). Despite being insulin resistant, AKR had better glucose tolerance at an early time point, and glucose-stimulated insulin secretion was significantly higher in AKR (Fig. 1, E and F). Higher glucose-stimulated insulin secretion in vivo strongly implies that AKR have better islet function compared with Bl6 and serve as an ideal model to analyze genetic factors that improve islet function.
Next, AKR mice and Bl6 mice were placed on NC (4% fat kcal) or HF (45% fat kcal) and monitored for changes in their body weight, blood glucose levels, and serum insulin levels. After 8 wk on the HF, both AKR and Bl6 mice substantially gained weight. AKR weighed 31.3 ± 1.3 g on NC and 46.4 ± 1.3 g on HF (means ± SE, n = 8, P < 0.0001 between NC and HF), while Bl6 weighed 29.7 ± 0.4 g on NC and 35.5 ± 1.5 g on HF (means ± SE, n = 8, P < 0.0001 between NC and HF). The body weight after HF was significantly higher in AKR compared with Bl6 (P < 0.0001), indicating that AKR are more prone to obesity than Bl6. Glucose tolerance test showed that AKR mice have similar glucose tolerance compared with BL6 after HF challenge even though they are more prone to obesity (Fig. 2A). To better delineate differences in response to HF, the relationships between % body fat, glucose levels, and insulin levels from AKR and Bl6 on NC and HF were plotted (Fig. 2, B–D). AKR mice tend to have lower blood glucose levels compared with Bl6 matched for % body fat, indicating that AKR are protected from diet-induced hyperglycemia (Fig. 2B). Moreover, AKR maintained lower blood glucose levels accompanied by higher insulin levels (Fig. 2C). The correlation between serum insulin levels and % body fat indicated that AKR increase insulin levels more effectively when challenged with HF (Fig. 2D).
Pancreatic islets from AKR mice have higher glucose-stimulated insulin secretion and larger islet mass.
Since AKR showed higher insulin secretion in vivo compared with Bl6, we investigated whether ex vivo insulin secretion and islet mass are different in Bl6 and AKR. Here, islets from AKR and Bl6 on NC were compared, considering that higher insulin secretion in vivo was seen even on NC (Fig. 1). Thus we aimed to address whether islet characteristics differ between the two strains at minimally challenged status. Perifusion of pancreatic islets from AKR and Bl6 revealed that glucose-stimulated insulin secretion was significantly higher in AKR (Fig. 3A, two-way ANOVA, P <0.05). The threshold concentration of glucose was 10.5 mM for AKR islets and 12.2 mM for Bl6 islets. We did not observe a significant difference in DNA content (15.6 ± 2.4 ng/islet in Bl6 vs. 12.4 ± 1.0 ng/islet in AKR) or in insulin content between the two strains (3.4 ± 0.1 ng/ng of DNA in Bl6 vs. 3.6 ± 0.7 ng/ng of DNA in AKR). Morphometric analysis of pancreatic islets showed that β-cell area was significantly larger in AKR than in Bl6 (Fig. 3B). Islet weight adjusted for body weight did not reach statistical significance but also showed a trend of increased mass in AKR compared with Bl6 (P = 0.052).
Differential expression profile of pancreatic islets from AKR and Bl6.
Since higher serum insulin levels in AKR were associated with elevated insulin secretion in vivo and ex vivo and larger islet mass, we compared islet gene expression profile in DIO AKR and Bl6 (on 45% fat kcal diet) to screen for the factors involved in islet adaptation to obesity. To maximize the differential expression of genes involved in islet adaptation to insulin resistance, the analysis was performed using islets from AKR and Bl6 on HF.
Out of 14,000 distinct genes tested, we observed upregulation of 202 genes and downregulation of 270 genes in AKR islets using a fold change cutoff of ≥1.8 and a false discovery rate of 0.08% (Supplementary Table I).1 We did not observe a significant difference in the expression levels of insulin or pancreatic amylase between the two groups, indicating that the purity of islets preparation was comparable between AKR islets and Bl6 islets. Cell compartment assignment was done based on Gene Ontology and UniProt Knowledgebase Keyword classification through DAVID Bioinformatics Resources (http://david.abcc.ncifcrf.gov). As is shown in Fig. 4A, 42% of differentially expressed genes belong to extracellular space and 40% are associated with the cell membrane; 4.9% of genes belong to vesicles, cytosol, and endoplasmic reticulum (Fig. 4A). To elucidate functional features of differentially regulated genes, functional Annotation Clustering was performed using DAVID Bioinformatics Resources (http://david.abcc.ncifcrf.gov). The analyses showed that genes that have signal sequence (Table 1; category; SP_PIR_KEYWORDS, signal) are enriched in both upregulated (P = 0.0022, enrichment compared with background genes) and downregulated genes (P = 3.3E-11, enrichment compared with background genes). In addition, genes involved in extracellular matrix (Table 1; category; GOTERM_CC_ALL, extracellular region) and fatty acid desaturases (Table 1; category; INTERPRO_NAME, fatty acid desaturase) were enriched in upregulated genes in AKR islets (P = 0.0073 and 0.0046, respectively, enrichment compared with background genes). The representative genes from these clusters were validated using real-time PCR in islets from AKR and Bl6 on NC and HF. Stearoyl-CoA desaturase 1 gene (SCD1), stearoyl-CoA desaturase 2 (SCD2), and fatty acid desaturase-2 (FAD) catalyze desaturation of fatty acids and are implied in the regulation of lipid metabolism (5, 33). Asporin and periostin are extracellular proteins that are shown to play a role in cell-cell interactions in other organs (17, 26). These genes were confirmed to be upregulated in AKR compared with Bl6 in islets from both NC- and HF-fed mice (Fig. 4B).
To investigate factors involved in islet adaptation to insulin resistance, the expression profiles of islet genes were compared between inbred mice of high and low competence for islet compensation in response to HF. By using a cutoff of ≥1.8-fold change and a false discovery rate of 0.08%, 202 genes were found to be upregulated and 270 genes to be downregulated in hyperinsulinemic AKR islets. These genes are promising in the identification of new factors that aid islet adaptation to obesity and prevent development of diabetes.
In agreement with previous reports, AKR maintained lower glucose levels despite being more insulin resistant (31, 39). Although the expression of insulin genes in pancreatic islets was comparable between AKR and Bl6 islets, following observation supports the notion that elevated insulin secretion confers improved glucose homeostasis in AKR. First, AKR had higher glucose-stimulated insulin secretion in vivo (Fig. 1F). Second, AKR increased serum insulin levels effectively and maintained lower blood glucose levels on HF (Fig. 2, C and D). In addition, ex vivo insulin secretion and islet mass were increased in AKR compared with Bl6 (Fig. 3, A and B). Therefore, we used AKR islet as a model of pancreatic islets that show efficient adaptation to insulin resistance.
Previously, several quantitative trait locus analyses addressed candidate genes that confer vulnerability to hyperglycemia in inbred mice (37, 38). One of the genes identified was nicotinamide nucleotide transhydrogenese (NNT), a mitochondrial proton pump gene, which is functionally missing in Bl6 mice and proposed to impair glucose-stimulated insulin secretion (9). Aston-Mourney et al. (2) further demonstrated that NNT activity correlates with first-phase insulin secretion in five mouse strains. However, whether NNT plays any role in the development of diabetes in humans awaits further study. So far, candidate loci in human type 2 DM do not include NNT. A recent study also pointed out that NNT might not be solely responsible for lower insulin secretion in Bl6 compared with DBA/2 (2). Also, the mechanisms of the regulation of insulin secretion by NNT are not fully understood. Therefore, our study that identified multiple genes and gene clusters has the potential to elucidate additional genes that affect islet functions independently or in concert with NNT.
We used freshly isolated islets for gene expression profiling. Islet isolation has been shown to alter the expression of multiple genes including those involved in inflammation, hypoxia, stress, and apoptosis (24). However, the functional cluster analysis of our microarray did not show enrichment of genes in these categories. Since islets from both AKR and Bl6 were isolated by the same protocol, the effects of isolation on gene expression was likely cancelled when the comparison was made between the two.
The functional cluster analysis and cell compartment assignment indicated that the most prominent difference between AKR islets and Bl6 islets is in the category of peptides with signal sequences, which includes secreted proteins and receptors that reside in the cell membrane. Thus, response to external stimuli may yield improved islet function and mass in AKR. For example, a receptor for glucagon-like peptide-1 (GLP-1), an incretin hormone known to increase insulin secretion and islet mass, was upregulated in AKR islets (6). Furthermore, extracellular matrix (ECM) genes were among the enriched genes in AKR islets. Together, these findings indicate that the interaction between β-cells and the surrounding matrix may contribute significantly to islet adaptation to insulin resistance. Interestingly, the two independent microarray analyses of rat models of diabetes, Goto-Kawasaki rats and Zucker Diabetes Fatty rats, also detected the major changes in the expression of ECM genes (11, 42). The importance of ECM has been shown for the maintenance of insulin secretion in cultured islets and for the proper development of islets (4, 12, 13, 23). As such, further studies that address the role of islet architecture in islet adaptation to insulin resistance are warranted.
The elevation of circulating fatty acid (FA) is proposed to link obesity with changes in islet functions ranging from adaptive hypersecretion to islet dysfunction (27). Although there are conflicting data indicating both positive and detrimental effects of FA on islet function and size, the significant impact of lipid metabolism on islet function is well recognized (41). In the current study, three fatty acid desaturases were upregulated in AKR islets, while acetyl-coenzyme A dehydrogenase and CDP-diacylglycerol synthase 1 were downregulated. SCD-1 and -2 are the rate-limiting enzymes in monounsaturated fatty acid synthesis, and they negatively affect fatty acid oxidation and increase lipogenesis in liver (5). However, the deficiency of SCD-1 promotes the development of diabetes in leptin-deficient mice due to islet dysfunction, indicating a protective role for SCD in insulin secretion (8). Ultimately, comparison of lipid metabolism in AKR and Bl6 islet should address whether differential regulation of these genes has functional consequence. However, modification of lipid metabolism is one plausible mechanism that results in better islet function in AKR islets.
Insulin-like growth factor I (IGF-I) was downregulated in AKR islets in our microarray analysis (Table 1). The studies using various islet-specific knockout mice have shown that the insulin signaling plays a critical role in islet growth and islet compensation to insulin resistance (19). However, in contrast to insulin receptor and its signaling molecules that positively regulate islet mass, endogenous IGF-I in islets may inhibit islet growth. Pancreatic-specific inactivation of IGF-I gene results in islet enlargement and confers resistance to high fat-induced diabetes (22). Thus downregulation of IGF-I seen in AKR islet may aid adaptation to HF in AKR islets.
In addition to genes mentioned above, the list of differentially expressed gene contains noble candidate genes that may modulate islet functions. For example Rab3C, a GTP-binding protein, was upregulated 58.9 times in AKR islets compared with Bl6 in microarray analysis and was one of the most prominently differentiated genes (Supplementary Table Ia). Rab3C is known to be colocalized to secretory vesicles and play a role in exocytosis (32). In PC12 cells, Rab3C inhibited Ca2+-triggered exocytosis (32). Further studies may reveal that Rab3C regulates insulin release from β-cells. Dipeptidyl aminopeptidase 6 (DPP6) is another differentially expressed gene that might be a noble regulator of insulin secretion (Supplementary Table Ia). DPP6 is the single transmembrane protein that is associated with a voltage-gated potassium channel subtype 4 (Kv4) and alters its biophysical properties (34). Since Kv channels have a functional role in the process of glucose-stimulated insulin secretion, DPP6 may regulate insulin secretion by modifying the property of a Kv4 channel (7, 40).
As presented above, our analysis revealed many new candidate genes for the islet adaptation to obesity that were previously not connected with insulin secretion nor islet function (Supplementary Table I). On the other hand, limited number of genes previously implicated in islet growth and islet compensation are found in the list (30). GLP-I receptor, IGF-I, SCD1, and peroxisome proliferator-activated receptor γ are examples of such genes (6, 8, 22, 30). However, transcription factors associated with islet growth such as pancreatic and duodenal homeobox 1 (PDX-1), mafA, and NeuroD were not detected in the present analysis (Supplementary Table I). This may be in part due to stringent selection criteria used in the study (a cutoff of ≥1.8-fold change and a false discovery rate of 0.08%).
In summary, cDNA microarray analyses compared gene expression profiles in pancreatic islets from two strains of mice that both develop DIO but show differences in islet adaptation to obesity. The differentially expressed genes identified may serve as molecular markers of islet adaptation to DIO. Future studies should test whether the candidate genes are functionally linked to different propensities for islet dysfunction in the two strains.
The study was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-071536 and Institute for Diabetes Obesity and Metabolism at Penn (IDOM) pilot and feasibility grant to Y. Imai, DERC Mouse Metabolic Phenotyping and Islet Biology Cores (P30-DK-19525), the Penn Genome Frontiers Institute, and a grant with the Pennsylvania Department of Health to J. W. Tobias.
The Department of Health specifically disclaims responsibility for any analyses, interpretations, or conclusions.
We thank Edem Abotsi and Dr. Yong Qi for technical support and David Faleck for critical reading of the manuscript. The technical supports for the study were provided by Penn Microarray Facility, Morphology Core at Center for Molecular Studies in Digestive and Liver Disease, and Radioimmunoassay and Biomarker Core in Diabetes and Endocrinology Research Center (DERC) at University of Pennsylvania School of Medicine.
↵1 The online version of this article contains supplemental material.
Address for reprint requests and other correspondence: Y. Imai, Dept. of Medicine, Div. of Endocrinology, Diabetes and Metabolism, Univ. of Pennsylvania School of Medicine, Philadelphia, PA (e-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society